| Title: | Mathematics at DEC |
| Moderator: | RUSURE::EDP |
| Created: | Mon Feb 03 1986 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 2083 |
| Total number of notes: | 14613 |
Any optimization gurus out there? I have been working on a
'real-life' problem which involves getting maximum likelihood
estimates for an n-paramater model. At present, I am working
with n on the order of 3 to 5. I have been going off on a
tangent and have developed what I hope is a new and better
algorithm for getting to the top of complex likelihood
functions, independent of starting point. In order to complete
the algorithm, I need to find a way to solve a system of
n first order differential equations. Does anybody know of
any packages available for solving diffrential equations?
thanks in advance - Dave Clark
p.s. if you want to discuss this (very interesting) problem
with me, send me mail or call me at 276-9164.
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 800.1 | MATRIX::ROTH | May you live in interesting times | Thu Dec 10 1987 18:59 | 14 | |
Are you trying to use some sort of homotopy continuation technique?
Or steepest descent of some kind?
There is a large literature on optimization methods for nonlinear
problems, with and without constraints. I've used some of them for
circuit optimization, and could see them being applicable to your type
of problem too.
If your equations are not 'stiff' just about anything would work well
for starters; I'd think most of the work is setting up the function
calls. A simple RK4(5) Runge Kutta variable step size method would
be easy and fast enough to experiment with.
- Jim
| |||||