<p dir="ltr">I'll give it a try.</p>
<p dir="ltr">John Sibert<br>
Emeritus Researcher, SOEST<br>
University of Hawaii at Manoa<br>
(Sent from the N7; please excuse the typos.)</p>
<div class="gmail_quote">On Jul 20, 2014 11:25 AM, "dave fournier" <<a href="mailto:davef@otter-rsch.com">davef@otter-rsch.com</a>> wrote:<br type="attribution"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
On 07/20/2014 02:03 PM, John Sibert wrote:<br>
<br>
Did you use something like<br>
<br>
<br>
feenableexcept(FE_DIVBYZERO | FE_INVALID | FE_OVERFLOW );<br>
<br>
with that you shoud be able to find out where the problem occurs suing the debugger.<br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Ugh.<br>
<br>
The variable I'm trying to profile has a pretty low estimated standard deviation. From the .std file:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
87 natural_mortality 1.5239e-01 7.8301e-03<br>
</blockquote>
By fiddling with the model structure and the lprof control variables, it will sometimes work, but the resulting profile plot is pretty strange. See attached.<br>
<br>
I'm working to track down the problem, but even simplifying the model it is slow going because of the need to compute the Hessian every time.<br>
<br>
The contents of diags is:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
f$ cat diags<br>
tempint1<br>
inf inf inf inf inf inf inf inf inf<br>
m(2)<br>
inf inf inf inf inf inf inf inf inf<br>
m(3)<br>
0.882 0.956 0.989 0.999 1 0.999 0.989 0.956 0.882<br>
xdistance<br>
1 1 1 1 1 1 1 1 1<br>
</blockquote>
<br>
Changing exp to mfexp for the computation of tempint1 in newmodmn.cpp seem to help a bit, but does not solve the problem. I think the cause occurs earlier, but I'm having trouble tracking it down.<br>
<br>
Does anyone really use this option? It seems like a good option to diagnose an ill-defined parameter when MCMC would take too long.<br>
<br>
Would it discomfort other users to change the way the procedure composes the file name from the variable name?<br>
<br>
John<br>
<br>
John Sibert<br>
Emeritus Researcher, SOEST<br>
University of Hawaii at Manoa<br>
Honolulu HI (GMT-10)<br>
<a href="tel:808-294-3842" value="+18082943842" target="_blank">808-294-3842</a> (mobile)<br>
<br>
Visit the ADMB projecthttp://<a href="http://admb-project.org/" target="_blank">admb-project.<u></u>org/</a><br>
<br>
On 07/17/2014 06:03 PM, dave fournier wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
Almost anything might go wrong with lprof. It works by maximizing the log-likelihood subject to a constraint<br>
on the value of the variable being profiled. This process drags the values into what might be areas where the<br>
function becomes unstable in some way.<br>
<br>
Hard to say more about a particular case without seeing it<br>
<br>
Dave<br>
<br>
</blockquote>
<br>
</blockquote>
<br>
</blockquote></div>