Csim affords simulation and output visualization / analysis for layered networks of spiking neurons. The fundamental entities are cell types, cell grids, probabilistic and/or radial distance-determined connections, and output modules. Memory and time resource constraints are kept small to maximize the network sizes and lengths of time that can be simulated. For example, individual connections need not be stored, and a simulation timestamp of 0.1 msec can be used while still modeling important cellular properties such as bursting, short and long inhibition, after-hyperpolarization (AHP), etc..
LatticeConn is a utility for generating connectivity specification input for Csim starting from a higher-level format.
 |
 |
Cmatrix is a tool for determining complete cortical connectivity rules given incomplete specification data and confidence levels in its components. It implements a constraint-satisfaction network and uses simulated annealing to arrive at global optima.
Csim was developed as part of my PhD dissertation research. Thanks to the Neuron group's maintenance of InterViews, it has no external dependencies and should run on all three major platforms (Mac OS X, Windows, and Linux), though building will be easiest on unix systems. The source code is released under a BSD-style license. It is unsupported, but if you have any questions, contact me at arobert at interstitiality dot net and if I'm not too tied up I'll try to answer them.
You can download it (documentation included) here.
Back to Adrian's "serious" page.