Parallel neural Circuit SIMulator PCSIM Tutorial Dejan Pecevski

Parallel neural Circuit SIMulator (PCSIM) - Tutorial Dejan Pecevski Institute for Theoretical Computer Science Graz University of Technology FIAS Theoretical Neuroscience Summer School, Frankfurt, August, 2008

Outline of the Tutorial • • • General Info: What is PCSIM? Features: What is PCSIM useful for? Network elements: neurons and synapses Scalability: Using clusters for parallel simulation. Python Interface PCSIM basic operations § Creating and connecting neurons § Simulating § Recording signals § Populations and Projections • Summary Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 2/40

Who am I? • Ph. D. Student at the Institute for Theoretical Computer Science, Graz University of Technology • One of the developers of PCSIM. • Research Interests § Plasticity and Learning mechanisms in neural circuits § Spiking Neural Networks § Simulation technologies/algorithms for biological neural networks Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 3/40

PCSIM in a nutshell • Simulator for parallel simulation of spiking and analog neural networks with point neuron models • Implemented in C++, with Python and Java interfaces • High-level definition of neural networks • Library of built-in neuron and synapse models • Part of FACETS standardization efforts: the py. NN interface http: //www. neuralensemble. org • Open-source, free, released under the GPL license, web page at: http: //www. igi. tugraz. at/pcsim Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 4/40

Developers Thomas Natschläger Ph. D. Software Competence Center Hagenberg, Austria Dejan Pecevski DI Institute for theoretical computer science Graz University of Technology Graz, Austria Other Contributors: Klaus Schuch DI (IGI), Christian Ernstbrunner DI (SCCH), Walter Hargassner DI (SCCH) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 5/40

Feature Overview • Object Oriented Design § The user interface/view is object oriented • General Communication System § Analog and spiking messages § Network elements with multiple input and output ports § Supported in distributed/multi-threaded mode • Flexible construction of more complex network architectures § Populations of simulated objects § Projections for connectivity specification Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 6/40

Feature Overview • Parallel simulation § Mixed multi-thread and distributed § Easy to use, transparent to the user • Easily extensible § A compilation tool for creating PCSIM extension packages • Usable as backend component in C++, Python, Java, … § Easier and faster setup of simulations in Python and Java • Runs on Linux or other Unix-like platforms § can be compiled under Windows (still not tested). Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 7/40

PCSIM high-level structure Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 8/40

Type of models PCSIM is suitable for • Large recurrent networks with simple point neurons § Distributed and multi-threaded capabilities § Efficient simulation engine in C++ § More than 105 spiking neurons and 108 synapses • Hybrid models § Abstract modules (filters etc. ) combined together with circuits of analog neurons and spiking neurons § Extensible with new custom elements § Closed loop/feedback models Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 9/40

Type of models PCSIM is suitable for • Structured models § Networks composed of smaller subnetworks (Populations) § Probabilistic connectivity patterns based on neuron’s attributes • Diversity of neurons and synapses § Different neuron types in the neuron populations § Specifying random distribution for parameter values § New neuron and synapse types can be implemented Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 10/40

Examples of PCSIM usage • Spiking neural network model of the V 1 area in the visual cortex of mammals (Schuch & Rasch) • Testing the computational performance of laminar cortical models based on experimental data. (Schuch & Haeusler) • Experiments to examine the learning capabilities of rewardmodulated spike-timingdependent plasticity (Legenstein, Pecevski, Maass) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 11/40

A simple model represented in PCSIM Model equations • The equations can be decoupled into separate simulated elements. • Presynaptic neurons send spikes to the synapses. • Synapses inject currents (or modify conductances) in the postsynaptic neuron. Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 12/40

Network element: basic building block • Nodes of the network (dynamical systems) • Advanced/integrated each time step of the simulation • Represented by class Sim. Object § Neurons, synapses, recorders are derived from this class • Multiple input and output ports, either analog or spiking. • Connections are formed from output to input ports of the same type. Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 13/40

Neurons and Synapses as Network Elements • Synapses as network elements are attached to the postsynaptic neuron • Synapses have one input port, no output ports • Neurons have only one output, no inputs • The spiking and analog connections can connect neurons on different nodes Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 14/40

Built-in Neuron Models Spiking Models Input Neurons • Lif. Neuron, Cb. Lif. Neuron Leaky integrate-and-fire • HHNeuron Hodgkin-Huxley Neuron • Izhi. Neuron (Izhikevich, 2004) • ae. IFNeuron Adaptive Exponential I&F (Brette and Gerstner, 2005) • Linear. Poisson. Neuron Leaky integrate with Poisson output • Poisson. Input. Neuron • Spiking. Input. Neuron Analog Neurons • Linear. Analog. Neuron Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 15/40

Built-in Synapse Models • Type § current based § conductance based • PSR Kernel § exponential, alpha, double-exponential, Square • Short-term Plasticity (Markram et al. 1998) • Other Mechanisms § NMDA (Gabbiani et al. 1994) § Gaba. B (Mainen et al. 1994) § Reward-modulated STDP (Izhikevich, 2007) § Homeostatic plasticity (Buonomano et al. 2005) § Ornstein-Uhlenbeck noise synapse • Spike-timing-dependent Plasticity § (Froemke and Dan, 2002) § (Gütig et al, 2003) § each pair and nearest (Destexhe et al. 2001) • Analog Synapse neighbour Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 16/40

Simulation Strategy • Clock-driven simulation with a fixed time step. • Hybrid Integration Strategy § Neurons are integrated every time step. § Synapses are event driven, i. e. they are integrated only after receiving a spike. • Numeric Integration Algorithms § Exponential Euler Method § ODE solvers from GSL Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 17/40

Parallel Simulation • • Each MPI process advances subset of neurons MPI as underlying communication layer § • spiking and analog messages Each process can have multiple-threads • Transparent to the user § Default round-robin distribution of the neurons over nodes § Custom distribution strategies § Easy retrieval of recordings (as in singlethread case) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 18/40

Parallel Simulation • Why? § Speed-up simulation of an existing model by using more processors § Enables simulation of larger models: no memory limit problem § Scale up the number of neurons while keeping the simulation time the same. • Comparison between single-threaded and distributed simulation § < 104 neurons and 107 synapses on a single machine § > 105 neurons and 108 synapses on clusters § > 106 neurons and 1010 synapses on supercomputers? Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 19/40

Scalability of PCSIM Hardware: 3. 4 Ghz, 512 kb cache, 4 GB RAM Model: • 50000 Lif. Neurons, 4. 0 Hz average firing rate • 50 106 synapses, 1. 5 ms transmission delay Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 20/40

Scalability of PCSIM Hardware: 3. 4 Ghz, 512 kb cache, 4 GB RAM Cuba Model: • 5000 Lif. Neurons per node, 4. 0 Hz average firing rate • 103 synapses per neuron, 1. 0 ms transmission delay, broad distribution of time synaptic constants Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 21/40

Extending PCSIM • What are extensions? § User implemented classes that are plugged into the PCSIM OO framework § usualy coded in C++ (can be coded also in Python) • Compilation of additional separate Python extension packages, without the need to recompile the main PCSIM code. • The extension classes are automatically wrapped in Python • Usual extensions for PCSIM § new neuron/synapse types § other user defined network elements § custom rules for specific network construction Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 22/40

Python Interface What is Python? (taken from www. python. org) • Python is a dynamic object-oriented programming language • offers strong support for integration with other languages and tools. • comes with extensive standard libraries (“batteries included”). • can be learned in a few days. • Many Python programmers report substantial productivity gains and feel the language encourages the development of higher quality, more maintainable code. • many scientific packages: scipy, numpy, matplotlib, ipython, Rpy etc. Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 23/40

Python Interface • PCSIM can be imported as a package within Python from pypcsim import * • The model is represented as a network object: net = Single. Thread. Network() • Individual neurons and synapses are accessible as python objects. • Population and Projection objects for high-level construction of networks composed of populations • Together with the other Python packages for scientific computing provides a complete working environment for neural simulations. Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 24/40

Creating and connecting neurons # Creates one LIF neuron with default parameter values nrn_model = Lif. Neuron() nrn_id = net. create( nrn_model ) # Creates array of 10 LIFneurons nrn_array = net. create( Lif. Neuron(), 10) # Connects the first and the second neuron in the array syn_id = net. connect( nrn_array[0], nrn_array[1], Static. Spiking. Synapse()) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 25/40

Specifying neuron/synapse parameters nrn = net. create(Lif. Neuron( Cm=2 e-10, Rm=1 e 8, Vthresh=-50 e-3, Vresting=-60 e-3, Vreset=-60 e-3, Trefract=5 e-3, Vinit=-60 e-3 )) ; syn = net. connect( pre_nrn, post_nrn, Static. Spiking. Synapse( W=1 e-10, tau= 5 e-3, delay=1 e-3 )) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 26/40

Recording # record the output spikes of a neuron rec_spk = net. record( nrn, Spike. Time. Recorder() ) # record the membrane potential rec_vm = net. record( nrn, “Vm”, Analog. Recorder() ) # record the weight of a synapse rec_syn = net. record( syn, “W”, Analog. Recorder() ) # record any field in an element rec = net. record( my. Element, “any. Field. Name”, Analog. Recorder()) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 27/40

Simulate the model # simulate the network for 1 second net. simulate(1. 0) # or advance it for 200 time steps net. advance(200) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 28/40

Accessing individual neurons/synapses # creates return an ID of the neuron nrn_id = net. create( Lif. Neuron() ) # get the actual neuron object from the network nrn_object = net. object(nrn_id) # Manipulate the neuron object nrn_object. Vthresh = -59 e-3 # getting the recorded values from a recorder values = list(net. object(rec_id). get. Recorded. Values()) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 29/40

Synapse classes nomenclature • Static. Curr. Exp. Synapse (Static. Spiking. Synapse) Current/conductance based? • Dynamic. Cond. Dbl. Exp. Synapse Short-term plasticity Shape of PSR kernel • Static. Stdp. Cond. Alpha. Synapse • Dynamic. Curr. Alpha. Synapse • Dynamic. NMDAAlpha. Synapse Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 30/40

Hello, world! from pypcsim import * net = Single. Thread. Network() pre_nrn = net. create( Lif. Neuron(Iinject = 5 e-8) ) post_nrn = net. create( Lif. Neuron(Iinject = 5 e-8) ) net. connect( pre_nrn, post_nrn, Static. Spiking. Synapse()) rec = net. record( post_nrn, Spike. Time. Recorder() ) net. simulate(0. 2) print “spike times: ”, list(net. object(rec). get. Recorded. Values()) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 31/40

Specifying inputs # Create input neuron which spikes at specific times net. create(Spiking. Input. Neuron( [0. 1, 0. 2, 0. 4])) # Create input neuron with Poisson process spike times net. create( Poisson. Input. Neuron(rate = 10, duration = 10) ) # Create analog input neuron with an # output analog signal specified as array net. create( Analog. Input. Neuron( arange(0, 1 e-4) % 1 ) ) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 32/40

Creating and connecting many neurons • • Population of 1000 neurons randomly interconnected with 0. 1 probability Population of recorders: one for each neuron in the population # create a population of 1000 LIF neurons popul = Sim. Object. Population(net, Lif. Neuron(), 1000) # create the projection proj = Connections. Projection( popul, Static. Spiking. Synapse(), Random. Connections( 0. 1 ) ) # create a recorder population rec_popul = popul. record( Spike. Time. Recorder() ) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 33/40

Random distributions for parameter values • Factory – an object that generates PCSIM elements • Neuron and synapse objects (used as models for creation) are simple clone factories • Sim. Object. Variation. Factory – factory that attaches random distributions to parameter names nrn_factory = Sim. Object. Variation. Factory( Lif. Neuron() ) nrn_factory. set( “Vinit”, Normal. Distribution( -55 e-3, 0. 1 ) ) popul = Sim. Object. Population(net, nrn_factory, 1000) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 34/40

Spatial heterogenous populations • • Each neuron has coordinates in 3 D space The population is composed of different families of neurons (that can be of different type) exc_nrn = Lif. Neuron( Cm = 2 e-10, Inoise = 1 e-10) inh_nrn = Lif. Neuron( Cm = 3 e-10 ) popul = Spatial. Family. Population(net, [ exc_nrn, inh_nrn], Ratio. Based. Families((4, 1) ), Cuboid. Integer. Grid 3 D(20, 10)) exc_popul, inh_popul = popul. split. Families() Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 35/40

Distance dependent random connections Creates random connections with probability where D is the euclidean distance between the two neurons proj = Connections. Projection( popul, Static. Spiking. Synapse(), Euclidean. Distance. Random. Connections( C, lambda ) ) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 36/40

Saving recorded data from pypcsimplus import * # Create the recording class r = Recordings() # Add the recorder populations as attributes to r r. spikes = popul 1. record( Spike. Time. Recorder() ) r. vm_traces = popul 2. record( “Vm”, Analog. Recorder() ) # Some additional variables to be saved r. v = 5 # Save the Recordings class to a HDF 5 file r. save. In. One. H 5 File(“results. h 5”) Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 37/40

Running a parallel simulation • Change the network class from Single. Thread. Network to either: § Multi. Thread. Network( n. Threads = 4 ) § Distributed. Single. Thread. Network § Distributed. Multi. Thread. Network • then run the script with mpirun $ mpirun –n 10 python my_experiment. py Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 38/40

Summary • PCSIM is a tool for simulation structured random neural networks composed of spiking and analog point neuron models • Has various built-in neuron and synapse models • Supports parallel simulation: distributed and multi-threaded • Flexible high-level construction of networks based on probabilistic rules • The Object-oriented framework is designed to support easy extensions • Can be used as a package within the Python interpreting programming language, together with other useful scientific packages Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 39/40

PCSIM Resources • Web page http: //www. lsm. tugraz. at/pcsim • On Sourceforge http: //www. sourceforge. net/projects/pcsim • Mailing list http: //sourceforge. net/mailarchive/forum. php? forum_name=pcsim-users • User manual http: //www. lsm. tugraz. at/pcsim/usermanual/html/index. html • C++ class reference http: //www. lsm. tugraz. at/pcsim/cppclassreference/html/hierarchy. html Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 40/40

Thank you for your attention! Dejan Pecevski, FIAS Theoretical Neuroscience Summer School, August, 2008 41/40