Re: [Help-glpk] Getting exact solution and starting from infeasible basis

[Andrew Makhorin]

> I started using glpk a couple of weeks ago and there are some
> questions whose answers I couldn't find in the reference manual. In
> case I overlooked it, please feel free to point me to the relevant
> sections avoiding long answers.
> 
> - Is there any way to get the exact solution computed by glp_exact?
> For big numbers, getting the double 1.2345 doesn't necessarily mean
> that the exact solution was 12345/10000...

Unfortunately, you cannot obtain the exact solution in rationals on api
level. The only way could be changing the api routine glp_exact (see
file src/glpapi07.c) for your needs. 

> 
> - In case the initial basis provided with the basis constructing
> methods corresponds to an unfeasible solution, is this basis used in
> Phase I of the simplex method as a the starting point?
> 

Yes, glp_simplex always starts from the initial basis specified in the
problem object. Feasibility is not required, however, the initial basis
should be correct in the following sense: the total number of basic
auxiliary and structural variables is equal to the number of rows, and
the basis matrix composed of the columns of basic variables (including
unity columns of auxiliary variables) is non-singular.