question about point processes

Moderator: wwlytton

Post Reply
bhalterm
Posts: 52
Joined: Wed Mar 08, 2006 10:43 am
Location: University of Pennsylvania

question about point processes

Post by bhalterm »

I'm writing a method to make synapses that creates both the target point process and the netcon that connects to it. Is there a way to test for the presence of a point process already inserted at the location of the synapse to connect the netcon to, so I don't have a bunch of overlapping duplicate point processes?
ted
Site Admin
Posts: 5784
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Post by ted »

The MechanismType class has several methods that could be useful for this--see
file:///home/ted/ms/neuron/nrnhelp/help/neuron/neuron/classes/mechtype.html#MechanismType
especially count, is_netcon_target, pp_begin, pp_next

An alternative strategy that might be faster, although at the cost of increased storage
requirement, is to keep track of all PointProcesses and their locations as they are created.
Which is best for any particular situation would depend on network architecture and the
biophysical complexity of the cells.

Perhaps most efficient is to decide connectivity in advance of spawning any cells or
synaptic mechanisms. Then use this information during the "create cells" phase of network
instantiation to place synaptic mechanisms, and once again during the "connect cells"
phase to govern the NetCons. Of course, this means abandoning the idea of having a proc
that creates the synaptic target and its NetCon at the same time.
Raj
Posts: 220
Joined: Thu Jun 09, 2005 1:09 pm
Location: Hanze University of Applied Sciences
Contact:

Post by Raj »

If you are going to connect several sources to the same target it is probably because you are using synapses with a linear dynamics for the state variables. In that case it might be easier to create a synapse at all relevant locations and assume these exist when connecting to them.

Only with a very sparse connectivity and a large multicompartment model there would be simulation time gain in the more complicated strategy you are trying to implement, although because of the low overlap you would only eliminate a small percentage of the synaptic differential equations.

If your goal is to make complicated synapses were for example a dopaminergic neuron is synapsing close to the presynaptic terminal of a glutamatergic neuron and thus gating its activity, then I expect your more complicated strategy might be required.

In another words I think combining synapse and netcon in one object is most likely not a good idea for linear synapses and if the syanpses are nonlinear there is no need for illiminating multiple synapses in one segment.
Post Reply