Warning: PARAMETER ESTIMATE IS NEAR ITS BOUNDARY

From: Shankar Lanke Date: August 21, 2014 technical Source: mail-archive.com
Dear nmusers, I have a trouble in solving this warning error despite of going through all old post in this forum. I have used the initial estimates from WINONLIN parameter estimates, Tried median and means as initial estimates (IEs). $EST METHOD=1 INTERACTION PRINT=5 MAX=9999 SIG=3 POSTHOC MSFO=1cm_mixture.msf NOABORT As mentioned earlier by Tom Ludden & Scot Van Wart, I have tried everything they suggested, but still no luck $EST METHOD=CONDITIONAL INTER MAXEVAL=9999 PRINT=5 NOTHETABOUNDTEST NOOMEGABOUNDTEST NOSIGMABOUNDTEST 1. I have tried the initial estimates of the last run as initial estimate for the next run 2. I have changed the initial estimates around the IEs from winonlin run. 3. I have changed the upper bounds, omega block (several times) I appreciate your suggestions, thank you very much in advance. Regards, Shankar Lanke Ph.D. $PK IF (MIXNUM.EQ.1)THEN TVCL=THETA(1) CL=TVCL*EXP(ETA(1)) ELSE TVCL=THETA(2) CL=TVCL*EXP(ETA(2)) ENDIF IF (MIXNUM.EQ.1)THEN TVV=THETA(3) V=TVV*EXP(ETA(3)) ELSE TVV=THETA(4) V=TVV*EXP(ETA(4)) ENDIF S1=V $MIX NSPOP=2 P(1)=THETA(5) P(2)=1-P(1) $ERROR DEL=0 IF (F.LE.0.0001) DEL=1 IPRE=F W= IPRE +DEL IRES= DV-IPRE IWRE=IRES/W Y = F + W*ERR(1) $EST METHOD=CONDITIONAL INTERACTION PRINT=5 MAX=9999 NOTHETABOUNDTEST NOOMEGABOUNDTEST NOSIGMABOUNDTEST SIG=3 POSTHOC MSFO=1cm_mixture.msf NOABORT $THETA (0,22,1000);[CL MIXNUM-1] (0,105,1000);[CL MIXNUM-2] (0,20,1000);[V MIXNUM-1] (0,44,1000);[V MIXNUM-2] (0,0.5,1) ;[MIXEST] $OMEGA 0.0279 ;[P] omega(1,1)S 0.1 ;[P] omega(2,2) 0.1 ; 0.1 ; 0.1 ; $SIGMA 0.0755 ;[A] sigma(1,1)