Revision as of 11:21, 12 July 2010

In 1994, Josef Skrzypek wrote a book titled "Neural Network Simulation Environments." Unfortunately, there is no modern equivalent and you couldn't sort the simulators using javascript. This is a GFDL-licensed project to document a variety of modern neural simulators along a number of useful dimensions. Given the size of such a task we need your help! If you have experience with any simulator (especially those not listed), feel a dimension is missing, would like to elaborate in a sub-article or simply want to check facts, please feel free to edit, leave a comment on the discussion page or e-mail the primary maintainer. Basically, Wikipedia's Be Bold policy applies here.

Simulators

Simulators by feature

In principle, most simulators can mimic the functionality provided by another simulator. This table tries to reflect the feature set that the simulator intends to provide, including plugins when they are commonly used, as in the case of neuroConstruct.

Simulator	Primary focus	User programming style	Programming interface	Virtual environment	Cluster computing	GPU support	PyNN support	Point neurons	3D neurons	Biological neurons	Backpropagation	Self Organizing	Constraint Satisfaction
Emergent	Networks	Visual Scripts C++		Rigid body physics 3D graphics 3D model file import Full robotics support	MPI	In development	No	Yes	No	Integrate and fire Ion channels	Cross-entropy, Feedforward (SRN), Recurrent, Almeida-Pineda, Real-time continuous	Competitive Learning (hard/soft), Hebbian (soft/hard/ZSH, Kohonen	Boltzmann, Binary/continuous Hopfield, Interactive Activation and Competition, GRAIN
FANN	Networks	Visual Dozens of language bindings		No	No	In development	No	Yes	No	No	RPROP iRPROP Quickprop	Kohonen	No
NEURON	Neurons Networks	Visual Scripts (interpreted) Scripts (compiled to DLL)		No	MPI	No	Yes	Yes	intrinsic, also neuroConstruct	Yes	No, but possible	No, but possible	No, but possible
Neuron-C	Neurons Networks	Visual Scripts (interpreted) Scripts (compiled to DLL)		No	MPI	No	Yes	No		Yes	No, but possible	No, but possible	No, but possible
GENESIS	Neurons	Visual Scripts	No images provided.	No	PGENESIS	No	No	Yes	neuroConstruct	Yes	No, but possible	No, but possible	No, but possible
SNAAP	Neurons Networks	Unknown		No	Parallel SNAAP	No	No	Yes	No	Yes	No	No	Hebbian
LENS	Networks	Unknown		No	No	No	No	Yes	No	No	Quickprop Feed-forward recurrent RBPTT/CRBPTT	Kohonen	Boltzmann
Catacomb2	Neurons Networks	Unknown	No images provided.	Walls Sound sources Food dispeners Sensors Robot effectors	No	No	No	Unknown	Unknown	Yes	Unknown	Unknown	Unknown
Topographica	Networks, Maps	Python scripts Visual C++		Blender interface	In development	In development	No	Yes	No	Integrate and fire	No	Hebbian (soft/hard/covariance/CPCA) SOM LISSOM BCM	No
iNVT	Network	Programming		No	MPI	No	No	Yes	No	No	No	No	No
KInNeSS	Unknown	Unknown		Unknown	Unknown	Unknown	No	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown
SNNS	Networks	Visual		No	Unknown	No	No	Yes	No	No	Quickprop RPROP Backpercolation Counterpropagation DLVQ BPTT CC PCC RCC TDNN CGM	Hebbian, Kohonen, Delta	Unknown
Synapse	Data mining	Visual		No	Unknown	Yes	No	Yes	No	No	Quickprop	Normalized Hebbian Kohonen Competitive learning	No
XNBC	Neurons Networks Analysis tools	Visual		No	No	No	No	Yes	No	[Leaky] Integrate and fire Condition. burster Phenomenologic Ion conductance (14 ion currents) Hybrid networks	No	Hebbian	No
PCSIM	Large networks	Python Scripts Java	Non-graphical	No	MPI	No	No	Yes	No	Yes	No, but possible	STPD, etc...	No, but possible
NEST	Large networks	Programming Scripts	Non-graphical	No	MPI	No	Yes	Yes	No	Yes	Unknown	Unknown	Unknown
Brian	Neurons, networks	Python scripts	Non-graphical	No	Parallel python	No	No	Yes	Yes	Integrate and fire, Hodgkin-Huxley; Equations are user specified	No	STDP, short-term plasticity	Unknown
Simbrain	Networks, Visualization	Visual, Scripts (interpreted)	File:SimbrainScreenshot	Simple 2-d world	In development	No	No	Yes (partial implementation)	No	Integrate and fire, Izhikevich	In development	Hebbian, SOM, Short Term Plasticity	Hopfield, IAC

TODO

• We are also working on a survey to send out to the author of each simulator. You can help by adding new questions and clarifying existing ones.
  
• Fill in all "Unknown" fields in the tables
  
• Add all simulators from the paper "Simulation of networks of spiking neurons: A review of tools and strategies" by BretteEtAl07
  
• How can these tables better compare spiking simulators?