Implicit filtering is a steepest descent
algorithm for noisy optimization
problems with bound constraints. The method is intended for
problems with optimization landscapes that look like
the image above ( i.e. with many local minima).
The details are in the new
Implicit Filtering Book , SIAM 2011.
Check out the codes for noisy problems in
The group working on implicit filtering and related
algorithms (eg Nelder-Mead, DIRECT, Hooke-Jeeves) and their applications is
and a host of alumni. You may see a help wanted ad here soon.
- Paul Gilmore (firstname.lastname@example.org)
- Tony Choi (email@example.com)
- Owen Eslinger (Owen.J.Eslinger@erdc.usace.army.mil)
- Joerg Gablonsky (firstname.lastname@example.org)
- Alton Patrick (email@example.com)
- Dan Finkel (firstname.lastname@example.org)
- Chris Kees (Christopher.E.Kees@erdc.usace.army.mil)
- Vincent Bannister (Vincent.Bannister@microsoft.com)
- Jill Reese (email@example.com)
- Karen Dillard (firstname.lastname@example.org)
We are working with
Cass T. Miller and Greg Characklis from UNC, Katie Fowler from Clarkson,
and Karen Dillard from the Air Force Institute of Technology on
problems in water resource management.
We've produced two FORTRAN codes, but no longer support them. Please
use the new matlab code for your implicit
- IFFCO V2 (Implicit Filtering for Constrained Optimization)
Choi, Eslinger, Gilmore, Kelley, Gablonsky, Patrick
Code: a nice tarball for y'all
IFFCO.tar.gz contains serial, PVM, and MPI versions of the code.
DOCUMENTATION BEFORE MESSING WITH THE PARALLEL CODE!
- DIRECT (Gablonsky, v2.0.4)
DIRECTv204.tar.gz FORTRAN code for DIRECT with documentation
Last modified: July 17, 2005.