NEOS Server Home

NEOS Interfaces to sdplr

WWW Form & Sample Submissions
E-mail
XML-RPC

sdplr





The NEOS Server offers SDPLR (version 1.02) for the solution of semidefinite programming problems in SeDuMi format, in sparse SDPA format, or in an extension to specify low-rank matrices, the SDPLR format.

SDPLR extends and improves on the first order method of Burer&Monteiro. Source code, binaries, and documentation are available from the SDPLR Homepage

SDPLR was developed by Samuel Burer, Renato Monteiro, and Changhui Choi (U Iowa).

This solver was implemented by Hans Mittelmann and executes at under


Using the NEOS Server for SDPLR

The user must submit a model in either SeDuMi (Matlab binary), sparse SDPA, or SDPLR format to solve a semidefinite programming problem. Examples of models in the latter formats can be found here while models in SeDuMi format are here.
Note that when submitting via e-mail or XML-RPC empty tokens need to be deleted

An initial solution to the problem may also be supplied. If non-standard parameter settings are required, the user may also submit this parameter file. While this file was generated by the SDPLR code, a careful editing without format changes is possible.

The code prints the 6 error measures according to the DIMACS 7th Challenge, see the benchmarking paper. This facilitates comparison with other SDP solvers.

Very important: The header of the file should only have the numbers, especially no {..}!


Enter the complete path to the sparse SDPA format file (SEE SAMPLE SUBMISSIONS)
SDPA data:


Alternatively, enter the complete path to the SDPLR format file (SEE SAMPLE SUBMISSIONS)
SDPLR data:


Alternatively, enter the complete path to the SeDuMi format (At,b,c,K) file
SeDuMi data:


Enter the data file containing the initial solution (optional)
Initial Point:


Enter the parameter file (optional).
Parameter File:



Comments:


e-mail address:

Please do not click the 'Submit to NEOS' button more than once.


NEOS Server Home
Submit comments and questions


DOE disclaimer
DOE Web privacy policy