splitneuron
Object Tabs
SplitNeuron is a completely new approach to large-scale, biologically plausible, neural-network simulation
Purpose
SplitNeuron is a library written in C for data structures and functions extending SQLite to simulate large-scale networks of Izhikevich Simple Model compartments.
SplitNeuron answers a fundamental issue in large-scale simulation, data transfer between storage and functional software: it uses database not only for data storage but also as simulation engine, moving computation to data rather than using storage systems only for data holding. This choice offers more features with less code to write and a unique way of accessing data for further analysis.
A specific way of treating connections as addresses enables explicit parallel data spreading onto clusters for HPC and a MapReduce approach for large-breath analysis. Features under development are direct import and cell/network creation from NeuroML, MPI and OpenCL extension.
Until now, splitneuron has been tested successfully on Linux and Darwin (OSX) with randomly connected network of up to 10^6 units and > 10^9 connections.
SplitNeuron answers a fundamental issue in large-scale simulation, data transfer between storage and functional software: it uses database not only for data storage but also as simulation engine, moving computation to data rather than using storage systems only for data holding. This choice offers more features with less code to write and a unique way of accessing data for further analysis.
A specific way of treating connections as addresses enables explicit parallel data spreading onto clusters for HPC and a MapReduce approach for large-breath analysis. Features under development are direct import and cell/network creation from NeuroML, MPI and OpenCL extension.
Until now, splitneuron has been tested successfully on Linux and Darwin (OSX) with randomly connected network of up to 10^6 units and > 10^9 connections.
License
GNU General Public LicensePrerequirements
*nix gccEase of Use
IntermediateMaturity
IntermediateOperating system
Unix-like
Downloads
This tool has been downloaded 0 times.
- 1 external files
Ratings & Reviews
Topics
Keywords
conductance based model, database interface, C, Database, Izhikevich, computational neuroscience, SQLite, simulators, spiking neural networks, Algorithm or Reusable Library, Neural Network, GNU General Public License (GPL),External web site
http://www.mindupload.org- Registered:
- Dec 13, 2010
- Last Modified:
- Dec 28, 2010
