Trailing-Edge
-
PDP-10 Archives
-
tops10and20_integ_tools_v9_3-aug-86
-
tools/crc/help/subs.lbr
There are no other files named subs.lbr in the archive.
E/
sub .LBR.1 ;xF _
s BINOML ( %[urCLEOL , V%[urCLEOS ,W Z%[usCOPY -1 `%[usCPLOT 0 \%[usCPLOT2 5n k%[utCRBARS :Y %[utCRBELL >w %[utCRC151-2 ?w ]%[uuCRC153-156 DT 9%[uuCRC154-155 K
-%[uzICHOUT r7 l3]AQCRJOUT u# #3]ANCREMOT xF m%[u{CRERBR 4 k%[u{CRFLIN %[u{CRFNIN = w%[u|CRFREQ 4 :%[u}CRFUNC o k%[u~CRGAM Z *%[u~DATA-ENTRY-INDEX DEX F:u\QCRGRIT "o %[vCRGRLB $s ;%[vCRHACH ). }%[vCRHALT -, *%[vCRHIAX .V b%[vCRHIST 18 7%[vCRMTOC 8o 3]AOCRLBOP : &%[vCRLDAV @% %[vCRLHST B; n%[vCRLNFC H* l%[vCRLOGGAM L 2%[vCRMAIL OH %[v CRMATH2 Vb V%[v CRMNIM Y8 |%[v CRMNX-2 c5 %[v CRMOVE gL %[v
CRUSER ie k3]APCRMVCH kP 3%[v CRNIN o %[v CRNTOI r
%[v
CRON-CROFF-CRTEST u, \%[v
1%[vCRPOLY A @%[vICHIN x k%[vCRSINT d )%[vCRSIZE
}%[vCRTWMY /f %[vCRVCR 4~ I%[vCRWAIT :G %[vCRWTST <] x%[vCRXPNG BU J%[vCURHOM H [%[vCURUP H{ s%[vDATA-ENTRY In %[vDATE f -%[vDATEIN g/ J%[vDOJSYS uy
G%[vEASTER A ]%[vCRUPER 34-GETADR I%[vGETDDT Q ;%[vHOSPNO
C%[vICOMP O 4%[vICRCH2 i%[vICRCH3 #l O%[vICRCH4 &< o%[vICRCH7 )+ m%[vICRNTM , ^%[vICRSCH .w d%[vICRTLU 4[ %[vDATA-ENTRY-INDEX 7n z MONTH INDEX :i ^3;=OINPUT-TERMINAL-INDEX <G F%kX9CRCLR CRZERO TING > q 0%[v!MW s g%[v"NELDER ~w %[v"NINT
A%[v#NPH11 N X%[v#NPH2
& A%[v$NPH3 o %[v$NPLOT1 u %[v$NSCOR2 y %[v%NUMIN V%[v%PK ,b %[v&PNPLOT /a POLY 4z %[v&PUTCRS 7 t%[v'RADII 9
B%[v'RANDOM EJ }%[v'RANDUP HG q%[v'RANSET J9 '%[v(REALIN K` %[v(RIGHT Ya 2%[v(SBS [
%[v)SNDEV vP S%[v)SPLITN xs . :%[v,STAT7A h n%[v,STAT7M W
%[v,STAT8 !c h%[v,TABCLR $L J%[v-TABLE % g%[v-TABSET 0} g%[v-TBLOOK 1d %[v.TBRLOC 8h %[v.TEXTIN @ O%[v.VT-SUBROUTINE-CONVENTIONS OO %[v/VT132 Sf R%[v/VT80 T8 Q%[v/VTATR U .%[v0VTBIG W7 2%[v0VTCBL Xj W%[v0VTCBS YA V%[v0VTCLIN Z W%[v0VTCLS Zn L%[v0VTDWN [: m%[v1VTEND \' %[v1VTGOF ]9 d%[v1VTGON ^ &%[v1VTINIT _D %[v2VTIUP `O #%[v2VTJMP as J%[v2VTLED b= t%[v2VTLFT c2 m%[v3VTORG d %[v3VTRES f. [%[v3VTRIT g m%[v3VTSAV gw X%[v4VTSCRN hO h%[v4VTSMTH i7 L%[v4VTTY j 3%[v4VTWID m7 =%[v5WCOEF nt %[v5WEXT qw }%[v5YESNO vt
O%[v5LP-PLOTTING-INDEX D '%kX:NRMPLT k t%cd?CRSIL _ +NECRGRAF y
+c_CRAX5-F-G
C+ 29CRGRAX %I + LECRGRWO ERS-INDEX -h '6P CRAISE 3 +.|[kDATES-INDEX 5> Z3;=
CHKLET BINOML ROUTINE
CAT 8 {3=/MISCELLANEOUS-INDEX (N,M,PT,PROB)
LANG : 3=.tCRDTOI SCRIPTION THIS SUBROUTINE CAL ?# 3]ANCRLBRD Y OF 'M' SUCCESSES
IN 'N' D9 3]AOCRRSCN THE PROBABILITY OF A SINGLE
b{ HW |3]APSIMINP TURNS THETCALCULATED PROBABILITY MT l3]AQGENCHK Q@
CAT TM G4-pGRAPH-PLOTTING-INDEX N,M,PT,PROB)
LANG Z
6QOW SCRIPTION THIS SUBROUTINE CAL ^ m:{>}KYDALW Y OF 'M' SUCCESSES
IN 'N' f B:{? TIMEIN THE PROBABILITY OF A SINGLE
lN 3:{S`PNPLOT TURNS THE CALCULATED PROBABILITY | E;xFINVALID VALUES OF THE INPUT PARAMETERS(E.G. M>N ETC.)
CAUSE THE VALUE OF PROB TO BE RETURNED AS ZERO.
METHOD: PROB = N!/(M!*(N-M)!)*(PT**M)*(1-PT)**N-M
BUT THE CALCULATION IS CARRIED OUT IN TERMS OF THE LOG
OF THE FACTORIALS SO AS TO MINIMISE
THE POSSIBILITY OF OVERFLOW OR UNDERFLOW FOR LARGE
N OR M AND TO OPTIMISE THE COMPUTATION SPEED.
PROB=EXP[ LOGFAC(N)-LOGFAC(M)-LOGFAC(N-M)+M*LOG(PT)
+(N-M)*LOG(1-PT) ]
PARAMETERS N INTEGER VARIABLE
M INTEGER VARIABLE
PT FLOATING POINT VARIABLE 0.0.LE.PT.LE.1.0
PROB OUTPUT VALUE - FLOATING POINT VARIABLE
LEVEL: LEVEL 1
AUTHOR G.GIBBS/I.D.HILL
DATE: FEBRUARY 1981
CLEOL ROUTINE
CATEGORY: L6
NAME: CLEOL
PURPOSE: Clears from the cursor to the end of the line.
PARAMETERS: None
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
CLEOS ROUTINE
CATEGORY: L3
NAME: CLEOS
PURPOSE: Clears the screen from the cursor to the end of screen
PARAMETERS: None
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
COPY ROUTINE
CATEGORY: A2
TITLE: SUBROUTINE COPY(N,A,I,B,J)
LANGUAGE: FORTRAN
DESCRIPTION: COPY CHARACTERS
N CHARACTERS ARE COPIED FROM B TO A.
IF ANY OF N,I AND J ARE NON-POSITIVE,
THERE IS NO EFFECT.
PARAMETERS: N INTEGER INPUT NO OF CHARACTERS TO BE COPIED
A ANY INPUT ELEMENT TO WHICH TO COPY
I INTEGER INPUT CHAR POSITION IN A TO WHICH
COPYING IS TO START
(COUNTING FROM 1)
B ANY INPUT ELEMENT FROM WHICH TO COPY
J INTEGER INPUT CHAR POSITION IN B FROM WHICH
COPYING IS TO START
ROUTINES CALLED: CRMVCH
LEVEL: LEVEL 1
AUTHOR: G.GIBBS
DATE: MARCH 1981
CPLOT ROUTINE
CATEGORY: D9
TITLE: CPLOT (SSP,NVAR,NUM,NCHAN)
LANGUAGE FORTRAN
DESCRIPTION GIVEN A SYMMETRIC MATRIX OF CORRELATION
COEFFICIENTS IN SSP BASED ON NVAR VARIABLES
OBSERVED ON NUM CASES THIS SUBROUTINE USES
A GRAPHICAL TECHNIQUE TO DETERMINE WHICH OF
THE CORRELATION COEFFICIENTS ARE TOO LARGE
TO COME FROM ZERO POPULATION VALUES.
INTERPRETATION OF THE PRINTOUT RESTS ON
COMPARING THE HALF-NORMAL PLOT WHICH IS
OBTAINED FROM A GIVEN MATRIX WITH A
STRAIGHT LINE THROUGH THE ORIGIN OF THE
MATRIX AND A POINT MARKED AT A VALUE ON THE
X AXIS EQUAL TO 2.0. THIS LINE CORRESPONDS
TO THE HALF-NORMAL PLOT WHICH IS APPROPRIATE
FOR NUM CASES AND NVAR UNCORRELATED VARIABLES.
NOTE THAT ARRAY SSP IS SORTED BY CRC154 FIRST.
FURTHER EVIDENCE IS OBTAINED BY ANY SHARP
CHANGES IN SLOPE OF THE OBSERVED PLOT.
REFERENCES: THE HALF-NORMAL PLOT IS USED AS DESCRIBED BY
HILLS , BIOMETRIKA(1969),56,249.
PARAMETERS SSP REAL ARRAY SYMMETRIC MATRIX
NVAR INTEGER NO. OF VARIABLES
NUM INTEGER NO. OF CASES
NCHAN INTEGER OUTPUT CHANNEL
OWN VARIABLES: Y REAL ARRAY DIMENSIONED TO 500
I.E. 1/2*NVAR*(NAR-1)+1.LE.500
OR NVAR.LE.32
ROUTINES CALLED: CRC154, SNDEV, CRC561
LEVEL: LEVEL 2
AUTHOR KLIM MCPHERSON
DATE 19TH JULY 1972
CPLOT2 ROUTINE
CATEGORY: D10
TITLE: SUBROUTINE CPLOT2(X,IX,Y,IY,N,ICHAN,RESLT)
DESCRIPTION: SCATTER DIAGRAM WITH STATISTICS
X,Y ARE 1-DIM'L ARRAYS OF COORDINATES OF N POINTS.
A SCATTER DIAGRAM OF Y AGAINST X IS OUTPUT AS A NEW
PAGE ON CHANNEL ICHAN.IF N IS NOT >1000, THEN THIS IS
FOLLOWED BY A TITLE LINE AND
THEN BY THE MEANS, SD'S AND CORRELATION OF
X AND Y. IX AND IY ARE TEXT ARRAYS OF UP TO 20 CHARS
EACH DESCRIBING THE X'S AND Y'S RESPECTIVELY, SO THAT
THE TITLE READS -
X-VARIABLE IS (IX) Y-VARIABLE IS (IY)
IF TEXT CONSTANTS ARE USED, THEY SHOULD HAVE AT LEAST
16 CHARS.
THE STATISTICS ARE ALSO PLACED IN THE ARRAY RESLT
IN THE SEQUENCE XBAR, SDX, YBAR, SDY, R IF N NOT >1000 .
PARAMETERS: X,Y REAL ARRAY INPUT COORDINATES
IX,IY INTEGER ARRAY INPUT COORD. DESCRIPTIONS
N INTEGER INPUT NO OF POINTS =< 1000
ICHAN INTEGER INPUT OUTPUT CHANNEL
RESLT REAL ARRAY OUTPUT STATISTICS
COMMON BLOCKS: NONE
OWN VARIABLES: NONE
ROUTINES CALLED: CRC561,G01ABF
LEVEL: LEVEL 3
AUTHOR: M.J.R.HEALY
DATE: 30 JULY 1973
CRBARS ROUTINE
CATEGORY: C29
TITLE: SUBROUTINE CRBARS(X,N,XLOW,XHIGH,IBP,IEP,YMIN,MARK)
LANGUAGE: FORTRAN
DESCRIPTION: DRAWS HISTOGRAM BARS (N.B. OPEN AT BOTTOM)
PARAMETERS: X REAL ARRAY I/P ARRAY CONTAINING HEIGHTS OF
NAME BARS (IN REQUIRED ORDER)
N INTEGER I/P NUMBER OF BARS
XLOW REAL I/P LOWER LIMIT OR VALUE OF FIRST
BAR
XHIGH REAL I/P UPPER LIMIT OR VALUE OF LAST
BAR
IBP, INTEGER I/P PERCENTAGE POINTS IN EACH
IEP INTERVAL OF THE START AND FINISH
POINTS OF EACH BAR
YMIN REAL I/P MINIMUM POINT ON Y AXIS
MARK INTEGER I/P A TWO DIGIT INDICATOR -2ND DIGIT
INDICATES HATCHING REQUIRED IN
BAR (SEE CRHATCH C30). 1ST DIGIT
IF ZERO INTERVAL IS A RANGE, IF
ONE INTERVAL IS A CATEGORY (SEE
CRAXIS5 C32)
REFERENCE: DOCAS REPORT ON PLOTTING PACKAGE
ROUTINES CALLED: CRHACH+GINO-F ROUTINES
LEVEL: LEVEL 1
AUTHOR: L.J.DOBSON/G.GIBBS
DATE: DECEMBER 1981
CRBELL ROUTINE
Category: L31
Title: CRBELL(N)
Language: MACRO
Description: Rings your terminal bell N times.
If N is less than 1 the bell is not rung.
Parameter: N Integer Input Number of times bell rung
Author: G.Gibbs
Date: 2 June 81
CRC151-2 ROUTINE
CATEGORIES: E5 AND E6
TITLE: CRC151 (LIST,N)
CRC152 (LIST,N,NCHAN)
LANGUAGE: FORTRAN
DESCRIPTION: SETS UP LIST OF INTEGERS. CRC151 SIMPLY WRITES
1,2,....N INTO THE ARRAY LIST. CRC152 EXPECTS IN
LIST A LIST OF INTEGERS IN ABBREVIATED FORM.
M,-K STANDS FOR M,M+1,...,K WHEN K>M
M,-K STANDS FOR M,M-1,...,K WHEN K<M
IF K=M, THEN -K IS CHANGED TO K.
THE ROUTINE EXPANDS THE LIST.
N IS THE LENGTH OF THE EXPANDED LIST. IF N IS TOO
SMALL AN ERROR MESSAGE IS OUTPUT ON CHANNEL NCHAN,
AND EXPANSION IS HALTED.
PARAMETERS: CRC151
LIST INTEGER 1-ARRAY OUTPUT LOT OF CONSECUTIVE
INTEGERS.
N INTEGER INPUT LENGTH OF LIST
CRC152
LIST INTEGER 1-ARRAY INPUT ABBREVIATED LIST
OUTPUT EXPANDED LIST
N INTEGER INPUT LENGTH OF EXPANDED
LIST
NCHAN INTEGER INPUT OUTPUT CHANNEL NO.
FOR ERROR MESSAGE.
COMPRESSED INTEGERS? YES
FORTRAN TRACEABLE LEVEL 1 ONLY
ROUTINES USED NONE
LEVEL: LEVEL 0
AUTHOR M.J.R. HEALY
DATE 3 JUNE 1971
CRC153-156 ROUTINE
CATEGORIES: E7 AND E8
TITLE: CRC153 (X,LIST,C,N,NCHAN)
CRC156 (X,CN,N,NCHAN,Y)
***********************
** Running but **
** NOT definitively **
** tested **
***********************
LANGUAGE: FORTRAN
DESCRIPTION: LOG TRANSFORMATION OF VARIABLES.
X IS A VECTOR OF VARIATE VALUES. CRC153 TRANSFORMS
THE N ITEMS IDENTIFIED BY THE ELEMENTS OF LIST SO
THAT X(LIST(I))=100.0*ALOG10(X(LIST(I))-C(I))
CRC156 SETS Y(I) =100.0*ALOG10(X(I)-CN) FOR I=1,N.
IF A NON-POSITIVE ARGUMENT OF THE LOGARITHM IS
ENCOUNTERED, THE ARGUMENT IS PRINTED OUT ON CHANNEL
NCHAN, AND THE MODULUS OF THE ARGUMENT IS USED
UNLESS THE ARGUMENT IS ZERO, WHEN X(LIST(I)) OR
Y(I) IS SET TO ZERO.
PARAMETERS: X REAL 1-ARRAY INPUT VECTOR OF VARIATE
VALUES.
OUTPUT VECTOR WITH SELECTED
(CRC153) VALUES TRANSFORMED.
LIST INTEGER 1-ARRAY INPUT VECTOR OF VARIATE
(CRC153) NUMBERS.
C REAL 1-ARRAY INPUT VECTOR OF CONSTANTS
(CRC153)
N INTEGER INPUT NO. OF VARIATE VALUES
TO BE TRANSFORMED.
NCHAN INTEGER INPUT OUTPUT CHANNEL NO.
FOR ERROR MESSAGE.
CN REAL INPUT CONSTANT
(CRC156)
Y REAL 1-ARRAY OUTPUT VECTOR OF TRANSFORMED
(CRC156) VALUES.
Y MAY COINCIDE WITH X
COMPRESSED INTEGERS? YES
FORTRAN TRACEABLE? LEVEL 1 ONLY
ROUTINES USED NONE
LEVEL: LEVEL 0
LIBRARY: SUBGROUPCRCA (0)
AUTHOR M.J.R. HEALY
DATE 3 JUNE 1971.
CRC154-155 ROUTINE
CATEGORY: B1
TITLE: CRC154 (A,N)
CRC155 (A,N,IY)
LANGUAGE FORTRAN
DESCRIPTION SORT SUBROUTINES. A IS THE REAL ARRAY OF LENGTH N
TO BE SORTED IN ORDER OF ASCENDING MAGNITUDE.
CRC155 SORTS INTEGER ARRAY IY IN THE ORDER CORRESPONDING
TO THAT OF A.
METHOD BOOTHROYD'S PART SORT. ALGORITHM 27, COMP.J.10,308 1967.
PARAMETERS: A REAL 1-ARRAY INPUT/OUTPUT ARRAY TO BE SORTED.
N INTEGER INPUT SIZE OF A
IY INTEGER 1-ARRAY INPUT/OUTPUT AUXILIARY ARRAY.
ROUTINES USED NONE
LEVEL: LEVEL 0
AUTHOR M.J.R. HEALY
DATE 3 JUNE 1971.
CRC161 ROUTINE
CATEGORY: D1
TITLE: CRC161 (SMALL,BIG,AXLEN,DIV,START,FINISH,MARK1)
LANGUAGE FORTRAN
DESCRIPTION CALCULATES AXIS PARAMETERS FOR PLOTTING ROUTINE CRC561 TO
ACCOMODATE VALUES BETWEEN SMALL AND BIG
PARAMETERS: SMALL,BIG REAL INPUT EXTREME VALUES TO
BE PLOTTED.
AXLEN REAL INPUT NO. OF DIVISIONS
AVAILABLE ON PLOTTING
AXIS.
DIV REAL OUTPUT LENGTH OF ONE DIVISION
START,FINISH REAL OUTPUT COORDINATES OF
EXTREMES OF AXIS.
MARK1 INTEGER OUTPUT POSITION ON AXIS OF
FIRST POINT TO BE
LABELLED.
ROUTINES USED NONE
LEVEL: LEVEL 0
AUTHOR M.J.R. HEALY
DATE 3 JUNE 1971
CRC162 ROUTINE
CATEGORY: D2
TITLE: CRC162 (NCHAN)
LANGUAGE: FORTRAN
DESCRIPTION: PLOTS AXIS FOR PLOTTING ROUTINES (CRC362,CRC561)
ON CHANNEL NCHAN.
PARAMETERS: NCHAN INTEGER INPUT CHANNEL NO. FOR OUTPUT.
COMMON BLOCK /PSCOM/ NM,MARKX,NUM.
ROUTINES CALLED NONE
LEVEL: LEVEL 0
AUTHOR M.J.R. HEALY
DATE 3 JUNE 1971
CRC163 ROUTINE
CATEGORY: D3
TITLE: CRC163 (AXLEN,START,FINISH,DIV,MARK1)
LANGUAGE: FORTRAN
DESCRIPTION: CALCULATES AXIS PARAMETERS FOR PLOTTING ROUTINE
CRC362.
PARAMETERS AXLEN REAL INPUT NO. OF DIVISIONS REQUIRED ON
AXIS.
START REAL INPUT (COORDINATES OF EXTREMES
FINISH REAL INPUT (OF AXIS
DIV REAL OUTPUT LENGTH OF ONE DIVISION
MARK 1 INTEGER OUTPUT POSITION ON AXIS OF FIRST
POINT TO BE LABELLED.
ROUTINES CALLED NONE
LEVEL: LEVEL 0
AUTHOR S. CHINN
DATE 13 SEPT. 1971
CRC361 ROUTINE
CATEGORY: D4
TITLE: CRC361 (X,N,AXLEN,DIV,START,FINISH,MARK1)
LANGUAGE: FORTRAN
DESCRIPTION: X IS AN ARRAY OF LENGTH N. THE ROUTINE CALCULATES
PARAMETERS FOR PLOTTING ROUTINE CRC561 TO FIT ALL
VALUES OF X ONTO A PAGE.
PARAMETERS: X REAL 1-ARRAY INPUT VECTOR OF VALUES
TO BE PLOTTED.
N INTEGER INPUT LENGTH OF X
AXLEN REAL INPUT NO. OF DIVISIONS
AVAILABLE ON PLOTTING
AXIS
DIV REAL OUTPUT LENGTH OF ONE DIVISION
START REAL OUTPUT (COORDINATES OF ENDS
FINISH REAL OUTPUT (OF AXIS
MARK 1 INTEGER OUTPUT POSITION ON AXIS OF
FIRST POINT TO BE
LABELLED.
ROUTINES CALLED: CRC161
LEVEL: LEVEL 1
AUTHOR M.J.R. HEALY
DATE 3 JUNE 1971
CRC363 ROUTINE
CATEGORY: D5
TITLE: CRC363 (X,NX,NINT,XMIN,XINT,NAME,NCHAN)
LANGUAGE FORTRAN
DESCRIPTION: PLOTS HISTOGRAM. X IS AN ARRAY OF LENGTH NX CONTAINING
VALUES FOR HISTOGRAM. NINT IS THE NUMBER OF
INTERVALS USED, EACH OF LENGTH XINT, XMIN IS
THE MINIMUM VALUE ON THE AXIS. THE CHARACTERS
HELD IN THE FIRST 18 ELEMENTS (ONE IN EACH)
OF NAME ARE USED TO LABEL THE HISTOGRAM. THE
FIRST INTERVAL CONTAINS ALL POINTS LESS THAN
XMIN+NINT*XINT. IF THERE ARE MORE THAN 106
POINTS IN ANY INTERVAL THE HISTOGRAM WILL NOT
FIT ON THE LINE PRINTER PAGE. AN ERROR
MESSAGE IS PRINTED ON CHANNEL NCHAN, AND THE
CALLING ROUTINE RE-ENTERED.
N.B. THE ROUTINE SORTS X IN ASCENDING ORDER.
PARAMETERS X REAL 1-ARRAY INPUT/ VECTOR OF DIS-
OUTPUT TRIBUTION VALUES
NX INTEGER INPUT LENGTH OF X
NINT INTEGER INPUT NO. OF INTERVALS
XMIN REAL INPUT MINIMUM VALUE ON AXIS
XINT REAL INPUT LENGTH OF INTERVAL
NAME INTEGER 1-ARRAY INPUT HOLDS LABEL FOR HISTOGRAM
NCHAN INTEGER INPUT OUTPUT CHANNEL NUMBER
ROUTINES CALLED: CRC154
LEVEL: LEVEL 1
AUTHOR M.J.R. HEALY
DATE 3 JUNE 1971
CRC561-362 ROUTINE
CATEGORY: D7
TITLE: CRC561 (X,Y,N,NCHAN)
CRC362 (X,Y,N,XPMIN,XPMAX,YPMIN,
YPMAX,XDIV,YDIV,NCHAN)
LANGUAGE FORTRAN
DESCRIPTION: LINE PRINTER PLOT (2-WAY DISTRIBUTION). GIVEN ARRAYS X,Y
OF LENGTH N, A TWO-WAY LINE PRINTER PLOT OF Y AGAINST X IS
PRODUCED ON CHANNEL NCHAN.
AN INTEGER BETWEEN 1 AND 9 REPRESENTS THAT NO. OF
VALUES AT ONE POINT; IF THERE ARE MORE THAN 9 VALUES
AT A POINT,* IS PRINTED. X AND Y ARE NOT ALTERED
CRC561 CALCULATES EXTREME VALUES ON X AXIS OF LENGTH
100 AND Y AXIS OF LENGTH 45 SO THAT ALL POINTS ARE
PLOTTED.
CRC 362 USES AXES AS SPECIFIED (LENGTH OF X AXIS .LE.
100) THE NUMBER OF POINTS ( IF ANY ) THAT FALL OUTSIDE
THE RANGE OF THE AXES IS PRINTED.
PARAMETERS: X,Y REAL 1-ARRAY INPUT VECTORS OF COORDINATES
NCHAN INTEGER INPUT CHANNEL NUMBER
N INTEGER INPUT LENGTH OF X,Y
IF NEGATIVE THE GRID MARKINGS
ARE SUPPRESSED
XPMIN REAL INPUT (EXTREMES OF X AXIS
XPMAX REAL INPUT (
YPMIN REAL INPUT (EXTREMES OF Y AXIS
YPMAX REAL INPUT (
XDIV REAL INPUT (NUMBER OF DIVISIONS
YDIV REAL INPUT (ON EACH AXIS
COMMON BLOCKS: PSCOM, NM, MARK, X, NUM
ROUTINES CALLED CRC 561 CALLS CRC 361, CRC 162
CRC 362 CALLS CRC 162, CRC 163
LEVEL: CRC561 - LEVEL 2
CRC362 - LEVEL 1
AUTHOR M.J.R. HEALY, S.CHINN
DATE 13 SEPT. 1971.
CRCMD ROUTINE
CATEGORY: K13
NAME: CRCMD(COMND,FLAGS,JSERR)
PURPOSE: Allows a user program to execute EXEC commands while
running. The sequence is somewhat similar to typing 2 ctrl/c's, followed
by PUSH, the command, POP and CONTINUE. There are numerous differences,
however. The command string may either be in an array, terminated by a
zero word, or a quoted string.
The flag word is a sum of a number of separate numbers as
below ; if you want to use a certain combination of features, you
just add up the numbers corresponding to them, and the result is the
flag word.
Value Effect
----- ------
0 None
1 MEXEC is used instead of EXEC.
2 Command is echoed on the screen.
8 Performs a PUSH - see below
16 Execute COMAND.CMD on first call.
The value of 1 must be specified on the first call, if used.
Its value has no effect in subsequent calls. Similarly, 16 (to use
COMAND.CMD) should be used on the first call, if desired. Normally,
COMAND.CMD is temporarily renamed while the subroutine executes :
specifying this flag causes it to be run when the EXEC is first started.
The 8 value is used if you want to do a PUSH command from your
program. It merely suspends your program, and allows any EXEC commands
to be entered from your terminal. To return to your program, type POP.
This then effects a return from the CRCMD subroutine. When 8 is used,
the contents of the COMND array are ignored.
On return, JSERR will contain the JSYS error code , if any,
or zero for success.
PARAMETERS: COMND Array input Contains the command to be executed
May be a quoted string or array.
FLAGS Integer input Bit encoded flag to indicate whether
COMAND.CMD should be used, MEXEC, or
whether command should be echoed.
JSERR Integer output Contains the JSYS error code on exit
or zero. Error codes may be looked up
in the file <SOURCES>MONSYM.MAC.
RESTRICTIONS: The LOGOUT or BYE commands may not be executed either by
CRCMD, or by the user when using the PUSH feature. This is to
prevent accidental loss of files. Attempting to execute these
commands produces the error ?LOG capability required.
Fielding of CTRL/C interrupts while CRCMD is running is
not handled by the superior EXEC, but by CRCMD's inferior. Thus,
a CTRL/C struck while in CRCMD merely causes CRCMD to return
success instantly.
AUTHOR: K. Ashley 27-Oct-81
REVISED: 1-Nov-81,1-Dec-81
Language: Macro-20
CRDIXN ROUTINE
category: E43
title: Subroutine crdixn(x, n, p, ifail)
language: fortran
description: Dixon's test for a single outlier in a normal distribution.
p is the probability that x(n) is not a member of the sample
x(1) to x(n-1).
P-values are interpolated from the table
given by Dixon (Biometrics, 74-89, 9, 1953) using natural
cubic spline functions (note that if this routine is to
be called a very large number of times it will be more
efficient to group the calls by n).
If the data are supplied in ascending order, the extreme
upper value is tested as a putative outlier; if the data
are supplied in descending order, the extreme lower value
is tested.
parameters:
x real array input observations
n integer input number of observations
p real output probability that x(n) is an outlier
ifail integer output failure flag:
-1: success, but p outside the range
given by Dixon (0.005<p<0.3).
p therefore set to 0.005 or 0.3,
as appropriate.
