Support for CPLEX

[nathanncohen]

This patch enables support for Cplex, when the new package #8171 for CBC is installed. It also includes several lines of documentation to explain how CPLEX can be used, and fixes a bad docstring which I noticed while testing this package :-)

Nathann does have access to 't2'. Minh has made it clear the last alpha (4.3.4.alpha1) builds ok on 't2', so testing this should be possible for Nathann.

CC: @malb @haraldschilly

Component: numerical

Author: Nathann Cohen

Reviewer: Dmitrii Pasechnik, Martin Albrecht

Merged: sage-4.4.1.alpha2

Issue created by migration from https://trac.sagemath.org/ticket/8172

[malb]

comment:2

You are doctesting solve_coin in solve_cplex

[nathanncohen]

comment:3

OOps :-)

The code from mip_cplex is a pure copy of mip_coin ( I only needed to change the types from OsiCbcSolverInterface to OsiCplexSolverInterface, etc .... )

This code has already been reviewed, by the way.....

Fixed ! Thank you :-)

Nathann

[malb]

comment:4

I wonder if we can factor out the differences and this way would make it very very easy to hook up new solvers (with say, some function pointers and such) The redundancy of copying a function does not seem very elegant to me.

[nathanncohen]

comment:5

I do not like it either, but this way it answers two other difficulties :

• Coin me be compiled independently from the presence of CPLEX. So I can not be assume in mip_coin that CPLEX libraries are available and that this file can actually be compiled.

• Testing for the presence of a CPLEX solver amounts to try to import solve_cplex, which is nice and can only be done if the file has been compiled successfully

Nathann

[malb]

comment:6

We could still reduce the amount of code duplication. Say we replace the functions solve_* by instances of a class OSISolver which would get function pointers pointing to the correct function to call for each solver and have a call method which implements the current solve_* function. So for each new solver we'd only need to add a new instance. All the points you raise above would be addressed.

[nathanncohen]

comment:7

Hmm.. It is almost done, but I have a few problems with cimporting in mip_cplex the c_solve function defined in mip_coin.... :-)

[nathanncohen]

comment:8

Ok, I have spent 3 or 4 hours dealing with this bug, if you want to give it a try it is almost finished : we but have to find a way to import this *********** function...

I'm sorry for my rudeness, but I can not stand these stupid error messages anymore !

Nathann

[nathanncohen]

comment:9

If you can fix it, I will obviously finish to clean the patch....

Nathann

[sagetrac-drkirkby]

comment:10

What platforms has this been tested on? It needs to work on OS X, Linux and Solaris. It's not clear from reading this that anyone has actually checked it works on Solaris.

There's general information about building on Solaris at

http://wiki.sagemath.org/solaris

Information specifically for 't2' at

http://wiki.sagemath.org/devel/Building-Sage-on-the-T5240-t2

Both the source (4.3.0.1 is the latest to build on Solaris) and a binary
which will run on any SPARC can be found at http://www.sagemath.org
/download-source.html

Dave