README_Tests.md

Tests

The parameter files for the tests are found in the "Input"-folder. Run the code as follow:

.\main $(path_to_output_folder) $(path_to_parameter_file) 

The respective commands are found in Input/run.sh.

The output files can then be analyzed with the R-markdown file that can be found under Test_Results/AnalyseCode or with files in the respective subfolders of Test_Results/.

In brackets, we give rough estimates for computation times.

Test 1: Showcase (6 secs)

Tests the basic functions of the spiking network.

To analyse, run the R-markdown in "Test_Results/Test1". For result, see here.

Test 2: Four regimes of sparse spiking network activity (4 x 1 minute)

Reproduces 4 panels of Fig. 8 in Brunel (2000) "Dynamics of Sparsely Connected Networks of Excitatory and Inhibitory Spiking Neurons" J Comp Neurosci 8, 183.

To analyse, copy the R-files from Test_Results/AnalyseCode into the output folder and run the R-markdown file. See results here for cases a, b, c, d.

Test 3: Small network Bistability (10 min on cluster)

Reproduces Bistability Figure (Fig3) from Ekelmans et al. manuscript; in preparation.

To analyse, run the R-markdown in "Test_Results/Test3". For result, see here.

Test 4: Working Memory (< 1 minute)

This test reproduces Fig.2 B of Mongillo, G., Barak, O. and Tsodyks, M., 2008. Synaptic theory of working memory. Science, 319(5869), pp.1543-1546. The model uses STP synapses.

To analyse, copy the R-files from Test_Results/AnalyseCode into the output folder and run the R-markdown file. For result, see here.

Test 5: Transient working memory (< 6 minutes)

Reproduces Fig. 2 a (2) of Becker, S., Nold, A. and Tchumatchenko, T., 2020. Formation and synaptic control of active transient working memory representations. bioRxiv. The model employs STP synapses with a retrograde signaling pathway.

To analyse, copy the R-files from Test_Results/AnalyseCode into the output folder and run the R-markdown file. For result, see here.

Test 6: Convergence to mean-field equations and balanced state (4 x 15 minutes on cluster)

Compares simulation for N = 10000...60000 with mean-field limit and balanced state results. These are four simulations.

To analyse, copy the Matlab-code Test_Results/Test6/MatlabAnalysis into the Output folder, and run the file Run_CheckConvergence_N.m. For result, see here.

Test 7: Large network Bistability (37h on cluster)

Reproduces Fig 4 A of Mongillo, G., Hansel, D., & Van Vreeswijk, C. (2012). "Bistability and spatiotemporal irregularity in neuronal networks with nonlinear synaptic transmission". PRL, 108(15), 158101. The model simulates a network of 250000 neurons with STP synapses.

To analyse, copy the R-files from Test_Results/AnalyseCode into the output folder and run the R-markdown file. For result, see here.

Test 8: Current covariance (2 x 2h on cluster)

Reproduces Fig. 3B and 4B of Rosenbaum, R., Smith, M.A., Kohn, A., Rubin, J.E. and Doiron, B., 2017. The spatial structure of correlated neuronal variability. Nature neuroscience, 20(1), pp.107-114.

To analyse, run the Matlab Live Script in "Test_Results/Test8". For resuts, see here.

Test 9: Contrast invariance (2 x 5h on cluster)

Reproduces contrast invariant selectivity and nonlinear sensitivity from Bernaez Timon et al. manuscript (Fig7); in preparation.

To analyse, run the Matlab Live script in "Test_Results/Test9". For resuts, see here.

Test 10: Heterosynaptic plasticity framework

Reproduces part of Saif Ahmed's masters thesis (Heterosynaptic Cooperativity in Pair-based Spike-Timing-Dependent Plasticity (STDP), 2021). This test is built to test the basic heterosynaptic framework works as intended in its original version. We replicate the behaviour of heterosynaptic cooperativity when we increase alpha and beta parameters in the weight spread while varying the A factor(LTD_pre_factor) from Ahmed's model.

IMPORTANT

Put the ten generated folders from the simulation's iteration in the same folder as this notebook. The five folders' data is necessary for this analysis to replicate the figure.

Test 11: DictatNeuronPop

This test acts as a unit test of the DictatNeuronPop feature, allowing the user to dictate the activity of the population through an external text file. The test requires an extra file apart from Parameters.txt, which contains the neuron instructions for spiking. The analysis is performed by the script found in the test's folder in Test_Results/.

Test 12: Cooperativity-based heterosynaptic plasticity with STDP and limited branch resources

This test acts as a unit test of the model named BranchedResourcesHeteroSTDP (name subject to change), a model that attempts to repoduce the logic found empirically in heterosynaptic plasticity experiments from a pairing framework, while retaining regular STDP in the model. This test also uses the DictatNeuronPop feature, which could be a source of error if Test 11 has not been reproduced.

The model consists of pairing of synapses through presynaptic spikes (effect decays in time and space) and a potentiation of the pairing by the postsynaptic spike. STDP is identical but without the time and space decay of the synaptic spine pairing. The weights are also constrained following the equation:

$w_{i}$=$\beta$*$\frac{\alpha_{base, i} + \alpha_{stim, i} *e^{-dt/\tau}}{\omega+\sum_{i=0}^{N}(\alpha_{base, i}+\alpha_{stim, i})}$

Where $\alpha_{stim, i}$ is the only changing variable during the simulation.

The test requires two extra files apart from Parameters.txt, which contains the neuron instructions for spiking. The analysis is performed by the script found in the test's folder in Test_Results/.