0 : success
1 : n<3 or n>25
2 : observations not ordered
3 : all/too many observations the same
4 : internal (fit) failure
5 : internal (interpolation) failure
routines called: x02acf, e01baf (calling e02baf & p01aaf), e02bbf (Nag library)
author: R.Wootton
date: 28-Jan-83
ICHOUT ROUTINE
category: M6
title: subroutine ichout(letter)
language: macro
description: send a character to the screen, see also ichin.
example: CALL VTINIT !initialise screen control
CALL CRSIL(1) !stop echoing of characters
DO 100 I=1,80
LETTER=ICHIN(IERROR) !get next character
IF (IERROR.NE.0) GO TO 999
IF (LETTER.EQ.ITAB) CALL TABS !check for a tab
CALL ICHOUT(LETTER) !echo character
100 CONTINUE
CALL CRSIL(0)
CALL VTEND
...
parameters: letter integer input Ascii value of the character to output.
or
character i/p The character to output.
routines called: none
level: level 0
author: G.Gibbs
date: July 1982, character variables July 1984
CRJOUT ROUTINE
Category: J22
Title: CRJOUT (ICDATE,JDAY,IFMT)
Language: MACRO
Descrption: Outputs date in character form, to one of an array, a character
variable or your terminal. In the latter case the name of
the array should be replaced by the number 1,
e.g. CALL CRDOUT(1,JDAY,IFMT)
The degree of verbosity used is defined by IFMT, increasing
from 27-Nov-81 for IFMT=1
to Friday, 27 November 1981 for IFMT=4
Sufficient space must be allowed in the array, if used, to
allow for the maximum number of characters you are likely
to get +1.
Parameters: ICDATE array or OUTPUT Date in character form
character
JDAY integer OUTPUT The Julian day number
IFMT integer INPUT The verbosity of output
from 1 to 4
Author: G.Gibbs
Date: November 1981, character variable July 1984
CREMOT ROUTINE
CATEGORY: K10
TITLE: CREMOT(FILNAM,UNIT,ERROR)
LANGUAGE: MACRO-20/MACRO-11
DESCRIPTION: The purpose of this routine is to allow a programmer on the 20
to spool a file to the OKI dot matrix plotter or the Calcomp plotter
connected to our 11/34 system. This subroutine is both FORTRAN and
MACRO callable. The routine communicates with our own network spooler,
FLXSYS, which maintains the queue of files to be printed, in the same
manner as the RPLOT program.
The plot spooler (FLXSYS) will delete the file once it has
been plotted. To remove files from the plot queue, use the PLDEL
program (see HELP RPLOT for details about this.)
To find out how long your file may take before it
is plotted, type the IPQ command. This will give information similar to
INFORMATION (ABOUT) OUTPUT-REQUESTS, and show you where you are in the
queue.
METHOD: The FORTRAN routine merely reformats the arguments, and then calls
the routine at the MACRO entry point. This routine uses the IPCF system
to communicate with [SYSTEM]FLXSYS, using a message format described in
FLXSYS.DOC . The error (if any) returned to the user is that received
FLXSYS. These are very few, and apart from an illegal unit number, the
only one likely to be recieved is F$$QFU (queue full). This will occur
if there are more than 64 files waiting to be plotted. Most errors
will be handled by the spooler, which will recover, and need not concern
the user.
PARAMETERS:
FORTRAN: FILNAM - An array containing the filename in ASCIZ or
a quoted string, e.g. 'OKI.DAT' . The file must not be open
at the time of the call.
UNIT - An integer constant or variable to specify the unit
number to plot on - 0 for the OKI, 1 for the Calcomp
ERROR - An integer variable to receive the error/completion code
(These are described below.) This variable MUST be zero on
entry to the routine. It is reserved for future use as an
input parameter.
MACRO: Enter with T1/ JFN of file to be plotted (must be closed)
T2/ Unit number (as above)
T3/ 0 (As above)
Returns: +1 always with
T3/ Error/completion code, as follows:
ERRORS:
Defined error numbers:
0/ Call completed succesfully
5/ Unknown system error - this indicates a bug in the spooler, and
should be reported to Kevin.
6/ Couldn't find filename (Fortran only)
7/ Illegal unit number
8/ Couldn't open file for read
AUTHOR: K. Ashley
DATE: 21-AUG-81
REVISED: 15-SEP-81,2-Dec-81
CRERBR ROUTINE
category: C41
title: subroutine crerbr(x,y,vminus,vplus,width,log)
language: fortran
description: Draws error bars around the indicated position.
If one of the uncertainties is zero then that "half" of
the bar will not be drawn.
parameters: x }real input { co-ordinates to draw error bars
y } { around, in the user's units.
vminus }real input { uncertainties in the -ve and +ve
vplus } { directions in the user's units.
width real input width of the error bars in mm.
log logical input .true. gives a vertical bar
.false. gives a horizontal one
routines called: gino-f and ginograf routines
level: level 0
author: G.Gibbs
date: May 1982
CRFLIN ROUTINE
Category: M3
Title: CRFLIN(A,MESAGE,IFT)
Language: MACRO
Description: Outputs MESAGE to your terminal as a prompt and accepts a
real number from the keyboard,terminated with a new line or 80
characters, full ^R,^W,^U facilities are available.
IFT is 0 if successful and 1 if erroneous.
Parameters: A Real Output The number entered(if successful)
MESAGE Array Input The prompt to be sent to terminal
This should be as a quoted string
eg:- 'PLEASE ENTER NUMBER >>'
If it is neccessary to have a
variable prompt it must terminate
with a <null>.This may be inserted
using PK , thus if the message has
24 characters you would require
CALL PK(MESAGE,0,25)
IFT integer Output 0 =success
1 = fail
Author: G.Gibbs
Date: 5th June 81
CRFNIN ROUTINE
Category: M4
Title: CRFNIN(MESAGE,MODE,DEFNAM,DEFEXT,DFILEN,IPPN,IFT)
Language: MACRO
Description: Outputs MESAGE to your terminal as a prompt and accepts a
file name from the keyboard,terminated with a new line.
Recognition is available using the escape key, and defaults
may be provided for the name and/or the extension. The file
name is returned together with the 'project programmer number'
which is required for opening files in a different directory.
^R,^W,^U facilities are available, although ^U will erase
the prompt, however, ^R will restore it.
The file may be required to exist or not-exist or don't care
and a flag is returned to indicate success or not.
Parameters: MESAGE Array Input The prompt to be sent to terminal
This should be as a quoted string
eg:- 'PLEASE ENTER FILE NAME >>'
If it is neccessary to have a
variable prompt it must terminate
with a <null>.This may be inserted
using PK , thus if the message has
24 characters you would require
CALL PK(MESAGE,0,25)
MODE integer input 0 = the file may or may not exist
1 = the file must not exist
2 = the file must exist
DEFNAM integer input The default file name in quotes
e.g. 'FRED', or 0 if there is to
be no default
DEFEXT integer input The default file extension in quotes
e.g. 'DAT', or 0 if there is no default
DFILEN double output The complete file name with extension
precision in up to 10 characters
IPPN integer output The project programmer number for
array(2) the directory. This may be passed
staight to the OPEN statment
IFT integer Output 0 =success
1 = fail
Example: DIMENSION IPPN(2)
DOUBLE PRECISION DFILEN
100 CONTINUE
CALL CRFNIN ('FILE NAME>',2,'FRED','DAT',DFILEN,IPPN,IFL)
IF (IFL .NE. 0) GO TO 100
OPEN(UNIT=8,DEVICE='DSK',FILE=DFILEN,ACCESS='SEQIN'
+ ,DIRECTORY=IPPN)
========= OR ===========
DIMENSION IPPN(2)
DOUBLE PRECISION DFILEN
100 CONTINUE
CALL CRFNIN ('FILE NAME>',0,0,0,DFILEN,IPPN,IFL)
IF (IFL .NE. 0) GO TO 100
OPEN(UNIT=9,DEVICE='DSK',FILE=DFILEN,ACCESS='SEQOUT')
Author: G.Gibbs
Date: 5th June 81
CRFREQ ROUTINE
CATEGORY: G14
TITLE: SUBROUTINE CRFREQ(X,N,Y,NO,XLOW,XHIGH)
LANGUAGE: FORTRAN
DESCRIPTION: GIVEN AN ARRAY (X) OF VALUES ASSIGNS THEM TO THE
REQUIRED INTERVAL, WHERE THERE ARE N INTERVALS BETWEEN
XLOW AND XHIGH SO THAT THE FIRST INTERVAL IS FROM XLOW
UPTO XLOW + (XHIGH-XLOW)/N. FREQUENCY COUNTS FOR EACH
INTERVAL ARE OUTPUT IN THE ARRAY Y. Y(N+1) AND Y(N+2)
CONTAIN THE NUMBER LT XLOW, AND GE XHIGH RESPECTIVELY
PARAMETERS: X REAL ARRAY I/P DATA TO BE ASSIGNED
NAME
N INTEGER I/P NO. OF VALUES
Y REAL ARRAY O/P ARRAY CONTAINING FREQUENCY COUNTS
NAME
NO INTEGER I/P NO.OF INTERVALS + 2
XLOW REAL I/P LOWER LIMIT
XHIGH REAL I/P UPPER LIMIT
ROUTINES CALLED: CRCLR
LEVEL: LEVEL 1
AUTHOR: L.J.DOBSON
DATE: DECEMBER 1977
CRFUNC ROUTINE
CATEGORY: G12
TITLE: FUNCTION CRFUNC(NF,C,NC,X)
FUNCTION CRUSFN(NF,C,NC,X)
FUNCTION CRMLFN(NF,X,NARG,C,NC)
LANGUAGE: FORTRAN
DESCRIPTION: RETURNS VALUE OF REQUIRED FUNCTION OF X, IF NF IS .GT.
0-9 THEN THE FUNCTION CRUSFN IS CALLED, THIS IS A
USER DEFINED FUNCTION OF X. THE USER CAN VARY THE
CONSTANTS SUPPLIED TO A STANDARD FUNCTION OR HIS OWN
FUNCTION. N.B. THERE IS NO CHECK THAT THE USER HAS
SUPPLIED THE CORRECT NO. OF CONSTANTS. THE FUNCTION
CRUSFN BELOW IS THE ONE USED IF THE USER HAS NOT
SUPPLIED HIS OWN:- FUNCTION CRUSFN(NF,C,NC,X)
HER DIMENSION C(NC)
CRUSFN=X
RETURN
END
IF NF IS -VE THEN CRMLFN IS CALLED, IF THE USER HAS NOT
SUPPLIED HIS/HER OWN VERSION THEN THE FOLLOWING IS USED
:- FUNCTION CRMLFN(NF,X,NARG,C,NC)
DIMENSION X(NARG),C(NC)
CRMLFN=0.0
RETURN
END
PARAMETERS: NF INTEGER I/P FUNCTION NO. REQUIRED. FOR
FUNCTIONS 1-9 SEE BELOW
C REAL ARRAY I/P CONSTANTS IN FUNCTION
NAME
NC INTEGER I/P NUMBER OF CONSTANTS
X REAL I/P X VALUE TO CALCULATE FUNCTION FOR
NF=0 CRFUNC= X
1 = C(1)+C(2)*X
2 = C(1)+C(2)*X+C(3)*X**2
3 = C(1)+C(2)*X+C(3)*X**2+C(4)*X**3
4 = C(1)+C(2)*LOG10(X+C(3))
5 = C(1)+C(2)*LOG(X+C(3))
6 = C(1)+C(2)*EXP10(X+C(3))
7 = C(1)+C(2)*EXP(X+C(3))
8 = C(1)+C(2)*(X+C(3))**C(4)
9 = C(1)/(X+C(2))
ROUTINES CALLED: ALOG10,ALOG,EXP10,EXP,CRUSFN,CRMLFN (CRFUNC)
LEVEL: CRFUNC -LEVEL 1, CRUSFN -LEVEL 0, CRMLFN -LEVEL 0
AUTHOR: L.J.DOBSON
DATE: DECEMBER 1977
CRGAM ROUTINE
category: E42
title: Function crgam(x,alpha,beta,upper,ifault)
language: fortran
description: incomplete gamma integral, i.e. tail area of the gamma
distribution wtih parameters alpha and beta. Upper should
be set .TRUE. to get upper tail area, use .FALSE. for lower
tail area.
method: slightly modified version of Applied Statistics algorithm
references: AS 32 by G.P.Bhattacharjee (Appl. Statist.,19,285,1970)
parameters: x real input gamma variate
alpha real input parameter 1 (shape)
beta real input parameter 2 (scale)
upper logical input tail area selector
ifault integer output fault indicator :-
0 if ok
1 if alpha =< 0
2 if x < 0
common blocks: none
own variables: no
routines called: s14abf (nag)
level: level 0
author: J.P.Royston
date: October 1982
DATA-ENTRY
CATEGORY N
Question/answer data entry
---------------------------
N1 TEXTIN
Reads free text from terminal after prompt
N2 HOSPNO
Reads a hospital number after a prompt
N3 DATEIN
Reads a free-format date after prompt, converts to Julian
N4 KYWORD
Reads one of a list of keywords or phrases, with recognition
N5 SEXIN
Reads a sex
N6 YESNO
Reads yes or no
N7 NUMIN
Reads an integer after a prompt
N8 REALIN
Reads a real after a prompt
N9 TABLE
Builds a keyword table for KYWORD
N10 CROPEN
Reads a filename, with recognition, and opens the file
N11 TBLOOK
Looks up a character string in a table created by TABLE
N12 TBRLOC
Relocates lookup table pointers after reading from disk.
N13 KYALOW
Allows special characters to be entered as part of keywords
N14 KYDALW
Prevents certain characters being considered part of keywords
o 5 normal plots on the plotter
41 CRERBR
draws error bars
CRGRIT ROUTINE
CATEGORY: C34
TITLE: SUBROUTINE CRGRIT
LANGUAGE: FORTRAN
DESCRIPTION: GRAPH DRAWING ROUTINE NO. 1
THIS INITIALISES AND SETS THE GRAPH DRAWING PARAMETERS
TO THEIR DEFAULT VALUES AND SO MUST BE THE FIRST
ROUTINE CALLED. SUBSEQUENT CALLS JUST RESET THE
PARAMETERS TO THEIR DEFAULT VALUES.
PARAMETERS: NONE
REFERENCE: DOCAS REPORT ON PLOTTING PACKAGE
ROUTINES CALLED: CRMOVE
LEVEL: LEVEL 1
AUTHOR: L.J.DOBSON
DATE: DECEMBER 1977
CRGRLB ROUTINE
CATEGORY: C35
TITLE: SUBROUTINE CRGRLB(I,LAB)
LANGUAGE: FORTRAN
DESCRIPTION: GRAPH DRAWING ROUTINE NO. 2
THIS SETS UP LABELS OF THE REQUIRED TYPE FOR USE BY
CRGRAPH OR CRHIST; THE DEFAULT IS NO LABELS. EACH
LABEL WILL ONLY BE USED ONCE. THE LABEL ITSELF MUST
BE A CHARACTER STRING OF UPTO 80 CHARACTERS, INCLUDING
THE STRING TERMINATOR '*.' IT CAN ALSO INCLUDE SWITCHES
SO THAT UPPER AND LOWER CASE LETTERS CAN BE PLOTTED, IF YOU
CHOSE NOT TO USE LOWER CASE CHARACTERS IN 'LAB'.
THE DEFAULT TO START IS UPPER CASE, THEN '*L' WILL SWITCH
TO LOWER CASE AND '*U' WILL SWITCH TO UPPER CASE. TO
OUTPUT AN '*' JUST PUT 2 TOGETHER, '**'.
PARAMETERS: I INTEGER I/P TYPE OF LABEL. 1= X AXIS
2= Y AXIS, 3= GRAPH TITLE,
4= LINE,POINT OR HISTOGRAM
BAR LABEL
LAB INTEGER ARRAY I/P LABEL CHARACTER STRING
OR TEXT CONSTANT
REFERENCE: DOCAS REPORT ON PLOTTING PACKAGE
ROUTINES CALLED: CRGRER,IPK,PK,CRMOVE
LEVEL: LEVEL 1
AUTHOR: L.J.DOBSON
DATE: DECEMBER 1977
CRHACH ROUTINE
CATEGORY: C30
TITLE: SUBROUTINE CRHACH(XMIN,XMAX,YMIN,YMAX,MARK)
LANGUAGE: FORTRAN
DESCRIPTION: HATCHES OR SHADES A RECTANGULAR AREA IN THE CURRENT
LINE MODE. THE PERPENDICULAR DISTANCE BETWEEN LINES IS
THE CYCLE LENGTH OF THE LINE TYPE BEING USED, IN THE
CASE OF SOLID LINES IT IS THE DOTTED LINE CYCLE.
N.B. HATCHING IS EXPENSIVE ON PLOTTING TIME AND
THEREFORE SHOULD ONLY BE USED WHEN IT IS ESSENTIAL.
PARAMETERS: XMIN,XMAX, REAL I/P LIMITS OF RECTANGLE TO BE
YMIN,YMAX HATCHED
MARK INTEGER I/P HATCHING TYPE REQUIRED:-
0- NONE
1- 45 DEG. DIAGONAL LINES
2- 1 AND 3 COMBINED
3- 135 DEG. DIAGONAL LINES
4- HORIZONTAL LINES
5- 4 AND 6 COMBINED
6- VERTICAL LINES
REFERENCE: DOCAS REPORT ON PLOTTING PACKAGE
ROUTINES CALLED: GINO-F ROUTINES
LEVEL: LEVEL 0
AUTHOR: L.J.DOBSON/G.GIBBS
DATE: DECEMBER 1981
CRHALT ROUTINE
Category: K9
Title: CRHALT
Language: FORTRAN calling MACRO routine
Description: Halts program.
Any open files are closed and then the program is halted
without producing the usual STOP &c. on the terminal screen.
Routines called: CRHLTF
Level: 1,CRHLTF Level 0
Author: G.Gibbs
Date: 17-June-1981
CRHIAX ROUTINE
CATEGORY: C38
TITLE: SUBROUTINE CRHIAX(XLOW,XHIGH,IBP,IEP)
LANGUAGE: FORTRAN
DESCRIPTION: GRAPH DRAWING ROUTINE NO. 5
THIS SETS UP THE EXTRA PARAMETERS REQUIRED FOR A
HISTOGRAM AND SO MUST BE CALLED BEFORE CALLING CRHIST.
PARAMETERS: XLOW REAL I/P LOWER LIMIT OR VALUE OF FIRST BAR
XHIGH REAL I/P UPPER LIMIT OR VALUE OF LAST BAR
IBP, INTEGER I/P PERCENTAGE POINTS IN EACH INTERVAL
IEP OF THE START AND FINISH POINTS OF
EACH BAR.
REFERENCE: DOCAS REPORT ON PLOTTING PACKAGE
ROUTINES CALLED: CRGRER
LIBRARY: SUBGROUP-PLT
LEVEL: LEVEL 1
AUTHOR: L.J.DOBSON
DATE: DECEMBER 1977
CRHIST ROUTINE
CATEGORY: C39
TITLE: SUBROUTINE CRHIST(I,X,N,ISEL)
LANGUAGE: FORTRAN
DESCRIPTION: GRAPH DRAWING ROUTINE NO. 6
THIS DRAWS GRAPH AXES AND HISTOGRAMS OF VARIOUS TYPES.
THE GRAPH AXES ARE DRAWN WHEN NO AXES HAVE PREVIOUSLY
BEEN DRAWN OR THE PREVIOUS GRAPH HAS BEEN FINISHED,
I.E. ANY CALL TO CRHIST CAN DRAW THE AXES. THE Y AXIS
WILL BE 'BEAUTIFIED' UNLESS STATED OTHERWISE (SEE
CRGRAF). A GRAPH IS FINISHED BY SETTING THE HISTOGRAM
TYPE OR GRAPH TYPE (IN CRGRAF) NEGATIVE.
N.B. A CALL TO CRHIAX MUST BE MADE BEFORE CRHIST.
PARAMETERS: I INTEGER I/P HISTOGRAM TYPE. THE TYPES ARE:-
1=DRAW AXES ONLY (Y AXIS NOT
BEAUTIFIED
2=DRAWS A HISTOGRAM GIVEN ACTUAL
DATA WHICH NEEDS ASSIGNING TO
INTERVALS, CALCULATING THE
FREQUENCY COUNTS AND PLOTTING THEM
3=DRAWS A HISTOGRAM GIVEN THE ACTUAL
FREQUENCY COUNTS FOR EACH INTERVAL
OR CATEGORY IN THE REQUIRED ORDER.
11=DRAWS AXES ONLY (Y AXIS IS
BEAUTIFIED)
X REAL ARRAY I/P INPUT DATA ARRAY
N INTEGER I/P NO. OF POINTS IN INPUT ARRAY
ISEL INTEGER I/P THIS IS A TWO DIGIT SELECTOR.
FOR HISTOGRAMS WHERE ONLY AXES ARE
DRAWN (I=1,11) THE 1ST DIGIT IS
REDUNDANT AND THE 2ND DIGIT INDICATES
THE TYPE OF AXIS OFFSET REQUIRED
(SEE CRGRAF).
FOR OTHER HISTOGRAMS THE 1ST DIGIT
INDICATES THE LINE TYPE USED FOR THE
HATCHING (SEE CRGRAF) AND THE 2ND
DIGIT INDICATES THE TYPE OF
HATCHING REQUIRED (SEE CRHACH C30)
REFERENCE: DOCAS REPORT ON PLOTTING PACKAGE
ROUTINES CALLED: CRFREQ,CRMNX,CRSCAL,CRGRAF,CRBARS,
+GINO-F AND GINOGRAF ROUTINES
LEVEL: LEVEL 4
AUTHOR: L.J.DOBSON/G.GIBBS
DATE: DECEMBER 1981
CRMTOC ROUTINE
Category: J18
Title: CRMTOC ( IMONTH,MONTH)
Language: FORTRAN
Description: Convert integer month to characters. A month in integer form
is converted to a 3 character abbreviation of the month's
name, thus 1 becomes 'Jan'.A month of 0 is assumed unknown
and is returned as '000'. An erroneous month, i.e.
0>MONTH>12 returns 'XXX'.
Parameters: IMONTH INTEGER INPUT month as integer
MONTH INTEGER OUTPUT month as characters
See also: the character function MONTH
Author: G.Gibbs
Date: April 1981
CRLBOP ROUTINE
Category: K16
Title: CALL CRLBOP(LIBRARY,MODULE,JFN,START,ERR)
Purpose: Opens a LBR-format library file for reading
Language: MACRO-20
Description: This routine can be called from FORTRAN to read
files that are stored in LBR-format libraries directly. CRLBOP is used
to "open" a specific module in the library, and CRLBRD to read records
from the file. LIBRARY is the library file name, either a quoted string
or an A5 array terminated by a null ; MODULE is the name of the specific
module desired, in a similar format. JFN and START are values returned
by this routine which must be passed to the CRLBRD routine to enable
it to read the file. The user should not modify these words. ERR is
returned zero on success, else is positive to indicate an error
condition.
Parameters:
LIBRARY Input Array or quoted string Name of library
(type .LBR assumed)
MODULE Input Array or quoted string Name of module to retrieve
JFN Output Integer Job file number for CRLBRD
START Output Integer File address for CRLBRD
ERR Output Integer Return status (see below)
Errors: 0 Success
1 - Cannot find library
2 - Cannot open library
3 - File is not a library, or format incorrect
4 - Cannot locate module
5 - general JSYS error
Level: 0
Author: Kevin Ashley
Date: August 1982
CRLDAV ROUTINE
CATEGORY: K13
NAME: CRLDAV(LOAD1,LOAD5,LOAD15)
PURPOSE:
Returns the 1-minute, 5-minute and 15-minute load averages (as
displayed by the SYSTAT command) as three single-precision floating
point numbers. This can be useful for a program that requires fast
terminal response to be useful, ie with an interactive display.
PARAMETERS: LOAD1 Real output 1-minute load average
LOAD5 Real output 5-minute load average
LOAD15 Real output 15-minute load average.
Author: K. Ashley 19-Jan-82
Language: Macro-20
Level: 0
CRLHST ROUTINE
Category: D12
Title: CRLHST (IFREQ,MM,DATA,N,LENG,NCHAN,IND)
Language: FORTRAN
Description: plots histogram on the line printer across the page
This routine will accept the data in DATA(N) if
IND = 1, if the frequencies are available they should be
supplied in IFREQ(MM) with IND = 0,otherwise they will be
calculated and returned.The length of the plot is controlled
by LENG (in lines maximum)if .LE.10 a value of 50 is used.
Failure is indicated by IND set to 2
Reference: Algorithm AS 45 Applied Statistics (J.R.Statist.Soc. C)
(1971) vol.20,No.3
Parameters: IFREQ integer input vector of frequencies from which
array output lists printed
MM integer input number of classes into which the
data are to be grouped, when given
in DATA ,and size of array IFREQ.
DATA real input the original data from which the
array frequencies are to be obtained.
If these are supplied in IFREQ then
DATA is not used.
N integer input size of array DATA
LENG integer input maximum number of lines of print
which output will occupy.A minimum
of 11 is required & if less will be
set to 50.
NCHAN integer input channel number to receive output
IND integer input type of data input and
output failure indicator
=0 frequencies supplied
=1 data only supplied
=2 returned if all values equal
Routines called: CRSCL2
Level: 1,CRSCL2 Level 0
Author: D.N.Sparks
Date: 1971
CRLNFC ROUTINE
CATEGORY: G10
TITLE: FUNCTION CRLNFC(J,IFAULT)
LANGUAGE: FORTRAN
DESCRIPTION: NATURAL LOGARITHM OF FACTORIAL FOR NON-NEGATIVE
ARGUMENT. THE LOG-FACTORIAL IS MUCH EASIER TO HANDLE
THAN THE FACTORIAL ITSELF, SINCE 58! IS TOO BIG FOR
THE MACHINE TO HOLD, WHEREAS LOG(8388607!) IS WELL
WITHIN ITS RANGE. AN EXPRESSION SUCH AS (M+N)!/M!N!
MAYBE CALCULATED AS:
EXP(CRLNFC(M+N,IFA)-CRLNFC(M,JFA)
-CRLNFC(N,KFA))
IF(IFA+JFA+KFA.NE.0) GO TO ....
WITHOUT RISK OF OVERFLOW.
METHOD: TABLE LOOK-UP FOR SMALL ARGUMENT.
STIRLING'S FORMULA FOR LARGE ARGUMENT.
PARAMETERS: J INTEGER INPUT ARGUMENT
IFAULT INTEGER OUTPUT 0 FOR VALID ARGUMENT
1 FOR NEG.ARGUMENT.
COMMON BLOCKS: NONE
OWN VARIABLES: NONE
COMPRESSED INTEGERS: YES
FORTRAN TRACEABLE: YES
ROUTINES CALLED: NONE
LIBRARY: SUBGROUP-CRC
AUTHOR: I.D.HILL
DATE: JULY 77
CRLOGGAM ROUTINE
CATEGORY: G11
TITLE: FUNCTION CRLOGGAM(X,IFAULT)
LANGUAGE: FORTRAN
DESCRIPTION: NATURAL LOGARITHM OF GAMMA FUNCTION FOR POSITIVE
ARGUMENT. THE ADVANTAGE OF USING THE LOGARITHM,
RATHER THAN THE GAMMA FUNCTION ITSELF, IS AS GIVEN
FOR G10.
METHOD: STIRLING'S FORMULA FOR LARGE ARGUMENT. RECURRENCE
RELATION (GAMMA(X+1)=XGAMMA(X)), FOLLOWED BY
STIRLING'S FORMULA FOR SMALL ARGUMENT.
PARAMETERS: X REAL INPUT ARGUMENT
IFAULT INTEGER OUTPUT 0 FOR VALID ARGUMENT
1 FOR NEGATIVE OR
ZERO ARGUMENT.
2 FOR ARGUMENT GREATER
THAN 3.2E74
COMMON BLOCKS: NONE
OWN VARIABLES: NONE
COMPRESSED INTEGERS: YES
FORTRAN TRACEABLE: YES
ROUTINES CALLED: NONE
LIBRARY: SUBGROUP-CRC
AUTHOR: I.D.HILL
DATE: JULY 77
CRMAIL ROUTINE
CATEGORY: K14
TITLE: CRMAIL(FILNAM,USRLST,ERROR,ERQUAL)
LANGUAGE: MACRO-20
DESCRIPTION: This routine allows a FORTRAN or MACRO programmer to
transmit a text file to another user or list of users as mail. Any
errors which occur are reported. Arguments defined as array or string
must be in ASCIZ, ie a string of characters terminated by a null byte.
If you use a quoted string, this occurs. If you build up a string in an
array, you must be sure to place a zero word or byte at the end of the
string.
PARAMETERS:
FILNAM Integer array or string : The name of the file containing the
mail. This will be transmitted as is,
The system mail program will add the
header lines which say "From:"...
and "To:"... and also the date and time.
USRLST: Integer array or string : An ASCIZ string specifying a list of
user names to receive the mail,
separated by commas.
ERROR: Integer output : Must be 0 on entry. Contains 0 on exit
for success, or an error number (see
below).
ERQUAL: Integer output : Must be 0 on entry. If ERROR is 0 on
exit, the contents of this variable are
undefined. Else, ERQUAL contains a JSYS
error code, or some other number which
identifies the error more exactly.
