I've started to work on this again, and can confirm there seems to be some
sort of bug in
the gradient test at the beginning of the current R-forge version of optimx. It
is not
something obvious, and looks like a mixup in arguments to functions, which have
been an
issue since I've been trying to trap NaN and Inf returns.
Worse, making the control starttests = FALSE fails because there I inadvertently
put the
initial function calculation inside the block that does the tests. Sigh.
Will try to get something done by end of this week. (This will be R-forge
version.)
JN
On 08/31/2011 09:31 AM, Ravi Varadhan wrote:> Hi Reuben,
>
>
>
> I am puzzled to note that the gradient check in ?optimx? does not work for
you. Can you
> send me a reproducible example so that I can figure this out?
>
>
>
> John ? I think the best solution for now is to issue a ?warning? rather
than an error
> message, when the numerical gradient is not sufficiently close to the
user-specified
> gradient.
>
>
>
> Best,
>
> Ravi.
>
>
>
> -------------------------------------------------------
>
> Ravi Varadhan, Ph.D.
>
> Assistant Professor,
>
> Division of Geriatric Medicine and Gerontology School of Medicine Johns
Hopkins University
>
>
>
> Ph. (410) 502-2619
>
> email: rvaradhan at jhmi.edu <mailto:rvaradhan at jhmi.edu>
>
>
>