[Clp] Infinite loop

[Michael Hennebry]

On Fri, 21 Sep 2012, Victor Miller wrote:

> I have a number of LP problems that I'm solving with Clp.  In all of them I
> have a fixed polytope, and given a point, x, I want to find the closest (in
> L1) point in the polytope.  I do this by adding a bunch of variables y[i]
> to the original variables u[i] , and setting up the inequalities -x[i] <=
> y[i] - u[i], x[i] <= y[i] + u[i].  I set up a new problem once, and then
> for every point x I encounter I just change the lower bounds of the
> appropriate inequalities and then resolve.  For the most part this works
> fine, except in one problem it seems to go into an infinite loop, endlessly
> repeating a line like
>
> Clp0029I End of values pass after 497 iteration
> clp0006I 497 Obj 7.1249e+08 Primal inf 6.20756e+276 (330) Dual inf
> 5.25822e+17 (286) w.o. free dual inf (240)
>
> Is it somehow encountering an ill conditioned problem?  How can I track
> this down?

It seems to me that this problem could not be ill-
conditioned unless the polytope is ill-conditioned.
Even in that case, I'm suprised at the difficulty
in getting a feasible solution.
Perhaps that problem could be side-stepped by
providing an initial basic feasible solution.
Make u a vertex of the polytope.
y becomes the minima allowed by the inequalities involved.

-- 
Michael   hennebry at web.cs.ndsu.NoDak.edu
"On Monday, I'm gonna have to tell my kindergarten class,
whom I teach not to run with scissors,
that my fiance ran me through with a broadsword."  --  Lily