ERRORS:
Defined error numbers:
0/ Call completed succesfully
-1/ Could not find named file.
-2/ Could not open named file
-3/ Name in username list invalid (ERQUAL contains position of
first invalid name. NO mail has been sent.)
-4/ MAILER returned errors (ERQUAL contains further info.)
-5/ Some IPCF error occurred. (MAILER has crashed, or you have
exceeded your PID or message quota - unlikely.)
-6/ Failed opening temporary file
-7/ Failed writing temporary file
AUTHOR: K. Ashley
DATE: 26-Apr-82
CRMATH2 ROUTINE
category: G3
title: crmth2(i,j,iq,ir,iover)
language: macro
description: quotient and remainder.
iq, ir are the quotient and remainder on dividing
i by j. ir = i - iq*j, and ir has the same sign
as i. iover is normally set to 0, but if overflow
occurs (e.g. if j = 0), iover is set to 1 and iq,
ir are undefined.
parameters i integer input dividend
j integer input divisor
iq integer output quotient
ir integer output remainder
iover integer output overflow indicator
routines called: none
level: level 0
author: G.Gibbs
date: May 1981
CRMNIM ROUTINE
CATEGORY: G15
*NEW*
TITLE: SUBROUTINE CRMNIM(F,STEP,NOP,FUNC,MAX,IPRINT,NCHAN,
STOPCR,NLOOP,IQUAD,SIMP,VAR,FUNCTN,IFAULT)
LANGUAGE: FORTRAN
DESCRIPTION: FUNCTION MINIMISATION. WORKS ON THE SAME METHOD
AS G5 (NELDER), BUT WITH THE ADDITIONAL OPTION
OF ESTIMATING THE VARIANCE-COVARIANCE MATRIX OF
THE ESTIMATES BY FITTING A QUADRATIC SURFACE TO
THE MINIMUM.
PARAMETERS: F REAL ARRAY (NOP) INPUT: INITIAL VALUES OF THE
VARIABLES
OUTPUT: VALUES AT THE MINIMUM
STEP REAL ARRAY (NOP) INPUT: INITIAL STEP SIZES
NOP INTEGER INPUT: NUMBER OF VARIABLES,
INCLUDING THOSE TO BE
HELD FIXED.
1 < NOP < 21
FUNC REAL OUTPUT: FUNCTION VALUE AT
MINIMUM
MAX INTEGER INPUT: MAXIMUM ALLOWABLE NUMBER
OF FUNCTION EVALUATIONS
IPRINT INTEGER INPUT: <0 FOR NO PRINTED
OUTPUT WANTED
=0 FOR PRINTED OUTPUT
(ON UNIT NCHAN) OF FINAL
VALUES
>0 AS FOR IPRINT=0, BUT
WITH ADDITIONAL
PROGRESS REPORTS
DURING MINIMISATION
EVERY IPRINT FUNCTION
EVALUATIONS
**NEW** NCHAN INTEGER INPUT: OUTPUT CHANNEL NUMBER
STOPCR REAL INPUT: A SMALL NUMBER USED AS
CONVERGENCE CRITERION
NLOOP INTEGER INPUT: CONVERGENCE IS TESTED
FOR AFTER EVERY NLOOP
ITERATIONS
IQUAD INTEGER INPUT: =0 FINDS MINIMUM AND
RETURNS
#0 ALSO FITS QUADRATIC
SURFACE TO FIND
VAR-COV MATRIX
SIMP REAL INPUT: A SMALL NUMBER USED IN
FITTING QUADRATIC
SURFACE
VAR REAL ARRAY (NOP) OUTPUT: DIAGONAL ELEMENTS OF
INVERSE OF INFORMATION
MATRIX
FUNCTN REAL FUNCTION NAME OF THE FUNCTION
TO BE MINIMISED. THE
NAME MUST ALSO BE
DECLARED AS EXTERNAL IN
THE CALLING PROGRAM.
THE FUNCTION MUST HAVE
ONE ARGUMENT (A REAL
ARRAY OF DIMENSION NOP)
IFAULT INTEGER OUTPUT: 0 - NO FAULT
1 - NUMBER OF FUNCTION
EVALUATIONS EXCEEDS
MAX
2 - INFORMATION MATRIX
NOT POSITIVE
SEMI-DEFINITE
3 - NOP<2 OR NOP>20
4 - NLOOP<1
ROUTINES CALLED: CRC301 (WHICH CALLS CRC120)
AUTHOR: MODIFIED BY I.D.HILL FROM A SUBROUTINE BY D.E.SHAW
AND R.W.M.WEDDERBURN IN PROGRAM LIBRARY OF
ROTHAMSTED EXPERIMENTAL STATION
DATE: 3 MAY 1979
CRMNX-2 ROUTINE
CATEGORY: G13
TITLE: CRMNX(X,N,XMIN,XMAX)
CRMNX2(X,N,XMIN,XMAX,IMIN,IMAX,A)
LANGUAGE: FORTRAN
DESCRIPTION: FIND THE MINIMUM AND MAXIMUM OF N ELEMENTS IN
ARRAY X. IN CRMNX2 THE LOCATIONS OF THESE
ARE STORED IN IMIN AND IMAX RESPECTIVELY, AND
EXTREME VALUES (EG MISSING VALUE CODES) CAN BE
EXCLUDED FROM CONSIDERATION BY MEANS OF VALUE
OF A.
PARAMETERS: X REAL ARRAY INPUT DATA ARRAY
N INTEGER INPUT LENGTH OF ARRAY
XMIN REAL OUTPUT MINIMUM VALUE
XMAX REAL OUTPUT MAXIMUM VALUE
IMIN INTEGER OUTPUT POSITION OF MINIMUM
VALUE WITHIN X
IMAX INTEGER OUTPUT POSITION OF MAXIMUM
VALUE WITHIN X
A REAL INPUT A=1.0 ALL VALUES EXAMINED
A<1.0 VALUES .LE. A NOT
CONSIDERED
A>1.0 VALUES .GE. A NOT
CONSIDERED
ROUTINES CALLED: AMIN1,AMAX1
LEVEL: LEVEL 0
AUTHORS: L.J.DOBSON,D.G.ALTMAN
DATE: DECEMBER 1977,APRIL 1978
CRMOVE ROUTINE
CATEGORY A4
TITLE: CRMOVE (IA,IB,N)
LANGUAGE: MACRO
DESCRIPTION: HIGH-SPEED MOVE. IA,IB ARE ARRAYS
N WORDS STARTING AT IA ARE MOVED TO THE AREA STARTING
AT IB. THE BLock Transfer INSTRUCTION IS USED.
PARAMETERS: IA ANY INPUT START OF SOURCE AREA.
IB ANY INPUT START OF DESTINATION
AREA
N INTEGER INPUT NO. OF WORDS
TO BE MOVED.
ROUTINES CALLED NONE
LEVEL: LEVEL 0
AUTHOR G.GIBBS
DATE MARCH 1981.
CRUSER ROUTINE
Category: K8
Title: CRUSER(USER)
Language: MACRO
Description: Places in the array or character variable USER the current
user's name. If it is an array USER MUST be long enough to
receive this, n.b. 8 elements will hold the longest allowable
name of 39 characters.
Parameters: USER Array or output User's name
character
Routines called: None
Level: 0
Author: G.Gibbs
Date: 16 June 1981, character variables July 1984
CRMVCH ROUTINE
CATEGORY: A1
TITLE: SUBROUTINE CRMVCH(IA,I,JA,J,N)
LANGUAGE: MACRO
DESCRIPTION: MOVE CHARACTERS
N CHARACTERS ARE MOVED FROM THE I'TH POSITION OF IA
(COUNTING FROM 1) AND ONWARDS TO THE J'TH POSITION
OF JA AND ONWARDS. IF N , I OR J ARE NOT +VE,
THERE IS NO EFFECT.
THIS IS EQUIVALENT TO THE ICL ROUTINE COPY (A2) WHICH NOW
CALLS THIS ROUTINE. NOTE THE DIFFERENT SEQUENCE OF
THE PARAMETERS.
PARAMETERS: IA ANY INPUT AREA TO COPY FROM
I INTEGER INPUT CHARACTER NO. IN 'IA'
JA ANY INPUT AREA TO COPY TO
J INTEGER INPUT CHARACTER NO. IN 'JA'
N INTEGER INPUT NO OF CHARACTERS TO
BE MOVED
COMMON BLOCKS: NONE
OWN VARIABLES: NO
ROUTINES CALLED: NONE
LEVEL: LEVEL 0
AUTHOR: G.GIBBS
DATE: MARCH 1981
CRNIN ROUTINE
Category: M2
Title: CRNIN(I,MESAGE,IFT)
Language: MACRO
Description: Outputs MESAGE to your terminal as a prompt and accepts an
integer from the keyboard,terminated with a new line or 80
characters, full ^R,^W,^U facilities are available.
IFT is 0 if successful and 1 if erroneous.
Parameters: I Integer Output The number entered(if successful)
MESAGE Array Input The prompt to be sent to terminal
This should be as a quoted string
eg:- 'PLEASE ENTER NUMBER >>'
If it is neccessary to have a
variable prompt it must terminate
with a <null>.This may be inserted
using PK , thus if the message has
24 characters you would reqiure
CALL PK(MESAGE,0,25)
IFT integer Output 0 =success
1 = fail
Author: G.Gibbs
Date: 5th June 81
CRNTOI ROUTINE
Category: J19
Title: CRNTOI (IDATE,IDAY,MONTH,IYEAR)
Language: FORTRAN
Description: Convert character date to 3 integers.
A date in character form, with the month expressed as a number
(e.g. '22.11.80') is converted to 3 integer numbers
(i.e. 22 11 1980 )
Any non-digit may separate day and month, but a * separateing
month and year indicates a date between 1800 and 1899;
any other non-digit separator is taken to mean a 20th
centuary date.
This routine checks the validity of the date, and
returns values of 3 zeroes for a non valid date.
Parameters: IDATE INTEGER ARRAY INPUT Date in character form
IDAY INTEGER } {The 3 parts
MONTH INTEGER } OUTPUT {of the date
IYEAR INTEGER } {
Author: G.Gibbs
Date: April 1981
CRON-CROFF-CRTEST ROUTINE
CATEGORY: A24
TITLE: SUBROUTINE CRON(I,J)
SUBROUTINE CROFF(I,J)
LOGICAL FUNCTION CRTEST(I,J)
LANGUAGE: MACRO
DESCRIPTION: BIT SWITCHING ROUTINES
CRON SETS BIT J OF WORD I
CROFF UNSETS BIT J OF WORD I
CRTEST RETURNS THE STATE OF BIT J IN WORD I. THE
RESULT IS TRUE IF SET AND FALSE IF UNSET
BIT POSITIONS ARE NUMBERED FROM 0 TO 35, STARTING
WITH THE MOST SIGNIFICANT
PARAMETERS: I INTEGER VARIABLE INPUT/OUTPUT WORD CONTAINING BIT
OR ARRAY ELEMENT
J INTEGER INPUT POSITION OF BIT
ROUTINES CALLED: NONE
LEVEL: LEVEL 0
AUTHOR: G.GIBBS
DATE: APRIL 1981
CROOM ROUTINE
category: K16
title: subroutine CROOM(DIRLFT,SYSLFT)
language: MACRO-20
description: Returns the space left in your connected directory and connected
structure. The space left in your directory is calculated by subtracting
the number of pages used from your working quota, and is returned in
DIRLFT. SYSLFT receives the total number of pages free on the disk.
Note that if SYSLFT is below 200 pages, then the system prevents
everyone from writing to the disk, so that even if DIRLFT is large,
you will not be able to create any files.
This routine is useful for a program that is about to create
a large file (for plotting, say) and wishes to ensure that space is
available before doing so.
method: The GTDAL and GDSKC JSYSs are used.
parameters: DIRLFT Integer output Pages free in connected directory
SYSLFT Integer output Pages free on whole structure.
common blocks: none
own variables: no
routines called: none
level: level 0
author: Kevin Ashley
date: August 1982
CROPEN ROUTINE
category: N10
title: subroutine cropen(prompt,unit,status,fail,
help,defnam,defext,defdir,defdev,option)
language: macro
description: the routine prompts at the terminal for a file name and then
opens the file. The routine allows the user to make full use of
the tops-20 facilities for recognition of file names using the
escape key, and providing default names, &c. Only the first 4
arguments have to be supplied. Other arguments may be omited by
either giving a value of zero or by simply omitting adjacent,
unwanted, trailing arguments,viz:-
CALL CROPEN ('File name >>',30,'old',fail)
CALL CROPEN ('File 2 =',31,'unknown',fail,
+ 0,'fred','dat',0,0,'/dispose:delete')
The arguments consisting of text should be given in quotes as
shown above, or may be passed in an array but must terminate
with a null.
The routine will prompt repeatedly until an acceptable name,
as defined by the status argument, is input. If it is wished
to change this then the second word of the common block/cmdprm/
is set to the number of retries to be allowed (-1 is infinite).
If you wish to access the file name then it is stored in the
common block /cmdstg/ starting at word 171, together with any
options given in the final argument. This array may be given
to a later open statement using dialog=. See defbfr(56) under
common blocks below.
parameters: prompt any input the prompt to be output by the routine
unit integer input the fortran logical unit number for the
file to be opened on
status any input a keyword as supplied to the status
field in an open statement,'old''new'&c
fail integer output a failure indicator, 0= success
a failure in opening the file returns
the iostat error code, or enters dialog
help any input *optional* help message displayed in
reply to a ? .
defnam any input *optional* default file name (not
including the extension)
defext any input *optional* default filename extension
(not including the .).
defdir any input *optional* default directory
(including the < >)
defdev any input *optional* default device
option any input *optional* any further options you wish
to give to an open statement in the
format required for 'dialog='
common blocks: /cmdstg/cmdbfr(70),atmbfr(70),fdb(5),hlpbfr(25),
defbfr(56),gjfblk(15),argblk(5)
/cmdprm/initf,retrys,tried,endnse,raise,savret,
savp,exilab,useext,spare(3),CMDBLK(10)
(all variables integer)
routines called: none
level: level 0
author: G.Gibbs
date: August 1982
CRPOLY ROUTINE
CATEGORY: E28
TITLE: CRPOLY(X,Y,M,N,C,S,IER)
LANGUAGE: FORTRAN
DESCRIPTION: FITS THE POLYNOMIAL REGRESSION, (OF ORDER 1, 2 OR 3)
OF Y(I) ON X(I)
METHOD: THE REQUIRED REGRESSION COEFFICIENTS ARE FOUND BY
DIRECT SOLUTION OF THE RELEVANT SIMULTANEOUS
EQUATIONS.
IN INDETERMINATE CASES (E.G. A. CUBIC REQUESTED
WHERE THE DATA LIE ON A STRAIGHT LINE) THE HIGH
ORDER COEFFICIENTS ARE SET TO ZERO.
PARAMETERS: X REAL ARRAY(M) INPUT THE 'INDEPENDENT' VARIABLE
Y REAL ARRAY(M) INPUT THE 'DEPENDENT' VARIABLE
M INTEGER INPUT NUMBER OF DATA POINTS
N INTEGER INPUT DEGREE OF POLYNOMIAL
(MUST BE 1,2 OR 3 AND LESS
THAN M)
C REAL ARRAY(4) OUTPUT (1) CONSTANT TERM
(2) COEFFICIENT OF X
(3) COEFFICIENT OF X**2
(OR 0 IF N=1)
(4) COEFFICIENT OF X**3
(OR 0 IF N=1 OR 2)
S REAL ARRAY(4) OUTPUT (1) SUM OF SQUARES DUE
TO LINEAR FIT
(2) ADDITIONAL SUM OF
SQUARES DUE TO
QUADRATIC FIT
(OR 0 IF N=1)
(3) ADDITIONAL SUM OF
SQUARES DUE TO CUBIC
FIT (OR 0 IF N=1
OR 2)
(4) RESIDUAL SUM OF SQUARES
IER INTEGER OUTPUT 0 FOR SUCCESS
1 FOR ILLEGAL VALUE OF N
2 FOR ZERO VARIANCE
IN X ARRAY
3 FOR INDETERMINATE CASE
- LINEAR FIT WHERE
QUADRATIC OR CUBIC WAS
REQUESTED.
4 FOR INDETERMINATE CASE
- QUADRATIC FIT WHERE
CUBIC WAS REQUESTED.
AUTHOR: I.D.HILL
DATE: 21 OCTOBER 1977
ICHIN ROUTINE
Category: M5
Title: INTEGER FUNCTION ICHIN(IERR)
CHARACTER FUNCTION ICHIN(IERR)
Language: MACRO
Description: Obtains the next character from your terminal without waiting
for a <return> to be typed. It should however be noted that
a <return> will give a <line feed> as well, and that some
control characters will still have their usual effect viz:-
^C,^O,^S,^Q, and ^T.
The character is returned as the ascii value by the function,
if declared as integer, or as a character if so declared
It is necessary to call VTINIT before using ICHIN and you may
wish to call CRSIL to turn off echoeing of the characters as
they are entered e.g.
CHARACTER*1 ICHIN
CALL VTINIT
CALL CRSIL(1)
IF (ICHIN(IERR).EQ.'A') CALL ALPHA
CALL VTEND
Parameters: ICHIN integer OUTPUT The ascii value of the next character.
character O/P The next character.
IERR INTEGER OUTPUT An error indicator, 0=success
Level: level 0
Author: G.Gibbs
Date: 5 April 1982, character variables July 1984
CRRUN ROUTINE
category: K22
title: subroutine CRRUN(PRGNAM,COMAND,ERROR)
language: macro-20
description: Runs the program (expected to be in an EXE file)
specified by the first argument. Passes the command line specified
by the second argument via the 'rescan buffer'. This simulates running
a program called FRED, say, by typing
@FRED input,output
FRED could use CRRSCN to read this command if you typed it in.
Calling CRRUN('FRED.EXE','FRED input,output',ERROR)
has the same effect. Error is 0 on success, 1 if the program cannot
be found, or 2 if it cannot be run.
Alternatively, if you wanted to invoke an editor from your program, you
could use
CALL CRRUN('SYS:TV.EXE','EDIT myfile.txt',ERROR)
or
CALL CRRUN('SYS:SED.EXE','EDIT myfile.txt',ERROR)
if you prefer to use SED as your editor. The routine returns when the
program exits.
method: RSCAN% and GET% are used.
parameters: PRGNAM Array or text literal input Name of program file
COMAND Array or text literal input Command line
ERROR Integer output Completion code
common blocks: none
own variables: no
routines called: none
level: level 0
author: Kevin Ashley
date: April 1982
CRSHPE ROUTINE
CATEGORY: E30
TITLE: CRSHPE(X,N,SKEW,AKURT,GEARY)
LANGUAGE: FORTRAN
DESCRIPTION: CALCULATES USUAL COEFFICIENTS OF SKEWNESS & KURTOSIS;
ALSO MEAN DEVIATION/STD. DEVIATION, ANOTHER MEASURE OF
KURTOSIS KNOWN AS GEARY'S A STATISTIC. UPPER AND
LOWER PERCENTAGE POINTS OF GEARY'S A HAVE BEEN TAB-
ULATED FOR NORMAL SAMPLES OF SIZE 11-2000.
METHOD: USES STAT7B. A DIVISOR OF N (NOT N-1) IS USED IN
THE CALCULATION OF STANDARD DEVIATION.
REFERENCE: FOR SIGNIFICANCE LEVELS AS TESTS OF DEPARTURE FROM
NORMALITY, SEE BIOMETRIKA TABLES, VOL.1,P.207-8.
PARAMETERS: X REAL ARRAY(N) INPUT SAMPLE VALUES
N INTEGER INPUT SAMPLE SIZE
SKEW REAL OUTPUT COEFF. OF SKEWNESS
AKURT REAL OUTPUT COEFF OF KURTOSIS
GEARY REAL OUTPUT GEARY'S A STATISTIC
LEVEL: LEVEL 1
ROUTINES CALLED: STAT7B
AUTHOR: J.P.ROYSTON
DATE: 3 MAY 1978
CRSINT ROUTINE
category: A31
title: subroutine crsint(array,chars)
language: Fortran 77
description: Move characters from an array to a character variable.
The number of characters moved is defined by the length of the
character variable, as declared in the calling routine. If,
however, a null is encountered in the array, then the rest of
the character variable is padded out with blanks.
parameters: array any input area to copy from
chars character o/p area to copy to
common blocks: none
routines called: crmvch
author: G.Gibbs
date: August 1983
CRSIZE ROUTINE
category: K19
title: subroutine crsize('file.name',pages,bytes,bytsiz,err)
language: macro-20
description: Returns a file's size in pages and bytes, and returns the size
of the bytes. These are the three numbers seen in a VDIRECTORY command.
The byte size is the number in brackets, as in
FILE.NAME 4 11023(7)
means the file has 4 pages, 11023 bytes (characters) and the bytes
are each 7 bits (ie they are ASCII text). Can only be used for disk
files.
method: SIZEF% and GTFDB%
parameters: FILNAM Array or text literal input Name of file
PAGES Integer Output Pages in file
BYTES Integer Output Bytes in file
BYTSIZ Integer Output Bits per byte
ERR Integer Output Error code
0=Success
1=File not found
2=File exists, but
cannot find size (eg
file was TTY:)
common blocks: none
own variables: no
routines called: none
level: level 0
author: Kevin Ashley
date: August 1982
CRSOUT ROUTINE
category A30
title: crsout (array)
language: macro
description: Send a string of characters, terminated by a null, to the
terminal. The string may be in qoutes, e.g.:-
CALL CRSOUT('?ERROR ABANDON RUN')
or you may set up an array containing the characters,
which may include escape sequences and control characters,
provided that it is terminated with a null. You will have
to call VTINIT if you want the control characters to be
passed to the screen intact.
parameters: array any input the string of
characters
routines called none
level: level 0
author G.Gibbs
date February 1983.
CRTIMR ROUTINE
category: K26
title: SUBROUTINE CRTIMR(INTVAL,ROUTIN,FAIL)
language: macro-20
description: Calls a specified routine when a time limit expires
This subroutine is used to generate "interrupts" from Fortran. The
caller specifies a number of milliseconds, and either a subroutine
name, or a statement label in the second argument. CRTIMR returns
immediately, letting your program continue, but when the specified
number of milliseconds has elapsed, your program is suspended, and the
subroutine you named in the call is called. When it returns, your
program is allowed to continue. This can be useful for timeing out
a request to read from the terminal, for instance. Maybe you have a
program that prompts at the terminal to read something. It is possible
that sometimes, you may have left the program running, and not noticed
when it prompts again. Using CRTIMR, you can ring the terminal bell if
you do not answer the question within, say, 10 seconds. See the example
below.
To eliminate a request for an interrupt due to a previous
call to CRTIMR, call CRTIMR with an elapsed time of 0. You can only
have one outstanding timer request.
Unless it is disarmed by calling CRTIMR with a time of 0,
the timer continues to trigger at every time interval specified.
Thus, in the example below, the bell continues to ring every 10
seconds until the user answers the question.
NOTE: if you supply a subroutine name as the second parameter, you
must declare the subroutine EXTERNAL in the calling program.
You specify a statement label by prefixing it with a *, as in *100
Restrictions: The called subroutine must NOT perform any Fortran I/O.
If you need to perform I/O, call one of the CRC subroutines written
in Macro-20, or use DOJSYS to do it.
If you supply a statement label to be called when the timer triggers,
the statement cannot be in the main program.
The time interval cannot be more than one day in the future
(ie 1000*60*60*24 milliseconds.)
Example: To ring the bell every 5 seconds until the user answers
a question.
EXTERNAL WAKEUP !Subroutine to call to wake up user
.
.
.
CALL CRTIMR(10000,WAKEUP,FAIL) !Arm the timer to go off in 10 seconds
CALL CHECK(FAIL) !Make sure it worked
CALL NUMIN('Enter your new x value:',NX,FAIL)
CALL CHECK(FAIL) !Make sure that worked
CALL CRTIMR(0,0,FAIL) !Disarm the timer - they answered us
.
.
.
SUBROUTINE WAKEUP
CALL CRBELL(3) !Ring the bell 3 times every 10 seconds
RETURN !Leave interrupt, hope they answer soon
END
method: The TIMER% Jsys is used. All accumulators are saved over the
interrupt call, and restored before leaving interrupt context.
Parameters: INTVAL Integer Input Number of milliseconds to wait
before calling interrupt routine.
ROUTIN Routine/ Input Name of EXTERNAL routine to call
Label when time interval expires.
FAIL Integer Output Error code, 0=success
-1: You already have an outstanding
call to CRTIMR. Only one is
permitted.
-2: FOROTS could not allocate an
interrupt channel.
-3: System error: No timer blocks.
This should not occur.
common blocks: none
level: level 0
author: Kevin Ashley
date: January 1983
CRTWMY ROUTINE
category: G19
title: subroutine crtwmy(n,a,g,gamma,w,lw,f,rms,ifail)
language: Fortran
description: Twomey's method for the regularized solution of a
Fredholm integral equation of the first kind
(J.Ass.Comp.Mach. 97-101, 10, 1963).
The integral equation is expressed as a quadrature
and the smoothest new solution f(n), is obtained in
comparison with any existing solution f(n), in the
sense of minimizing its second differences.
parameters:
n integer input no of quadrature coefficients
a(n) real array input system vector
g(n) real array input response vector
gamma real input regularization parameter
w real array --- workspace
lw integer input length of w, at least n(3n+2)
f real array in/out solution vector
rms real output residual mean square
ifail integer output failure flag:-
0=success
1=not enough workspace
2=bad parameters (n<5, gamma<0)
3=internal error (f01clf)
4=internal error (f01caf)
5,6=ill-conditioning (f01abf)
7-10=internal error (f01ckf)
11-14=internal error (f01ckf)
15-18=internal error (f01ckf)
level: 2
routines called: rwtwm1 (calling crmove)
rwtwm2 (calling f01abf [calling f01acf], f01caf, f01ckf,
f01clf and f01def)
author: R.Wootton
date: 25-apr-83
CRVCR ROUTINE
CATEGORY: G17
TITLE: SUBROUTINE CRVCR(FSUMSQ,M,N,S,V,LW,W,IFAIL)
LANGUAGE: FORTRAN
DESCRIPTION: COMPUTES THE VARIAANCE-COVARIANCE AND CORRELATION MATRICES
FOR THE PARAMETERS ESTIMATED BY THE 'COMPREHENSIVE' LEAST-
SQUARES NAG ROUTINES E04FCF, E04HEF, E04GDF OR E04GBF.
PARAMETERS:
FSUMSQ REAL INPUT VALUE OF OBJECTIVE FUNCTION AT FINAL
POINT.
M INTEGER INPUT NUMBER OF RESIDUALS.
N INTEGER INPUT NUMBER OF PARAMETERS.
S REAL ARRAY(N) INPUT DIAGONAL MATRIX CONTAINING THE
SINGULAR VALUES OF THE ESTIMATED
JACOBIAN MATRIX FOLLOWING ITS
SINGULAR VALUE DECOMPOSITION.
V REAL ARRAY(N,N) INPUT ORTHOGONAL MATRIX CONTAINING THE
RIGHT-HAND SINGULAR VECTORS OF THE
ESTIMATED JACOBIAN AFTER ITS S.V.D.
(NOTE THAT PARAMETERS 1-5 ARE OUTPUT BY THE ABOVE NAG ROUTINES)
LW INTEGER INPUT LENGTH OF ARRAY W AS DECLARED IN THE
CALLING SEGMENT. MUST BE AT LEAST
N(2N+1).
W REAL ARRAY(LW) OUTPUT WORKSPACE/RESULTS. ON SUCCESSFUL
EXIT, THE FIRST N*N ELEMENTS OF W
CONTAIN THE V-C MATRIX AND THE NEXT
N*N THE CORRELATION MATRIX.
IFAIL INTEGER OUTPUT FAILURE FLAG: 0=SUCCESSFUL EXIT;
1=FAIL ON FIRST MATRIX MULTIPLICATION;
2=FAIL ON SECOND; 3=NOT ENOUGH
WORKSPACE SUPPLIED.
LEVEL: 1
ROUTINES CALLED:RWVCR (AT LEVEL 0) (CALLING F01CKF AND P01AAF)
AUTHOR: R.WOOTTON
DATE: 25-JAN-82
CRWAIT ROUTINE
CATEGORY: K15
TITLE: CRWAIT(MILLIS)
LANGUAGE: MACRO-20
DESCRIPTION: This routine suspends program execution for the specified
number of milliseconds. The accuracy of the routine is limited to the
nearest 20 milliseconds - the maximum period that may be specified is
2^27 milliseconds (18 hours, 38 minutes, 28 seconds and 864 milliseconds
). No errors are possible.
PARAMETERS:
MILLIS Input integer variable or constant:
Time to wait in milliseconds.
ERRORS:
None.
