optimaltree
hangs on edge case with all costs equal to 1
#207
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fix for #206
@Jutho correct me if I'm wrong, but I think what is going on here is that the initial guess for the
maxcost
is wrong for these cases:From what I can tell we use:
Which evaluates to
1
in this case, while the actual maximal cost would be(n-1) * 1
withn
the number of tensors.I patched the specific case of all sizes equal to 1, but I am not sure if this now captures all possible problems. (I'm also not entirely sure where the initial formula came from?)
I could also replace this
maxcost
with the cost of simply multiplying in the order that is supplied, but I'm not sure what the implications are for the algorithm.