Depends on you problem.
But you said you had a non-convex NLP…
So i’d need more info to say anything, because non-convex NLP is almost as general as it gets.
Your problem might just be very hard anyway, and the you need som special algorithm, decomposition, relaxation or approximation.
1 Like
If I want to change the tolerance what function do I use with this?
I check this https://github.com/jump-dev/AmplNLWriter.jl
but the link to options was broken
Here’s the link: https://github.com/coin-or/Couenne/blob/master/src/couenne.opt
1 Like
Thank you @odow
1 Like
Hi, @mayar, where did you get the couenne.exe? I downloaded from Open Source Solvers - AMPLAMPL, the 64-bit Windows version. However, when I run couenne.exe like you, nothing happened and no information was displayed.
2 Likes
Figured the reason out. One or more DLLs are missing for couenne on Windows 10. However, if one only runs couenne in the command line or with JuMP, no error will be thrown. But if you double click the “couenne.exe” file directly, an error message will pop out. To fix the missing DLLs, install the visual studio distributable here.
2 Likes
any idea how to use couenne solver options in JuMP. branching options are not working.
Did you try using an opt file?