If what you really want is a reliable inverse gamma function, instead of
why Excel's isn't very reliable, then download Ian's probability library

http://members.aol.com/iandjmsmith/Examples.xls

Jerry

wrote:

zimmerp wrote:

why does =GAMMAINV(0,5;389;1) return #NUM! ? This function is supposed to
return num only when parametres are less than or equal 0. Excel solves it
numerically, but in case it doesn't find a numeric solution, Excel is
supposed to return #N/A error value.


Excel can evaluate GAMMADIST(x,389,1) for 0<= x <= 6E+153. Outside of
that range it returns #NUM! So my guess is that during its repeated
attempts to solve the equation GAMMADIST(x,389,1) = 0.5, it comes up
with an estimate outside this region.

A similar example is =GAMMADIST(1E-290,4.4,1E-190,FALSE) where it
returns #DIV/0! Again the most plausible explanation is that EXCEL has
problems with intermediate calculations.

However, only someone with access to the code in Excel could really
tell you what the problem is.


Ian Smith