Re: [Help-glpk] Question about the behavior of lpx_exact

[Marc Lanctot]

Andrew Makhorin wrote:

> Using the exact solver to solve lp from the very beginning is not
> a good idea. It is better to call it *after* the floating-point solver
> has found optimal solution.

Ah, thanks.  :)

> However, to check if the solution obtained by glp_simplex is accurate
> enough you can use the api routine lpx_check_kkt.

Great! I didn't know of this method, and its existing implementation in 
GLPK.  I still have one more question: the GLPK guide mentions that 
high-quality solutions found when using an internally scaled problem 
means that it is quite accurate, but doesn't say what it means for 
unscaled problems. Should I just always scale my LPs? And if so, is the 
best way to scale every row and column by the same value?

Thanks,
Marc

--
"From down here we can make the whole wall collapse!"
"Uh, yeah, boss, but how do we get out?"
  -- Goblin Digging Team (MTG)