AUTHOR: K. Ashley
DATE: 15-Jul-82
CRWTST ROUTINE
CATEGORY: E34
TITLE: CRWTST(X,N,AVE,SD,W,PW,IOUT)
LANGUAGE: FORTRAN
DESCRIPTION: CALCULATES SHAPIRO AND WILK'S W STATISTIC, A TEST OF
NORMALITY OF THE N SAMPLE VALUES X. THIS STATISTIC
INDICATES HOW CLOSELY A NORMAL PLOT OF X APPROXIMATES
TO A STRAIGHT LINE. IT IS MEANT TO BE USED AS AN ADJUNCT
TO A NORMAL PLOT, AND HAS ONLY BEEN WORKED OUT FOR
SAMPLES OF SIZE BETWEEN 3 AND 50. THE APPROIMATE SIG-
NIFICANCE LEVEL PW OF W IS CALCULATED BY INTERPOLATION
IN THE RANGE .01.LE.PW.LE.0.99,VALUES BEING EXACT AT
.01,.02,.05,.10,.50,.90,.95,.98 AND .99. SMALL VALUES
OF W AND PW INDICATE DEVIATION FROM NORMAL DISTRIBUTION.
IF N IS LESS THAN 3 OR GREATER THAN 50, W AND PW ARE SET
TO ZERO, BUT AVE AND SD ARE STILL CALCULATED.
REFERENCE: 'AN ANALYSIS OF VARIANCE TEST FOR NORMALITY (COMPLETE
SAMPLES)' BY S.S.SHAPIRO AND M.B.WILK, BIOMETRIKA,1965,
52,PAGE 591.
PARAMETERS: X REAL ARRAY(N) INPUT SAMPLE DATA
N INTEGER INPUT SAMPLE SIZE
AVE REAL OUTPUT MEAN OF DATA
SD REAL OUTPUT STANDARD DEVIATION
W REAL OUTPUT W STATISTIC
PW REAL OUTPUT SIG. LEVEL OF W
IOUT INTEGER OUTPUT 1 = PW<.01
2 = .01<PW<.99
3 = PW>.99
4 = N<3 OR N>50
LEVEL: LEVEL 1
ROUTINES CALLED: STAT7,CRC154,SQRT,MOD,ALOG10
AUTHOR: J.P.ROYSTON
DATE: 22 AUGUST 1978
CRXPNG ROUTINE
category: K21
title: subroutine crxpng('directory-name',error)
language: macro-20
description: Expunges (ie removes all deleted files from) the directory
specified in the first argument. All space used by deleted files is
released. This can be useful in a program that produces many temporary
output files (ie plot files).
The directory name may either be in the form PS:<A-USER>, or
may be a logical name (such as HERE:). If in the first form, the
structure ID (PS: or US: or MELON:) must always be present, as must
the angle or square brackets. Wildcards (* or %) may not be used.
If in the second form, you may supply the logical name either
with or without its terminating colon. Examples:
CALL CRXPNG('PS:<C-CHAPLIN>',IER)
CALL CRXPNG('HERE:',IER)
CALL CRXPNG('DP',IER) !DP is assumed to be a logical name
To use this routine effectively, you must either be connected to
the directory, or have expunge access (which means that you can connect
without giving a password.)
method: DELDF%, LNMST% and RCDIR% are used
parameters: DIRNAM Array or text literal input Name of directory
ERR Integer Output Completion code:
0: Success
1: Cannot find directory
2: Directory cannot be expunged.
common blocks: none
own variables: no
routines called: none
level: level 0
author: Kevin Ashley
date: October 1982/August 1983
CURHOM ROUTINE
CATEGORY: L2
NAME: CURHOM
PURPOSE: Moves the cursor to the top left of the screen ("home")
PARAMETERS: None
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
CURUP ROUTINE
CATEGORY: L1
NAME: CURUP(N)
PURPOSE: Moves the cursor up the number of times specified.
PARAMETERS: N Integer Input Number of places to move cursor (+ve)
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
DATA-ENTRY ROUTINES
Category: N0
A number of data entry routines are available to make it easier
for a user to write a questionaire-type input program. The suite of
routines provide a consistent interface both the programmer and user,
allowing for such facilities as retyping of prompts after errors,
range checking, defaults, and recognition of the special terms
EXIT and UNKNOWN for any question, and ^ to indicate the desire to
go back and re-enter a previous question. The method by which these
are handled takes much of the tedium of such matters away from the
programmer.
To see the range of routines available, type
INFO INDEX DATA-ENTRY to the SUBS program, which will give a list of
routines ; writeups on each one may be obtained by typing INFO followed
by the name of the routine.
Each of the routines usually has a set of compulsory
parameters which must be supplied, and a set of optional parameters.
Optional parameters which come after any arguments you supply may
simply be omitted from the call. Optional parameters which come
between parameters you wish to supply should be replaced by 0. One of
these optional parameters to every routine is BACK, which is a
statement label to go to if the user types "^", indicating that they
wish to re-enter their answer to the previous question. You supply
this parameter as noted in the FORTRAN reference manual, by placing a
* in front of the statement label in the parameter list, as in
CALL YESNO(' More ?',YEA,FAIL,*10)
to jump to label 10, or by using ASSIGN to a variable, and then using
the variable in the call. Constructs such as this make it much easier
for you to write a data-entry program which is fairly flexible from
the users' point of view, but avoids too much repetetive coding on
your part. The jump to the statement label (usually a label for the
call to the routine that asked the last question) is done automatically.
It avoids lots of IF-type statements occurring after each question/
answer sequence in the program, and avoids the usual problem of trapping
out-of-range errors, or checking that a word typed in belongs to one of
a short list of commands.
Each of these routines has a specific sort of answer
it will accept, and some also have ranges in which they must fall. An
answer that does not satisfy the criteria is in error, and one of two
things may be done - either an error message can be typed, and the user
reprompted for the answer, or an error code can be returned to the
calling program to do with as it wishes. Under normal circumstances,
an error message is output until valid input is received. Thus, your
program can be sure that what is input satisfies certain conditions.
Alternatively, you can tell the routines only to reprompt a certain
number of times, and then accept the answer anyway. The number of times
can be 0 (thus, no error is ever typed by the routines, but your program
receives an error code if an error occurs.) This communication is
established via a common block CMDPRM, described below.
COMMON /CMDPRM/initf,retrys,tried,endnse,raise,savret,
savp,exilab,useext,spare(3),cmdblk(10)
INITF - **internal** Non-zero if data entry routines have been
initialized. Do not alter this word.
RETRYS - Number of times to reprompt the user if an error occurs.
If this number is negative, reprompt for ever until a
correct answer is input. Set to -1 initially. May be
altered at any time, inspected at each call.
TRIED - **internal** Number of attempts made.
ENDNSE - If non-zero, indicates that the caller is willing to
allow garbage input to occur after the end of a field,
i.e. if we are reading a number, we are willing to
accept "3 eggs" as input as well as just "3". Initially
set to 0, in which case the error code 2 is returned, or
the user is reprompted if extraneous input should occur.
N O T E
One problem with this feature is that, if a
default has been supplied for a field, and the user
enters something totally incorrect (eg you wish to read
a number and they type WOT), the default for the field
is returned, without much indication to the calling
program. This is because the numeric field is terminated
by the first non-numeric character, so is defaulted,
and the text becomes part of the trailing text which you
have indicated our willingness to ignore. This text is,
however, returned in the CMDBFR.
RAISE - If non-zero, indicates that all input text should be
raised to upper case. This does NOT affect what is on
the screen, only what is stored in the command buffer,
and any text strings returned to your program.
This is useful in areas where letter case is important,
i.e. in key fields to be inserted into RAPPORT data
files, which consider text of differing cases to be
different. It does not affect the user typing at the
terminal, who is unaware of it.
SAVRET - **internal** Return address for reparse
SAVP - **internal** Saved stack pointer
EXILAB - If non-zero, indicates the statement to jump to if
the user types EXIT. To cause control to transfer to
statement 9000 when EXIT is typed, use code such as
the following:
ASSIGN 9000 TO EXILAB
This can be changed as the program continues, to
transfer control to different statements at different
points. Note that if this is 0, an error code of -2
will be returned instead if EXIT is typed.
USEEXT - Indicates whether the routines should allow EXIT or
UNKNOWN to be be typed in answer to a question.
Normally, EXIT or UNKNOWN can always be typed as an
answer, returning special fail codes to the calling
program, or jumping to a special label. If this
variable is set non-zero, the keywords EXIT and UNKNOWN
will not be permitted as answers. Can be changed at any
point in the program.
SPARE - Unused
CMDBLK - **internal** Command state block (see monitor calls
reference manual if you're interested.)
This common block need only be declared if you wish to alter values
of variables in the block. Another common block, CMDSTG, is used mainly
by the routines internally. However, one field in it may be of use if
you wish to access the actual text entered by the user as it was typed
(which may be especially useful if you want to examine their input in
case of an error).
COMMON /CMDSTG/cmdbfr(70),atmbfr(70),fdb(5),hlpbfr(25),
defbfr(56),gjfblk(15),argblk(5)
CMDBFR - Stores exact text of last line typed in.
ATMBFR - Stores exact text of last field typed in.
Usually the same as CMDBFR, unless rubbish was
entered after the real input (in which case
this contains the extraneous input.)
FDB - COMND function descriptor block
HLPBFR - Used to construct help messages if user does
not supply them.
DEFBFR - Used to contruct text for defaults.
GJFBLK - GTJFN argument block, zeroed on each call.
ARGBLK - FOROTS argument block for CROPEN.
Note: A copy of the writeup for all routines is in HLP:DATA-ENTRY.DOC
Example
=======
DATE ROUTINE
category: J4
title: subroutine date(text)
language: macro
description: returns today's date in character form, e.g.14-Aug-83 , in
either a 2 element array, or a character variable*10
references: Tops 10/20 Fortran language manual
parameters: text array/ output the date in character form
character
author: Dec
DATEIN ROUTINE
category: N3
title: subroutine DATEIN (PROMPT,JULIAN,FAIL,BACK,LOWER,UPPER,HELP,DEFALT)
language: MACRO-20
purpose: The routine prompts the terminal user with a supplied string, and then
reads a date, typed in any reasonable format. Dates may be numeric (in
which case day/month/year is assumed) or text, such as Feb 8 82 or
Wednesday, September 1st 1982. In general, anything unambiguous is
accepted.
The corresponding julian day number is returned. The program can
supply both upper and lower limits for the date (which are inclusive)
independently of each other, or supply neither.
A default can be supplied by the program( as a julian day no.).
This routine is part of the terminal data entry routine suite -
full facilities for recognition of defaults, retyping of prompts,
automatic recognition of UNKNOWN, EXIT and ^ are available. This
is more fully explained in the DATA-ENTRY section of this program,
which should be read in conjunction with this.
method: The COMND% Jsys is used.
parameters: prompt Array or input The prompt for the user
string
Julian Integer output Returned julian day number
fail Integer output Error code (see below)
back Label input *Optional*
Statement to jump to if
user uses ^ to go back one
question, or 0 to ignore this.
Lower Integer input *Optional*
Lower limit for date (inclusive)
or 0 if not desired.
Upper Integer input *Optional*
Upper limit for date (inclusive)
or 0 if not desired.
help Array or input *Optional*
string Help message to be output if
user types question mark, or
0 to ignore argument.
If no help is supplied, a
message will be constructed of
the form "Enter a date, before
1-jul-82, after 1-jul-81", with
limits only mentioned if they
are supplied in the call.
defalt Integer input *Optional*
Default to be used if user
types escape, or 0 if not
desired.
NOTE: Optional input arguments can be omitted if they are
all trailing, so the call
CALL DATEIN('Date:',JULIAN,FAIL) is valid, as is
CALL DATEIN('Date:',JULIAN,FAIL,*10,0,JTODAY(K),'Enter date
& of birth')
which supplies a label to return to if the user types ^ (the
label is 10),stores the date of birth in JULIAN, specifies no
lower limit, but specifies an upper limit of today's date, and
outputs a help message if a question mark is typed. No default
has been supplied, however.
You may use 0 for an optional argument if you wish to supply
some arguments that follow it, but you wish that argument to
be ignored (ie if you want to supply a default, but not a help
message.)
Error codes: 0 Success
1 Invalid input
2 Not confirmed (this is only possible if RETRYS is set
to a non-negative value and the user types EXIT,
UNKNOWN or ^ and then types rubbish after the command.)
5 Date is not in valid format
6 Date is out of range.
-1 User typed UNKNOWN.
-2 User typed EXIT (this code is only returned if you did
not initialize location EXILAB in common block CMDPRM
with the label to jump to).
-3 User typed "^" to go back a question (this code is only
returned if you did not supply a label to jump to on
this condition.)
common blocks: /cmdstg/cmdbfr(70),atmbfr(70),fdb(5),hlpbfr(25),
defbfr(56),gjfblk(15),argblk(5)
/cmdprm/initf,retrys,tried,endnse,raise,savret,
savp,exilab,useext,spare(3),cmdblk(10)
(All integer, see DATA-ENTRY for more info)
own variables: no
routines called: none
level: level 1
author: G.Gibbs / Kevin Ashley
date: August 1982
DOJSYS ROUTINE
category: K24
title: SUBROUTINE DOJSYS(JSNUM,ARG1,ARG2,ARG3,ARG4,LABEL)
language: macro
description: Executes any TOPS-20 monitor call.
This allows the high-level programmer (in ALGOL,PASCAL or FORTRAN)
to do ANYTHING that the system is capable of, providing that he or
she can work out what arguments to supply. This is not always easy.
A description of the monitor calls available may be found in the
TOPS-20 Monitor Calls Reference Manual. There are about 300-400 of
them, providing access to all system facilities. An introduction can
be found in the TOPS-20 Monitor calls User's Guide.
Monitor calls usually accept arguments in the machine's
accumulators when called from a MACRO program. When using this
subroutine, ARG1 contains the argument usually supplied in AC1,
ARG2 usually contains that supplied in accumulator two, etc. JSNUM
is the number of the JSYS required. Note that the numbers in the
reference manual are OCTAL. Label is an fortran statement label which
will be jumped to if the monitor call fails. Most calls return an error
code in ARG1. The error can always be obtained with the GETER jsys (
number 10 (decimal)). ERSTR (JSYS 9) can be used to print an informative
message on failure.
All arguments except JSNUM are optional. If an error occurs, and
the LABEL argument is not supplied, the program is halted, and an
informative message is printed giving the reason for the error.
N O T E
=======
Arguments ARG1 to ARG4 are all input/output. This is because
monitor calls often use them to return values as well as accept them.
The implication of this is that these arguments may NOT be literals,
constants or expressions. All of them must be assigned to variables or
array elements before using these in the call. The LABEL and JSNUM
arguments do not suffer from this restriction.
Hints: Functions LEFT and RIGHT can be very useful in retrieving
arguments that are returned in right or left halves of acs.
Function GETADR can be useful when one of the ac arguments
requires the address of a storage location. These routines are
all described in this library.
method: Horrific
parameters: JSNUM Integer Input Monitor call to execute
ARG1 Any Input/output Contents of first accumulator
ARG2 Any Input/output Contents of second accumulator
ARG3 Any Input/output Contents of third accumulator
ARG4 Any Input/output Contents of fourth accumulator
LABEL Label Input Statement to return to on error
Example: We wish to find out which terminal number our program is running
at. Perhaps we want to know if it is feasible to use graphics on
it. The GJINF% JSYS is most useful in this regard. Its number is
13 (octal).
CALL DOJSYS("13,USRNUM,DIRNUM,JOBNUM,TERNUM)
This call never fails. We also receive our user number,
number and job number, which may or may not be useful.
N O T E
A pint of beer (or other suitable replacement) is offered
to the first person to get COMND (Jsys 544(octal)) to work
using this method.
common blocks: none
level: level 0
author: Dan Murphy/Kevin Ashley
date: November 1982
EASTER ROUTINE
category: J23
title: Integer function easter(year)
language: fortran
description: returns the julian day number of Easter Sunday for the given
year. It is valid between 1583 and 4200, although if a year
of less than 100 is given, then it is assumed to be in the 20th
century. An invalid year returns a value of 0.
method: The method is due to Gauss.
references: Programming in Standard Fortran 77 by Balfour and Marwick
parameters: year integer input the year for which the date of Easter
Sunday is sought.
common blocks: none
own variables: no
routines called: none
level: level 0
author: Balfour/Marwick/G.Gibbs
date: December 1982
CRUPER ROUTINE
category: A33
title: subroutine cruper(text)
language: macro
description: put text into upper case
method: Uses the MOVe String Translated instruction.
references:
parameters: text character i/o text to be raised to capitals
common blocks: none
routines called: gtbypt,gtbufa,ptbufa
author: G.Gibbs
date: July 1984
E. IN
A1 FORMAT.
LOOK-UP TABLE.
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
A B D E F H J K L M N P R S T V X
PARAMETERS: IUN INTEGER INPUT NPH UNIT NUMBER
ICHL INTEGER OUTPUT CHECK LETTER.
ROUTINES CALLED:
LEVEL: LEVEL 0
AUTHOR: G.GIBBS
DATE: 2nd JUNE 1981
GETADR ROUTINE
Category: A28
Title: INTEGER FUNCTION GETADR(Thing)
Language: Macro-20
Description: Returns the address of its argument. The argument can be a
variable of any type, an array, a literal string, or a routine
or label. This is most useful in conjunction with DOJSYS.
Parameters: THING Any Input The address of this will be
returned.
Example: We wish to discover the account we are running under.
We can store the account string in an array, but the GACCT
jsys requires a byte pointer to the array, which consists
of a word containing -1 in the left half, and the array address
in the right. We may do this as follows:
MINUS=-1
ACCPTR="777777000000.OR.GETADR(ACOUNT)
CALL DOJSYS("546,MINUS,ACCPTR)
Array ACOUNT will receive a string of characters representing
the account of the current job.
Level: 0
Author: Kevin Ashley
Date: Nov 82
GETDDT ROUTINE
category: K25
title: SUBROUTINE GETDDT
language: macro-20
description: Loads DDT into a process, and causes a breakpoint
This subroutine provides a method for a program to
invoke the dynamic debugger (DDT) at will, without it being permanently
resident in a program. Thus, a program could have a special DEBUG
command, which loads DDT, or it could load DDT whenever a serious error
occurs which requires attention. Alternatively, DDT could be loaded
asynchronously by an interrupt routine, invoked, say by CTRL/D. The
possibilities are endless. On each call to GETDDT, a breakpoint is
simulated by jumping to $0BPT in DDT. One can proceed as from a normal
breakpoint. Note that DDT is liable to be useful only to those with
a knowledge of assembler programming.
method: SYS:UDDT is merged into the process, and the symbol table
address (from .JBSYM) is placed in the DDT user symbol pointer.
The user ACs are restored, and a jump is made to the unsolicited
breakpoint in DDT.
common blocks: none
level: level 0
author: Terry Pettit/Kevin Ashley
date: January 1983
HOSPNO ROUTINE
category: N2
title: subroutine HOSPNO (PROMPT,HOSNUM,LETTER,FAIL,BACK,HELP,DEFALT)
language: MACRO-20
purpose: This routine reads a hospital number with its attached check
letter from the terminal. An error is returned if the check letter does
not match the digits, and the user is usually reprompted to try again.
The letter and the number are returned separately. This routine is part
of the general data entry suite, and so automatic recognition of
UNKNOWN, EXIT and ^ are available. This is more fully explained in the
DATA-ENTRY section of this program, which should be read in
conjunction with this.
method: The COMND% Jsys is used.
parameters: prompt Array or input The prompt for the user
string
hosnum Integer output The returned hospital number
Letter Any output The ascii code for the check
letter (ie R1 format)
fail Integer output Error code (see below)
back Label input *Optional*
Statement to jump to if
user uses ^ to go back one
question, or 0 to ignore this.
length Integer output *Optional*
Length of text string.
help Array or input *Optional*
string Help message to be output if
user types question mark, or
0 to ignore argument.
A help message of
"hospital number" will be used
if you do not supply one.
defalt Array or input *Optional*
string Default to be used if user
types escape, or 0 if not
desired.
NOTE: Optional input arguments can be omitted if they are
all trailing, so the call
CALL HOSPNO('Number:',PATNUM,PATLET,FAIL) is valid, as is
CALL HOSPNO('Number:',PATNUM,PATLET,FAIL,*10,'Enter patient''s
& Number')
which supplies a label to return to if the user types ^ (the
label is 10) and outputs a help message if a question mark is
typed. No default has been supplied, however.
You may use 0 for an optional argument if you wish to supply
some arguments that follow it, but you wish that argument to
be ignored (ie if you want to supply a default, but not a help
message.)
Error codes: 0 Success
1 Invalid input (should not occur)
2 Not confirmed (this is only possible if RETRYS is set
to a non-negative value and indicates that extraneous
information was entered after the check letter.)
3 Hospital number is not a number, or is out of range
(ie is negative, contains too many digits, etc.)
4 Check letter does not match patient number.
-1 User typed UNKNOWN.
-2 User typed EXIT (this code is only returned if you did
not initialize location EXILAB in common block CMDPRM
with the label to jump to).
-3 User typed "^" to go back a question (this code is only
returned if you did not supply a label to jump to on
this condition.)
common blocks: /cmdstg/cmdbfr(70),atmbfr(70),fdb(5),hlpbfr(25),
defbfr(56),gjfblk(15),argblk(5)
/cmdprm/initf,retrys,tried,endnse,raise,savret,
savp,exilab,useext,spare(3),cmdblk(10)
(All integer, see DATA-ENTRY for more info)
own variables: no
routines called: none
level: level 1
author: G.Gibbs / Kevin Ashley
date: August 1982
ICOMP ROUTINE
CATEGORY: A3
TITLE: FUNCTION ICOMP (N,A,I,B,J)
LANGUAGE: MACRO
DESCRIPTION: COMPARE CHARACTER STRINGS.
CHARACTER STRINGS OF LENGTH N STARTING AT THE I'TH
CHARACTER OF A AND THE J'TH CHARACTER OF B (COUNTING
FROM 1). RESULT IS +1,0 OR -1 ACC AS THE A-STRING IS
>,= OR< THE B-STRING. N IS ALSO SET TO THE NO OF
EQUAL CHARACTERS AT THE BEGINNING OF THE STRINGS.
IF N, I OR J ARE NON-POSITIVE,N IS SET TO 0 AND ICOMP
RETURNS A VALUE OF 2. A,B MAY BE TEXT CONSTANTS.
PARAMETERS N INTEGER INPUT/OUTPUT NO OF CHARACTERS TO BE
COMPARED
A,B ANY INPUT STRINGS TO BE COMPARED
I,J INTEGER INPUT STARTING POSITIONS IN
STRINGS
EXAMPLE: N=4
IF (ICOMP(N,BUFFER,1,'FRED',1).EQ.0) GO TO 200
ROUTINES CALLED: NONE
AUTHOR: G.GIBBS
DATE: MARCH 1981
ICRCH2 ROUTINE
CATEGORY A6
TITLE: FUNCTION ICRCH2 (I,J)
LANGUAGE: FORTRAN
DESCRIPTION: CHARACTER EXTRACTION(ICL CODE). I IS A SCALAR OR (USUALLY)
AN ARRAY ELEMENT (NOT AN ARRAY NAME). RESULT IS THE INTEGER
VALUE OF THE JTH CHARACTER IN ICL CODE, COUNTING THE FIRST
CHARACTER IN I AS NO. 1. IF, FOR EXAMPLE, THE 3-WORD ARRAY
I CONTAINS THE CHARACTERS M.J.R.HEALY, THEN THE VALUE
OF ICRCH2 (I(2),4) IS 33 (EQUIVALENT TO A).
IF YOU DO NOT SPECIFICALLY REQUIRE ICL CODE, IPK (WHICH
THIS CALLS) IS PREFERED AS IT IS QUICKER AND GIVES ASCII CODE
A LOWER CASE LETTER RETURNS THE VALUE FOR THE UPPER CASE
LETTER AND A CONTROL CHARACTER RETURNS THE VALUE -1.
PARAMETERS I INTEGER INPUT START OF CHARACTER STRING.
J INTEGER INPUT POSITION OF REQUIRED
CHARACTER.
ROUTINES CALLED NONE
LEVEL: LEVEL 1
AUTHOR G.GIBBS
DATE 31st MARCH 1981
ICRCH3 ROUTINE
CATEGORY: A7
TITLE: ICRCH3 (K,I,J)
LANGUAGE FORTRAN
DESCRIPTION CHARACTER INSERTION(IN ICL CODE).
THE LEAST SIGNIFICANT 6 BITS OF K (TAKEN AS ICL CODE) ARE
INSERTED IN CHARACTER POSITION J (COUNTING FROM 1) OF I.
IF YOU DO NOT SPECIFICALLY REQUIRE ICL CODE,PK (WHICH THIS
CALLS) IS PREFERED AS IT IS QUICKER,AND USES ASCII CODE.
PARAMETERS K INTEGER INPUT CHARACTER TO BE INSERTED
I INTEGER OUTPUT CHARACTER STRING
J INTEGER INPUT POSITION IN CHARACTER STRING
ROUTINES CALLED PK
LEVEL: LEVEL 1
AUTHOR G.GIBBS
DATE 31st MARCH 1981
ICRCH4 ROUTINE
CATEGORY: A8
TITLE: FUNCTION ICRCH4 (I,J,K)
LANGUAGE: MACRO
DESCRIPTION: DECIMAL TO BINARY CONVERSION.
VALUE RETURNED IS THAT OF THE POSITIVE INTEGER IN THE
K NUMERIC CHARACTERS FROM THE J TH CHARACTER POSITION
OF I (COUNTING FROM 1). ANY NON-DIGIT,A VALUE OF
>34359738368 OR K< 1 CAUSES -1 TO BE RETURNED. LEADING
SPACES ARE IGNORED BUT IF ALL K CHARACTERS ARE SPACES,
-2 IS RETURNED.
PARAMETERS: I ANY INPUT CHARACTER STRING
J INTEGER INPUT POSITION OF FIRST CHARACTER
K INTEGER INPUT NO. OF CHARACTERS
ROUTINES CALLED NONE
LEVEL: LEVEL 0
AUTHOR G.GIBBS
DATE APRIL 1981
ICRCH7 ROUTINE
CATEGORY: A21
TITLE: ICRCH7(I,J,K,L)
LANGUAGE: MACRO
DESCRIPTION: BINARY INTEGER TO DECIMAL CONVERSION
THE VALUE OF I IS CONVERTED TO A STRING OF DECIMAL
DIGITS, AND THIS IS PLACED IN THE L CHARACTER
POSITIONS STARTING FROM THE K'TH CHARACTER OF ARRAY J
IF L IS NEGATIVE LEADING ZEROES ARE OUTPUT AS SPACES.
L ASTERISKS WILL BE OUTPUT IF L IS TOO SMALL
PARAMETERS: I INTEGER INPUT BINARY VALUE
J ARRAY OUTPUT START OF CHARACTER STRING
K INTEGER OUTPUT POSITION OF START IN J
L INTEGER INPUT NO. OF DECIMAL DIGITS
ROUTINES CALLED: NONE
LEVEL: LEVEL 0
AUTHOR: G.GIBBS
DATE: APRIL 1981
ICRNTM ROUTINE
category: K23
title: function icrntm(dummy)
language: macro
description: returns the current run time of the current fork
This may be used for timing sections of code and seeing if
changes improve the speed. It should be noted that your program
will take different lengths of time, depending on the load
etc. so that you should time sections of code several times
to get a figure for comparison. Simply reference the function
before and after the section of interest, and obtain the
difference.
method: uses the monitor call RUNTM%
parameters: dummy for syntax only
common blocks: none
level: level 0
author: G.Gibbs
date: November 1982
ICRSCH ROUTINE
category: A25
title: function icrsch(record,i,lenrec,word,j,numcha)
language: macro
description: Search the array "record" (starting at the ith character) for
the characters in "word" (starting at the jth character), the
number of characters in the search string being numcha.
The function returns the character position in "record"
where the match starts, regardless of the starting position (i)
A value of 0 is returned if there is no match and -1 if
erroneous arguments are given.
method: Comparing the search string against each allowable position
until a match is found, using the hardware compare strings
instruction.
parameters: record any input array to be searched for match
i integer input the starting position in characters,
counting from 1.
lenrec integer input the length of the array "record",
starting from the begining to the
last character to be considered in the
search, measured in characters.
word any input the string of text to be searched for
j integer input the starting position in characters,
counting from 1, in "word" for the
comparison.
numcha integer input The length of the string being sought.
Note that this is not necessarily the
length of the array "word", especially
if you do not start from character 1.
Measured in characters.
routines called: none
level: level 0
author: G.Gibbs
date: November 1982
ICRTLU ROUTINE
CATEGORY: A16
TITLE: FUNCTION ICRTLU (I,JA,L)
LANGUAGE: MACRO
DESCRIPTION: TABLE LOOK-UP BY BINARY SEARCH.
JA IS A TABLE CONSISTING
OF L ONE-WORD ENTRIES ARRANGED IN ASCENDING
ORDER (IN THE CHARACTER SENSE, SO THAT NEGATIVE
INTEGERS EXCEED POSITIVE INTEGERS). THE RESULT
IS THE POSITION OF I IN THE TABLE (COUNTING FROM
1); IF I DOES NOT OCCUR IN THE TABLE, THE RESULT
IS ZERO. (CF. NPH2).
PARAMETERS: I INTEGER INPUT VALUE TO BE LOOKED UP.
JA INTEGER INPUT FIRST LOCATION OF TABLE.
