From: Justin Wilkins jwilkins@uctgsh1.uct.ac.za
Subject: [NMusers] Trouble increasing ETA capacity in NMV 1.1
Date: Thu, April 29, 2004 9:49 am
Hi all,
Following on from the discussion by Joost DeJongh and William Bachman
( http://www.cognigencorp.com/nonmem/nm/99aug212000.html) I've been
attempting to increase NONMEM's ETA capacity to cope with a particularly
demanding IOV problem.
However, the amended source is getting stuck on LPAR2 in NSIZES:
0WARNING: PARAMETER LPAR2 NOT SET CORRECTLY
The comment says
C LPAR2: I*(I+1)/2 WHERE I=SQRT(FLOAT(2*LPAR))
but with an LPAR input value of 100, I make this 107.07 (assuming FLOAT
converts (2*LPAR) to a real number)... which doesn't work. And besides,
the manual gives examples:
LPAR LPAR2
30 28
70 66
...implying an LPAR : LPAR2 ratio of 1 : 0.95. However, punching in
LPAR2=95 where LPAR=100 doesn't work either.
Adjusted parameters:
LTH=40
LVR=60
MTHVR=60
LWS1=79000
Have I missed something? Can anyone see where I'm going wrong?
Justin
--
-------------------------------------------
Justin Wilkins
-------------------------------------------
Division of Pharmacology
Department of Medicine
University of Cape Town
-------------------------------------------
K45 Old Main Building
Groote Schuur Hospital
Observatory 7925
South Africa
-------------------------------------------
Email: jwilkins@uctgsh1.uct.ac.za
Phone: +27 (0)21 406 6448
Fax: +27 (0)21 448 1989
Web: http://www.uct.ac.za/depts/pha/
Trouble increasing ETA capacity in NMV 1.1
From: slaporte silvy.laporte@chu-st-etienne.fr
Subject: RE: [NMusers] Trouble increasing ETA capacity in NMV 1.1
Date: Thu, April 29, 2004 10:17 am
Justin,
You can use this modified code (in NSIZES).
With LAPR=81 (15+66+10) then I=SQRT(162)=12
then LPAR2=12*13/2=78.
Hope this helps.
Silvy Laporte
code:
C LTH: MAX. NO. OF THETA'S
PARAMETER (LTH=15)
C LVR: MAX. NO. OF ETA'S + EPS'S
PARAMETER (LVR=66)
C LVR2: MAX. NO. OF ETA'S PERMITTED WHEN LAPLACIAN METHOD IS USED
PARAMETER (LVR2=10)
C LPAR: MAX. NO. OF PARAMETERS THAT MAY BE ESTIMATED
PARAMETER (LPAR=81)
C LPAR2: I*(I+1)/2 WHERE I=SQRT(FLOAT(2*LPAR))
PARAMETER (LPAR2=78)
C LPAR3: MAX(2850,LPAR*(LPAR+1)/2)
PARAMETER (LPAR3=3321)
C MTHVR: MAX(LTH,LVR)
PARAMETER (MTHVR=66)
C NO: MAX NO. OF OBSERVATION RECORDS / INDIVIDUAL RECORD
PARAMETER (NO=350)
C LNP4: SIZE OF COMMON NMPRD4
PARAMETER (LNP4=1000)
C LWS1: SIZE OF WORKING SPACE 1
PARAMETER (LWS1=42000)
C LADD: SHOULD EXCEED NO*(NO+1)/2 * (NUMBER OF DIAGONAL BLOCKS OF SIGMA)
From: Leonid Gibiansky
Subject: RE: [NMusers] Trouble increasing ETA capacity in NMV 1.1
Date: Thu, April 29, 2004 10:21 am
My guess (just looking on the formulas that you wrote) is that I=x operator
takes the floor (minimum integer less than x) operation. Then LPAR2 is
given by the S+ code:
LPAR <- 100
I <- floor(sqrt(2*LPAR))
LPAR2<- I*(I+1)/2
LPAR2
105
It gives correct values for the examples, and the value 105 for LPAR=100
Leonid
From: Justin Wilkins jwilkins@uctgsh1.uct.ac.za
Subject: RE: [NMusers] Trouble increasing ETA capacity in NMV 1.1
Date: Fri, April 30, 2004 5:30 am
Dear Silvy, Leonid and Bill,
Thanks for the advice - problem neatly solved!
Justin
-------------------------------------------
Justin Wilkins
-------------------------------------------
Division of Pharmacology
Department of Medicine
University of Cape Town
-------------------------------------------
K45 Old Main Building
Groote Schuur Hospital
Observatory 7925
South Africa
-------------------------------------------
Email: jwilkins@uctgsh1.uct.ac.za
Phone: +27 (0)21 406 6448
Fax: +27 (0)21 448 1989
Web: http://www.uct.ac.za/depts/pha/
_______________________________________________________