Pat,
On my Mac it takes 36 iterations to converge the first time (and
subsequently 4+/-1). So it looks like I should increment NITMAX to 50
until the strange initialisation problem is fixed.
Tim
On Fri, 2 Sep 2005, Patrick Wallace wrote:
> Tim Jenness's sla_PERTEL program can be made to work by a change to the
> sla_UE2PV routine.
>
> Replace:
>
> PARAMETER (TEST=1D-13,NITMAX=20)
>
> with:
>
> PARAMETER (TEST=1D-13,NITMAX=25)
>
> It turned out that exactly one more iteration was needed to reach the TEST
> criterion, at which stage further convergence did not occur. Subsequent
> passes through the code required far fewer iterations, typically four, so
> there's obviously something pathological about the first time through.
> I've set the iteration limit to five more than the previous value; the
> TEST criterion looks reasonably conservative as it is.
>
> Patrick Wallace
> ____________________________________________________________________________
> SSTD/HMNAO Internet: [log in to unmask]
> Rutherford Appleton Laboratory Tel: +44-1235-445372
> Chilton, Didcot, Fax: +44-1235-446362
> Oxon OX11 0QX, UK
> ____________________________________________________________________________
>
--
Tim Jenness
JAC software
http://www.jach.hawaii.edu/~timj
|