This is a program by Masashi Kiyomi
to solve linear programming problems exactly using GNU MP.
Very slow ;-p

From the version 0.6.0, the licence is changed to LGPL 2.1 or later.

0.5.3 is mostly same to 0.5.2, but now the dual simplex routine has steepest
edge pricing routine. And some bugs of 0.5.2 are fixed.
0.5.2 is mostly same to 0.5.1, but --dual option now works correctly(,I hope).
I have not implemented steepest edge routine for dual version, so it's very slow.
exlp-0.6.0 2006/1/26
exlp-0.5.6-2 2005/10/17bug fix
exlp-0.5.6-1 2005/5/17very small fix
exlp-0.5.6 2004/5/7bug fix release
exlp-0.5.5 2004/4/29
exlp-0.5.4 2003/8/28
exlp-0.5.3 2003/8/1
exlp-0.5.2 2003/7/31
exlp-0.5.1 2003/7/21
exlp-0.5.0 2003/7/1
exlp-0.4.0 2003/4/13

benchmark(netlib) with 0.5.6
benchmark(netlib, --dual option) with 0.5.6


If you are looking for free practical LP/IP solver, I recommend you glpk.
It is very fast, and it returns right answers for most problems.

lp_solve is also a very famous solver.
From the version 6.0, Kjell Eikland, the maintener of lp_solve, is planning
to make lp_solve able to call exlp as an external routine.

Glpk or lp_solve is a good tool. However it uses floating point arithmetic,
thus it cannot do exact calculation.
For example this problem is actually infeasible.
But, glpk (ver. 4.8 on x86 Linux) finds an 'optimal' solution.
Or, there exists more simple example.
This problem is very small and does not contain very large values.
The optimal value is clearly 10000, however glpk says the problem
has unbounded solution.