-
Notifications
You must be signed in to change notification settings - Fork 96
Assertion errors in DistanceFromCut::opera
8000
tor[]
#200
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
Comments
Hey @twesterhout, thanks for reaching out. Could you provide us with the input hypergraph and the configuration you used? @larsgottesbueren this looks like some violation in the hyperflowcutter code. Could you take a look? |
So here's the configuration that kahypar displays (this is just the
I'm using kahypar through the Python interface. Here's the kwargs dict that I'm passing to the
|
Hi, I fixed the out-of-bounds operator[] (PR is linked). Can you try running it again? As the memory backing the out-of-bounds access was actually there, just the size was wrong, I suspect that the second assertion failure can still pop up: In my test, it doesn't, but I also had to test through the binary :) |
Hello @larsgottesbueren, thanks for the quick fix! I tried your PR, and like you suspected, the original assertion failure is replaced with various failures of the form:
I'll try to create a complete reproducer for you. I'm using Cotengra to optimize some tensor network contractions, and it's calling KaHyPar many times with different hyperparameters, so it's not always trivial to trace back the exact parameters used. |
@larsgottesbueren could you give this a try: https://gist.github.com/twesterhout/cab4dda947aa7aad21968ac184e44379 ? Does it fail on your side too? |
Any updates on this? I'd be happy to do testing of any work-in-progress PRs. |
Sorry for the delay. Please give #201 another try :) |
Hm... still failing, but with yet another error:
Do you happen to have an idea what the error is related to? |
Can you please share the instance where this assertion fails. |
Sure thing @larsgottesbueren! I've updated the gist: https://gist.github.com/twesterhout/cab4dda947aa7aad21968ac184e44379#file-other_assert-py ( |
This is in the initial partitioning code. After some quick debugging it seems that the max part weights aren't set up correctly. @kittobi1992 Can you take a look at this one? You'll know this code best. |
@kittobi1992 / @TobiasHeuer Do you have an idea what could go wrong here? |
@kittobi1992 @TobiasHeuer @SebastianSchlag do you happen to have any new insights into the failures? |
@larsgottesbueren Since we don't see these failures in the current master branch, could it be that something is off with #201? |
For me, these failures appear on master. |
Hello,
I'm getting the following assertion failures in v1.3.5:
Do you have any idea what could be causing it?
The text was updated successfully, but these errors were encountered: