Exponential overflow error

When Python is the interpreter, what is a good
design for the interface to the basic NEURON
concepts.

Moderator: hines

Post Reply
maxwellphenderson
Posts: 22
Joined: Mon Mar 07, 2011 12:17 am

Exponential overflow error

Post by maxwellphenderson » Wed Apr 09, 2014 9:33 am

I was having an issue similar to the one mentioned in:

http://www.neuron.yale.edu/phpbb/viewto ... =31&t=2130

I am not doing any parallelization so I think that the issue must be stemming from an overflow in a mod file. I am using simplified exc. and inh. neurons from ModelDB (#3817 http://senselab.med.yale.edu/modeldb/sh ... tex\README). However, when the error message comes up, it doesn't refer to a particular mod file... additionally, the error comes up semi-randomly (if I change stimulus for the network, on certain values an overflow will occur but NOT on others, if the network has 100 neurons vs 1000 neurons, etc etc). Is there a way to find search issues within neuron mod files, or is this more a guess-and-check issue?

ted
Site Admin
Posts: 5567
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: Exponential overflow error

Post by ted » Wed Apr 09, 2014 10:16 am

First one must be able to reproduce the error. What to do next depends on the nature of the error message; sometimes that is sufficient to identify the cause. If not, it may be useful to run a simulation that stops a few steps before the message is generated, then advance in single steps while monitoring state and assigned variables. If you are able to fix the problem based on these hints, good for you. If not, zip up just enough code (hoc, ses, and mod files only; no dll, .o, or .c files) to reproduce the problem, and email the zip file to
ted dot carnevale at yale dot edu
and I'll tell you what I discover.

Post Reply