Date: Mon, 12 Jul 1999 09:44:14 -0400 (EDT)
From: SRIRAM KRISHNASWAMI <Sriram.Krishnaswami@dupontpharma.com>
Subject: Error message
Dear All
Could anybody help me with the following error message
0PROGRAM TERMINATED BY OBJ
ERROR IN NCONTR WITH INDIVIDUAL 1 ID=0.10000000E+01
NUMERICAL HESSIAN OF OBJ. FUNC. FOR COMPUTING CONDITIONAL ESTIMATE
IS NON POSITIVE DEFINITE
MESSAGE ISSUED FROM ESTIMATION STEP
AT 0TH ITERATION, UPON EVALUATION OF GRADIENT WITH RESPECT TO OMEGA
0TRY SUBSTANTIALLY INCREASING "INITIAL ESTIMATE OF SIGMA"
Stop - Program terminated
I am trying to use the FOCE method to analyze the data when this error message appears.
FO method works very well for the same initial sigma and omega estimates
And no matter how much the omegas and sigmas are , this error message appears with the FOCE method.
Thanks in advance
Sriram
Error message
Date: Tue, 13 Jul 1999 10:52:26 -0700 (PDT)
From: ABoeckmann <alison@c255.ucsf.edu>
Subject: Re: Error message
A question was asked about the message
NUMERICAL HESSIAN OF OBJ. FUNC. FOR COMPUTING CONDITIONAL ESTIMATE IS NON POSITIVE DEFINITE
In general, this message indicates that an estimate of the individual's eta could not be obtained at the specified theta value. There is probably something rather wrong with the intraindividual model for the individual's data.
Since this message appears with the first individual and the initial estimate of theta, the modeling difficulty may apply to the data from other individuals as well.
-- Alison Boeckmann
Date: Tue, 17 Aug 1999 09:41:01 -0400 (EDT)
From: SRIRAM KRISHNASWAMI <Sriram.Krishnaswami@dupontpharma.com>
Subject: Error message
Dear NMusers
I would appreciate if someone could help me with the following error message.
forrt1: severe <33> : ENDFILE error, Unit 4, file MPIIIC2N01
The syntax on the control file looks like this
ESTIMATION METHOD=1 PRINT=5 POSTHOC
MAXEVAL MSFO=MPIIIC2N01 NOABORT
Thanks
Sriram
From: "Gibiansky, Ekaterina" <gibianse@globomax.com>
Subject: RE: Error message
Date: Tue, 17 Aug 1999 11:07:36 -0400
This looks very much like the error that you get trying to use Digital 5 compiler without the patches. It can give you the error on some files and not on others. You can update your compiler by downloading the patch from the Digital Web page and following the instructions.
Ekaterina Gibiansky
GloboMax LLC
7250 Parkway Drive, Suite 430
Hanover, MD 21076
Voice (410) 782-2234
FAX (410) 712-0737
E-mail: gibianse@globomax.com
Date: Tue, 17 Aug 1999 09:19:38 -0700 (PDT)
From: ABoeckmann <alison@c255.ucsf.edu>
Subject: Re: Error message
Like Ekaterina (see below), I assume that Siriam is using Digital Fortran, probably version 5.0.
It turns out that there are known bugs when the "fl32" command is used (DIGITAL's "PowerStation I/O compatibility mode"). These bugs do not exist when the "df" command is used (DIGITAL's "native DIGITAL Fortran I/O"). I don't know if the bugs were fixed in later versions.
Rather than patching the compiler, I suggest that with DIGITAL Visual Fortran, all compiles of NONMEM, PREDPP, and NM-TRAN be done using the "df" command rather than "fl32".
E.g., reinstall NONMEM with this command:
a:\setup a c nmv df y
Alison Boeckmann