Building random network with different seeds

Moderator: wwlytton

Post Reply
psychic
Posts: 3
Joined: Fri Apr 02, 2010 11:33 am

Building random network with different seeds

Post by psychic » Wed Sep 29, 2010 12:51 am

Hi,
I'm trying to build a network with random connections.
The trick is I would like to average my results over several different random seeds.

The problem is I'm not sure how to clear the previous network created before I reassign a new seed and re-run the simulation.

I cannot also envolope the code creating the connections into a procedure because I'm instantiating new objects inside the function.

What I want to do is this:
Loop {
Reassign random seed
BuildConnections()
Run simulation ()
Get results()
Purge previous connections()
}

Any advice on how I can structure my code to accomplish this.
THanks

ted
Site Admin
Posts: 5566
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: Building random network with different seeds

Post by ted » Wed Sep 29, 2010 8:18 am

The general answer is to exit neuron and start anew for each net architecture. This is the only way to ensure that nothing potentially troublesome will persist from run to run. Not knowing your level of expertise, or that of others who might subsequently read this thread, and having seen many examples of "expedient programming" (read "clever hacks, not-so-clever hacks, and just plain stupid tricks"), I hesitate to suggest anything else.

Post Reply