L INTEGER INPUT LENGTH OF TABLE
COMMON BLOCKS: NONE
OWN VARIABLES: NO
ROUTINES CALLED: NONE
LEVEL: LEVEL 0
AUTHOR: G.GIBBS
DATE: APRIL 1981
DATA-ENTRY
CATEGORY N
Question/answer data entry
---------------------------
N1 TEXTIN
Reads free text from terminal after prompt
N2 HOSPNO
Reads a hospital number after a prompt
N3 DATEIN
Reads a free-format date after prompt, converts to Julian
N4 KYWORD
Reads one of a list of keywords or phrases, with recognition
N5 SEXIN
Reads a sex
N6 YESNO
Reads yes or no
N7 NUMIN
Reads an integer after a prompt
N8 REALIN
Reads a real after a prompt
N9 TABLE
Builds a keyword table for KYWORD
N10 CROPEN
Reads a filename, with recognition, and opens the file
N11 TBLOOK
Looks up a character string in a table created by TABLE
N12 TBRLOC
Relocates lookup table pointers after reading from disk.
MONTH ROUTINE
category: J24
title: character function month(number)
language: Fortran 77
description: returns the month in characters, 0 returns 'Unknown'
and 'Error' is return for other out of range numbers.
parameters: number integer input the n'th month of the year
routines called: none
example: character*9 month
do i=0,13
type (fmt='1x,a') month(i)
end do
author: G.Gibbs
date: June 1984
INPUT-TERMINAL
CATEGORY M
INPUT FROM THE TERMINAL
-----------------------
J20 CRDTOI
CONVERT CHARACTER DATE TO 3 INTEGERS
M2 CRNIN
INPUT INTEGER FROM TERMINAL
M3 CRFLIN
INPUT REAL NUMBER FROM TERMINAL
M4 CRFNIN
INPUT A FILE NAME FROM TERMINAL
M5 ICHIN
INPUT NEXT CHARACTER FROM THE TERMINAL
M6 ICHOUT
OUTPUT A CHARACTER TO THE TERMINAL
CRCLR ROUTINE
category a19
title: crclr(i,j,k)
language: macro
description: crclr sets 'j' words starting at 'i' to the value of 'k'
k must be of the same type as i but not double precision
nor complex.
parameters: i any input start of area
j integer input length of area
k any input value to fill area
example: dimension buffer (30)
call crclr (buffer,30,' ')
own variables: no
routines called: none
level: level 0
author: G.Gibbs
date: April 81
MATHEMATICS
CATEGORY: *G
G MATHEMATICS
-----------
3 CRMATH2
QUOTIENT AND REMAINDER.
5 NELDER
FUNCTION MINIMISATION.
10 FUNCTION CRLNFC
NATURAL LOGARITHM OF FACTORIAL FOR NON-NEGATIVE ARGUMENT
12 CRFUNC,CRUSFN,CRMLFN
RETURNS VALUE OF REQUIRED FUNCTION GIVEN X
13 CRMNX,CRMNX2
FIND THE MINIMUM AND MAXIMUM OF N ELEMENTS IN ARRAY X.
14 CRFREQ
CALCULATES HISTOGRAM FREQUENCY COUNTS
15 CRMNIM
FUNCTION MINIMISATION
17 CRVCR
COMPUTES MATRICES FOR INPUT TO NAG LEAST-SQUARES ROUTINES
18 NINT
ROUNDS REALS TO INTEGERS
19 CRTWMY
Twomey's method for Fredholm integral equation of 1st kind
CRDOUT ROUTINE
Category: J21
Title: CRDOUT (ICDATE,IDAY,MONTH,IYEAR,IFMT)
Language: MACRO
Descrption: Outputs date in character form, to one of an array, a character
variable or your terminal. in the latter case the name of
the array should be replaced by the number 1,
e.g. CALL CRDOUT(1,IDAY,MONTH,IYEAR,IFMT)
The degree of verbosity used is defined by IFMT, increasing
from 27-Nov-81 for IFMT=1
to Friday, 27 November 1981 for IFMT=4
Sufficient space must be allowed in the array, if used, to
allow for the maximum number of characters you are likely
to get +1.
Parameters: ICDATE array or OUTPUT Date in character form
character
IDAY } {The 3 parts
MONTH} integer INPUT {of the date
IYEAR} {
IFMT integer INPUT The verbosity of output
from 1 to 4
Author: G.Gibbs
Date: November 1981, character variables July 1984
SORTING
CATEGORY: *B
B SORTING
-------
1 CRC154,CRC155
IN-CORE SORT(REAL VARIABLES)
STATISTICS
CATEGORY: *E
E STATISTICS
----------
1 BINOML
BINOMIAL PROBABILITY.
4 SNDEV
CALCULATES A NORMAL EQUIVALENT DEVIATE.
5&6 CRC151,CRC152
SET UP LIST OF INTEGERS.
7&8 CRC153,CRC156
LOG TRANSFORMATION.
12 STAT7,STAT7B
MEAN AND POWER SUMS ABOUT MEAN
13 STAT8
NORMAL TAIL AREAS.
20 STAT13
MARGINAL MEANS OF MULTI-WAY TABLE
21 STAT14
ANALYSIS OF MULTI-WAY TABLE
22 STAT7A
AS CRSTAT7 BUT IGNORING NEGATIVE AND ZERO VALUES.
27 STAT19
MEAN,STANDARD DEVIATION FROM CENSORED SAMPLE
28 CRPOLY
FITS THE POLYNOMIAL REGRESSION
30 CRSHPE
CALCULATES USUAL COEFFICIENTS OF SKEWNESS AND KURTOSIS.
34 CRWTST
CALCULATES SHAPIRO AND WILK'S STATISTIC.
37 STAT7M
MEAN AND SD ALLOWING FOR MISSING VALUES.
39 RANDOM
RETURNS A RANDOM NUMBER
40 RANDUP
STARTS THE RANDOM NUMBER GENERATOR PSEUDO-RANDOMLY
41 RANSET
STARTS THE RANDOM NUMBER GENERATOR REPEATABLY
42 CRGAM
INCOMPLETE GAMMA INTEGRAL
43 CRDIXN
Dixon's test for a single outlier
44 WEXT
W test for normality
45 WCOEF
Set up WEXT for a given sample size
46 NSCOR2
Calculates rankits
47 POLY
Calculates a polynomial
48 SPLITN
Fits a mixture of Normal distributions.
49 MW
Test of multivariate normality.
50 RADII
Test of multivariate normality.
51 SBS
Testing multivariate normality.
TERMINAL-CONTROL
CATEGORY L
TERMINAL CONTROL
----------------
L0 VTINIT
Initializes the terminal for use
L1 CURUP
Moves the cursor up the number of times specified.
L2 CURHOM
Moves the cursor to the top left of the screen ("home")
L3 CLEOS
Clears the screen from the cursor to the end of screen
L4 SCRRG
Alters the scrolling region so that only certain areas of
L5 PUTCRS
Positions the cursor to an absolute position on the screen.
L6 CLEOL
Clears from the cursor to the end of the line.
L7 VTLED
Lights the specified LED above the keyboard, or clears all leds
L8 TABCLR
Clears all tab settings
L9 TABSET
Places a tab stop at the specified column
L10 VTCBL
Clears from the beginning of line to the cursor
L11 VTCLIN
Clears the entire line containing the cursor
L12 VTCBS
Clears from the start of screen to the cursor
L13 VTCLS
Clears the entire screen
L14 VTLFT
Moves the cursor left the specified number of places
L15 VTRIT
Moves the cursor right the specified number of places
L16 VTDWN
Moves the cursor down the specified number of places
L17 VTSAV
Saves the current cursor position and attributes
L18 VTRES
Restores cursor position and attributes saved by VTSAV
L19 VTIUP
Performs a "reverse index"
L20 VT80
Sets the terminal to 80-column width
L21 VT132
Sets the terminal to 132-column width
L22 VTSCRN
Sets the screen background to light or dark
L23 VTORG
Alters the setting of "origin mode".
L24 VTJMP
Sets jump scroll mode
L25 VTSMTH
Sets smooth scroll mode
L26 VTATR
Sets or clears the character attributes.
L27 VTWID
Types a text string in double width characters.
L28 VTBIG
Types a text string in double height, double width chars.
L29 CRSIL
Turns terminal echoing of typein on or off
L30 VTEND
Called at the end of a program to reset terminal parameters
L31 CRBELL
RING TERMINAL BELL
L32 VTGON
Turns VT100 graphics mode on.
L33 VTGOF
Turns VT100 graphics mode off
L34 VTTY
Changes terminal for screen routines
IPK ROUTINE
CATEGORY: A11
TITLE: FUNCTION IPK (IARRAY, J)
LANGUAGE: MACRO
DESCRIPTION: ASCII CHARACTER EXTRACTION. IARRAY IS THE AREA CONTAINING
TEXT OR A TEXT CONSTANT. THE RESULT IS THE ASCII
INTEGER VALUE OF THE J'TH CHARACTER, COUNTING THE
FIRST CHARACTER IN IARRAY AS 1. IF, FOR EXAMPLE
IARRAY CONTAINED THE CHARACTERS 'PENTONVILLE ROAD',
THE VALUE RETURNED BY IPK (IARRAY, 7)WOULD BE 86 -
EQUIVALENT TO V. THIS ROUTINE IS PREFERED TO ICRCH2.
PARAMETERS: IARRAY ANY INPUT AREA HOLDING
CHARACTER STRING
J INTEGER INPUT POSITION OF REQUIRED
CHARACTER
ROUTINES CALLED: NONE
LEVEL: LEVEL 0
AUTHOR: G.GIBBS
DATE: 30th MARCH 1981
JDATE ROUTINE
CATEGORY: J15
TITLE: JDATE (J,IDAY,MONTH,IYEAR)
LANGUAGE: FORTRAN
DESCRIPTION: CONVERTS JULIAN DAY NUMBER TO CALENDAR DATE. THE DAY NUMBER
IS CONVERTED TO CALENDAR DATE, GREGORIAN CALENDAR,HELD AS
3 INTEGER NUMBERS. IF THE DAY NUMBER CORRESPONDS TO A DATE
PRIOR TO THE ADOPTION OF THE GREGORIAN CALENDAR IN BRITAIN,
14 SEPT 1752, THE DATE WILL BE RETURNED AS 0 , 0 , 0 .
Reference: Collected algorithms from CACM 199
(R.G.Tantzen 1963)
Parameters: J INTEGER INPUT Julian day number
IDAY INTEGER OUTPUT} The 3 parts
MONTH INTEGER OUTPUT} of the date
IYEAR INTEGER OUTPUT}
Authors: R.G.Tantzen/G.Gibbs
Date: April 1981
JDAY ROUTINE
Category: J16
Title: FUNCTION JDAY (IDAY,MONTH,IYEAR)
Language: FORTRAN
Description: Convert calendar date to Julian day number. A calendar date
Gregorian calendar, held as 3 integer numbers, is used to
compute the Julian day number. Non valid dates, e.g.
29, 2, 1973 or 32, 13, 1981 return a Julian day number of 0,
as will a date prior to the adoption of the Gregorian
calendar in England, 14 September 1752. A year of less than
100 will be assumed to be in the 20th Centuary.
Calls JDATE to check validity of date.
Reference: Collected algorithms from CACM 199
(R.G.TANTZEN 1963)
Parameters: IDAY INTEGER INPUT} The 3 parts
MONTH INTEGER INPUT} of the date
IYEAR INTEGER INPUT}
Routines called:JDATE
Author: G.Gibbs
Date: April 1981
JTODAY ROUTINE
Category: J17
Title: FUNCTION JTODAY (K)
Language: MACRO
Description: Today's Julian day number.
the function returns today's Julian day number. K is a
dummy argument.
Method: Today's Smithsonian day number is retrieved from the internal
date and time , and is converted to the Julian day number.
Parameters: K DUMMY Required by FORTRAN syntax
Author: G.Gibbs
Date: April 1981
KYWORD ROUTINE
category: N4
title: subroutine KYWORD (prompt,ktable,keynum,fail,back,help,defalt,string)
language: MACRO-20
purpose: The routine prompts the terminal user with a supplied string, and then
accepts one of a supplied set of keywords or phrases from a previously
constructed table. Any valid abbreviation of a keyword or phrase is
accepted (ie one which is unique among all words or phrases), and
the escape key can be used to complete a phrase. The question mark
key can be used to obtain a list of valid inputs at any point during the
entering of the phrase.
A default string can be supplied by the program, which will be
used if the user just types escape or return. The tables are set up
using the TABLE routine (see separate entry), and contain both the
text of the phrases, and a value to associate with each. It is this
value that is returned, although the actual text of the phrase may
be returned as well if desired. The case (upper or lower) of input is
not significant when matches are being considered.
This routine is part of the terminal data entry routine suite -
full facilities for recognition of defaults, retyping of prompts,
automatic recognition of UNKNOWN, EXIT and ^ are available. This
is more fully explained in the DATA-ENTRY section of this program,
which should be read in conjunction with this.
This routine is unable to read a keyword
which is an abbreviation of the words EXIT or UNKNOWN.
method: The COMND% Jsys is used.
parameters: prompt Array or input The prompt for the user
string
ktable Integer array input Table of pointers and values
for keyword strings (set up by
TABLE subroutine.)
keynum Integer output Returned number associated by
the user with the keyword or
phrase.
fail Integer output Error code (see below)
back Label input *Optional*
Statement to jump to if
user uses ^ to go back one
question, or 0 to ignore this.
help Array or input *Optional*
string Help message to be output if
user types question mark, or
0 to ignore argument.
A help message of
"one of the following:" followed
by a list of all possible words
or phrases in alphabetical order
will be supplied if the user
does not supply one. It is
recommended that the default
help is used.
defalt Array or input *Optional*
string Default to be used if user
types escape, or 0 if not
desired.
answer Array output *Optional*
Returned text string of
phrase or keyword.
NOTE: Optional input arguments can be omitted if they are
all trailing, so the call
CALL KYWORD('Disease:',DISTAB,DISEASE,FAIL) is valid, as is
CALL KYWORD('Disease:',DISTAB,DISEASE,FAIL,*10,
& 'Enter patient''s complaint')
which supplies a label to return to if the user types ^ (the
label is 10) and outputs a help message if a question mark is
typed. No default has been supplied, however.
You may use 0 for an optional argument if you wish to supply
some arguments that follow it, but you wish that argument to
be ignored (ie if you want to supply a default, but not a help
message.)
Error codes: 0 Success
1 Phrase or word did not match any keyword in the list
2 Not confirmed (only if endnse is non-zero, which is the
usual condition. It indicates superfluous input after
keyword or phrase.)
17 Input was ambiguous (ie matched more than one word)
-1 User typed UNKNOWN.
-2 User typed EXIT (this code is only returned if you did
not initialize location EXILAB in common block CMDPRM
with the label to jump to).
-3 User typed "^" to go back a question (this code is only
returned if you did not supply a label to jump to on
this condition.)
common blocks: /cmdstg/cmdbfr(70),atmbfr(70),fdb(5),hlpbfr(25),
defbfr(56),gjfblk(15),argblk(5)
/cmdprm/initf,retrys,tried,endnse,raise,savret,
savp,exilab,useext,spare(3),cmdblk(10)
(All integer, see DATA-ENTRY for more info)
own variables: no
routines called: none
level: level 1
author: G.Gibbs / Kevin Ashley
date: August 1982
LCRBAT ROUTINE
category: k20
title: logical function lcrbat(dummy)
language: macro-20
description: returns true if job is running in batch and false if interactive
method: uses monitor call getji% (get job info)
parameters: DUMMY for syntax only
common blocks: none
own variables: no
routines called: none
level: level 0
author: G.Gibbs
date: October 1982
LEFT ROUTINE
Category: A27
Title: INTEGER FUNCTION LEFT(WORD)
Language: Macro-20
Description: Returns the Left half (ie highest 18 bits) of its argument
sign extended. This is most useful with the DOJSYS routine.
Parameters: WORD Any Input Word to return Left half of
Level: 0
Author: Dan Murphy/Kevin Ashley
Date: November 1982
MW ROUTINE
category: E49
title: subroutine MW (X,MAN,MA,MMA,N,N2,XBAR,A,EPS,SSP,
MB,MMB,WORK,XWORK1,XWORK2,ZW,H,E,PH,IFAULT)
language: FORTRAN 66
description: MW provides a test of multivariate normality based on
the combination of univariate Shapiro-Wilk W statistics.
Allowance is made for the observed correlations in the data.
The resulting test statistic is a chi-square with possibly
non-integral degrees of freedom, depending on the said
correlations; however its P value is given also, so no problem.
For data with correlations less than about 0.7, the test
is the equivalent of combining the significance levels of the
individual univariate (one-tailed) W tests.
NOTE that the input array X is required to be ONE dimensional
in row order, and that the SSP matrix (SSP) is also ONE-d
arranged as lower triangle. The routine CRC135 is designed
to produce just such an SSP matrix row-by-row from the input
data.
method: Requires Royston's method for evaluating the P value from W.
references: Royston JP (1983). Some techniques for assessing multivariate
normality based on the Shapiro-Wilk W.
Appl. Statist. 31, 121-133.
parameters: X REAL INPUT multivariate data set,
ARRAY (MAN) arranged in row order
MAN INTEGER INPUT total number of data values (MA*N)
MA INTEGER INPUT number of variates
MMA INTEGER INPUT length of sums of squares and products
matrix
N INTEGER INPUT number of observations per variate
(i.e. sample size)
N2 INTEGER INPUT N/2 if N even, (N-1)/2 otherwise
XBAR REAL INPUT vector of variate means
ARRAY (MA)
A REAL INPUT coefficients for calculation of W
ARRAY (N2) (use S/R WCOEF)
EPS REAL INPUT minimum possible value of W
(use S/R WCOEF)
SSP REAL INPUT sums of squares and products (SSP)
ARRAY (MMA) for X, arranged as lower triangle in
row order
LIST INTEGER INPUT variates to be included in analysis
ARRAY (MB)
MB INTEGER INPUT number of variates in LIST (<=MA)
MMB INTEGER INPUT (MB*(MB+1))/2
WORK REAL OUTPUT )
ARRAY (MMB) )
XWORK1 REAL OUTPUT ) work space
ARRAY (MB) )
XWORK2 REAL OUTPUT )
ARRAY (N)
ZW REAL OUTPUT array of normal equivalent deviates
from MB univariate W statistics
ARRAY (MB)
H REAL OUTPUT chi-square test statistic from ZW
E REAL OUTPUT equivalent degrees of freedom for H
PH REAL OUTPUT error indicator, equal to
7 if any P value for a univariate
W test is zero or one
6 if a zero SD is found for any variate
in LIST
5 if N<3 or N>2000
4 if items in LIST are not in
ascending order
3 if any of the items in LIST are <1
or >MA
2 if MMB<>(MB*(MB+1))/2 or MB<1 or MB>MA
1 if MMA<>(MA*(MA+1))/2 or MA<1
0 otherwise (success!)
common blocks: none
own variables: no
routines called: SUBTRI,SSPCOR,CRC154,WEXT,SNDEV,MULTW
level: level 3
author: P.Royston
date: September 19, 1983
NELDER ROUTINE
CATEGORY: G5
TITLE: NELDER (F,N,START,STEP,ACC,MAXIT,RESULT,ARG,IT)
LANGUAGE: FORTRAN
DESCRIPTION: FUNCTION MINIMISATION. THE ROUTINE FINDS THE MINIMUM OF
A FUNCTION F OF N VARIABLES(N MUST BE .LE.25) BY A SEARCH
PROCEDURE THAT DOES NOT USE DERIVATIVES.
SEE NELDER,J.A. & MEAD,R.,COMPUTER J,7,308-313,1965,
F SHOULD BE WRITTEN AS A FORTRAN FUNCTION WITH A SINGLE
(VECTOR) ARGUMENT HOLDING THE N VARIABLE VALUES. IT MUST
BE DECLARED AS EXTERNAL IN THE CALLING SEGMENT.
PARAMETERS: F REAL FUNCTION FUNCTION TO BE MINIMISED
N INTEGER INPUT NO.OF VARIABLES OF F
(MUST BE .LE.25)
START REAL ARRAY INPUT INITIAL VALUES OF VARIABLES
STEP REAL ARRAY INPUT INITIAL STEP LENGTHS
ACC REAL INPUT ACCURACY CRITERION. THE
FINAL VALUE WILL USUALLY
BE WITHIN THIS AMOUNT OF
TRUE MINIMUM.
MAXIT INTEGER INPUT MAX. ALLOWED NUMBER OF
FUNCTION EVALUATIONS.
THE ROUTINE WILL BE LEFT
WHEN THIS NO. OF EVALUA-
TIONS HAS BEEN COMPLETED,
EVEN IF THE ACCURACY CRI-
TERION HAS NOT BEEN
REACHED.
RESULT REAL OUTPUT VALUE OF F AT THE MINIMUM.
ARG REAL ARRAY OUTPUT VALUES OF VARIABLES AT
THE MINIMUM.
IT INTEGER OUTPUT NO. OF FUNCTION EVALUA-
TIONS ACTUALLY PERFORMED.
ROUTINES CALLED: NONE
LEVEL: LEVEL 0
AUTHORS: E. CHAMBERS/M.J.R. HEALY.
DATE: 9TH SEPTEMBER, 1971.
NINT ROUTINE
category: G18
title: function nint(real)
language: macro
description: round a real number to give an integer. It will round halves
in the positive direction.
method: it uses the FIXR instruction
parameters: real real input the number to be rounded
common blocks: none
routines called: none
level: level 0
author: G.Gibbs
date: November 1982
NPH11 ROUTINE
CATEGORY: A14
TITLE: FUNCTION NPH11 (I, J, K)
LANGUAGE: MACRO
DESCRIPTION LOGICAL AND SHIFT OPERATIONS. J AND K ARE THE OPERANDS
AND THE VALUE FOR I = 1-5 IS
I = 1 J.AND.K
2 J.OR.K
3 J.NEQ.K
4 J SHIFTED LEFT K PLACES LOGICALLY
5 J SHIFTED LEFT K PLACES CIRCULARLY
FOR I = 1-3, THE RESULTS ARE GIVEN BY THE TABLE BELOW
J K .AND. .OR. .NEQ.
0 0 0 0 0
0 1 0 1 1
1 0 0 1 1
1 1 1 1 0
FOR I = 4-5,NEGATIVE VALUES OF K GIVE RIGHT SHIFTS
FOR I OUT OF RANGE THE FUNCTION IS SET TO ZERO.
PARAMETERS: I INTEGER INPUT TYPE OF OPERATION
J, K INTEGER INPUT ARGUMENTS
OTHER ROUTINES CALLED: NONE
LEVEL: LEVEL 0
AUTHOR: G.GIBBS
DATE: APRIL 1981
NPH2 ROUTINE
CATEGORY: A12
TITLE FUNCTION NPH2 (I,JA,K)
LANGUAGE MACRO
DESCRIPTION TABLE LOOK-UP. JA IS AN ARRAY OF K ITEMS.
THE RESULT IS THE POSITION OF THE FIRST ITEM IN THE
ARRAY EQUAL TO I, IF THERE IS ONE; IF THERE IS NOT,
THE RESULT IS ZERO.
NOTE THAT FASTER METHODS OF LOOK-UP SHOULD BE USED
FOR LARGE TABLES, ESPECIALLY IF THESE CAN BE SORTED
INTO ASCENDING ORDER.
IF K IS NOT +VE THE RESULT IS ZERO.
THE ITEM TO BE LOOKED UP MUST BE OF THE SAME TYPE AS
THE ARRAY E.G. INTEGER , AND MUST NOT BE DOUBLE PRECISION
OR COMPLEX.
PARAMETERS I ANY INPUT ITEM TO BE LOOKED UP
JA ARRAY INPUT LOOK-UP TABLE
K INTEGER INPUT NO. OF ITEMS IN TABLE.
ROUTINES CALLED: NONE
LEVEL: LEVEL 0
AUTHOR G.GIBBS
DATE APRIL 1981
NPH3 ROUTINE
CATEGORY: A13
TITLE: FUNCTION NPH3 (I,J,K)
LANGUAGE: MACRO
DESCRIPTION: BIT EXTRACTION. I IS A SCALAR. RESULT IS BITS
J THROUGH K OF I, RIGHT JUSTIFIED. BIT-POSITIONS
ARE NUMBERED FROM 0 TO 35, STARTING WITH THE
MOST SIGNIFICANT.
THUS IF I = '47610', THE VALUE OF NPH3 (I,18,20)
IS 6 AND NPH3(I,14,20) IS 54,THE ASCII VALUE OF 6.
-1 IS RETURNED UNLESS 0<J<K<35
NOTE THAT IPK IS QUICKER IF A CHARACTER IS
TO BE EXTRACTED.
PARAMETERS: I INTEGER INPUT DATA WORD.
J,K INTEGER INPUT FIRST AND LAST BIT
POSITIONS OF PORTION TO
BE EXTRACTED.
ROUTINES CALLED: NONE
LEVEL: LEVEL 0
AUTHOR: G.GIBBS
DATE: APRIL 81
NPLOT ROUTINE
CATEGORY: D8
TITLE: NPLOT (X,N,NCHAN)
LANGUAGE: FORTRAN
DESCRIPTION THIS SUB-ROUTINE PLOTS (ON CHANNEL NCHAN) NORMAL
PLOTS FOR THE N VALUES OF THE FLOATING POINT
ARRAY X. THE VALUES Y= (X-XBAR)/S (WHERE XBAR AND
S ARE THE SAMPLE ESTIMATES OF THE MEAN AND STANDARD
DEVIATION) ARE PLOTTED ON THE Y-AXIS. ON THE
X-AXIS IS PLOTTED THE NORMAL EQUIVALENT DEVIATE
CORRESPONDING TO THE PROPORTION PI = (I-0.5)/N
FOR THE I'TH RANKED ELEMENT OF X. PLOTTING IS
DONE BY SUB-ROUTINE CRC 561 AND THE NORMAL
EQUIVALENT DEVIATES CALCULATED BY FUNCTION SNDEV.
VALUES OF XBAR AND S ARE PRINTED AFTER THE PLOT ON
CHANNEL NCHAN.
PARAMETERS: X REAL ARRAY VECTOR OF OBSERVATIONS
N INTEGER NUMBER OF OBSERVATIONS
UNTIL GREATER THAN 1000.
NCHAN INTEGER OUTPUT CHANNEL NUMBER.
ROUTINES CALLED: CRC154,CRC561,FUNCTION SNDEV
LEVEL: LEVEL 2
AUTHOR KLIM MCPHERSON
NPLOT1 ROUTINE
CATEGORY: D11
TITLE: NPLOT1(X,N,NCHAN)
LANGUAGE: FORTRAN
DESCRIPTION: NORMAL PLOT ***REVISED VERSION OF NPLOT (AUG.78)***
NORMAL EQUIVALENT DEVIATES ARE CALCULATED FOR THE
ADJUSTED SAMPLE CUMULATIVE PROPORTIONS P(I) VIA
P(I)=(I-1/3)/(N+1/3). THESE ARE PLOTTED AGAINST THE
RANKED SAMPLE VALUES X (WHICH MAY BE INPUT UNSORTED) WITH
THE NED'S ON THE HORIZONTAL AXIS. OUTPUT IS DIRECTED TO
CHANNEL NCHAN.
THE PROGRAM ALSO OUTPUTS THE FOLLOWING STATISTICS WHICH
AUGMENT THE NORMAL PLOT AS TESTS OF NORMALITY:-
1) MEAN AND STANDARD DEVIATION OF X;
2) COEFFICIENTS OF SKEWNESS AND KURTOSIS OF X(FROM
CRSHAPE [E29]);
3) SHAPIRO AND WILK'S W-STATISTIC,AND ITS APPROXIMATE
SIGNIFICANCE LEVEL.(USES CRWTEST[E35]).
REFERENCE: FOR W-STATISTIC, DESCRIPTION OF SUBROUTINE CRWTEST[E35]
AND BIOMETRIKA 52 (1965),P.591.
PARAMETERS: X REAL ARRAY(N) INPUT SAMPLE DATA (IN ANY ORDER)
N INTEGER INPUT SAMPLE SIZE (LESS THAN 1000
NCHAN INTEGER INPUT OUTPUT CHANNEL NUMBER
LEVEL: LEVEL 2
ROUTINES CALLED: CRWTST,CRC154,FLOAT,SNDEV,CRC561,CRSHPE
AUTHOR: J.P.ROYSTON
DATE: 22 AUGUST 1978.
NSCOR2 ROUTINE
category: E46
title: subroutine nscor2(s,n,n2,ifault)
language: fortran
description: Calculates expected normal order statistics
for sample size N (also known as 'rankits'). S(1) contains
the largest positive rankit; the array stops at S(N2),
i.e. the negative symmetric rankits are not supplied in S,
but can be found via S(N-J+1) = S(J) for J=1,...,N2. The
array S is needed in the calculation of the W test (E44).
method: uses a polynomial approximation based on
Blom's (1958) method. Accurate to 0.0001 and
usually much better.
references: Applied Statistics Vol.31 no.2 1982, algorithm AS 177
parameters: s real output contains n2 largest rankits
array
n integer input sample size
n2 integer input largest integer .le. 1/2 n
ifault integer output fault indicator, 3 if n2.ne.[n/2]
2 if n.gt.2000
1 if n.le.2
0 otherwise
common blocks: none
own variables: no
routines called: correc,sndev
level: level 1
author: P.Royston
date: February 1983
NUMIN ROUTINE
category: N7
title: subroutine NUMIN (prompt,answer,fail,back,lolim,hilim,help,defalt)
language: MACRO-20
purpose: The routine prompts the terminal user with a supplied string, and then
reads an integer number from the terminal, optionally checking if
it falls within specified limits. A default may be used which is
supplied by the calling program, which will be used if the user types
return or escape. The limits can be both specified, or only one
specified, or neither. They are inclusive. If it is wished to include
arguments following the limits, but not the limits themselves, one may
supply the largest or smallest integer (as appropriate) as limit values.
