Re: Issues with Simulation
Hi Matt,
Recently, we had a similar error with a model simulation. The model was a
complex with multiple conditions, covariates etc.
We asked Bob Bauer, and per his reply....
There is a feature in NM7.3, where you may set NOHABORT and it will
generate the TABLE by ignoring some errors during the $TABLE step. He
warned us to use the results at our own risk as they may be incorrect.
We utilized this and saw that the CL was a function of 2-3 covariates of
which one was missing in the data file. The CL output in the TABLE file was
all NaN. One of the required covariates was missing and hence not being
available to generate the results.
Not sure if this option works in NM7.2 or if there is any other cause for
the error.
Kind regards
Navin Goyal
Quoted reply history
On Fri, Feb 7, 2014 at 12:11 PM, Matt Hutmacher <[email protected]>wrote:
> Dear All,
>
>
>
> Recently we were running some simulation/estimation using SUBPROBLEMS in
> NONMEM 7.2. In these runs we were generating covariates using a random
> distribution such as the following:
>
>
>
> IF(ICALL.EQ.4) THEN
>
> COV1 = function + range*PHI(ETA(5))
>
> COV2 = function(COV1)*EXP(ETA(4))
>
> ENDIF
>
> TVKA = THETA(1)
>
> KA = TVKA*DEXP(ETA(1))
>
> TVCL = THETA(2)*(COV2/X)**THETA(7)
>
> TVV = THETA(3)*(COV2/X)**THETA(8)
>
>
>
> The NONMEM run completes, but a TABLE file is not output and the following
> error is reported
>
>
>
>
>
> 0PROGRAM TERMINATED BY NP4F
>
> PRED RETURNS DIFFERENT OUTPUTS WITH THE SAME INPUTS
>
> MESSAGE ISSUED FROM TABLE STEP
>
>
>
> How do we avoid this error and get a table file?
>
> We looked at the archive, and found that Dennis Fischer had asked this
> question previously, but we did not see any follow-up on how to solve the
> problem. I do not remember having this issue with previous versions of
> NONMEM.
>
>
>
> Can anybody provide a solution to this problem? We would greatly
> appreciate it.
>
> Best,
> Matt
>
>
>
--
Navin Goyal