Typical number of compartments for neuron discretization?

Post Reply
mldichter
Posts: 4
Joined: Thu Nov 17, 2011 4:28 pm

Typical number of compartments for neuron discretization?

Post by mldichter » Fri Apr 14, 2017 2:17 pm

I was able to find many posts on how to pick a good number of segments with the d_lambda rule, and also symptoms of the discretization being too coarse for the model parameters, but what I really want to know is what is the typical number of compartments necessary to model a typical neuron. I don't need anything too exact. Just an order of magnitude would be wonderful. Is it 10, 100, 1000, 10000, 100000, etc?

I ask because I'm looking into the feasibility of a project idea I've had for a while and the determining factor is the number of compartments. Essentially, my idea is to solve for some of the compartment voltages implicitly in the algorithm, and only explicitly compute and store the remaining compartments. Hopefully this can cut down on memory access and decrease computation time.


Also, I know that an academic with free time is almost an oxymoron, but if anyone is interested in what I'm trying to do, I'd love to have a chat!

ted
Site Admin
Posts: 5091
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: Typical number of compartments for neuron discretization

Post by ted » Sun Apr 16, 2017 3:49 pm

There is no typical model neuron. Not being coy, just stating a fact. Some modelers use single compartment models, others use models with simplified branching structure that may require from 2 to a dozen or so compartments, and others focus on detailed morphologies that require hundreds or thousands of compartments, depending on the cell type. It all depends on the hypotheses that drive their research.

mldichter
Posts: 4
Joined: Thu Nov 17, 2011 4:28 pm

Re: Typical number of compartments for neuron discretization

Post by mldichter » Tue Apr 18, 2017 3:05 pm

That's exactly what I needed to know. Some sort of estimate for good upper and lower bounds.

Thank you much,
Mitchal

ted
Site Admin
Posts: 5091
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: Typical number of compartments for neuron discretization

Post by ted » Tue Apr 18, 2017 11:20 pm

ModelDB contains source code for > 1200 models implemented with a bunch of different simulators and general purpose programming languages. In some of those models, the number of compartments will be a hard-wired constant, and in others (especially models implemented with NEURON that use the d_lambda discretization strategy) it will be calculated during execution of model setup code. Have fun.

Post Reply