LOG...

6 messages 5 people Latest: Sep 20, 2001

LOG...

From: Beautyemy Date: September 19, 2001 technical
From: "=?big5?B?YmVhdXR5ZW15?=" <beautyemy@sinamail.com> Subject: LOG... Date: Wed, 19 Sep 2001 20:23:55 +0800 Dear NONMEM users: As a beginer of NONMEM, I have a problem of using LOG(natral log). It seems that NONMEM can't regnize LOG...... What should I do?? Thanks for any suggestion.....

Re: LOG..

From: Luann Phillips Date: September 19, 2001 technical
From: Luann Phillips <luann.phillips@cognigencorp.com> Subject: Re: LOG.. Date: Wed, 19 Sep 2001 11:35:27 -0400 I have used LOG in NONMEM frequently so I'm curious about how you're using it and what error messages you are receiving. I know that NONMEM will accept Y=LOG(F). However, some platforms (depending on the FORTRAN compiler and its options) will terminate the run when F=0 (which frequently happens on the first dose record) because LOG(0) = negative infinity. To prevent termination of a run because F=0 for dose records, you can try the following: IF(AMT.NE.0) TPRED=F+1 ;dose record IF(AMT.EQ.0) TPRED=F ;event record Y = TPRED + ESP(1) Luann Phillips Cognigen Corporation Manager of Pharmacometrics R&D 395 Youngs Rd. Williamsville, NY 14221 email: Luann.Phillips@cognigencorp.com Voice: (716) 633-3463 ext. 236 Fax: (716) 633-7404

RE: LOG..

From: Yuying Gao Date: September 19, 2001 technical
From: Yuying Gao <ygao@pharsight.com> Subject: RE: LOG.. Date: Wed, 19 Sep 2001 09:18:55 -0700 You can try the following: $ERROR TY=3DF IF(F.GT.0) THEN TY=3DLOG(F) ELSE TY=3D0 ENDIF Y=3DTY+EPS(1) -------------------------------- Christine Y. Gao, M.D., Ph.D. Consulting Group Pharsight Corporation 800 El.Camino Real, Suite 200 Mountain View, CA 94040 Phone: (650)314-3848 Fax: (650)314-3810 E-mail: ygao@pharsight.com http://www.pharsight.com

Re: LOG..

From: Luann Phillips Date: September 19, 2001 technical
From: Luann Phillips <luann.phillips@cognigencorp.com> Subject: Re: LOG.. Date: Wed, 19 Sep 2001 13:09:23 -0400 The code below can alter the objective function value because it could change the predicted value for an event record. Luann Phillips > Yuying Gao wrote: > > You can try the following: > > $ERROR > TY=F > IF(F.GT.0) THEN > TY=LOG(F) > ELSE > TY=0 > ENDIF > Y=TY+EPS(1)

Re: LOG..

From: Alan Xiao Date: September 19, 2001 technical
From: Alan Xiao <Alan.Xiao@cognigencorp.com> Subject: Re: LOG.. Date: Wed, 19 Sep 2001 17:09:36 -0400 Be careful to use ".EQ." function in Fortran. You might want to use (ABS(AMT - ????) .LT. ??? ) instead of (AMT.EQ.0). Alan. -- ***** Alan Xiao, Ph.D *************** ***** PK/PD Scientist *************** ***** Cognigen Corporation ********** ***** Tel: 716-633-3463 ext 265 ******

RE: LOG..

From: Vladimir Piotrovskij Date: September 20, 2001 technical
From: "Piotrovskij, Vladimir [JanBe]" <VPIOTROV@janbe.jnj.com> Subject: RE: LOG.. Date: Thu, 20 Sep 2001 08:22:32 +0200 $ERROR CALLFL =3D 0 requests $ERROR to be called with observation records only (AMT=3D0) Best regards, Vladimir