| Author |
Comment/Response |
david silverman
|
 04/17/01 6:56pm
Hi,
Here's my next problem with the F-distribution. I used:
ftemp=First[fcut/.Solve[1-gamma==CDF[FRatioDistribution[1,Length[x]-Length[used]-1],fcut],fcut]];
as suggested, with Length[x]=920, Length[used]=0. Works great most of the time, but occassionally all I get is an error:
First::''normal'': ''Nonatomic expression expected at position 1 in First[fcut].''
So I ran the expression alone (without the ''ftemp=First['') to see what was being returned and it returned the word 'fcut' if gamma was less than 0.08. So I tried fooling with SetPrecision around the CDF portion, setting it to 30. Now I get answers for 0.05 and 0.06 but still only get 'fcut' for 0.08 and 0.07, everything's OK for 0.09 and above.
In case you need it, I'm using v4 on a pentium III xeon, $MachinePrecision=16
Thanks again,
David
URL: , |
|
