Closed rschwarz closed 2 years ago
@rschwarz With version 1.13.0 I could not reproduce your behavior. Could you please confirm?
I tried again with mip-1.13
and observed conflicting behavior: When I ran the script the first time, it crashed again. When I repeated the call to capture the log output, it did not crash. I had to try 4 more times to see the crash, so it seems that the behavior is not deterministic?! (See OK and crash for the two cases.)
What I found curious is that at the top of the log, it still says:
Python: 3.9.2 (default, Feb 28 2021, 17:03:44)
[GCC 10.2.1 20210110]
mip: 1.13.0
[...]
Welcome to the CBC MILP Solver
Version: devel
Build Date: Nov 15 2020
So the Cbc build date is the same between mip
versions 1.12 and 1.13? That doesn't seem to be right.
I tried this again using version 1.14
(with a lot of runs) and was not able to reproduce the crash, so I'm closing as fixed (probably because of updated Cbc).
I create a MIP model, including a SOS2 set and provide a MIP start on a subset of variables. During solve, there is an error in Cbc.
With the Cbc that is shipped with python-mip, I get:
If I use Cbc built with debug mode instead, I get:
I've created a gist with all files needed to recreate the issue as well as recorded outputs. In particular, see run.py. That gist also has some comments on what I think went wrong.
I thought that the issue lies within Cbc, not the python-mip wrapper, but I was unable to recreate the issue using Cbc from the command-line (reading problem and MIP start from files).
Versions: