Leland McInnes
Leland McInnes
I believe it should be possible, but I think the data structures are likely currently backwards for what you want to do -- instead of each datapoint knowing which leaf...
This sounds like great work -- I never imagined this would be easy at all, and it sounds like you have made a great deal of progress. I don't think...
That is the big catch. There are very few places where distances are computed in large batches. Almost everything is in relatively small batches. In terms of the main loop...
Yes, but it is probably possible to make a decent estimate of an upperbound, so as long as memory is relatively cheap you can overallocate and leave a sentinel value...
The number on a single call is often surprisingly small; anywhere from 400 to 3600 total distance computations (i.e. between 20 and 60 points all to all). That may be...
I have fixed the aspect that @jamestwebber pointed out -- we shouldn't alter the users data, that's definitely a bug. With regard to "dot" as a metric -- it is...
@troilus-canva : That makes sense I guess; could you suggest a naming convention you would like>
I agree, that's a bug. I think there are some somewhat sensible ways I can fix that, but I'm not sure when I can get to it.
I just pushed some fixes for this to the master branch. If you have the time to check it out and see if it resolves your problem I would greatly...
Ah yes, I think I see what the problem there would. I think I can fix that; hopefully I'll get to it in the next few days.