[ADMB Users] inv_cumd_gamma parameters
Mollie Brooks
mbrooks at ufl.edu
Wed Jan 19 08:05:29 PST 2011
I retract my question.
Ben suggested that I try some numbers and compare them to what I would
expect.
I see that this gamma takes the quantile, and shape parameter and
always has has a rate=1 or scale=1.
To get the value at the z quantile of a gamma with shape=k,
scale=theta or rate=1/theta I can use
theta*inv_cumd_gamma(z,k)
Maybe this will save the next person a little time.
-Mollie
On 18-Jan-11, at 8:24 PM, Mollie Brooks wrote:
> Hi,
> I'm guessing Hans is the best person to ask, but I thought this
> could benefit others as well.
>
> I'm having trouble relating inv_cumd_gamma() to the gamma
> distribution I know which has 2 parameters. I've seen the example at
> http://otter-rsch.com/admbre/examples/gamma/gamma.html
> which uses it as follows:
> theta*inv_cumd_gamma(z, 1/theta)
>
> If z is an observation of the cumulative density, and inv_cumd_gamma
> should map z back to the gamma distribution, then I'm wondering why
> the gamma distribution only has one parameter. Do I need to do
> another transformation to get k in there?
>
> thanks,
> Mollie
>
>
> _______________________________________________
> Users mailing list
> Users at admb-project.org
> http://lists.admb-project.org/mailman/listinfo/users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.admb-project.org/pipermail/users/attachments/20110119/ca8ac22b/attachment.html>
More information about the Users
mailing list