LBFGS for use in large maximum likelihood problem

Robert Dodier wrote:
> On 8/19/08, dlakelan <dlakelan at street-artists.org> wrote:
> 
>>  It seems to me that it takes longer to do each step of the LBFGS when I use
>> the "sum" noun than when I construct the huge expression, although the sum
>> noun is of course much faster to construct.
> 
> I'm not seeing that behavior. I've attached a script which solves an easy
> maximum likelihood problem (computing mean and standard deviation)
> via LBFGS. The log likelihood function is either a symbolic sum or a literal
> sum (i.e. an expression which has operator = "+"). For the symbolic sum,
> the execution time grows a little bit faster than linear. For the literal sum,
> the execution time grows somewhat faster than linear, and except for
> small numbers of data, the symbolic sum is faster than the literal sum.

Hi Robert,

I think what was happening is that in my main problem I was triggering 
complex arithmetic because of some steps where I used logexpand=super 
and wound up with logarithms of negative numbers... In theory they 
should all cancel out in the end, but with varying branch cuts or some 
other problem, the complex portion wasn't canceling out and so all the 
arithmetic was more expensive... I detected it yesterday evening and 
seem to have fixed it this morning...

Thanks for all the help on this problem.

Dan