Is the bias parameter in LSH actually necessary?

[alexklibisz]

I can't think of a good motivation for the LSH bias parameter, the b in the formula below:

I'll review the paper Locality sensitive hashing scheme based on p-Stable distributions, Datur, et. al., and I'll post my notes in this discussion.

This seems like a relevant part of the paper:

[alexklibisz]

This article also seems useful: http://tylerneylon.com/a/lsh1/

[alexklibisz]

~2.5 years later.. I'm still not really sure why this parameter is really necessary. Out of curiosity, I tried removing it from the L2LshModel and found that all the tests still pass. I didn't benchmark it. So I'm not sure of the effect on performance.

[alexklibisz]

I ran a benchmark in #666 and noticed that removing the b parameter causes the latency to increase substantially. More here: #666 (comment)

Still it would be interesting to know why.