std.deviation yielding NaN when it should yield zero

[hespora]

Hi there,


please refer to report definition against the samples attached.

when using cognos statistical function standard-deviation-pop, the system sometimes yields Not a number as a result. It seems it only does so when the result should be zero (or infintessimally small); on the example report attached, it happens for product numbers '18110' and '123150', for example.

Has anybody seen something like that or even knows a solution?

[hespora]

for now, I'm solving it by calculating variance and standard deviation manually.