IMP method: covariance matrix was absent from output file even though estimation succeeded.
Hi all,
I am running a model using IMP method. Model optimization was successful but no
covarance matrix was output unless I specify UNCONDITIONAL in the $COV block.
Any idea why?
When I read the user guide, it said that UNCONDITIONAL is needed if we want the
covariance step to be implemented even when model minimization fails. It did
not mention anything about successful minimization but no covariance output. I
also checked NONMEM output file for my model and there was not error message
either, so I am not sure the missing covariance matrix was due to $COV failure
or any other reason.
Following is a snapshot of my model output
iteration 90 OBJ= 6617.1481876524931 eff.= 93. Smpl.=
300. Fit.= 0.88856
#TERM:
OPTIMIZATION WAS COMPLETED
ETABAR IS THE ARITHMETIC MEAN OF THE ETA-ESTIMATES,
AND THE P-VALUE IS GIVEN FOR THE NULL HYPOTHESIS THAT THE TRUE MEAN IS 0.
ETABAR: 4.9637E-03 9.3967E-03 3.7603E-05 -1.9003E-04
SE: 4.9923E-02 9.6445E-02 2.7522E-02 2.6685E-03
N: 56 56 56 56
P VAL.: 9.2080E-01 9.2238E-01 9.9891E-01 9.4323E-01
ETASHRINKSD(%) 2.7783E+00 1.7267E+01 5.3092E+00 6.9977E+01
ETASHRINKVR(%) 5.4794E+00 3.1553E+01 1.0337E+01 9.0986E+01
EBVSHRINKSD(%) 2.7482E+00 1.7184E+01 5.6691E+00 7.0131E+01
EBVSHRINKVR(%) 5.4209E+00 3.1415E+01 1.1017E+01 9.1078E+01
EPSSHRINKSD(%) 1.2374E+01 1.2374E+01
EPSSHRINKVR(%) 2.3217E+01 2.3217E+01
TOTAL DATA POINTS NORMALLY DISTRIBUTED (N): 578
N*LOG(2PI) CONSTANT TO OBJECTIVE FUNCTION: 1062.2929443846017
OBJECTIVE FUNCTION VALUE WITHOUT CONSTANT: 6617.1481876524931
OBJECTIVE FUNCTION VALUE WITH CONSTANT: 7679.4411320370946
REPORTED OBJECTIVE FUNCTION DOES NOT CONTAIN CONSTANT
TOTAL EFFECTIVE ETAS (NIND*NETA): 224
#TERE:
Elapsed estimation time in seconds: 28.83
Number of Negative Eigenvalues in Matrix= 1
Most negative value= -7.5944835791791885
Most positive value= 29811.764851258835
Forcing positive definiteness
Root mean square deviation of matrix from original= 6.6243382481674822E-006
Elapsed covariance time in seconds: 0.28
Thank you in advance for your feedback