Alexander van der Grinten
Alexander van der Grinten
Why did you close this issue? Did you find the reason for these `NaN`s?
Alright, thanks for letting us know. There should probably be an error raised when trying to access the pairwise commute distance if the algorithm did not `run()` yet.
Hi Michael, thanks for your work. What I do not completely understand from the documentation is how replace objects are pushed/pulled. According to [some stackoverflow question](https://stackoverflow.com/questions/44028493/how-to-push-history-edited-with-git-replace-to-remote), replace objects are not...
I agree that the first variant is our best option for a first step. The advantage of the last variant would be that existing commit identifiers remain correct. In other...
I look at your fork. It works fine. Now we need to decide on which option to implement. Option 1 has no downsides, so I will implement it as a...
No option was implemented yet.
But an easy way to move forward is to fetch Michael's history into a `history` branch without doing any further work for now.
What about passing a (mutable) reference to a nk::Vector to the RHS producer function? That way we could avoid memory allocations. This would be a good addition, if we can...
Thank you for the PR. I assume that you are still debugging the CI failures? Let us know when it is ready to be reviewed.
This PR is huge. Would it be possible to break it into smaller parts? I am also not convinced that all auxiliary data structures are needed. It makes sense to...