It's the absolute value of y that should be bounded,
not just y itself.  Thus:

tcdf=: 4 : 0
 assert. (%:x)>|y
 0.5 + y * (!-:x-1) * ((0.5,-:1+x) H. 1.5 x%~-*:y) % (%:o.x) * !<:-:x
)



----- Original Message -----
From: Roger Hui <[EMAIL PROTECTED]>
Date: Tuesday, August 12, 2008 8:20
Subject: Re: [Jprogramming] general Gamma distribution
To: Programming forum <[email protected]>

> x tcdf y is the CDF at y of the t-distribution with x degrees of 
> freedom
> tcdf=: 4 : 0
>  assert. y<%:x
>  0.5 + y * (!-:x-1) * ((0.5,-:1+x) H. 1.5 x%~-*:y) % 
> (%:o.x) * !<:-:x
> )
> 
>    5 tcdf 2.01505 1.47588
> 0.95 0.899999
> 
> The y<%:x restriction is rather limiting.  Perhaps Ewart Shaw
> or others knowledgeable on this subject know of a better
> formulation for the CDF.
> 
> 
> 
> ----- Original Message -----
> From: Brian Schott <[EMAIL PROTECTED]>
> Date: Tuesday, August 12, 2008 6:43
> Subject: Re: [Jprogramming] general Gamma distribution
> To: Programming forum <[email protected]>
> 
> > Robert,
> > 
> >     I may have misunderstood your original question
> > regarding the student's T distribution. I thought you were
> > asking for a way to generate random variates from the T, but
> > you may have been asking for a way to compute probabilities
> > from the student's T. If the latter was your question, I am
> > afraid that computing probabilities is much more difficult
> > and I have no thoughts about that.
> > 
> >     OTOH if you are thinking about constructing
> > confidence intervals around sample means using the student
> > T, then you need random normal variates and a table lookup
> > of a student's T to build the interval(s) and it would be
> > easier to read in a table of student's T values or just the
> > ones you need for the sample sizes and levels of confidence
> > you are simulating.
> > 
> > 
> > On Tue, 12 Aug 2008, Robert Cyr wrote:
> > 
> > + Bran Schott said:
> > + T_k = Z/sqrt(U/k)
> > +
> > + I have seen the usual formulas for building the distribution 
> > function,but+ none like this one. It seems very interesting.
> > +
> > + As for the cdf required for the usual confidence 
> > interval,  accuracy comes
> > + at a price.  Perhaps less accurate but useful is a 
> simple 
> > list of a few
> > + dozen factors and a simple formula.
> > +
> > + Robert Cyr
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to