Skip to content
This repository was archived by the owner on Jan 11, 2023. It is now read-only.
This repository was archived by the owner on Jan 11, 2023. It is now read-only.

"ANGULAR" distance defined differently than in main annoy package #22

Description

@jrhaberstroh

This Java implementation defines ANGULAR distance as "cosine similarity", while the C++/python implementation defines ANGULAR distance as "approximate theta".

Evidence:

Java implementation -- dot(u,v), the value of cosine

publicstaticfloatcosineMargin(finalfloat[] u, finalfloat[] v) {

floatmargin = (INDEX_TYPE == IndexType.ANGULAR) ? cosineMargin(v, queryVector)

C++ / python implementation -- sqrt(2 * (1 - dot(u,v))), an approximation of theta
https://github.com/spotify/annoy/blob/6f6b0c84ab413337eb4d2e850a4cba637f52ccbc/src/annoylib.h#L473
https://github.com/spotify/annoy/blob/6f6b0c84ab413337eb4d2e850a4cba637f52ccbc/src/annoylib.h#L502

Related issue defining "angular" method in the other repo: spotify/annoy#530

Because this would be a significant breaking change, I did not put a PR together. Would the lead devs provide guidance on the appropriate way to name a new distance that implements angular correctly ("angular-cpp" or "angular-correct"?).

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions