Date:         Fri, 1 May 2009 13:58:00 -0700
Reply-To:     Bhuvan <bhuvan.pathak@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Bhuvan <bhuvan.pathak@GMAIL.COM>
Organization: http://groups.google.com
Subject:      Convergence criteria met but final hessian is not positive
              definite.
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset=ISO-8859-1

Hi,

I am Bhuvan Pathak. I am working on effect of Ca in peanut seed
development. I was  working on the Ca analysis data with following
data. However, in the output and log it gives

"
NOTE: Convergence criteria met but final hessian is not positive
definite.
NOTE: Estimated G matrix is not positive definite."


proc mixed data=Ca76dap_stage2pod08;
   class rep subP whP;
   model ca = whP subP whP*subP;
   random rep rep*whP;
   lsmeans whP/pdiff;
   lsmeans subP/pdiff;
   lsmeans whP*subP/pdiff;
   run;
Hence in the output statement, it tries to normalize the variance
statement on whole plot and does not give any value for it.....I do
not know if I need to switch to proc GLM or is there anyway to remove
the effect.

Pl. guide