[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Help-glpk] Linear IP solve instability

From: Laurence Davies
Subject: [Help-glpk] Linear IP solve instability
Date: Mon, 7 Oct 2013 12:14:05 +0000

Hi, I have a very simple sparse LP which is a L1 norm minimisation. It solves 
fine using the simplex solver but terminates early in the IP solver with a 
numerical instability error.

I've attached the file that causes this.

Unnecessary history: This isn't the first solver I've tried - I've had success 
with CVX-Matlab and Mathematica solvers using various formulations. I tried to 
write my own solver using Dikin's method without success (seems to never 
converge) and I've written my own simplex solver which works fine.

The KKT conditions report high quality except for KKT.PB which reports medium 
quality for the simplex solver and primal infeasible for the IPM.

I've placed nonneg bounds on the residuals u+v and X is bounded by +/- 1e6 for 
the sake of sanity. The latter seemed to be useful in Mathematica. Could anyone 
please make a suggestion of what to try next?



Attachment: test_glpk_prob.txt
Description: test_glpk_prob.txt

reply via email to

[Prev in Thread] Current Thread [Next in Thread]