[Coin-lpsolver] Using Clp under Neos

Math407 Summer05 m407s05 at u.washington.edu
Wed Jun 8 14:01:35 EDT 2005


Introduction:
=============

The support people at Neos have suggest that I contact the Clp help 
directly for my questions below. I am having some trouble understanding 
the input and output to the Neos Clp solver for a linear programming case.
On the web page
   	http://www-neos.mcs.anl.gov/neos/solvers/LP:CLP-MPS/
it states "To solve a linear programming problem using MPS as input, you 
need only submit the MPS file". I was able to find something close to a 
specification of this format in the file
   	http://www.sbsi-sol-optimize.com/manuals/Minos%20Manual.pdf

Question I:
==========
Where is the authoritative specification for MPS format ?

Still the Clp examples seem to have other commands in the submitted job. For
example,
   	BEGIN.PAR
   	primalS
   	sol -
   	END.PAR

   	BEGIN.COMMENT
   	Solve using primal simplex, and print solution value
   	END.COMMENT

It may be that these commands are listed on the web page
   	http://www-neos.mcs.anl.gov/neos/solvers/LP:CLP-MPS/
but there is no documentation there (or link to documentation) describing what
the commands do.

Question II:
============
Where are the specifications for these other commands (output the MPS input
section) and exactly what they do ?

For example, I am not sure how to get Clp to choose which of the rows 
corresponds to the objective . For another example, I am not sure if Clp 
is attempting to maximize or minimize the objective. (These do not seem to 
be part of the MPS input format but perhaps I am wrong in this.)

Question III:
============
Where are the specifications for the output that Neos e-mails back to and
what it means ?

For example, what does the following mean:
   	Optimal objective 0 - 0 iterations time 0.002, Presolve 0.00
   	Clp:      0 R1                   20                      0
   	1 R2                    0                      0
   	2 R3                   20                      0
   	0 X1                    0                      1
   	1 X2                   20                      0
   	Clp:



More information about the Clp mailing list