[Help-glpk] Re: glpk mip

[Andrew Makhorin]

>This is surprising.  The presolve routines I'm aware of are just as
> valid for
>integer variables as for continuous variables.  Basically, they
> progressively
>tighten bounds on variables, and when a variable's two bounds become
> equal, the
>variable is fixed and dropped from the problem.

You are absolutely right. Nevertheless, although from the user's
standpoint lp and mip presolvers do similar job, they differ in the goal
to be achieved (i.e. simplifying lp is not the same as simplifying mip).
That's why it was decided to implement lp and mip presolvers as separate
components. However, mip presolver is not implemented yet (I plan to use
it as a preprocessor within a new mip solver which itself is still under
development).

Andrew Makhorin