parameter in HIM distance

[jkbren]

we should understand how the default value of 1.0 affects the distance

[sdmccabe]

The paper does not give much guidance on what a default value of ζ should be. In its absence I think using 1.0 is fine. Maybe we could add a little more documentation, i.e., explicitly say if ζ=0 it's Hamming and as ζ approaches infinity it becomes IM, but I'm not sure what we can do past that.

[leotrs]

I don't think there's much more to say other than the fact that HIM is the square root (why?) of the weighted average of H² and IM², and ξ is the ratio of the weights. ξ=1 corresponds to equal weights.

Does anyone know (1) why they take the square root, or (2) why @sdmccabe wrote ζ instead of ξ?

[sdmccabe]

For (2), it’s because I can’t read.

…

On May 13, 2019, at 2:16 PM, Leo Torres ***@***.***> wrote: I don't think there's much more to say other than the fact that HIM is the square root (why?) of the weighted average of H² and IM², and ξ is the ratio of the weights. ξ=1 corresponds to equal weights. Does anyone know (1) why they take the square root, or (2) why @sdmccabe wrote ζ instead of ξ? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[hartle]

My guess is that the square root is to make it (proportional to) the Euclidean length of a vector with components H and (√ξ)IM. Another way to read it that looks a bit better is (1+ξ)(HIM)²=H²+ξ(IM)².