[ADMB Users] very small number math

[Martell, Steven]

I would advise against using the min function, as this will create a discontinuity in your objective function. You could try other approaches, such as rescaling your data etc, but hard to help without seeing the code.
Steve
On 2013-03-06, at 11:36 AM, "Fowler, Mark" <Mark.Fowler at dfo-mpo.gc.ca<mailto:Mark.Fowler at dfo-mpo.gc.ca>>
 wrote:


Hello,

I think I’ve hit a precision limit, looking for a solution along the same sort of lines as posfun or mfexp. My program stops (no error message) in a loop when a particular equation attempts to divide 19616.8 by 7.57411e-154. The preceding iteration worked with 21666/5.73085e-132. These are silly starting estimates of a value which will get reasonable with a little fitting, I just want to accept a maximal result instead of bombing. In this case I can just use a min function on the denominator, but wonder if there isn’t a conventional approach for this situation, say defining a minimum number threshold.

Mark Fowler

Population Ecology Division

Bedford Inst of Oceanography

Dept Fisheries & Oceans

Dartmouth NS Canada

B2Y 4A2

Tel. (902) 426-3529

Fax (902) 426-9710

Email Mark.Fowler at dfo-mpo.gc.ca<mailto:Mark.Fowler at dfo-mpo.gc.ca>

Home Tel. (902) 461-0708

Home Email mark.fowler at ns.sympatico.ca<mailto:mark.fowler at ns.sympatico.ca>


_______________________________________________
Users mailing list
Users at admb-project.org<mailto: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/20130307/76969447/attachment.html>