12/14/2007, 07:20 AM

So let me get this straight. On the first graph you are plotting arg-vs-log(err) and on the second graph you are plotting log(size)-vs-log(err), is that right? I think that would mean the direct quantities would have a power relationship, if then which means you may have found an upper bound for the error.

Wow!

Andrew Robbins

Wow!

Andrew Robbins