These may be obtained using NAG functions, or just input as constants
(which are octal 377777777777 and 400000000000 for upper and lower
limits respectively.)
This routine is part of the terminal data entry routine suite -
full facilities for recognition of defaults, retyping of prompts,
automatic recognition of UNKNOWN, EXIT and ^ are available. This
is more fully explained in the DATA-ENTRY section of this program,
which should be read in conjunction with this.
method: The COMND% Jsys is used.
parameters: prompt Array or input The prompt for the user
string
answer Integer output Returned value
fail Integer output Error code (see below)
back Label input *Optional*
Statement to jump to if
user uses ^ to go back one
question, or 0 to ignore this.
lolim Integer input *Optional*
Lower limit for input, inclusive
(ie this number is valid), or
"400000000000 to ignore.
hilim Integer input *optional*
Upper limit for input, inclusive
or "377777777777 to ignore.
help Array or input *Optional*
string Help message to be output if
user types question mark, or
0 to ignore argument.
A help message of the form
"Enter a number, at least 0 and
not more than 120"
will be used if you do not
supply one (with limits set
only if supplied.)
defalt Integer input *Optional*
Default to be used if user
types escape, or 0 if not
desired.
NOTE: Optional input arguments can be omitted if they are
all trailing, so the call
CALL NUMIN('Age:',AGE,FAIL) is valid, as is
CALL NUMIN('Age:',AGE,FAIL,*10,0,120,'Enter patient''s age at
& at examination')
which supplies a label to return to if the user types ^ (the
label is 10),stores the integer age answer in AGE, and
outputs a help message if a question mark is typed.
You may use 0 for an optional argument if you wish to supply
some arguments that follow it, but you wish that argument to
be ignored (ie if you want to supply a default, but not a help
message.) (But note exceptions for LOLIM and HILIM)
Error codes: 0 Success
1 Invalid input
2 Not confirmed (this indicates superfluous input
after the word.)
7 Not a number
8 Number too small
9 Number too large
-1 User typed UNKNOWN.
-2 User typed EXIT (this code is only returned if you did
not initialize location EXILAB in common block CMDPRM
with the label to jump to).
-3 User typed "^" to go back a question (this code is only
returned if you did not supply a label to jump to on
this condition.)
common blocks: /cmdstg/cmdbfr(70),atmbfr(70),fdb(5),hlpbfr(25),
defbfr(56),gjfblk(15),argblk(5)
/cmdprm/initf,retrys,tried,endnse,raise,savret,
savp,exilab,useext,spare(3),CMDBLK(10)
(All integer, see DATA-ENTRY for more info)
own variables: no
routines called: none
level: level 1
author: G.Gibbs / Kevin Ashley
date: August 1982
PK ROUTINE
CATEGORY: A15
TITLE: PK (IARRAY,K,J)
LANGUAGE: FORTRAN
DESCRIPTION: CHARACTER INSERTION (IN ASCII CODE).
IARRAY IS THE NAME OF AN ARRAY INTO THE JTH CHARACTER
OF WHICH (COUNTING FROM 1) IS WRITTEN A CHARACTER
WHOSE VALUE IS OBTAINED FROM THE LEAST
SIGNIFICANT 7 BITS OF K, IE. K MODULO 128.
THIS ROUTINE IS PREFERED TO ICRCH3
PARAMETERS: IARRAY ARRAY OUTPUT ARRAY INTO WHICH
CHARACTER IS TO
BE INSERTED.
K INTEGER INPUT CHARACTER TO BE
INSERTED.
J INTEGER INPUT POSITION IN
CHARACTER STRING.
ROUTINES CALLED: NONE
LEVEL: LEVEL 0
AUTHOR: G.GIBBS
DATE: APRIL 1981
PNPLOT ROUTINE
CATEGORY: C40
TITLE: PNPLOT(X,N,NDS,IDS,NCHAN)
LANGUAGE: FORTRAN
DESCRIPTION: PRODUCES UP TO 5 NORMAL PLOTS ON EITHER THE OKI DOT MATRIX
PRINTER OR THE CALCOMP PEN PLOTTER (WHEN READY). STANDARD-
IZED DATA VALUES(X-XBAR)/SD ARE PLACED ON THE Y-AXIS;
OTHERWISE METHOD AND OUTPUT ARE THE SAME AS THE LINE-
PRINTER NORMAL PLOT [D8]. MEAN,STANDARD DEVIATION,
SKEWNESS, KURTOSIS AND (WHEN 2<N<51) THE W-STATISTIC
(SEE CRWTST [E34]) FOR EACH DATA SET ARE OUTPUT TO
CHANNEL NCHAN. THE PLOT IS SENT TO THE OKI
UNLESS NCHAN IS SET NEGATIVE, WHEN IT IS SENT TO THE
CALCOMP.
PARAMETERS: X REAL ARRAY(N) INPUT CONTAINS UP TO 5 DATA SETS
N INTEGER INPUT TOTAL NUMBER OF X-VALUES
NDS INTEGER INPUT NUMBER OF DATA SETS TO
BE ANALYSED (UP TO 5)
IDS INTEGER ARRAY(5)INPUT SAMPLE SIZES OF DATA
SETS IN X (IN SAME ORDER)
NCHAN INTEGER INPUT OUTPUT CHANNEL NUMBER.
IF SET NEGATIVE THE PLOT
WILL BE SENT TO THE CALCOMP
RATHER THAN TO THE OKI
LEVEL: LEVEL 4
ROUTINES CALLED: CRWTST,CRC154,CRSHPE,SNDEV
CRGRIT,CRGRAX,CRGRLB,CRGRAF
AUTHOR: J.P.ROYSTON/G.GIBBS
DATE: 12 JANUARY 1982.
POLY ROUTINE
category: E47
title: function poly(c,nord,x)
language: fortran
description: Calculates the algebraic polynomial of order NORD-1
for input value X. C(NORD) are the polynomial coefficients.
E.g. for NORD = 3 POLY will return
C(1) + X*C(2) + X*X*C(3).
references: Applied Statistics Vol.31 no.2 1982, algorithm AS 181
parameters: c real input coefficients
nord integer input order
x real input
common blocks: none
own variables: no
routines called: none
level: level 0
author: P.Royston
date: February 1983
PUTCRS ROUTINE
CATEGORY: L5
NAME: PUTCRS(ROW,COLUMN)
PURPOSE: Positions the cursor to an absolute position on the screen.
The next write will occur at that position.
PARAMETERS: ROW Integer input Row number
COLUMN Integer input Column number
Row and column numbers start at (1,1) although (0,0) is read
as (1,1). The maximum value of row is 24, and that of column is
80 or 132, depending on screen setting.
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
RADII ROUTINE
category: E50
title: subroutine RADII (X,MAN,MA,MMA,N,N2,XBAR,A,EPS,SSP,
MB,MMB,WORK,XWORK1,D,OMEGA,P,IFAULT)
language: FORTRAN 66
description: The square of the scaled Euclidean distance of a
datum (sampled at random from a multivariate normal dist-
ribution) from the sample centroid of the whole data has a
beta distribution. These 'squared radii' may be used
as the basis of a test of multivariate normality by
transforming them to a standard normal variate and then
subjecting the resulting vector to the usual W test.
The present routine performs this procedure, called
the 'omega test'(to distinguish it from the W test). It also
outputs the squared radii (array D) for plotting if the
user desires. (I recommend this.)
NOTE: it has been found that it is very useful to examine
subsets of VARIATES for departure from normality using the
squared radii method, because subsets can reveal anomalies
not apparent in the full data. Routine SBS has been included
in the Library for this purpose. See also Routine MW.
method: Requires Royston's method for evaluating the P value from W.
references: Royston JP (1983). Some techniques for assessing multivariate
normality based on the Shapiro-Wilk W.
Appl. Statist. 31, 121-133.
Healy, MJR (1968). Multivariate normal plotting.
Appl. Statist. 17, 157-161.
parameters: X REAL INPUT multivariate data set,
ARRAY (MAN) arranged in row order
MAN INTEGER INPUT total number of data values (MA*N)
MA INTEGER INPUT number of variates
MMA INTEGER INPUT length of sums of squares and products
matrix
N INTEGER INPUT number of observations per variate
(i.e. sample size)
N2 INTEGER INPUT N/2 if N even, (N-1)/2 otherwise
XBAR REAL INPUT vector of variate means
ARRAY (MA)
A REAL INPUT coefficients for calculation of W
ARRAY (N2) (use S/R WCOEF)
EPS REAL INPUT minimum possible value of W
(use S/R WCOEF)
SSP REAL INPUT sums of squares and products (SSP)
ARRAY (MMA) for X, arranged as lower triangle in
row order
LIST INTEGER INPUT variates to be included in analysis
ARRAY (MB)
MB INTEGER INPUT number of variates in LIST (<=MA)
MMB INTEGER INPUT (MB*(MB+1))/2
WORK REAL OUTPUT )
ARRAY (MMB) )
XWORK1 REAL OUTPUT ) work space
ARRAY (MB) )
D REAL OUTPUT transformed squared radii
ARRAY (N)
OMEGA REAL OUTPUT normal score from W test applied to D
P REAL OUTPUT P value for OMEGA
IFAULT INTEGER OUTPUT error indicator, equal to
7 if any P value for a univariate
W test is zero or one
6 if the submatrix of SSP corresponding
to LIST is not positive semi-
definite or has pos. nullity
5 if N<3 or N>2000
4 if items in LIST are not in
ascending order
3 if any of the items in LIST are <1
or >MA
2 if MMB<>(MB*(MB+1))/2 or MB<1 or MB>MA
1 if MMA<>(MA*(MA+1))/2 or MA<1
0 otherwise (success!)
common blocks: none
own variables: no
routines called: SUBTRI,CRC301,TSDIST,STAT7,CRC154,SNDEV,WEXT
level: level 3
author: P.Royston
date: September 19, 1983
RANDOM ROUTINE
category: E39
title: function random (dummy)
language: fortran
description: returns a random number
method: The routine maintains 3 separate generators and produces the
random number from the fractional part of the sum of them.
The state of the generator may be saved for restarting later
by saving the 3 integers and 1 logical in the common block.
The generator is started in a pseudo random way unless a prior
call has been made to ranset for a repeatable start.
references: B.A.Wichmann and I.D.Hill Applied Statistics 1982
parameters: dummy any input dummy required for syntax
common blocks: /crrand/ix,iy,iz,notset
routines called: randup
level: level 2
author: I.D.Hill/G.Gibbs
date: July 1982
RANDUP ROUTINE
category: E40
title: subroutine randup
language: fortran
description: starts the random number generator in a pseudo random way
method: The separate generators are started with three numbers, today's
Julian day number, the elapsed time since start-up, and the
current time and date in internal machine form.
parameters: none
common blocks: crrand
routines called: jtoday,mtime,meltime
level: level 1
author: G.Gibbs
date: July 1982
RANSET ROUTINE
category: E41
title: subroutine ranset (seed)
language: fortran
description: starts the random number generator in a repeatable way
parameters: seed real input number to set up generator with
common blocks: crrand
routines called: none
level: level 0
author: I.D.Hill/G.Gibbs
date: July 1982
REALIN ROUTINE
category: N8
title: subroutine REALIN (prompt,answer,fail,back,lolim,hilim,help,defalt)
language: MACRO-20
purpose: The routine prompts the terminal user with a supplied string, and then
reads an real number from the terminal, optionally checking if
it falls within specified limits. A default may be used which is
supplied by the calling program, which will be used if the user types
return or escape. The limits can be both specified, or only one
specified, or neither. They are inclusive. If it is wished to include
arguments following the lintis, but not the limits themselves, one may
supply the largest or smallest real (as appropriate) as limit values.
These may be obtained using NAG functions, or just input as constants.
This routine is part of the terminal data entry routine suite -
full facilities for recognition of defaults, retyping of prompts,
automatic recognition of UNKNOWN, EXIT and ^ are available. This
is more fully explained in the DATA-ENTRY section of this program,
which should be read in conjunction with this.
method: The COMND% Jsys is used.
parameters: prompt Array or input The prompt for the user
string
answer Real output Returned value
fail Integer output Error code (see below)
back Label input *Optional*
Statement to jump to if
user uses ^ to go back one
question, or 0 to ignore this.
lolim Real input *Optional*
Lower limit for input, inclusive
(i.e. this number is valid)
hilim Real input *optional*
Upper limit for input, inclusive
help Array or input *Optional*
string Help message to be output if
user types question mark, or
0 to ignore argument.
A help message of the form
"Enter a number, at least 0.
and not more than 120."
will be used if you do not
supply one (with limits set
only if supplied.)
defalt Real input *Optional*
Default to be used if user
types escape, or 0.0 if not
desired.
NOTE: Optional input arguments can be omitted if they are
all trailing, so the call
CALL REALIN('Age:',AGE,FAIL) is valid, as is
CALL REALIN('Age:',AGE,FAIL,*10,0.,120.,'Enter patient''s age at
& at examination')
which supplies a label to return to if the user types ^ (the
label is 10),stores the Real age answer in AGE, and
outputs a help message if a question mark is typed.
You may use 0 for an optional argument if you wish to supply
some arguments that follow it, but you wish that argument to
be ignored (ie if you want to supply a default, but not a help
message.) (But note exceptions for LOLIM and HILIM)
Error codes: 0 Success
1 Invalid input
2 Not confirmed (this indicates superfluous input
after the word.)
7 Not a number
8 Number too small
9 Number too large
-1 User typed UNKNOWN.
-2 User typed EXIT (this code is only returned if you did
not initialize location EXILAB in common block CMDPRM
with the label to jump to).
-3 User typed "^" to go back a question (this code is only
returned if you did not supply a label to jump to on
this condition.)
common blocks: /cmdstg/cmdbfr(70),atmbfr(70),fdb(5),hlpbfr(25),
defbfr(56),gjfblk(15),argblk(5)
/cmdprm/initf,retrys,tried,endnse,raise,savret,
savp,exilab,useext,spare(3),CMDBLK(10)
(All integer, see DATA-ENTRY for more info)
own variables: no
routines called: none
level: level 1
author: G.Gibbs / Kevin Ashley
date: August 1982
RIGHT ROUTINE
Category: A26
Title: INTEGER FUNCTION RIGHT(WORD)
Language: Macro-20
Description: Returns the right half (ie lowest 18 bits) of its argument
sign extended. This is most useful with the DOJSYS routine.
Parameters: WORD Any Input Word to return right half of
Level: 0
Author: Dan Murphy/Kevin Ashley
Date: November 1982
SBS ROUTINE
category: E51
title: subroutine SBS (X,MAN,MA,MMA,N,N2,XBAR,A,EPS,SSP,
MB,MMB,WORK,XWORK1,XWORK2,ICLIST,IC,MC,P,MBCMC,CHAN,IFAULT)
language: FORTRAN 66
description: Please read the description of routine RADII for the motivation
of this routine. SBS allows you examine the test of squared
radii for all subsets of variates of given size (you input
this quantity as MC). The input parameter MBCMC contains
the number of combinations represented by your choice. For
example, if MA=12 (twelve variates in entire data set),
MB=6 (LIST of six variates being analysed), MC=3 (examine all
variate-subsets of size 3), then MBCMC=6!/(3!3!)=20, and
you will get 20 P values in array P, corresponding to
W tests on squared radii for each combination. See
paper by Royston (1983) for further discussion.
method: Requires Royston's method for evaluating the P value from W.
references: Royston JP (1983). Some techniques for assessing multivariate
normality based on the Shapiro-Wilk W.
Appl. Statist. 31, 121-133.
Healy, MJR (1968). Multivariate normal plotting.
Appl. Statist. 17, 157-161.
parameters: X REAL INPUT multivariate data set,
ARRAY (MAN) arranged in row order
MAN INTEGER INPUT total number of data values (MA*N)
MA INTEGER INPUT number of variates
MMA INTEGER INPUT length of sums of squares and products
matrix
N INTEGER INPUT number of observations per variate
(i.e. sample size)
N2 INTEGER INPUT N/2 if N even, (N-1)/2 otherwise
XBAR REAL INPUT vector of variate means
ARRAY (MA)
A REAL INPUT coefficients for calculation of W
ARRAY (N2) (use S/R WCOEF)
EPS REAL INPUT minimum possible value of W
(use S/R WCOEF)
SSP REAL INPUT sums of squares and products (SSP)
ARRAY (MMA) for X, arranged as lower triangle in
row order
LIST INTEGER INPUT variates to be included in analysis
ARRAY (MB)
MB INTEGER INPUT number of variates in LIST (<=MA)
MMB INTEGER INPUT (MB*(MB+1))/2
WORK REAL OUTPUT )
ARRAY (MMB) )
XWORK1 REAL OUTPUT ) work space
ARRAY (MB) )
XWORK2 REAL OUTPUT )
ARRAY (N) )
ICLIST INTEGER OUTPUT working array containing changing set
ARRAY (MC) of variates entering S/R RADII
IC INTEGER OUTPUT working array representing the same
list as ICLIST but with binary
entries (0=exlude, 1=include)
MC INTEGER INPUT size of subsets of MB variates to test
P REAL OUTPUT P values for omega tests for all
ARRAY (MBCMC) combinations of MC variates from LIST
MBCMC INTEGER OUTPUT number of combinations of MC variates
out of MB [= MB!/(MC!(MB-MC)!)]
CHAN INTEGER INPUT channel number to print current list of
variates and P values for omega test.
Set CHAN to zero to suppress printing.
IFAULT INTEGER OUTPUT error indicator, equal to
8 if MBCMC is incorrect
7 if any P value for a univariate
W test is zero or one
6 if MC<1 or MC>MB
5 if N<3 or N>2000
4 if items in LIST are not in
ascending order
3 if any of the items in LIST are <1
or >MA
2 if MMB<>(MB*(MB+1))/2 or MB<1 or MB>MA
1 if MMA<>(MA*(MA+1))/2 or MA<1
0 otherwise (success!)
common blocks: none
own variables: no
routines called: RADII,NXCBN,MCN
level: level 4
author: P.Royston
date: September 19, 1983
SCRRG ROUTINE
CATEGORY: L4
NAME: SCRRG(TOP,BOTTOM)
PURPOSE: Alters the scrolling region so that only certain areas of
the screen can be scrolled off.
PARAMETERS: TOP Integer input Top line of region (0<TOP<25)
BOTTOM Integer input Bottom line of region
An error will occur if the parameters are out of range, or if
BOTTOM <= TOP.
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
SEXIN ROUTINE
category: N5
title: subroutine SEXIN (prompt,answer,fail,back,help,defalt)
language: MACRO-20
purpose: The routine prompts the terminal user with a supplied string, and then
reads MALE or FEMALE from the terminal. A default may be
supplied by the calling program, which will be used if the user types
return or escape.The returned value is 1 for male, -1 for female, 0 for
unknown (which also sets an error.) Any abbreviation can be used for
MALE and FEMALE.
A default can be supplied by the program, which is used if the
user types escape or return.
This routine is part of the terminal data entry routine suite -
full facilities for recognition of defaults, retyping of prompts,
automatic recognition of UNKNOWN, EXIT and ^ are available. This
is more fully explained in the DATA-ENTRY section of this program,
which should be read in conjunction with this.
method: The COMND% Jsys is used.
parameters: prompt Array or input The prompt for the user
string
answer Integer output Returned value
fail Integer output Error code (see below)
back Label input *Optional*
Statement to jump to if
user uses ^ to go back one
question, or 0 to ignore this.
help Array or input *Optional*
string Help message to be output if
user types question mark, or
0 to ignore argument.
A help message of "one of
the following:
FEMALE MALE"
will be used if you do not
supply one.
defalt Array or input *Optional*
string Default to be used if user
types escape, or 0 if not
desired.
NOTE: Optional input arguments can be omitted if they are
all trailing, so the call
CALL SEXIN('Sex:',TYPE,FAIL) is valid, as is
CALL SEXIN('Sex:',TYPE,FAIL,*10,'Enter sex','MALE')
which supplies a label to return to if the user types ^ (the
label is 10),stores the answer in TYPE, and
outputs a help message if a question mark is typed. MALE
is assumed if nothing is entered.
You may use 0 for an optional argument if you wish to supply
some arguments that follow it, but you wish that argument to
be ignored (ie if you want to supply a default, but not a help
message.)
Error codes: 0 Success
1 Invalid input (ie not MALE FEMALE or UNKNOWN)
2 Not confirmed (this indicates superfluous input
after the word.)
11 Does not match, or ambiguous
-1 User typed UNKNOWN.
-2 User typed EXIT (this code is only returned if you did
not initialize location EXILAB in common block CMDPRM
with the label to jump to).
-3 User typed "^" to go back a question (this code is only
returned if you did not supply a label to jump to on
this condition.)
common blocks: /cmdstg/cmdbfr(70),atmbfr(70),fdb(5),hlpbfr(25),
defbfr(56),gjfblk(15),argblk(5)
/cmdprm/initf,retrys,tried,endnse,raise,savret,
savp,exilab,useext,spare(3),CMDBLK(10)
(All integer, see DATA-ENTRY for more info)
own variables: no
routines called: none
level: level 1
author: G.Gibbs / Kevin Ashley
date: August 1982
SNDEV ROUTINE
CATEGORY: E4
TITLE: FUNCTION SNDEV(PROB)
LANGUAGE: FORTRAN
DESCRIPTION: CALCULATES THE NORMAL EQUIVALANT DEVIATE
ASSOCIATED WITH PROB. THAT IS THE AREA
OF A NORMAL CURVE WITH MEAN ZERO AND
VARIANCE UNITY, FROM MINUS INFINITY TO
SNDEV, IS PROB.
METHOD: APPLIED STATISTICS ALGORITHM AS 111
(J.D.BEASLEY AND S.G.SPRINGER, 1977)
THE INTERNAL CALCULATIONS ARE PERFORMED IN DOUBLE PRECISION
ALTHOUGH THE ARGUMENTS ARE PASSED IN SINGLE PRECISION
PARAMETER: PROB REAL INPUT
(0.0 .LT. PROB .LT. 1.0
ROUTINES CALLED: NONE
AUTHOR: I.D.HILL
DATE: 6 AUGUST 1979
SPLITN ROUTINE
category: E48
title: subroutine SPLITN
(ND,MODEL,ISTART,X,N,NCH,XBAR,SD,ALPHA,CHI,PCHI,IFAIL)
language: FORTRAN 66
description: Fits a mixture of Normal (Gaussian) distributions
by the method of maximum likelihood. Up to 10 distributions
may be fitted. Maximum sample size is 2000. Outputs the
estimated means, SDs and proportions for the putative
populations, together with a Chi-square statistic which
represents the likelihood ratio test for whether the current
model (multiple Normal) is a better fit than one Normal. A
significant chi-square therefore rejects the hypothesis of
one Normal. You can also restrict the model to have
equal population variances, using the parameter MODEL.
Starting values will be provided automatically if ISTART
is set to zero, otherwise the routine will expect to find
ND positive values in the array SD and ND-1 values between
0 and 1 exclusive in the first ND-1 values of array ALPHA,
where ND is the number of Normals you are fitting. Note that
the solution may not be unique, due to the tricky nature of
the 'likelihood surface' for this problem, but experience
so far suggests that the routine is pretty robust and
'almost always' converges.
If you wish, you can get more information about the progress
of the fit by specifying output to a channel (NCH) between
1 and 63, e.g. 5 sends output to the terminal.
method: Uses the simplex method of function minimisation to find
the maximum likelihood solution.
references: Royston, Flecknell & Wootton (Biol. Neonate 100-104, 42, 1982)
parameters: ND INTEGER INPUT Number of Normal distributions
to be fitted
MODEL INTEGER INPUT Model: 0 = distinct variances
1 = equal variances
ISTART INTEGER INPUT Starting values: 0 = automatic
1 = user-supplied
X ARRAY IN/OUT Sample data to be analysed
n.b. the data are sorted by the routine.
N INTEGER INPUT Number of data values in X
NCH INTEGER INPUT Channel number for output. No output
if NCH<1 or NCH>64, else goes to NCH.
XBAR ARRAY IN/OUT IN: may contain starting values for
estimated means
OUT:ML estimates of population means
SD ARRAY IN/OUT IN: may contain starting values for
estimated SDs
OUT:ML estimates of population SDs
ALPHA ARRAY IN/OUT IN: may contain starting values for
estimated proportions
OUT:ML estimates of proportions
CHI REAL OUTPUT Likelihood ratio test statistic
PCHI REAL OUTPUT P value for CHI
IFAIL INTEGER OUTPUT Failure indicator:
0=SUCCESS
1=INVALID NO OF OBSERVATIONS (5<N<=2000)
2=INVALID NO OF DISTRIBUTIONS REQUESTED
(1<ND<11)
3=TOO FEW DEGREES OF FREEDOM
4=MINIMIZATION FAIL: PARAMETER OUT OF
RANGE
5=MINIMIZATION FAIL: TOO MANY ITERATIONS
6=INTERNAL: SHOULD NOT OCCUR
(ESTIMATION OF START VALUES)
7=INTERNAL: SHOULD NOT OCCUR (SORT)
8-10=INTERNAL: SHOULD NOT OCCUR
(CHI-SQUARE TAIL)
11=ZERO STARTING VALUES FOR ALPHA OR SD
(ISTART=1 ONLY)
common blocks: CCRE40 /XCOM(2000),NCOM,NDCOM,NCHCOM,NPCOM,MODCOM/
own variables: no
routines called: E04CCF (calling P01AAF & X02AAF), FUNCT (calling
X02AEF & X02AGF), G01BCF (calling
P01AAF, S15ABF & X01AAF), M01ANF, MONIT,
MSQ, RWSTRT (calling MSQ & X02ACF), X02AAF, X02ACF
level: level ??
author: P.Royston & R.Wootton
date: July 27, 1983
STAT13 ROUTINE
CATEGORY: E20
TITLE: SUBROUTINE STAT13(T1,NF1,L1,INC,T2,NF2,L2,SS,ICH)
LANGUAGE: FORTRAN
DESCRIPTION: MARGINAL MEANS OF MULTI-WAY TABLE
T1 IS A MULTI-WAY TABLE, STORED AS A 1-DIMENSIONAL
ARRAY WITH THE 1ST FACTOR MOVING FASTEST. IT HAS NF1
FACTORS (NF1<15) WHOSE NUMBERS OF LEVELS ARE STORED
IN THE ARRAY L1. THE ROUTINE CALCULATES A MARGINAL
TABLE OF MEANS, T2, WITH NF2 FACTORS WHOSE NUMBERS
OF LEVELS ARE STORED IN THE ARRAY L2. T2 IS DEFINED
BY THE LOGICAL ARRAY INC, WHOSE ELEMENTS ARE .TRUE.
IFF THE CORRESPONDING FACTOR OF T1 IS TO BE INCLUDED
IN T2. SS IS CALCULATED AS THE SUM OF SQUARES OF THE
ELEMENTS OF T2 MULTIPLIED BY THE NUMBER OF VALUES IN
EACH MEAN, AS APPROPRIATE FOR AN ANALYSIS OF VARIANCE
OF T1.
ICH IS A PRINT CHANNEL. IF IT IS SET NON-NEGATIVE,
T2 WILL BE OUTPUT IN L2(1) COLUMNS, HEADED BY THE
CONTENTS OF INC. IF THE NUMBER OF LEVELS OF A PARTICULAR
FACTOR IS SET NEGATIVE, PRINTING OF TABLES INVOLVING THAT
FACTOR IS SUPPRESSED.
PARAMETERS: T1 REAL ARRAY INPUT MULTI-WAY TABLE
NF1 INTEGER INPUT NO OF FACTORS IN T1
L1 INTEGER ARRAY INPUT FACTOR LEVELS OF T1
INC LOGICAL ARRAY INPUT FACTORS TO BE INCLUDED
IN T2
T2 REAL ARRAY OUTPUT MARGINAL MEANS TABLE
NF2 INTEGER OUTPUT NO OF FACTORS IN T2
L2 INTEGER ARRAY OUTPUT FACTOR LEVELS OF T2
SS REAL OUTPUT SUM OF SQUARES FOR T2
ICH INTEGER INPUT PRINT CHANNEL
ROUTINES CALLED: STAT7
LEVEL: LEVEL 1
AUTHOR: M.J.R.HEALY
DATE: 27 JULY 1973
STAT14 ROUTINE
CATEGORY: E21
TITLE: SUBROUTINE STAT14(T,NF,L,SS,TW,ICH)
LANGUAGE: FORTRAN
DESCRIPTION: ANALYSIS OF MULTI-WAY TABLE
T IS A MULTI-WAY TABLE, STORED AS A 1-DIMENSIONAL
ARRAY WITH THE 1ST FACTOR MOVING FASTEST. IT HAS NF
FACTORS (NF<15) WHOSE NUMBERS OF LEVELS ARE STORED
IN THE ARRAY L. THE SUMS OF SQUARES FOR THE MAIN
EFFECTS AND INTERACTIONS ARE PLACED IN SS IN THE
SEQUENCE (1),A,B,AB,C,AC,BC,ABC,D,... (SS(1)=0.0).
