GUSEK/GLPK - Run a IP Faster

Asked Nov 25 '18 at 14:43

Active Nov 25 '18 at 14:55

Viewed 220 times

I am running an IP problem with around 120 decision variables. I would expect this to take some time but it's taking >8 hours. I am using GUSEK and wondering if there is anyway to use more CPUs or run in parallel to speed on the solution. Or any other methods to make this run faster.

I am assigning blocks of time to different departments to different offices for each day of the week. There are constraints on how room types, and all types of rooms each department gets per day, and per week. Each department has a total time per week goal, and I'm trying to minimize the difference between the assigned time and the goal time.

I am curious is GUSEK has anyway to assign more cores to it for processing?

edited Nov 25 '18 at 14:55

asked Nov 25 '18 at 14:43

Yolo_chicken

1,221
2
12
24

As the general problem is NP-hard, there is no way of addressing this without incorporating *details*. – sascha Nov 25 '18 at 14:45
I will edit the question to outline the specific problem – Yolo_chicken Nov 25 '18 at 14:48
GLPK is single threaded. You could try a few state-of-the-art MIP solvers at [NEOS](https://neos-server.org/neos/). AMPL syntax is very close to what GLPK is using. – Erwin Kalvelagen Nov 26 '18 at 17:54

GUSEK/GLPK - Run a IP Faster

0 Answers0