Quoting david eddy <[EMAIL PROTECTED]>: > > Grr, that was a mistake. It should be 2*B+log2(N) and 2*(B-2)+log2(N) > > for ordinary and balanced representation, respectively.
Nitpick: my response should have used 2*(B-1)+log2(N) > > 56 bits is nearer the mark! > I assume statistics along the lines I suggested (normal distribution) > explains the 4 bit discrpancy. Or is it the varying integer length of > a weighted transform? Ernst Mayer assumed the output digits of the product had a size dictated by the statistics of a random walk. My understanding is that it's entirely possible to get a digit in the output that's larger than 53 bits, in which case the convolution will fail even with zero roundoff error. jasonp ------------------------------------------------------ This message was sent using BOO.net's Webmail. http://www.boo.net/ _______________________________________________ Prime mailing list [email protected] http://hogranch.com/mailman/listinfo/prime
