[Shootout-list] comments on the "statistics" test (reposted from moderation)
Randy Melton
rmelton@atmel.com
Wed, 16 Mar 2005 15:29:17 -0500
(reposted, as this was held up in moderation)
I looked at the statistics test.
I'm not sure the "skew" (or the Kurtosis) is calculated correctly
according the the definitions of skew that I found on:
http://www.itl.nist.gov/div898/handbook/eda/section3/eda35b.htm
But for the moment assuming the "C-gcc" is the golden standard (it
passes according to the web page) I found mistakes/differences in the
Tcl code:
line:
23 set skew [expr {$skew + pow($skew, 3)}]
probably should be:
23 set skew [expr {$skew + pow($deviation, 3)}]
I was looking at optimizations, but until I know the test passes I
don't want to attempt them.
thanks,
randy melton