Accuracy evaluation of numerical methods used in state-of-the-art simulators for spiking neural networks

Henker, Stephan; Partzsch, Johannes; Schüffny, René

doi:10.1007/s10827-011-0353-9

Cited by 19 publications

(14 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Depending on the numerical method and SN, the accuracy increases monotonically as the step size reduces [61,81,34,72,29]. This is consistent with the theory of numerical methods where the solution provided by a convergent method approximates the exact solution as the step size tends to zero [26,10,24,20,25].…”

Section: Introductionsupporting

confidence: 82%

“…This behavior is also dependent on the numerical method and SN. The most studied aspect of SN simulation is the discontinuity error produced by the resetting mechanism to generate the AP that plenty of algorithms have emerged, mainly that of the LIF [27,71,47,52,6,59,66,58,29] and IZH [80,79,34]. Even though there are several open issues in the single SN simulation, this paper focuses on the time span and firing rate influence.…”

Section: Introductionmentioning

confidence: 99%

“…The influence of the time span variability has been ignored and, in consequence, remains unknown. Most current works have carried out simulations just on a single time span [31,27,68,47,71,52,6,59,66,58,79,29,72,34,81].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

How the Accuracy and Computational Cost of Spiking Neuron Simulation are Affected by the Time Span and Firing Rate

Godinez

Azuela

Santiago-Montero

2018

CyS

View full text Add to dashboard Cite

It is known that, depending on the numerical method, the simulation accuracy of a spiking neuron increases monotonically and that the computational cost increases in a power-law complexity as the time step reduces. Moreover, the mechanism responsible for generating the action potentials also affects the accuracy and computational cost. However, little attention has been paid to how the time span and firing rate influence the simulation. This study describes how the time span and firing rate variables affect the accuracy, computational cost, and efficiency. It was found that the simulation is importantly affected by these two variables.

show abstract

Section: Introductionsupporting

confidence: 82%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

How the Accuracy and Computational Cost of Spiking Neuron Simulation are Affected by the Time Span and Firing Rate

Godinez

Azuela

Santiago-Montero

2018

CyS

View full text Add to dashboard Cite

show abstract

“…Nevertheless, all simulations approximate the same network at the macroscopic level which can be characterized by meaningful network statistics. The latter can be compared across simulators (Davison et al, 2009; Henker et al, 2012). …”

Section: Resultsmentioning

confidence: 99%

Limits to high-speed simulations of spiking neural networks using general-purpose computers

Zenke

Gerstner

2014

Front. Neuroinform.

View full text Add to dashboard Cite

To understand how the central nervous system performs computations using recurrent neuronal circuitry, simulations have become an indispensable tool for theoretical neuroscience. To study neuronal circuits and their ability to self-organize, increasing attention has been directed toward synaptic plasticity. In particular spike-timing-dependent plasticity (STDP) creates specific demands for simulations of spiking neural networks. On the one hand a high temporal resolution is required to capture the millisecond timescale of typical STDP windows. On the other hand network simulations have to evolve over hours up to days, to capture the timescale of long-term plasticity. To do this efficiently, fast simulation speed is the crucial ingredient rather than large neuron numbers. Using different medium-sized network models consisting of several thousands of neurons and off-the-shelf hardware, we compare the simulation speed of the simulators: Brian, NEST and Neuron as well as our own simulator Auryn. Our results show that real-time simulations of different plastic network models are possible in parallel simulations in which numerical precision is not a primary concern. Even so, the speed-up margin of parallelism is limited and boosting simulation speeds beyond one tenth of real-time is difficult. By profiling simulation code we show that the run times of typical plastic network simulations encounter a hard boundary. This limit is partly due to latencies in the inter-process communications and thus cannot be overcome by increased parallelism. Overall, these results show that to study plasticity in medium-sized spiking neural networks, adequate simulation tools are readily available which run efficiently on small clusters. However, to run simulations substantially faster than real-time, special hardware is a prerequisite.

show abstract

“…In [7] is proposed a model for the assessment of the accuracy of various numerical methods employed in spiking neural simulations, showing perturbation sources due to numerical effects. An initial investigation of the parallelization techniques on the dynamics behavior was carried out by the authors in [3], based on the Hodgkin-Huxley neural model.…”

Section: Related Workmentioning

confidence: 99%

Simulation of Biological Neural Microcircuits on Multi-core Systems

Muntean

Joldos

2012

2012 Sixth International Conference on Complex, Intelligent, and Software Intensive Systems

View full text Add to dashboard Cite

Our research focuses on the identification and quantification of the impact that multi-core parallelization strategies have on the stability of the result of spiking neural networks simulations. We investigated OpenMP-based implementations of the Spike Response Model and Spike Time-Dependent Plasticity for studying behaviors of biological neurons and synapses. The underlying neural microcircuits have small-world topologies. The simulation strategy is a synchronous one. The software development methodology we follow makes use of systematic unit testing and continuous integration, giving us a way to verify various perturbations of simulation results. We carried out investigations on systems having different multi-core processors. The processing speed (spikes/second) of our simulator scales well with the number of cores, but the parallel efficiency is moderate when all cores of the system are used in the simulation (0.57 for 12 cores e.g.). The primary outcomes of this work are twofold: One the one hand, the proposed parallel simulation strategies show a dynamic behavior unaltered by the use of multi-core specific technologies. On the other hand, we analyze issues met in our approach to multi-core simulations.

show abstract

Accuracy evaluation of numerical methods used in state-of-the-art simulators for spiking neural networks

Cited by 19 publications

References 40 publications

How the Accuracy and Computational Cost of Spiking Neuron Simulation are Affected by the Time Span and Firing Rate

How the Accuracy and Computational Cost of Spiking Neuron Simulation are Affected by the Time Span and Firing Rate

Limits to high-speed simulations of spiking neural networks using general-purpose computers

Simulation of Biological Neural Microcircuits on Multi-core Systems

Contact Info

Product

Resources

About