Re: [Help-glpk] LP-of a MILP solution with better properties

[Andrew Makhorin]

> often a Mixel Linear Integer Problem has as a LP a high dimensional
> optimal solution space. Furthermore I read somewhere that in general 
> for finding additionally cuts it is better that the chosen optimpal 
> solution of the LP is as much integer as it could be.
>
> But I never read that after having solved the first LP an algoritm like 
> the "Feasibility Pump" by Lodi et alii with an additive objective 
> restriction is used. 
> Is it really not implemented in glpk? Are there plans?

That algorithm is really not implemented in glpk.

> I think that an
> easy version of that algorithm can be made within 50 lines of coding. I 
> may implement it in the next 3 months.

If you have a wish, please do that. This would be very interesting.
For which classes of mip's the algorithm is expected to be mostly
efficient?

Andrew Makhorin