-
Notifications
You must be signed in to change notification settings - Fork 9
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Sporadic malloc errors with CSDP 4.1 #39
Comments
Thanks for the detailed report. I have also noticed sporadic failures with CSDP but haven't found the root cause yet. See for instance |
I am not sure if it is the same bug or not, but I also got a segfault using CSDP partway through an optimization with Pajarito, where CSDP is the continuous solver (using Gurobi as the MIP solver). It ran for a few minutes and I could see CSDP had solved many problems without incident during the course of the optimization, and then on one of the problems it immediately segfaulted. So it seems like a sporadic problem also. The same overall optimization problem worked fine when I substituted CSDP for Mosek. I'll include the stacktrace below in case it's helpful:
|
I should preface this by saying that I'm a novice when it comes to optimization in Julia and JuMP.
I ran into some problems when using CSDP to solve a set of LMIs related H-infinity synthesis in robust control. To illustrate the problem, I have created a short script (see CSDP_issue_malloc.txt) which solves the same problem over and over again in a for-loop. It generates the correct solution repeatedly, until it eventually fails due to one of two reasons: a malloc issue (see printout 1), or not being able to find a feasible solution (see printout 2).
Typically these failures occur after 50-100 correctly solved problems. I have solved the same problem using CVX, as well as the and the interior-point methods by Nesterov used in Matlab's LMI solvers, in both cases yielding a gain of gamma=1.45. So the CSDP solver seems to work nicely, apart from the two sporadically occurring errors. I have run the same experiments with using SCS in Julia, and this works wihthout malloc errors, so the problem seems to be related to CSDP.jl and not JuMP.jl.
The first error might be related to your issue #2 , with the malloc leading to a segfault in my case, but I know too little about the internals of your CSDP-wrapper to accurately debug it. I thought I'd alert you to this issue, and would be thankful for any ideas on what may be causing the problems.
printout 1
printout 2
The text was updated successfully, but these errors were encountered: