Dear All,
I have a question on termination testing during importance sampling.
I used SAEM and IMP sequentially as follows:
$EST METHOD=SAEM LAPLACE INTERACTION NBURN=3000 NITER=30000 PRINT=10
CTYPE=1 CINTERVAL=10 NOABORT
$EST METHOD=IMP LAPLACE INTERACTION NITER=30 CTYPE=1 PRINT=1
ISAMPLE=20000 NOABORT EONLY=1 MAPITER=0
The IMP step completes 30 iterations with the message #TERM:
OPTIMIZATION NOT COMPLETED
If I look at the 30 iterations it seems odd that the termination
criterium was not met - see table below. In addition, the root.cnv
output file does only include information about the SAEM Burn-in, which
was indeed terminated, but not on the IMP part.
Any ideas?
Kind regards,
Filip De Ridder
Janssen R&D
MONITORING OF SEARCH:
iteration 0 OBJ= -1040.79270825355 eff.= 18884. Smpl.=
20000. Fit.= 0.96728
iteration 1 OBJ= -1041.22721039882 eff.= 8139. Smpl.=
20000. Fit.= 0.90861
iteration 2 OBJ= -1040.90636254178 eff.= 7960. Smpl.=
20000. Fit.= 0.90691
iteration 3 OBJ= -1041.44226717287 eff.= 7998. Smpl.=
20000. Fit.= 0.90717
iteration 4 OBJ= -1040.95071538084 eff.= 8008. Smpl.=
20000. Fit.= 0.90724
iteration 5 OBJ= -1041.22419150397 eff.= 7998. Smpl.=
20000. Fit.= 0.90730
iteration 6 OBJ= -1041.57239188217 eff.= 8010. Smpl.=
20000. Fit.= 0.90739
iteration 7 OBJ= -1041.09660100136 eff.= 8016. Smpl.=
20000. Fit.= 0.90735
iteration 8 OBJ= -1041.25562857940 eff.= 7993. Smpl.=
20000. Fit.= 0.90713
iteration 9 OBJ= -1041.41360999108 eff.= 8019. Smpl.=
20000. Fit.= 0.90745
iteration 10 OBJ= -1041.56018094091 eff.= 8005. Smpl.=
20000. Fit.= 0.90728
iteration 11 OBJ= -1041.02520618588 eff.= 7990. Smpl.=
20000. Fit.= 0.90733
iteration 12 OBJ= -1041.45539992459 eff.= 8011. Smpl.=
20000. Fit.= 0.90739
iteration 13 OBJ= -1041.93095864456 eff.= 8001. Smpl.=
20000. Fit.= 0.90728
iteration 14 OBJ= -1041.52589661482 eff.= 8016. Smpl.=
20000. Fit.= 0.90760
iteration 15 OBJ= -1041.48177246391 eff.= 7985. Smpl.=
20000. Fit.= 0.90717
iteration 16 OBJ= -1041.41608562484 eff.= 8036. Smpl.=
20000. Fit.= 0.90746
iteration 17 OBJ= -1041.09827194526 eff.= 7990. Smpl.=
20000. Fit.= 0.90714
iteration 18 OBJ= -1041.41267372704 eff.= 8015. Smpl.=
20000. Fit.= 0.90739
iteration 19 OBJ= -1041.24437170124 eff.= 8003. Smpl.=
20000. Fit.= 0.90723
iteration 20 OBJ= -1041.38502319648 eff.= 8003. Smpl.=
20000. Fit.= 0.90725
iteration 21 OBJ= -1041.60595559575 eff.= 8018. Smpl.=
20000. Fit.= 0.90725
iteration 22 OBJ= -1041.07447863555 eff.= 7995. Smpl.=
20000. Fit.= 0.90725
iteration 23 OBJ= -1044.02744029407 eff.= 8024. Smpl.=
20000. Fit.= 0.90726
iteration 24 OBJ= -1041.26147186940 eff.= 7994. Smpl.=
20000. Fit.= 0.90749
iteration 25 OBJ= -1041.02132091032 eff.= 8010. Smpl.=
20000. Fit.= 0.90723
iteration 26 OBJ= -1041.69960672796 eff.= 8002. Smpl.=
20000. Fit.= 0.90728
iteration 27 OBJ= -1041.11216053314 eff.= 7989. Smpl.=
20000. Fit.= 0.90729
iteration 28 OBJ= -1041.18001151427 eff.= 8027. Smpl.=
20000. Fit.= 0.90741
iteration 29 OBJ= -1041.52777622052 eff.= 8009. Smpl.=
20000. Fit.= 0.90746
iteration 30 OBJ= -1041.76570085815 eff.= 8005. Smpl.=
20000. Fit.= 0.90726
NM7.2 Importance Sampling Termination Testing
This is an erroneous statement by NONMEM given that convergence is not tested
when EONLY=1, even when CTYPE>0, which you may ignore. It has been corrected
for NM7.2.
Robert J. Bauer, Ph.D.
Vice President, Pharmacometrics, R&D
ICON Development Solutions
7740 Milestone Parkway
Suite 150
Hanover, MD 21076
Tel: (215) 616-6428
Mob: (925) 286-0769
Email: [email protected]
Web: http://www.iconplc.com/
Quoted reply history
________________________________
From: [email protected] [mailto:[email protected]] On
Behalf Of De Ridder, Filip [JRDBE]
Sent: Wednesday, April 04, 2012 5:51 AM
To: [email protected]
Subject: [NMusers] NM7.2 Importance Sampling Termination Testing
Dear All,
I have a question on termination testing during importance sampling.
I used SAEM and IMP sequentially as follows:
$EST METHOD=SAEM LAPLACE INTERACTION NBURN=3000 NITER=30000 PRINT=10 CTYPE=1
CINTERVAL=10 NOABORT
$EST METHOD=IMP LAPLACE INTERACTION NITER=30 CTYPE=1 PRINT=1 ISAMPLE=20000
NOABORT EONLY=1 MAPITER=0
The IMP step completes 30 iterations with the message #TERM: OPTIMIZATION NOT
COMPLETED
If I look at the 30 iterations it seems odd that the termination criterium was
not met - see table below. In addition, the root.cnv output file does only
include information about the SAEM Burn-in, which was indeed terminated, but
not on the IMP part.
Any ideas?
Kind regards,
Filip De Ridder
Janssen R&D
MONITORING OF SEARCH:
iteration 0 OBJ= -1040.79270825355 eff.= 18884. Smpl.= 20000.
Fit.= 0.96728
iteration 1 OBJ= -1041.22721039882 eff.= 8139. Smpl.= 20000.
Fit.= 0.90861
iteration 2 OBJ= -1040.90636254178 eff.= 7960. Smpl.= 20000.
Fit.= 0.90691
iteration 3 OBJ= -1041.44226717287 eff.= 7998. Smpl.= 20000.
Fit.= 0.90717
iteration 4 OBJ= -1040.95071538084 eff.= 8008. Smpl.= 20000.
Fit.= 0.90724
iteration 5 OBJ= -1041.22419150397 eff.= 7998. Smpl.= 20000.
Fit.= 0.90730
iteration 6 OBJ= -1041.57239188217 eff.= 8010. Smpl.= 20000.
Fit.= 0.90739
iteration 7 OBJ= -1041.09660100136 eff.= 8016. Smpl.= 20000.
Fit.= 0.90735
iteration 8 OBJ= -1041.25562857940 eff.= 7993. Smpl.= 20000.
Fit.= 0.90713
iteration 9 OBJ= -1041.41360999108 eff.= 8019. Smpl.= 20000.
Fit.= 0.90745
iteration 10 OBJ= -1041.56018094091 eff.= 8005. Smpl.= 20000.
Fit.= 0.90728
iteration 11 OBJ= -1041.02520618588 eff.= 7990. Smpl.= 20000.
Fit.= 0.90733
iteration 12 OBJ= -1041.45539992459 eff.= 8011. Smpl.= 20000.
Fit.= 0.90739
iteration 13 OBJ= -1041.93095864456 eff.= 8001. Smpl.= 20000.
Fit.= 0.90728
iteration 14 OBJ= -1041.52589661482 eff.= 8016. Smpl.= 20000.
Fit.= 0.90760
iteration 15 OBJ= -1041.48177246391 eff.= 7985. Smpl.= 20000.
Fit.= 0.90717
iteration 16 OBJ= -1041.41608562484 eff.= 8036. Smpl.= 20000.
Fit.= 0.90746
iteration 17 OBJ= -1041.09827194526 eff.= 7990. Smpl.= 20000.
Fit.= 0.90714
iteration 18 OBJ= -1041.41267372704 eff.= 8015. Smpl.= 20000.
Fit.= 0.90739
iteration 19 OBJ= -1041.24437170124 eff.= 8003. Smpl.= 20000.
Fit.= 0.90723
iteration 20 OBJ= -1041.38502319648 eff.= 8003. Smpl.= 20000.
Fit.= 0.90725
iteration 21 OBJ= -1041.60595559575 eff.= 8018. Smpl.= 20000.
Fit.= 0.90725
iteration 22 OBJ= -1041.07447863555 eff.= 7995. Smpl.= 20000.
Fit.= 0.90725
iteration 23 OBJ= -1044.02744029407 eff.= 8024. Smpl.= 20000.
Fit.= 0.90726
iteration 24 OBJ= -1041.26147186940 eff.= 7994. Smpl.= 20000.
Fit.= 0.90749
iteration 25 OBJ= -1041.02132091032 eff.= 8010. Smpl.= 20000.
Fit.= 0.90723
iteration 26 OBJ= -1041.69960672796 eff.= 8002. Smpl.= 20000.
Fit.= 0.90728
iteration 27 OBJ= -1041.11216053314 eff.= 7989. Smpl.= 20000.
Fit.= 0.90729
iteration 28 OBJ= -1041.18001151427 eff.= 8027. Smpl.= 20000.
Fit.= 0.90741
iteration 29 OBJ= -1041.52777622052 eff.= 8009. Smpl.= 20000.
Fit.= 0.90746
iteration 30 OBJ= -1041.76570085815 eff.= 8005. Smpl.= 20000.
Fit.= 0.90726