[Help-glpk] Presolve with active good basis

[H . Georg Büsching]

Dear,

 

I start with one LP, let it solve to optimality by glpk. Now I introduce in some rows a new variable with positive costs. The normal basis should still be usable.

Anyhow I want to freeze 98% of all variables to its original values.

 

Idea is now to solve the LP very fast.

 

If I do a normal simplex it still uses the old basis, but doesn't really understand that the LP is already 98% solved (A lot of iterations of the Simplex methods).

If I use the presolve option instead, it really uses the fixed values, but my basis is gone, as the solver starts again from scratch.

 

Do I have to implement something on my own, like deletion of all fixed columns and manipulating the right-hand-sides?

 

Best regards

H. Georg Büsching.