Re: [Help-glpk] numerical instability

[Akhil langer]

Andrew,

Thanks, Andrew. The NaN problem has been resolved. However, I keep getting  the following warning:

Warning: numerical instability (primal simplex, phase I)

Please note that I have been using INT_MAX as the upper bound for some columns. Could that be the cause of the instability? I can change these columns also to be free but that would require some changes in the code. Just wanted to check with you if changing the code will remove the instability?

Thanks,
Akhil

On Thu, Jul 7, 2011 at 10:01 PM, Andrew Makhorin <address@hidden> wrote:

> Please see attached.
>

Thank you.

In your mps file many columns (variables) have incorrect upper bounds:

BOUNDS
 UP BND1 C0000121 1.797693E308
 UP BND1 C0000122 1.797693E308
 UP BND1 C0000123 1.797693E308
 UP BND1 C0000124 1.797693E308
 UP BND1 C0000125 1.797693E308
. . . . .

Please correct your code. Most probably you call glp_set_col_bnds
specifying column type as GLP_UP and upper bound as DBL_MAX. If you need
to make a column free (unbounded), its type should be GLP_FR. Unlike
some lp/mip packages glpk *does not* consider huge positive (or
negative) bound as if it were infinite. Currently glpk api routines do
not perform sanity checks against incorrect data, so using such huge
bounds in computations leads to floating-point overflow, and if this
kind of exception is disabled, to producing NaNs rather to abnormal
termination. (Sometimes IEEE 754 annoys me.)