Re: [Bug-glpk] [Fwd: simplex with exact arithmetic]

[Andrew Makhorin]

Sorry for a delay in the response.

> I am writing my master's thesis and therefore I am using the GLPK with  
> exact arithmetic for large problems with 995 rows and 558  
> columns/variables. My problem has 372 variables which have to be  
> between 0 or 1. I use the GLPK during my implemented Branch and Bound  
> algorithm to solve the LP-relaxed subproblems. At some point I fixed  
> about 30 variables to 0 or 1 and then the solver loops infinitely on  
> that data set.
> 
> Do you know something about such an bug?
> 

It may happen that the exact solver is stalling due to primal
degeneracy. Could you provide more information? Is there any output from
the solver? You may try first to solve lp with glp_simplex and then call
glp_exact using the current basis found by glp_simplex.