number of section, compartments and nseg

[chinhou]

Dear Friends,

please forgive me , if i'm asking simple questions for you. but for me it's confusing!

1- "nseg" determine spatial accuracy, right? when we increase nseg*=3
then we have more accuracy. but how we see this? should we check every time ,for a section.e.g. v(0.1), ..v(0.5),...v(0.9)?

2- suppose i have a section with length L. if i divide this section to three parts with length L/3,then connect these sections, does it mean i change simulation from one-compartment to three-compartments?



is there any criteria to decide about number of compartments in a model for example with 100 section, which have different "L" and "diam"?i guess more compartments will give better results, but do we have a minimum and sufficient number?



Best Regards,

[ted]

1- "nseg" determine spatial accuracy, right? when we increase nseg*=3
then we have more accuracy. but how we see this? should we check every time ,for a section.e.g. v(0.1), ..v(0.5),...v(0.9)?

Inadequate spatial discretization will produce errors that distort plots of voltage and
current as functions of time and space. Create a graph that contains plots of
membrane potential vs. time for several locations in your model cell. Run a simulation
and observe the results. Then at the oc> prompt type
forall nseg*=3
to triple nseg in all sections, and run another simulation. If the plots look different, you
know that the original value of nseg wasn't big enough, so you need to triple nseg
again and run another simulation. Repeat this until you see no further change in the
shape of the plots.

2- suppose i have a section with length L. if i divide this section to three parts with length L/3,then connect these sections, does it mean i change simulation from one-compartment to three-compartments?

Only if each section has nseg = 1.

is there any criteria to decide about number of compartments in a model for example with 100 section, which have different "L" and "diam"?i guess more compartments will give better results, but do we have a minimum and sufficient number?

Read about the d_lambda rule in chapter 5 of The NEURON Book, or get this article
Hines, M.L. and Carnevale, N.T. NEURON: a tool for neuroscientists.
The Neuroscientist 7:123-135, 2001
from
http://www.neuron.yale.edu/neuron/bib/nrnpubs.html
and read it.

The CellBuilder can apply the d_lambda rule to a model cell--go to
http://www.neuron.yale.edu/neuron/docs
and work through its online tutorial.

The d_lambda rule can also be applied from hoc--go to the FAQ list
http://www.neuron.yale.edu/neuron/faq/general-questions
and see the link called
What's a good strategy for specifying nseg?

[chinhou]

Thank you so much Ted,

Just for example, consider following code:

Code: Select all

oc>create A
oc>access A
oc>L=100
oc>diam=2
oc>insert hh
oc>insert pas
oc>nseg
        1 
oc>gnabar_hh(0:1)=2:5
oc>for(x) {print x,"",gnabar_hh(x)}
first instance of x
0 3.5 
0.5 3.5 
1 3.5 
oc>nseg=3
oc>gnabar_hh(0:1)=2:5
oc>for(x) {print x,"",gnabar_hh(x)}
0 2.5 
0.16666667 2.5 
0.5 3.5 
0.83333333 4.5 
1 4.5 

1-when nseg =1, we can not create a non-unifrom distribution, right?
then,dose nseg=1 for a section mean 1-compartment?

2-and since, for nseg=3, we have three different numbers for gnabar_hh, through the section, do we have 3-compartment model?

Best Regards,

[ted]

Yes. Please read chapter 5 of The NEURON Book, or go to the on-line Documentation page
http://www.neuron.yale.edu/neuron/docs
and start reading the "Key papers about NEURON".