# Re: var-cov matrix issue?

From: Ethan Wu <ethan.wu75>
Date: Thu, 26 Feb 2009 04:05:23 -0800 (PST)

Dear Kyun, thanks for your help. I don't know if I understand this one=
"Some caution is necessary to simulate omega matrix that is alwasy posi=
tive definite. " Could you explain a bit more? __________=
______________________ From: "kyunseop.bae
..com> To: Ethan Wu <ethan.wu75
Sent: Tuesday, February 24, 2009 3:07:30 PM Subject: RE: [NMusers] var-co=
v matrix issue? Hi, Ethan, I think your question can be reduced w=
hether pseudo-inverse matrix can be used instead of inverse matrix. I =
do not know quite different cases, but I suppose it can be used. To be=
ppropriate, if you use VAR-COV matrix output for simulation under normal =
distribution assumtion, If your data supports normal distribution as=
sumption, MATRIX=R option will not give much difference in SEs. Defau=
lt VAR-COV output in NONMEM is a kind of sandwich estimate, which is thou=
ght to be more robust (a little larger) than inverse Fisher's information=
matrix (given MATRIX=R option). Some caution is necessary to simula=
te omega matrix that is alwasy positive definite. This may help you=
.. Thanks, Kyun Seop Bae MD PhD Email: kyunseop.bae

obomaxnm.com [mailto:owner-nmusers

com Cc: nmusers
sue? Hi Justin, only ETA was estimated with high SE but, again, I=
guess it came back to the question: how trustful it is if such error mes=
sage appears ________________________________ From: "justin.wi=
lkins
m Sent: Tuesday, February 24, 2009 1:19:17 PM Subject: Fw: [NMusers] va=
r-cov matrix issue? Dear Ethan, Algorithmically singular matr=
ices are often a sign that that your model is ill-conditioned in some way=
; I would be careful in how I used the variance-covariance matrix in this=
scenario, and especially for simulation. Are there any parameters that a=
re being estimated with particularly high standard errors? This might sug=
gest overparamaterization. Not sure how helpful this is! Best =

rmacology) CHBS, WSJ-027.6.076 Novartis Pharma AG Lichtstrasse 35 C=
H-4056 Basel Switzerland Phone: +41 61 324 6549 Fax: +41 61 324 3039=

in.wilkins
vartis on 2009/02/24 07:15 PM ----- Ethan Wu <ethan.wu75

To     nmusers
ject     [NMusers] var-cov matrix issue?        =
Dear all, I recently encounter this error messa=
ge (below). My objective was to use nonmem var-cov output  for approxim=
ation of distribution of parameters for performing a simulation. if su=
ch error message occur, is the var-cov matrix  still OK to use? -- I k=
now that better way to figure out distribution of parameters is to do boo=
tstrap, but given limited time I have.....   thanks   "0MIN=
IMIZATION SUCCESSFUL NO. OF FUNCTION EVALUATIONS USED:  331 NO. OF SI=
G. DIGITS IN FINAL EST.:  3.3 ETABAR IS THE ARITHMETIC MEAN OF THE ETA=
-ESTIMATES, AND THE P-VALUE IS GIVEN FOR THE NULL HYPOTHESIS THAT THE TRU=
E MEAN IS 0. ETABAR:  0.11E-02 SE:      0.23E-01 P VAL.: =
0.96E+00 0S MATRIX ALGORITHMICALLY SINGULAR 0S MATRIX IS OUTPUT 0INV=
ERSE COVARIANCE MATRIX SET TO RS*R, WHERE S* IS A PSEUDO INVERSE OF S 1=
Received on Thu Feb 26 2009 - 07:05:23 EST

The NONMEM Users Network is maintained by ICON plc. Requests to subscribe to the network should be sent to: nmusers-request@iconplc.com.

Once subscribed, you may contribute to the discussion by emailing: nmusers@globomaxnm.com.