Tracing ms

["J. Roy Robertson" <rb0ujr@obiwan.rohmhaas.com>]

Simple example from Guide to Statistics using gradient and trace with S+6 on
Windows 2000.
No Trace?

> lprob2 <- deriv( ~log(1+ exp(D*alpha)) - D*alpha, c("alpha"),
function(D,alpha)  NULL)
> ms( ~lprob2(D,alpha),data=pingpong,start=list(alpha=.008),trace=T)
value: 1127.635
parameters:
      alpha
 0.01114251
formula:  ~  lprob2(D, alpha)
3017 observations
call: ms(formula =  ~ lprob2(D, alpha), data = pingpong, start = list(alpha
= 0.008), trac
e = T  )

Add Hessian with deriv3 from MASS and get trace. Is this normal behavior?
I used this simple example because I am getting same behavior with more
complicated likelihood functions.

> lprob2 <- deriv3( ~log(1+ exp(D*alpha)) - D*alpha, c("alpha"),
function(D,alpha) NULL)
> ms( ~lprob2(D,alpha),data=pingpong,start=list(alpha=.008),trace=T)
Iteration:  0 ,  1  function calls, F=  1161.12
Parameters:
[1] 0.008
Iteration:  1 ,  2  function calls, F=  1129.615
Parameters:
etc.

Thanks
Roy Robertson
Rohm and Haas, Co.

"Opinions expressed are mine and not those of Rohm and Haas Company".