Date:         Thu, 27 Jun 2002 15:12:01 -0400
Reply-To:     jimmy chan <jimmy.chan@HEC.CA>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         jimmy chan <jimmy.chan@HEC.CA>
Subject:      proc nlmixed...
Comments: To: stringplayer_2@yahoo.com
Content-Type: text/plain; charset=us-ascii

Sorry Mr. McLerran for another question...

I want to that if I use proc nlmixed for maximizing a likelihood
function with model L ~ general (l) at end of the program, where l is
the function, do I need to apply log to the function ? Because I see
that I have transformed the initail likelihood function to a log
likelihood function and it doesn`t seem correct.

proc nlmixed data=jimday.abce1;
parms alpha1=0, alpha2=0, beta=0, sigma=1;
l = dum1*log(1/sqrt(2*3.1415*sigma**2))-1/(2*sigma**2)*(return*dum1+
alpha1) - beta*dum1*tse)**2 (3)

+ dum2*log(1/sqrt(2*3.1415*sigma**2))-1/(2*sigma**2)*(return*dum2+
alpha2) - beta*dum2*tse)**2 (1)

+ (dum3*log(1/sqrt(2*3.1415*sigma**2)))-1/(2*sigma**2)*(alpha1-
beta*dum3*tse)**2 - dum3*log(1/sqrt(2*3.1415*sigma**2)))-
1/2*sigma**2)*(alpha2 - ta*dum3*tse)**2); (2)

model L ~ general (l);

by ticker year; run;

Thank you.

Jimmy