TW IS A WORKING ARRAY AT LEAST AS LARGE AS T.
ICH IS A PRINT CHANNEL. IF IT IS NON-NEGATIVE,
THE MARGINAL MEANS TABLES OF T WILL BE OUTPUT
FOLLOWED BY THE VALUES IN SS, THE LATTER IN 8 COLUMNS.
PARAMETERS: T REAL ARRAY INPUT MULTI-WAY TABLE
NF INTEGER INPUT NO OF FACTORS IN T
L INTEGER ARRAY INPUT FACTOR LEVELS OF T
SS REAL ARRAY OUTPUT SUMS OF SQUARES FOR T
TW REAL ARRAY WORKSPACE
ICH INTEGER INPUT PRINT CHANNEL
ROUTINES CALLED: STAT13
CRMOVE
LEVEL: LEVEL 2
AUTHOR: M.J.R.HEALY
DATE: 27 JULY 1973
STAT19 ROUTINE
CATEGORY: E27
TITLE: STAT19(X,D1,D2,N,N1,N2,A,S,ARMS,SRMS),STT19A
LANGUAGE: FORTRAN
DESCRIPTION: CALCULATES ESTIMATES OF MEAN AND STANDARD DEVIATION,AND
THEIR ROOT MEAN SQUARED ERRORS,OF A NORMAL DISTRIBUTION
GIVEN A TYPE I CENSORED SAMPLE.
METHOD: ESTIMATES ARE LINEAR IN SUM OF OBSERVATIONS AND
CENSORING POINTS. D1 IGNORED IF N1=0,D2 IGNORED
IF N2=0
REFERENCE: J.G.SAW: BIOMETRIKA(1961) 48,367-377
PARAMETERS: X REAL ARRAY I/P DATA VALUES NOT CENSORED
D1 REAL I/P LOWER CENSORING POINT
D2 REAL I/P UPPER CENSORING POINT
N INTEGER I/P NO. OF VALUES KNOWN I.E. IN X
N1 INTEGER I/P NO. OF VALUES LT D1
N2 INTEGER I/P NO. OF VALUES GT D2
A REAL O/P ESTIMATE OF MEAN
S REAL O/P ESTIMATE OF STANDARD DEVIATION
ARMS REAL O/P ROOT MEAN SQUARED ERROR OF A
SRMS REAL O/P ROOT MEAN SQUARED ERROR OF S
'OWN' VARIABLES: NO
ROUTINES CALLED: STAT7,STAT8,STT19A,SNDEV
LEVEL: LEVEL 2
AUTHOR: S.CHINN
DATE: 29 JULY 1976
STAT7-7B ROUTINE
CATEGORY: E12
TITLE: STAT7(X,N,AVE,SSQ)
STAT7B(X,N,AVE,SSQ,S1,S3,S4)
LANGUAGE: FORTRAN
DESCRIPTION: MEAN AND POWER SUMS ABOUT MEAN. IF X IS AN ARRAY OF
LENGTH N (.GE.1), CRSTAT7 & CRSTAT7B BOTH PRODUCE THE
MEAN OF THE X-VALUES (AVE) & SUM OF SQUARES ABOUT THE
MEAN (SSQ). IN ADDITION, CRSTAT7B PRODUCES THE SUM OF
ABSOLUTE DEVIATIONS ABOUT MEAN (S1), AND SUM OF THIRD
AND SUM OF FOURTH POWERS ABOUT MEAN(S3,S4).
THE STANDARD DEVIATION CAN BE ESTIMATED AS:
SQRT(SSQ/(N-1))
METHOD: FIRST THE MEAN (AVE) IS COMPUTED, THEN THE SUMS OF
POWERS ABOUT THE MEAN. THIS METHOD IS ABOUT TWICE AS
FAST FOR SSQ AND ABOUT 3 TIMES AS FAST FOR THE HIGHER
SUMS OF POWERS AS THE 'RUNNING SUM' METHOD, WHICH
UPDATES THE POWER SUMS FOR EACH NEW INPUT VALUE.
PARAMETERS: X REAL ARRAY(N) INPUT SAMPLE VALUES
N INTEGER INPUT SIZE OF SAMPLE
AVE REAL OUTPUT MEAN OF VALUES IN X
SSQ REAL OUTPUT SUM OF SQUARES ABOUT MEAN
S1 REAL OUTPUT SUM OF ABSOLUTE DEVIATIONS
ABOUT MEAN
S3 REAL OUTPUT SUM OF CUBES ABOUT MEAN
S4 REAL OUTPUT SUM OF FOURTH POWERS ABOUT
MEAN
LEVEL: LEVEL 0
ROUTINES CALLED: ABS
AUTHOR: J.P.ROYSTON
DATE: 5 APRIL 1978
STAT7A ROUTINE
CATEGORY: E22
TITLE: STAT7A (X,N,AVE,SSQ)
LANGUAGE: FORTRAN
DESCRIPTION: EXACTLY AS STAT7 EXCEPT THAT ZERO AND NEGATIVE
RESULTS ARE IGNORED. THE VARIABLE N IS RETURNED AS
THE NUMBER OF RESULTS GREATER THAN ZERO FOUND IN X
METHOD: AS STAT7
PARAMETERS: X REAL ARRAY INPUT SAMPLE VALUES
N INTEGER INPUT/ SIZE OF SAMPLE
OUTPUT NO. OF VALUES GREATER
THAN ZERO IN X
AVE REAL OUTPUT MEAN OF VALUES GREATER
THAN ZERO IN X
SSQ REAL OUTPUT SUM OF SQUARES ABOUT
MEAN OF VALUES GREATER
THAN ZERO IN X
LEVEL: LEVEL 0
AUTHOR: G.GIBBS
DATE: 15th DECEMBER 1981
STAT7M ROUTINE
CATEGORY: E37
TITLE: STAT7M(X,N,VM,AVE,SSQ,N1)
LANGUAGE: FORTRAN
DESCRIPTION: MEAN & S.D.,ALLOWING FOR MISSING VALUES. IF ANY DATA
VALUE IN ARRAY X EQUALS VM,IT IS OMITTED FROM THE MEAN(AVE)
SUM OF SQUARES(SSQ) AND NO.OF OBSERVATIONS(N1).
METHOD: BY DEVIATIONS AROUND FIRST NON-MISSING VALUE.
PARAMETERS: X REAL ARRAY(N) INPUT ARRAY OF DATA
N INTEGER INPUT NO.OF VALUES IN X
VM REAL INPUT MISSING VALUE CODE
AVE REAL OUTPUT MEAN (ZERO IF N.EQ.0)
SSQ REAL INPUT SUM OF SQUARES ABOUT MEAN
N1 INTEGER OUTPUT NO. OF VALUES IN X UNEQUAL
TO VM.
LEVEL: LEVEL 0
ROUTINES CALLED: NONE
AUTHOR: J.P.ROYSTON
DATE: 21.APRIL 1980
STAT8 ROUTINE
CATEGORY: E13
TITLE: SUBROUTINE STAT8 (X,Q,P)
LANGUAGE: FORTRAN
DESCRIPTION: Q,P ARE THE LOWER AND UPPER TAIL AREAS OF A UNIT NORMAL
DISTRIBUTION AT DEVIATION X.
THE INTERNAL CALCULATIONS ARE PERFORMED IN DOUBLE PRECISION
ALTHOUGH THE ARGUMENTS ARE PAST IN SINGLE PRECISION.
THE RESULTS ARE GOOD TO 9 SIGNIFICANT FIGURES FOR ANY X.
REFERENCE: ADAMS,A.G. (1969), COMP.J.,12, 197-198
PARAMETERS: X REAL INPUT DEVIATE
Q,P REAL OUTPUT TAIL AREAS
COMMON BLOCKS: NONE
OWN VARIABLES: NONE
ROUTINES CALLED EXP
LEVEL: LEVEL 0
AUTHOR: M.J.R.HEALY
DATE: 25TH JANUARY, 1972
TABCLR ROUTINE
CATEGORY: L8
NAME: TABCLR
PURPOSE: Clears all tab settings
PARAMETERS: None
AUTHOR: G Gibbs 27-Oct-81
Language: Macro-20
TABLE ROUTINE
category: N9
title: subroutine TABLE (KEYWRD,KEYNUM,POINTS,STRINGS,FAIL)
language: MACRO-20
purpose: This routine is used to build up a table of keywords or phrases
for use by the KYWORD routine. Each call to TABLE adds one new keyword
or phrase to a table stored in two arrays. One array is used to store
all the text strings for each keyword, and the other array is used to
store a list of pointers and associated values. The usual practice
with these routines is to call them in a loop once only during the
program's initialization phase, and then use the arrays without
further modification. The array POINTS is supplied to the KYWORD
routine, and it locates the strings array itself. Multiple tables
may be built, but they must use different POINTS and STRINGS arrays.
The tables may be used for interactive data input with KYWORD, or for
primitive data bases using TBLOOK and TBRLOC (see below), which allow
one to find strings in tables, or any combination of these three.
The text phrase is supplied in KEYWRD, and may be a quoted
string, or text in an array terminated by a zero word or byte. A
value to associate with the phrase is supplied in KEYNUM, and will
be returned by KYWORD to the caller when that phrase is typed.
The value is limited to an 18-bit value. Thus, only numbers in the range
-131072 to 131071 can be associated with keywords. Real numbers cannot
be stored at all. This should not, in general, be a restriction.
The points array is used to store the values and pointers. One word
of this array is used for each keyword, plus three extra words for
overhead. The STRINGS array is used to store the text strings
themselves, and should be constructed to be long enough to hold
each string end to end, plus one word for each string, plus 2 words of
overhead. It is probably easiest just to make this array much too big.
On the first call to TABLE, strings must be all zero, and POINTS
should contain the size of the POINTS array in its first element (which
will be altered.)
Example
=======
CALL TABLE('Aspirin',1,drugs,worksp,fail)
CALL TABLE('Paracetamol',2,drugs,worksp,fail)
CALL TABLE('Opium',3,drugs,worksp,fail)
or:-
CALL TABLE('Start',1,comand,worka,fail)
CALL TABLE('Continue',2,comand,worka,fail)
CALL TABLE('Stop',3,comand,worka,fail)
then:-
CALL KYWORD('Drug given >',drugs,drugno,fail)
CALL KYWORD('Next action>'comand,cmdnum,fail)
method: The TBADD% Jsys is used.
parameters: keywrd Array or input The keyword or phrase for the
string table
keynum Any input Value to associate with keyword
points Integer input/output Array of pointers and values
strings Integer array output String storage
fail Integer output Error code (see below)
Error codes: 0 Success
13 Entry is already in table
12 Table is full
own variables: no
routines called: none
level: level 1
author: G.Gibbs / Kevin Ashley
date: August 1982
TABSET ROUTINE
CATEGORY: L9
NAME: TABSET(COLUMN)
PURPOSE: Places a tab stop at the specified column
PARAMETERS: COLUMN Integer input Column number
AUTHOR: G. Gibbs 27-Oct-81
Language: Macro-20
TBLOOK ROUTINE
category: N11
title: subroutine TBLOOK(KEYWRD,TAB,KEYNUM,FAIL)
language: MACRO-20
purpose: This routine can be used to lookup a character string in a
keyword table created by the TABLE routine. It matches the string in
much the same wasy as the KYWORD routine does - a string matches an
entry if it is an unambiguous abbreviation of a keyword in the table.
Letter case is not considered when searching for matches. The search
is binary, and part of the operating system, and is thus extremely fast,
even for tables containing many thousands of entries. This can be used
to implement primitive (or not so primitive) data bases.
The text phrase to look for is supplied in KEYWRD, and may be
a quoted string, or an array or variable, with text terminated by a zero
word or byte. The maximum length of this string is 349 characters.
KEYNUM returns the keyword value associated with the keyword when
TABLE was used to build the table of pointers. If fail is non-zero,
KEYNUM will contain zero.
Example
=======
CALL TABLE('Aspirin',1,drugs,worksp,fail)
CALL TABLE('Paracetamol',2,drugs,worksp,fail)
CALL TABLE('Opium',3,drugs,worksp,fail)
CALL TBLOOK('OP',drugs,drugno,fail)
Returns 3 in drugno, as 'op' is an unambiguous abbreviation
for 'opium'.
method: The TBLUK% Jsys is used.
parameters: keywrd Array or input The keyword or phrase to search
string for
tab Array Input Points array from TABLE routine.
keynum Integer Output Value associated with keyword
fail Integer output Error code (see below)
Error codes: 0 Success
16 No such keyword in table
17 Keyword is ambiguous
own variables: no
routines called: none
level: level 1
author: Kevin Ashley
date: November 1982
TBRLOC ROUTINE
category: N12
title: subroutine TBRLOC(POINTS,STRINGS)
language: MACRO-20
purpose: This routine is used to relocate pointers in a keyword table
after that table has been written to disk and read back in again, or
after it has been copied from one array to another.
The tables constructed by TABLE and used by KYWORD and TBLOOK
contain absolute memory addresses as part of their pointers. Thus,
if you copy them from one array to another, or if one run of a program
writes the array out, and another reads it in again, these pointers may
be askew. The TBRLOC routine merely relocates these pointers, using
information stored there by the TABLE routine.
This allows a program to use a keyword table as an index to
a file or database. Say you have a number of records about patients,
which you wish to look up via hospital number or patient name. You
can keep a random access file sorted by hospital number, and use a
binary search or index to retrieve the record by number. You may also
use a keyword table, which stores patient names, and with them, the
first record number for that patient. The keyword arrays are written
out by the program whenever it exits, and read back in again when it
starts.
Example:
CALL TABLE(PATNAM,PATPNT,PATSTR,FAIL)
.... Many times
WRITE(20) PATPNT !unformatted write of pointers
WRITE(20) PATSTR !and strings
Later.. in another program perhaps
READ(20) PATPNT !Read pointers
READ(20) PATSTR !Read strings from disk also
CALL TBRLOC(PATPNT,PATSTR) !Relocate pointers.
parameters: points Array Input/output Points from TABLE routine
strings Array Input/output Strings from TABLE routine
own variables: no
routines called: none
level: level 1
author: Kevin Ashley
date: November 1982
TEXTIN ROUTINE
category: N1
title: subroutine TEXTIN (prompt,answer,fail,back,length,help,defalt)
language: MACRO-20
purpose: The routine prompts the terminal user with a supplied string, and then
reads an arbitrary text string. A default can be supplied by the
program, and the routine also returns the number of characters read.
It is possible to specify a maximum number of characters to read.
Text strings are terminated with a null character, and this null must
be allowed for in calculations of the array size of the answer.
If the first element of the ANSWER array is set to a non-zero integer
value less than 350 on entry to the routine, it is used as a limit on
the number of characters to accept. Otherwise, the limit is 350 char-
acters, which could overflow many arrays. The restriction on the limit
being non-zero and < 350 means that if the array contains left-over
text, this will not be interpreted as a meaningless limit.
This routine is part of the terminal data entry routine suite -
full facilities for recognition of defaults, retyping of prompts,
automatic recognition of UNKNOWN, EXIT and ^ are available. This
is more fully explained in the DATA-ENTRY section of this program,
which should be read in conjunction with this.
This routine is unable to read a string
which is an abbreviation of the words EXIT or UNKNOWN.
method: The COMND% Jsys is used.
parameters: prompt Array or input The prompt for the user
string
answer Array output Textmtypedrbygusergth
fail Integer output Error code (see below)
back Label input *Optional*
Statement to jump to if
user uses ^ to go back one
question, or 0 to ignore this.
length Integer output *Optional*
Length of text string.
help Array or input *Optional*
string Help message to be output if
user types question mark, or
0 to ignore argument.
A help message of "text string"
will be used if you do not
supply one.
defalt Array or input *Optional*
string Default to be used if user
types escape, or 0 if not
desired.
NOTE: Optional input arguments can be omitted if they are
all trailing, so the call
CALL TEXTIN('Name:',PATNAM,FAIL) is valid, as is
CALL TEXTIN('Name:',PATNAM,FAIL,*10,NAMLEN,'Enter patient''s
& name')
which supplies a label to return to if the user types ^ (the
label is 10), stores the length of the name in NAMLEN, and
outputs a help message if a question mark is typed. No default
has been supplied, however.
You may use 0 for an optional argument if you wish to supply
some arguments that follow it, but you wish that argument to
be ignored (ie if you want to supply a default, but not a help
message.)
Error codes: 0 Success
1 Invalid input (should not occur)
2 Not confirmed (this is only possible if RETRYS is set
to a non-negative value and the user types EXIT,
UNKNOWN or ^ and then types rubbish after the command.
Normally it cannot occur.)
15 Maximum length of input exceeded. This error can only
occur if a maximum length was specified by setting word
1 of array answer to a value >0 and < 350 before
calling TEXTIN.
-1 User typed UNKNOWN.
-2 User typed EXIT (this code is only returned if you did
not initialize location EXILAB in common block CMDPRM
with the label to jump to).
-3 User typed "^" to go back a question (this code is only
returned if you did not supply a label to jump to on
this condition.)
common blocks: /cmdstg/cmdbfr(70),atmbfr(70),fdb(5),hlpbfr(25),
defbfr(56),gjfblk(15),argblk(5)
/cmdprm/initf,retrys,tried,endnse,raise,savret,
savp,exilab,useext,spare(3),cmdblk(10)
(All integer, see DATA-ENTRY for more info)
own variables: no
routines called: none
level: level 1
author: G.Gibbs / Kevin Ashley
date: August 1982
VT-SUBROUTINE-CONVENTIONS ROUTINE
All terminal control routines may be called from MACRO as
well as high-level languages. The MACRO entry points are the same as the
FORTRAN entry points, but the names are preceded by a dollar sign, thus
$VTINIT, $VTEND, etc. The impure storage may be kept in the user's own
reserved area. To do this, place the following statements in your
program -
SEARCH VTMAC
$TITL==0
VTSTOR
This should preceed all relocatable code. Alternatively, call
the FORTRAN VTINIT, which reserves its own impure area. Parameters
are passed in the scratch acs 1,2 etc. as needed. VTATR may only have
one parameter when called from MACRO. Acs 1-4 may be destroyed as
usual by all routines. You are advised not to fiddle with the JFN mode
word or CCOC word during program execution, unless you leave the altered
bits alone.
Where references are made to quoted string arguments from other
languages, the macro programmer is expected to provide a byte pointer
(which may be in implicit form) to an ASCIZ string.
VT132 ROUTINE
CATEGORY: L21
NAME: VT132
PURPOSE: Sets the terminal to 132-column width
PARAMETERS: None.
AUTHOR: K. Ashley 27-Oct-80
Language: Macro-20
VT80 ROUTINE
CATEGORY: L20
NAME: VT80
PURPOSE: Sets the terminal to 80-column width
PARAMETERS: None.
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTATR ROUTINE
CATEGORY: L26
NAME: VTATR(N1,N2,....)
PURPOSE: Sets or clears the character attributes. These affect whether
characters should be typed in reverse video, blinking, bold
or underlined, or any combination of these. Each attribute is
given a type code - 1 for bold, 4 for underline, 5 for blink
and 7 for reverse. You may specify as many or few parameters as
you wish. Successive calls are cumulative. Specify 0 to clear
all attributes and return typeout to normal.
PARAMETERS: N1,N2,... Integer input Type codes
AUTHOR: K. Ashley 27-Oct-80
Language: Macro-20
VTBIG ROUTINE
CATEGORY: L28
NAME: VTBIG(' string of characters') or VTBIG(ARRAY)
PURPOSE: Types a text string in double height, double width chars.
PARAMETERS: ARRAY Input - either a quoted string ('string') or
an array name, which contains the characters terminated by a
zero byte or word.
AUTHOR: K. Ashley 27-Oct-80
Language: Macro-20
VTCBL ROUTINE
CATEGORY: L10
NAME: VTCBL
PURPOSE: Clears from the beginning of line to the cursor
PARAMETERS: None
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTCBS ROUTINE
CATEGORY: L12
NAME: VTCBS
PURPOSE: Clears from the start of screen to the cursor
PARAMETERS: None.
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTCLIN ROUTINE
CATEGORY: L11
NAME: VTCLIN
PURPOSE: Clears the entire line containing the cursor
PARAMETERS: None.
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTCLS ROUTINE
CATEGORY: L13
NAME: VTCLS
PURPOSE: Clears the entire screen
PARAMETERS: None.
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTDWN ROUTINE
CATEGORY: L16
NAME: VTDWN(N)
PURPOSE: Moves the cursor down the specified number of places
PARAMETERS: N Integer input Number of places to move
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTEND ROUTINE
CATEGORY: L30
NAME: VTEND
PURPOSE: Called at the end of a program to restore the JFN mode word,
so allowing TOPS-20 to continue to remember the terminal width.
Not essential, but recommended.
PARAMETERS: None.
AUTHOR: K. Ashley 27-Oct-80
Language: Macro-20
VTGOF ROUTINE
CATEGORY: L33
NAME: VTGOF
PURPOSE: Turns off VT100 graphics characters after VTGON has turned
them on.
PARAMETERS: None.
AUTHOR: K. Ashley 17-Dec-81
Language: Macro-20
VTGON ROUTINE
CATEGORY: L32
NAME: VTGON
PURPOSE: Used to turn on the VT100's graphics character set. These are
useful for drawing graphs or line figures, or displaying certain
mathematical symbols. See the VT100 guide for fuller
documentation.
PARAMETERS: None.
AUTHOR: K. Ashley 17-Dec-81
Language: Macro-20
VTINIT ROUTINE
CATEGORY: L0
NAME: VTINIT
PURPOSE: Sets up the environment for the other routines. This routine
MUST be called first, otherwise you will receive errors at link time
and at runtime.
PARAMETERS: None
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTIUP ROUTINE
CATEGORY: L19
NAME: VTIUP
PURPOSE: Performs a "reverse index" - the cursor is moved up one
place, but, unlike CURUP, if the top of the scrolling region is
reached, the screen is scrolled down. A reverse linefeed, in
effect.
PARAMETERS: None.
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTJMP ROUTINE
CATEGORY: L24
NAME: VTJMP
PURPOSE: Sets jump scroll mode
PARAMETERS: None.
AUTHOR: K. Ashley 27-Oct-80
Language: Macro-20
VTLED ROUTINE
CATEGORY: L7
NAME: VTLED(N)
PURPOSE: Lights the specified LED above the keyboard, or clears all leds
PARAMETERS: N Integer input LED number, or 0 to clear all
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTLFT ROUTINE
CATEGORY: L14
NAME: VTLFT(N)
PURPOSE: Moves the cursor left the specified number of places
PARAMETERS: N Integer input Number of places to move
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTORG ROUTINE
CATEGORY: L23
NAME: VTORG(FLAG)
PURPOSE: Alters the setting of "origin mode". This defines whether
cursor addresses are relative to the scrolling region. If origin
mode is off, cursor addresses are relative to the whole screen.
If origin mode is on, cursor addresses are relative to the
scrolling region, (1,1) is the top left of the scrolling region,
and the cursor cannot move outside the scrolling region.
PARAMETERS: FLAG Integer input 0=off, 1=on
AUTHOR: K. Ashley 27-Oct-80
Language: Macro-20
VTRES ROUTINE
CATEGORY: L18
NAME: VTRES
PURPOSE: Restores cursor position and attributes saved by VTSAV
PARAMETERS: None.
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTRIT ROUTINE
CATEGORY: L15
NAME: VTRIT(N)
PURPOSE: Moves the cursor right the specified number of places
PARAMETERS: N Integer input Number of places to move
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTSAV ROUTINE
CATEGORY: L17
NAME: VTSAV
PURPOSE: Saves the current cursor position and attributes
PARAMETERS: None.
AUTHOR: K. Ashley 27-Oct-81
Language: Macro-20
VTSCRN ROUTINE
CATEGORY: L22
NAME: VTSCRN(FLAG)
PURPOSE: Sets the screen background to light or dark
PARAMETERS: FLAG Integer input 0=light, 1=dark
AUTHOR: K. Ashley 27-Oct-80
Language: Macro-20
VTSMTH ROUTINE
CATEGORY: L25
NAME: VTSMTH
PURPOSE: Sets smooth scroll mode
PARAMETERS: None.
AUTHOR: K. Ashley 27-Oct-80
Language: Macro-20
VTTY ROUTINE
CATEGORY: L34
NAME: VTTY(TERMIN)
PURPOSE: This routine can be used to tell the terminal routines (PUTCRS,
CLEOS, etc. ) to use a terminal other than the one you are
logged in on. This is useful to debug a program which uses the
screen in a complex manner, as you can still use the job's own
terminal for debugging, while the screen input/output is
happening on another terminal. One supplies the terminal name
as a parameter to switch I/O, as in CALL VTTY('TTY24:'). If
no parameter is supplied, or the parameter consists of blanks,
then output is switched back to your controlling terminal.
This routine should be called before VTINIT.
PARAMETERS: TERMIN String constant or Input Name of terminal to
character variable. which output is to be
redirected
AUTHOR: K. Ashley 2-Aug-83
Language: Macro-20
VTWID ROUTINE
CATEGORY: L27
NAME: VTWID('TEXT string') or VTWID(array)
PURPOSE: Types a text string in double width characters.
PARAMETERS: text array or quoted string input. This can be a string
as 'hello there' or an array containing the characters. If
an array, it must be terminated by a zero byte or word.
AUTHOR: K. Ashley 27-Oct-80
Language: Macro-20
WCOEF ROUTINE
category: E45
title: subroutine wcoef(a,n,n2,eps,ifault)
language: fortran
description: Set up WEXT, must be called once for a given sample size,
before WEXT is called.
method:
references: Applied Statistics Vol.31 no.2 1982, algorithm AS 181
parameters: a real output coefficients for calculation of w, set
array by WCOEF
n integer input sample size
n2 integer input [n/2] 1/2 n if n even 1/2(n-1) if n odd
eps real output minimum possible value of w set by WCOEF
ifault integer output fault indicator, 3 if n2.ne.[n/2]
2 if n.gt.2000
1 if n.le.2
0 otherwise
common blocks: none
own variables: no
routines called: nscor2
level: level 2
author: P.Royston
date: February 1983
WEXT ROUTINE
category: E44
title: subroutine wext(x,n,ssq,a,n2,eps,w,pw,ifault)
language: Fortran
description: Calculates the Shapiro and Wilk W test for detecting departures
from normality. The data vector (X) must be ordered before
calling WEXT. For a given sample size N, call the auxiliary
subroutine WCOEF once before using WEXT. Calculations
are still performed if N>2000, but the significance level
PW for the test cannot be as accurate as for lower N.
references: Applied Statistics Vol.31 no.2 1982, algorithm AS 181
parameters: x real input ordered sample values
n integer input sample size
ssq real input sum of squares of data about mean
a real input coefficients for calculation of w, set
array by WCOEF
n2 integer input [n/2] 1/2 n if n even 1/2(n-1) if n odd
eps real input minimum possible value of w set by WCOEF
w real output W statistic
pw real output significance level of W
ifault integer output fault indicator, 4 if data not sorted
or are all equal
3 if n2.ne.[n/2]
2 if n.gt.2000
1 if n.le.2
0 otherwise
common blocks: none
own variables: no
routines called: stat8,poly
level: level 1
author: P.Royston
date: February 1983
YESNO ROUTINE
category: N6
title: subroutine YESNO (prompt,answer,fail,back,help,defalt)
language: MACRO-20
purpose: The routine prompts the terminal user with a supplied string, and then
reads a YES, NO or UNKNOWN from the terminal. A default may be
supplied by the calling program, which will be used if the user types
return or escape. The returned value may be interpreted as logical,
in which case YES is true, and NO or UNKNOWN are false, or integer in
which case YES is -1, NO is 1 and UNKNOWN is 0.
A default can be supplied by the program, which is used if the
user types escape or return.
This routine is part of the terminal data entry routine suite -
full facilities for recognition of defaults, retyping of prompts,
automatic recognition of UNKNOWN, EXIT and ^ are available. This
is more fully explained in the DATA-ENTRY section of this program,
help file, which should be read in conjunction with this.
method: The COMND% Jsys is used.
parameters: prompt Array or input The prompt for the user
string
answer Integer or output Returned value
logical
fail Integer output Error code (see below)
back Label input *Optional*
Statement to jump to if
user uses ^ to go back one
question, or 0 to ignore this.
help Array or input *Optional*
string Help message to be output if
user types question mark, or
0 to ignore argument.
A help message of "one of
the following:
NO UNKNOWN YES"
will be used if you do not
supply one.
defalt Array or input *Optional*
string Default to be used if user
types escape, or 0 if not
desired.
NOTE: Optional input arguments can be omitted if they are
all trailing, so the call
CALL YESNO('New call:',YEA,FAIL) is valid, as is
CALL YESNO('New call:',YEA,FAIL,*10,NAMLEN,'Enter whether this
& is a new call','YES')
which supplies a label to return to if the user types ^ (the
label is 10),stores the true or false answer in YEA, and
outputs a help message if a question mark is typed. YES
is assumed if nothing is entered.
