A Python test suite for statistical properties of probabilistic networks with and without spatial structure
-
1
Norwegian University of Life Sciences, Department of Mathematical Sciences and Technology, Norway
The advances in both computer development and simulation technology in computational neuroscience allow us to simulate ever larger and more complex networks of interacting spiking neurons that are in line with neuroanatomical findings (Helias et al, 2012). Even though reconstructions of brain circuitry become more detailed and complete (Potjans and Diesmann, 2012), the exact circuitry will vary between individuals. It is thus general connection rules, such as random connectivity according to measured connection probabilities for individual types of neurons, or also spatially dependent probabilistic connection profiles, that determine network structure in simulations.
As simulations become more complex it is hence of importance to be able to test whether the generated network instantiations are correct with respect to the underlying connectivity rule and neuron distribution. Such tests should be automated, so that they can be executed as part of automated test suites (Eppler et al, 2009; Zaytsev and Morrison, 2013). Randomized operations such as probabilistic network wiring can only be tested statistically and statistical tests may—indeed ought to—fail in a certain number of cases. To still be able to integrate such tests in automated test regimes, adaptive testing strategies have been proposed (L'Ecuyer and Simard, 2007).
Here we present a Python-based test suite for some of the most commonly used probabilistic network types, e.g. random connectivity with fixed probability, or distance-dependent connection profiles in two- and three-dimensional space with open or periodic boundary conditions. The tested factors are characteristic statistical quantities such as the degree-distribution or the distribution of pairwise distances, which are tested against the expected distributions obtained analytically. We discuss the implementation, performance and sensitivity of the tests with respect, e.g., to sample size, number of samples, or the quality of the underlying random number generator. The tests employed are chi-square tests for discrete distributions, such as degree distributions, and Kolmogorov-Smirnov tests for continuous distributions, e.g., the distribution of distances.
Though the test suite is primarily developed for connection routines implemented in the NEST simulation software (Gewaltig and Diesmann, 2007), we emphasize that it can be easily extended to work with data obtained from analogous connection routines in other simulators.
Acknowledgements
Partially funded by the Research Council of Norway (Grant 178892/V30 eNeuro) and EU Grant 269921 (BrainScaleS). Simulations were performed using NOTUR resources.
References
Eppler JM, Kupper R, Plesser HE and Diesmann M (2009). A testsuite for a neural simulation engine. Front. Neur. Conference Abstract: Neuroinformatics 2009. doi: 10.3389/conf.neuro.11.2009.08.042
Gewaltig M-O and Diesmann M (2007) NEST (NEural Simulation Tool). Scholarpedia 2:1430
Helias M, Kunkel S, Masumoto G, Igarishi J, Eppler JM, Ishii S, Fukai T, Morrison A and Diesmann M (2012) Supercomputers ready for use as discovery machines for neuroscience. Front Neuroinform 6:26
L'Ecuyer, P and Simard, R (2007) TestU01: A C Library for Empirical Testing of Random Number Generators. ACM Transactions on Mathematical Software 33:22
Potjans TC and Diesmann M (2012). The Cell-Type Specific Cortical Microcircuit: Relating Structure and Activity in a Full-Scale Spiking Network Model. Cereb Cortex Online first. doi: 10.1093/cercor/bhs358.
Zaytsev YV and Morrison A (2013) Increasing quality and managing complexity in neuroinformatics software development with continuous integration. Front. Neuroinform. 6:31. doi: 10.3389/fninf.2012.00031
Keywords:
Modeling and Simulation,
neuronal network,
Reliability of Simulations,
Software Validation,
statistical testing
Conference:
Neuroinformatics 2013, Stockholm, Sweden, 27 Aug - 29 Aug, 2013.
Presentation Type:
Poster
Topic:
Large scale modeling
Citation:
Hjertholm
D,
Kriener
B and
Plesser
H
(2013). A Python test suite for statistical properties of probabilistic networks with and without spatial structure.
Front. Neuroinform.
Conference Abstract:
Neuroinformatics 2013.
doi: 10.3389/conf.fninf.2013.09.00043
Copyright:
The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers.
They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.
The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.
Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.
For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.
Received:
25 Apr 2013;
Published Online:
11 Jul 2013.
*
Correspondence:
Dr. Hans Ekkehard Plesser, Norwegian University of Life Sciences, Department of Mathematical Sciences and Technology, Aas, 1432, Norway, hans.ekkehard.plesser@nmbu.no