On Monday 07 February 2005 02:22, I wrote: > I imagine (but did not prove) that the maximum error is quite similar > over all the intervals between .5 and 1, because the curvature of 1/x > doesn't change much in that interval...
This turned out to be wrong. The result for the left end of the interval is: >>> ((1/.501 + 2)/2 - 1/.5005) 1.9940139697194326e-06 or about one part in 500,000, i.e., 19 bits of accuracy. Just enough. It's kind of surprising that the maximum error varies by a factor of 8 over the interval [.5, 1]. Could somebody please double check my reasoning? Regards, Daniel _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