You may use 0 for an optional argument if you wish to supply
some arguments that follow it, but you wish that argument to
be ignored (ie if you want to supply a default, but not a help
message.)
Error codes: 0 Success
1 Invalid input (ie not YES NO or UNKNOWN)
2 Not confirmed (this indicates superfluous input
after the word.)
10 Does not match keyword
-1 User typed UNKNOWN.
-2 User typed EXIT (this code is only returned if you did
not initialize location EXILAB in common block CMDPRM
with the label to jump to).
-3 User typed "^" to go back a question (this code is only
returned if you did not supply a label to jump to on
this condition.)
common blocks: /cmdstg/cmdbfr(70),atmbfr(70),fdb(5),hlpbfr(25),
defbfr(56),gjfblk(15),argblk(5)
/cmdprm/initf,retrys,tried,endnse,raise,savret,
savp,exilab,useext,spare(3),CMDBLK(10)
(All integer, see DATA-ENTRY for more info)
own variables: no
routines called: none
level: level 1
author: G.Gibbs / Kevin Ashley
date: August 1982
LP-PLOTTING
CATEGORY: *D
D PLOTTING (LINE PRINTER)
-----------------------
1 CRC161
CALCULATES AXIS PARAMETERS FOR PLOTTING ROUTINE CRC561.
2 CRC162
PLOTS AXIS FOR PLOTTING ROUTINES CRC362 AND CRC561.
3 CRC163
CALCULATES AXIS PARAMETERS FOR PLOTTING ROUTINE CRC362.
4 CRC361
CALCULATES AXIS PARAMETERS FOR ROUTINE CRC561 TO FIT ON A PAGE.
5 CRC363
PLOTS HISTOGRAM.
7 CRC561,CRC362.
LINE PRINTER PLOT (TWO WAY DISTRIBUTION)
8 NPLOT
NORMAL PLOT.
9 CPLOT
HALF-NORMAL PLOT OF CORRELATION COEFFICIENTS.
10 CPLOT2
SCATTER DIAGRAM WITH STATISTICS
11 NPLOT1
NORMAL PLOT (REVISED VERSION OF NPLOT)
12 CRLHST
PLOTS HISTOGRAM ACROSS PRINTER PAGE
13 NRMPLT
Normal plot (latest vesion [1983])
NRMPLT ROUTINE
category: D8
title: NRMPLT (x,n,nplus,nchan,ifault)
language: Fortran
description ** note: updated version of nplot; parameter list changed **
This subroutine plots (on channel nchan) normal
plots for the n values of the floating point
array x. The ordered values of x are plotted
on the y-axis. On the x-axis is plotted the
normal equivalent deviate corresponding
to the proportion pi = (i-1/3)/(n+1/3)
for the i'th ranked element of x. Plotting is
done by sub-routine crc 561 and the normal
equivalent deviates calculated by function sndev.
Values of mean and sd are printed after the plot on
channel nchan, together with skewness, kurtosis,
shapiro and wilk's w statistic and their
p values as tests of departure from normality.
Note that the array x must be dimensioned to at
least 2.5 times n (it contains work space). The
first n elements of x contain the data to be used.
They will be sorted by the routine into ascending order.
parameters: x array(nplus) vector of observations
n integer number of observations
nplus integer at least 2.5*n, dimension of x
nchan integer output channel number.
ifault integer fault code: 0 - ok
1 - n<3
2 - n>2000 (not fatal,
no w-statistic)
3 - nplus too small
11-16 - error from pskew
21-26 - error from pkurt
routines called: crc154,crc561,stat7,wcoef,wext,pskew,pkurt,crshpe,
function sndev
level: level 2
authors Klim McPherson/Patrick Royston
date 18 Nov 1983
CRSIL ROUTINE
CATEGORY: L29
NAME: CRSIL(FLAG)
PURPOSE: Turns terminal echoing of typein on or off, depending on the
value of the FLAG parameter
PARAMETERS: FLAG Integer input 1=off,0=on
**NOTE** This used to be incorrect. It is now RIGHT.
AUTHOR: K. Ashley 1-Feb-81
Language: Macro-20
crgraf routine
category: c37
title: subroutine crgraf(i,x,y,n,isel)
language: fortran
description: graph drawing routine no. 4
This draws graph axes and graphs of various types.
Multiple plots may be drawn on one set of axes with mutiple
calls to this routine, with i set negative for the final plot
on a given set of axes.
The graph axes are drawn when no axes have previously
been drawn or the previous graph has been finished,
i.e. any call to crgraf can draw the axes. These axes
will be 'beautified' unless stated otherwise, that is
to say the number of intervals and the axes limits will
be altered to provide acceptable values along the axis from the
values supplied in x and y. These may be numbers calculated by
you (in an array of 2) with a call which only draws the axes,
if you wish.
A graph is finished by setting the graph type or
histogram type (in crhist) negative.
You have further control over the plotting with other routines
to modify the default settings :-
crgrit initialises the plotting
crgrax controls the type of axes
crgrlb inserts labels
crhist draws histograms
parameters: i integer i/p graph type. Set negative to terminate
the current graph. The types are:-
1=draws axes only (x and y axes not
beautified)
2=scatter plot
3=scatter plot with points joints
in order presented
4=scatter plot and linear regression
line of y on x fitted
5=scatter plot and quadratic
polynomial fitted
6=scatter plot and cubic polynomial
fitted
7=scatter plot and linear regression
line of x on y fitted
10=draws axes only (x axis only
beautified)
11=draws axes only (y axis only
beautified)
12=draws axes only (both axes are
beautified)
13-17 are the same as 3-7 but
without the scatter plot.
x,y real i/p input data arrays
arrays
n integer i/p no. of points in input arrays
isel integer i/p this is a two digit selector.
for graphs where only axes are drawn
(i=1,10,11,12) the 1st digit is
redundant and the 2nd digit indicates
the type of axis offset required;
0-no offset, 1-y axis pulled away,
2-x axis pulled away, 3-both axes
pulled away
for all other graphs the 1st digit
indicates the line type for any
lines that are to be drawn (not axes)
0-solid,1-dotted,2-dashed,3-chained.
the 2nd digit indicates the marker
to be used in the scatter plot.
reference: docas report on plotting package
routines called: crmnx,crgrer,crpoly,crmove,crfunc,crscal,
crax5,crax5f,crax5g,crax5l
+ gino-f and ginograf routines
level: level 3
author: L.J.Dobson/G.Gibbs
date: December 1981
crax5-f-g routine
category: c32
title: crax5(x,y,alen,angle,nint,ia,itype,ltick,i1,i2,
vmin,vmax)
crax5f(x,y,alen,angle,nint,ia,itype,ltick,i1,i2,
vmin,vmax,func)
crax5g(x,y,alen,angle,nint,ia,itype,ltick,i1,i2,
vmin,vmax,func)
crax5l(x,y,alen,angle,nint,ia,itype,ltick,i1,i2,
vmin,vmax)
language: fortran
description: alternative axis plotting routines, with different ways
of numerical labelling.
crax5 produces a normal linear axis
crax5f produces log spacing of ticks, with power labels
crax5g produces equal spacing of the ticks.
crax5l produces log spacing of ticks, with nicer labels
parameters: x,y real i/p coordinates, (in plotter space),
of start of axis
alen real i/p length of axis (mm in plotter sp.)
angle real i/p angle of axis (in degrees(0. or 90.))
nint integer i/p no. of intervals along axis (if
negative (crax5 only) gives a
category axis with nint categories)
ia int.array i/p text to be used as label along axis
or quoted (terminated with '*.' as for a
string. string in crgrlb -c35)
e.g. 'Serum t3*.'
itype integer i/p orientation of values against
ticks and position of label.
0- normal x-axis with end label
1- normal x-axis with centred label
2- normal y-axis with end label
3- normal y-axis with centred label
ltick integer i/p controls the orientation and
labelling of ticks along axis:
if +ve, ticks on natural side
if -ve, ticks on opposite.
i1 integer i/p field width (including any decimal
point) of numerical labels
i2 integer i/p no. of digits after decimal point
of numerical labels
vmin, real i/p minimum and maximum coorinate
vmax values to be plotted against axis
func real i/p user defined mapping function,
function taking as an argument a single
real value (must be declared as
external in the calling segment).
this function must be the inverse
of the transform used on the data
being plotted.(note exp10(x) is
available in the crc library)
reference: docas report on plotting package
compressed integers: yes
fortran traceable: level 1
routines called: crax5- crax5g,ipk,xnull
crax5f- ipk
crax5g- ipk
crax5l- ipk
+ gino-f and ginograf routines
level: crax5- level 2
crax5f,g & l - level 1
author: G.Gibbs
date: January 82
CRGRAX ROUTINE
category: C36
title: subroutine crgrax(i,alen,nint,ntick,nf,nd,itype)
language: fortran
description: graph drawing routine no. 3
this alters the axis parameters from their default or
previous values. the default values for the parameters
are shown in brackets, < >.
parameters: i integer i/p axis type. 1= x axis, 2= y axis
alen real i/p axis length in mm. <half available size>
If zero is supplied the default will
be used.
nint integer i/p specifies the no. of intervals, if
it is negative a category axis is
given <5>
ntick integer i/p controls the orientation and
labelling of ticks along the axis,
it will label every ntick'th tick.
0 will suppress the tick marks
if +ve ticks are on natural side
if -ve they are on the opposite <1>
nf integer i/p specifies the maximum numeric label
field width, including any decimal
point and sign.<6>
nd integer i/p specifies the no. of digits
required after the decimal point
for numeric labels. <1>
itype integer i/p specifies the type of axis required <0>
0= normal or linear
transformed } { 1= log10 (log spaced ticks)} always
by the } { 2= log (log spaced ticks) } beautified
user before } { 3= log10 (equal spaced ticks)
calling } { 4= log (equal spaced ticks)
} { 5= equally spaced tickes, values printed
defined by user's function, myaxfn. a
version is available in the library
suitable for hours of the day
REAL FUNCTION MYAXFN (X)
MYAXFN=AMOD(X,24.0)
RETURN
END
supply untransformed } { 6= log10 (log spaced ticks, nice labels)
data } { 7= log10 ( -- ditto -- power labels)
reference: DoCaS report on plotting package
routines called: crgrer
level: level 1
author: L.J.Dobson
date: December 1977
CRGRWO ROUTINE
category: C42
title: subroutine crgrwo
language: Fortran
description: Terminate a graph being drawn by crgraf or crhist.
This provides an alternative to setting the first argument
negative in a call to either crgraf or crhist, in the final
call to that routine. Calling this routine may be easier in
circumstances where many lines are being plotted on one set
of axes, and some may consider it to be better style anyway.
parameters: none
common blocks: plghp
routines called: crgrer
level: level 2
author: G.Gibbs
date: August 1984
1
LOGICAL AND SHIFT OPERATIONS.
15 PK
CHARACTER INSERTION(ASCII code).
16 ICRTLU
TABLE LOOK-UP BY BINARY SEARCH.
19 CRCLR
CLEARING AN ARRAY
21 ICRCH7
INTEGER TO CHARACTER DIGITS CONVERSION
24 CRON,CROFF,CRTEST
BIT SWITCHING ROUTINES
25 ICRSCH
SEARCH FOR A CHARACTER STRING
26 RIGHT
Returns the right half of a machine word, sign extended
27 LEFT
Returns the left half of a machine word, sign extended
28 GETADR
Returns the absolute address of its argument.
29 TBLOOK
Lookup a table of character strings by binary search.
30 CRSOUT
Send a character string to the terminal.
31 CRSINT
Move characters from an array to a character variable.
32 CRAISE
Raise a character string to upper case.
CRAISE ROUTINE
category: A32
title: character function craise(lower)
language: macro
description: Returns a character variable raised to upper case.
Fortran 77 only.
method: uses the hardware instruction MOVST (MOVe String Translated).
parameters: name type i/o description
lower character input character variable to be raised to
upper case
example: character*80 line,upper,craise
read (unit=5,fmt='a') line
write (unit=5,fmt=*) craise(line)
or...
upper=craise(line)
routines called: none
level: level 0
author: G.Gibbs
date: March 1981
DATES
CATEGORY: *J
J DATE HANDLING
---------------
4 DATE
Today's date in characters
15 JDATE
Convert Julian day no. to calendar date
16 JDAY
Convert calendar date to Julian day no.
17 JTODAY
Today's Julian day no.
18 CRMTOC
Convert integer month to characters
19 CRNTOI
Convert character date(numeric) to 3 integers
20 CRDTOI
Convert character date to 3 integers
21 CRDOUT
Output date from 3 integers
22 CRJOUT
Output date from Julian day number
23 EASTER
Julian day number of Easter Sunday in specified year
24 MONTH
Returns month in characters
CHKLET ROUTINE
category: K28
title: character function chklet(number)
language: Fortran 77
description: returns the check letter corresponding to the hospital
number supplied.
parameters: number integer input the hospital number
routines called: none
example: character*1 chklet,hoslet
integer hosnum
accept (fmt='i,a') hosnum,hoslet
if (hoslet.ne.chklet(hosnum))then
type *,'Check letter does not match'
end if
author: G.Gibbs
date: June 1984
MISCELLANEOUS
CATEGORY *K
MISCELLANEOUS
-------------
1 GENCHK
Generates check letters
8 CRUSER
User's name obtained
9 CRHALT
Halts program with no messages
10 CREMOT
Plots files on Oki printer or Calcomp plotter
11 CRRSCN
Obtains EXEC command line for program
12 CRCMD
Allows a program to execute EXEC commands
13 CRLDAV
Returns the 1-minute, 5-minute and 15-minute load averages.
14 CRMAIL
Allows a program to send MAIL to other users.
15 CRWAIT
Suspends program execution for a number of milliseconds.
16 CROOM
Returns space available in directory and on whole disk.
17 CRLBOP
Opens a LBR-format library file for reading
18 CRLBRD
Reads a record from a file opened by CRLBOP
19 CRSIZE
Returns the size of a file in pages and bytes.
20 LCRBAT
Returns whether running in batch or not.
21 CRXPNG
Expunges a disk directory
22 CRRUN
Runs a program and passes a command line.
23 ICRNTM
Returns a programs run time.
24 DOJSYS
Executes a TOPS-20 monitor call (!)
25 GETDDT
Loads DDT and causes an unsolicted breakpoint.
26 CRTIMR
Calls a specified routine when a time limit expires.
27 SIMINP
Simulate terminal input
28 CHKLET
Returns the check letter for a hospital number
CRDTOI ROUTINE
Category: J20
Title: CRDTOI (ICDATE,IDAY,MONTH,IYEAR)
Language: MACRO
Descripton: Converts TOPS 20 character date to 3 integers.
A date in character form (e.g. '22-Nov-80' ,as returned by
the DEC routine DATE ),is converted into 3 integer numbers
(i.e. 22 11 1980 ).
The routine will accept any reasonable format for the date,
including any unambiguous truncation of the month in characters
or the month as a number, in the latter case the month MUST
be after the day. Non valid dates will return 3 zero values
as will a date prior to 17 November 1858.
It will accept input directly from your terminal if the
first argument is set to 1. In this case it is recomended that
a prompt be given before calling the routine and a responce
is given afterwards, perhaps a call to CRDOUT q.v.
e.g.:- TYPE 50
50 FORMAT ('ENTER THE DATE >>')
CALL CRDTOI (1,IDAY,MONTH,IYEAR)
CALL CRDOUT (1,IDAY,MONTH,IYEAR,2)
Method: The routine uses the monitor call IDTNC%
(Monitor Call Reference Manual 3-177)
Parameters: ICDATE array or INPUT Date in character form
character
IDAY integer } The 3 parts
MONTH integer } OUTPUT of the date
IYEAR integer }
See also: DATEIN
Author: G.Gibbs
Date: April 1981, character variables July 1984
CRLBRD routine
Category: K17
Title: CALL CRLBRD(ARRAY,RECLEN,JFN,START,ERR)
Language: MACRO-20
Description: Reads a record from a library file opened by CRLBOP. The record
returned is padded with trailing blanks if a character variable is used,
but not otherwise, and includes the carriage return/linefeed combination.
If ARRAY is integer it receives the record in A5 format,
RECLEN the record length in characters (not including the CR/LF).
JFN and START should be preserved between calls - they are initially
supplied by CRLBOP, and are updated on each call to CRLBRD. ERR is 0
for success, else is positive.
Parameters:
ARRAY Output Array or Receives record from file
character
RECLEN Output Integer Length of record in characters
JFN Input Integer Job file number from CRLBOP
START Input/output Integer Address of next record from CRLBOP
ERR Output Integer Return status
Errors: 0 - OK
1 - end of file
2 - some error reading file
Author: Kevin Ashley
Level: 0
Date: 11 Aug 1982, character variables July 1984
CRRSCN ROUTINE
CATEGORY: K11
NAME: CRRSCN(COMAND,LENGTH)
LANGUAGE: MACRO-20
AUTHOR: K. Ashley 8-Oct-81
PURPOSE: The CRRSCN routine enables a program to "rescan" the TOPS-20
command that caused the program to be run. This means that a program can
be written to behave more like a TOPS-20 command and can be invoked from
indirect command files. It eases the burden on users of your program who
do not have to distinguish so much between programs and commands. For
instance, say you have a program called REDUCE which takes a number of
input data files and produces one output file. The user could invoke
your program with the command
@REDUCE LPT:=INPUT1.DAT,INPUT2.DAT
indicating that they wished the input to come from the two files on the
right of the equals sign, and the output to the lineprinter. On calling
CRRSCN, the array or character variable COMAND would receive that
command in full, and you can parse it as you wish. The second parameter,
LENGTH, is an integer variable that receives the length of the command
in characters.
PARAMETERS: COMAND array or output the line of text typed in to
character invoke the program calling this routine
LENGTH integer output number of characters in COMAND
LEVEL: level 0
SIMINP ROUTINE
category: K27
title: subroutine SIMINP(TEXT)
language: Macro 20
description: Simulate input from the terminal.
The text of the argument will apear to the program as if it
were typed in at the terminal. This may be of use if your
program is running another program, and wishes to send data to
it, or if some segment of code which you are unable to alter
always requires the same input. Any input which the user may
have typed ahead will be thrown away. Multiple lines should be
separated by the backslash character "\".
e.g. call siminp('LINE ONE\LINE TWO\LINE THREE')
N.B. Do not use from batch.
method: The routine uses the monitor calls CFBIF%, STI% and GETJI%
references: Monitor Calls Reference Manual
parameters: text array input text to be input from terminal
or
character
common blocks: none
own variables: no
routines called: none
level: level 1
author: Paul O'Riordan/Geoff Gibbs
date: Jan-83
GENCHK ROUTINE
category: K1
title: GENCHCK (IUN,ICHL)
language: Fortran
description: The check letter is derived by taking n-th
of a set of selected letters of the alphabet,
where n is the remainder on dividing the unit
number by 17. The check letter is stored in
the first character position of ichl i.e. in
a1 format.
Look-up table.
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
A B D E F H J K L M N P R S T V X
parameters: iun integer input nph unit number
ichl integer output check letter.
routines called: none
level: level 0
see also: the character function CHKLET in the CRC library
author: G.Gibbs
date: 2nd June 1981
CHARACTERS
CATEGORY: *A
A. BIT,CHARACTER AND WORD MANIPULATION.
-------------------------------------
1 CRMVCH
MOVE CHARACTERS
2 COPY
COPY CHARACTERS
3 ICOMP
COMPARE CHARACTER STRINGS
4 CRMOVE
COPY WORDS
6 ICRCH2
CHARACTER EXTRACTION(ICL code, limited life).
7 ICRCH3
CHARACTER INSERTION(ICL code, limited life).
8 ICRCH4
CHARACTER DIGITS TO INTEGER CONVERSION
11 IPK
CHARACTER EXTRACTION(ASCII code).
12 NPH2
TABLE LOOK-UP BY SERIAL SEARCH
13 NPH3
BIT PATTERN EXTRACTION.
14 NPH11
LOGICAL AND SHIFT OPERATIONS.
15 PK
CHARACTER INSERTION(ASCII code).
16 ICRTLU
TABLE LOOK-UP BY BINARY SEARCH.
19 CRCLR
CLEARING AN ARRAY
21 ICRCH7
INTEGER TO CHARACTER DIGITS CONVERSION
24 CRON,CROFF,CRTEST
BIT SWITCHING ROUTINES
25 ICRSCH
SEARCH FOR A CHARACTER STRING
26 RIGHT
Returns the right half of a machine word, sign extended
27 LEFT
Returns the left half of a machine word, sign extended
28 GETADR
Returns the absolute address of its argument.
29 TBLOOK
Lookup a table of character strings by binary search.
30 CRSOUT
Send a character string to the terminal.
31 CRSINT
Move characters from an array to a character variable.
32 CRAISE
Raise a character string to upper case.
33 CRUPER
Raise a character string to upper case.
graph-plotting
category: *C
C Graph Plotter Routines.
------------------------
29 CRBARS
draws histogram bars
30 CRHATCH
hatches or shades a rectangular area
32 CRAX5, 5F, 5G and 5L
alternative axis plotting routines
34 CRGRIT (Graph drawing routine no. 1)
initialises graph drawing and resets to default values
35 CRGRLB (Graph drawing routine no. 2)
sets up a label of the required type
36 CRGRAX (Graph drawing routine no. 3)
alters axis parameters from default or previous values
37 CRGRAF (Graph drawing routine no. 4)
draws axes and various types of graphs
38 CRHIAX (Graph drawing routine no. 5)
sets up extra parameters for plotting histograms
39 CRHIST (Graph drawing routine no. 6)
draws axes and various types of histogram
40 PNPLOT
produces up to 5 normal plots on the plotter
41 CRERBR
draws error bars
42 CRGRWO
Terminates a plot drawn by crgraf or crhist
KYALOW ROUTINE
category: N12
title: subroutine KYALOW (allows)
language: MACRO-20
purpose: This routine is used only in conjunction with the KYWORD routine.
It allows special characters to be considered valid parts of keywords
that would otherwise be rejected. For instance, KYWORD does not usually
allow the character "^" in keywords (although they can be placed in
a table with the TABLE routine without problems.) However, attempting
to read one of these keywords using KYWORD could easily produce an
error, either because two keywords were only distinguishable by
an up-arrow, or because KYWORD complained of excess input on the line
(all the characters following the uparrow.)
However, a call to KYALOW, with the parameter being a text
string, character expression or integer array containing an uparrow
will make all following calls to KYWORD accept the "^" character.
See KYDALW for methods of preventing certain characters being
allowed as part of keywords (for instance, if you want to use the dot
or full stop character to separate keywords.)
Calling KYALOW with a null string , or a value of zero, causes
the keywords breakset (the characters allowed) to be reset to their
default settings. These are:
All alphanumerics are permitted, and also the following special
characters:
"-" (hyphen) "_" (underline "." (full stop)
"#" (hash or pound) "*" (asterisk) "(" (open round bracket)
" " (space) "/" (solidus) "%" (percent sign)
":" (colon) "'" (single quote) ")" (close round bracket)
parameters: allows Character expression Input The characters to
or array be added to those
permitted by KYWORD
or null to reset
defaults.
Examples:
CALL KYALOW('+=') !Permits plus and equals as
!part of strings
CALL KYALOW(0) !Resets defaults
routines called: none
level: level 1
author: Kevin Ashley
date: October 1984
KYDALW ROUTINE
category: N13
title: subroutine KYDALW (allows)
language: MACRO-20
purpose: This routine is used only in conjunction with the KYWORD routine.
It stops characters being considered valid parts of keywords that
would otherwise be allowed. For instance, KYWORD usually allows
the character "." in keywords. If you wish dots or full stops in your
input to delimit different keywords, you can call KYDALW with a string
containing dot as the parameter, and KYWORD will now terminate keyword
input on dot, as well as the usual characters.
See KYALOW for methods of allowing certain characters being
part of keywords (for instance, if you want to use the equals or
plus sign character as part of keywords.)
Calling KYDALW with a null string , or a value of zero, causes
the keywords breakset (the characters allowed) to be reset to their
default settings. These are:
All alphanumerics are permitted, and also the following special
characters:
"-" (hyphen) "_" (underline "." (full stop)
"#" (hash or pound) "*" (asterisk) "(" (open round bracket)
" " (space) "/" (solidus) "%" (percent sign)
":" (colon) "'" (single quote) ")" (close round bracket)
parameters: allows Character expression Input The characters to
or array be removed from those
permitted by KYWORD
or null to reset
defaults.
Examples:
CALL KYDALW('. ') !Stops dot and space being part
!of valid keywords
CALL KYDALW(0) !Resets defaults
routines called: none
level: level 1
author: Kevin Ashley
date: October 1984
TIMEIN ROUTINE
category: N14
title: subroutine TIMEIN (PROMPT,SECNDS,FAIL,BACK,LOWER,UPPER,HELP,DEFALT)
language: MACRO-20
purpose: The routine prompts the terminal user with a supplied string, and then
reads a time, typed in any reasonable format. Times may be 24 hour
format, separated by colons or spaces, or may be 12-hour with trailing
AM or PM, and possibly also 3-letter time zone (such as GMT or GDT for
greenwich daylight time, often known as BST.)
The routine returns the number of seconds since midnight, local
time. If a time zone is supplied, the seconds are relative to that time
zone (ie 00:01 always returns 60 seconds, even if it is 00:01 BST).
Upper and lower limits for the time, also in seconds since midnight,
may be supplied. They are inclusive and independant of each other.
A default can be supplied by the program (as seconds since
midnight.)
This routine is part of the terminal data entry routine suite -
full facilities for recognition of defaults, retyping of prompts,
automatic recognition of UNKNOWN, EXIT and ^ are available. This
is more fully explained in the DATA-ENTRY section of this program,
which should be read in conjunction with this.
method: The COMND% Jsys is used.
parameters: prompt Array or input The prompt for the user
string
Secnds Integer output Returned seconds since midnight
fail Integer output Error code (see below)
back Label input *Optional*
Statement to jump to if
user uses ^ to go back one
question, or 0 to ignore this.
Lower Integer input *Optional*
Lower limit for time (inclusive)
or 0 if not desired.
Upper Integer input *Optional*
Upper limit for time (inclusive)
or 0 if not desired.
help Array or input *Optional*
string Help message to be output if
user types question mark, or
0 to ignore argument.
If no help is supplied, a
message will be constructed of
the form "Enter a time, before
11:00 PM, after 1:00 PM", with
limits only mentioned if they
are supplied in the call.
defalt Integer input *Optional*
Default to be used if user
types escape, or 0 if not
desired.
NOTE: Optional input arguments can be omitted if they are
all trailing, so the call
CALL TIMEIN('Time:',SECNDS,FAIL) is valid, as is
CALL TIMEIN('Time:',SECNDS,FAIL,*10,0,3600,'Enter time
& of birth')
which supplies a label to return to if the user types ^ (the
label is 10),stores the time of birth in SECNDS, specifies no
lower limit, but specifies an upper limit of 1 AM, and
outputs a help message if a question mark is typed. No default
has been supplied, however.
You may use 0 for an optional argument if you wish to supply
some arguments that follow it, but you wish that argument to
be ignored (ie if you want to supply a default, but not a help
message.)
Error codes: 0 Success
1 Invalid input
2 Not confirmed (this is only possible if RETRYS is set
to a non-negative value and the user types EXIT,
UNKNOWN or ^ and then types rubbish after the command.)
18 Time is not in valid format
19 Time is out of range.
-1 User typed UNKNOWN.
-2 User typed EXIT (this code is only returned if you did
not initialize location EXILAB in common block CMDPRM
with the label to jump to).
-3 User typed "^" to go back a question (this code is only
returned if you did not supply a label to jump to on
this condition.)
common blocks: /cmdstg/cmdbfr(70),atmbfr(70),fdb(5),hlpbfr(25),
defbfr(56),gjfblk(15),argblk(5)
/cmdprm/initf,retrys,tried,endnse,raise,savret,
savp,exilab,useext,spare(3),cmdblk(10)
(All integer, see DATA-ENTRY for more info)
own variables: no
routines called: none
level: level 1
author: G.Gibbs / Kevin Ashley
date: August 1982
PNPLOT Routine
category: c40
title: PNPLOT(X,N,NDS,IDS,NCHAN)
language: Fortran
description: Produces up to 5 normal plots on either the Oki dot matrix
printer or the Calcomp pen plotter. Standardized data
values(x-xbar)/sd are placed on the y-axis;
otherwise method and output are the same as the line-
printer normal plot [d8]. Mean,standard deviation,
skewness, kurtosis and (when 2<n<51) the w-statistic
(see crwtst [e34]) for each data set are output to
channel nchan. The plot is sent to the Oki
unless nchan is set negative, when it is sent to the
Calcomp.
The routine is currently restricted to 1000 values, due
to the size of an internal array.
parameters: X real array(n) input contains up to 5 data sets
N integer input total number of x-values
NDS integer input number of data sets to
be analysed (up to 5)
IDS integer array(5)input sample sizes of data
sets in x (in same order)
NCHAN integer input output channel number.
if set negative the plot
will be sent to the calcomp
rather than to the oki
level: level 4
routines called: crwtst,crc154,crshpe,sndev
crgrit,crgrax,crgrlb,crgraf
author: J.P.Royston/G.Gibbs
date: 12 January 1982.