[WP34s] Regularized incomplete Beta function
|
05-08-2014, 10:58 PM
Post: #40
|
|||
|
|||
RE: [WP34s] Regularized incomplete Beta function
(05-08-2014 10:28 PM)Paul Dale Wrote: Given we're only promising accuracy for single precision, there are many orders of magnitude left before the risk of digit loss becomes significant. The user may enter dof values that lead to problems and strange results. At the moment I am looking at the Chi² quantile. For Chi² = n the CDF should be near 0,5 (slightly larger). Try this with n=2000 and you get 0,5042... (correct). Then try n=3000 and the 34s returns zero (!). Quote:For Student's-T and Chi² distributions we could approximate with a normal for very high degrees of freedom I guess. I don't remember the asymptotic properties of the F distribution when one or both degrees of freedom get large. I think that indeed very large dof values will be required until the Normal CDF matches the others only for SP accuracy. ;-) Dieter |
|||
« Next Oldest | Next Newest »
|
User(s) browsing this thread: 1 Guest(s)