rcauchy2d in package "fMultivar"

Martin Maechler maechler at stat.math.ethz.ch
Sat Jan 14 13:01:58 CET 2012


>>>>> Rolf Turner <r.turner at auckland.ac.nz>
>>>>>     on Sat, 14 Jan 2012 18:03:02 +1300 writes:

    > I have recently been trying to understand bivariate Cauchy
    > distributions, and to this end I've been perusing the
    > tools provided in the "fMultivar" package.

    > I have hit a snag, in that I don't understand what
    > rcauchy2d() is doing.

    > Basically it calls rt2d with the default degrees of
    > freedom, i.e. 4.

    > My understanding is that the bivariate Cauchy distribution
    > is the same as the bivariate t distribution with ***one***
    > degree of freedom.  So shouldn't rcauchy2d call

    >      rt2d(n=n,rho=rho,nu=1) ?

    > I note that dcauchy2d calls dt2d(x = x, y = y, rho = rho,
    > nu = 1), i.e.  the degrees of freedom are set equal to 1
    > as I would have expected.

    > Is this a bug in rcauchy2d(), or is there something that I
    > am misunderstanding?

Yes, this has been a bug, ... and I have fixed it now in the
(R-forge) development sources.

Thanks a lot, Rolf, for the report.

Best regards,
Martin



More information about the Rmetrics-core mailing list