doc/examples.html and lsh.py seem to have outdated information.

[yvesx]

line 1056 in lsh.py suggests to use N=100000 data points.
but the create procedure python2.6 lsh.py -d 5 -create needs to find out the nearest neighbors for N points using brutal force i.e. quadratic complexity. For N=100000, it would never finish. I tried just N=1000 and the create procedure finishes within 1 minute.

Another thing on doc/examples.html is also, I believe, inaccurate.

D=5;
N=100000;

clear results
for r=0:2
results(r+1) = CalculateMPLSHParameters(D, N, ... dnnHist, dnnBins, danyHist, danyBins, deltaTarget, r, uHash, uCheck);
end

But the matlab function CalculateMPLSHParameters isn't in the repo (did I miss something here?).
I changed that line to this and worked for me.
results(r+1) = CalculateLSHParameters( N, dnnHist, dnnBins, danyHist, danyBins);