How to find the sensitivity of the fitting on parameters?

Using the Multiple Run Fitter, praxis, etc..
Post Reply
Keivan
Posts: 127
Joined: Sat Apr 22, 2006 4:28 am

How to find the sensitivity of the fitting on parameters?

Post by Keivan » Thu Dec 15, 2011 2:13 am

Hi
I'm trying to publish an article in which I have used multirunfitter tool of NEURON. One of my reviewers has asked me to show sensitivities of the fitting on parameters. How can I do that?

Keivan
Posts: 127
Joined: Sat Apr 22, 2006 4:28 am

Re: How to find the sensitivity of the fitting on parameters

Post by Keivan » Fri Dec 16, 2011 11:23 am

Is there a solution for the above problem? Please, tell me if there is not.

Besides, latest alpha version of neuron has problem compiling in windows:

Code: Select all

./.libs/libivoc.a(ivocrand.o):ivocrand.cpp:(.text+0x6fa): undefined reference to `_nrnran123_getseq'

hines
Site Admin
Posts: 1517
Joined: Wed May 18, 2005 3:32 pm

Re: How to find the sensitivity of the fitting on parameters

Post by hines » Sat Dec 17, 2011 8:59 am

The principal axes and values of the praxis method are supposed to directly specify
the Hessian. However, I have found that the actual return values seem to vary quite
a bit and are useless for that. I've spent some effort to fix that but have been
unsuccessful. The only ways I know to compute the Hessian properly would be to use
a Python module or to explore the parameter space in the neighborhood of the minimum
and do your own computation of the principal axes and values.

I'll look into why the newest repository version has a problem linking ivoc and push
a fix.

Post Reply