What steps will reproduce the problem?
1. Build deal
2. Run: deal -x issue.tcl
With issue.tcl attached.
Expect output of a floating point number very close to zero.
Instead get -NaN.
The problem is that the standard deviation is the square root of a floating
point number, and, due to standard rounding errors, the square root is
taken of a negative number, rather than 0.
The hack solution is to return 0 for any negative value. I've implemented
that solution already.
Original issue reported on code.google.com by thomasoa on 29 Mar 2010 at 5:54
Original issue reported on code.google.com by
thomasoa
on 29 Mar 2010 at 5:54Attachments: