{ "cells": [ { "cell_type": "code", "execution_count": null, "metadata": { "collapsed": false }, "outputs": [], "source": [ "%matplotlib inline" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "\n# Clopath Rule: Spike pairing experiment\n\nThis script simulates one ``aeif_psc_delta_clopath`` neuron that is connected with\na Clopath connection [1]_. The synapse receives pairs of a pre- and a postsynaptic\nspikes that are separated by either 10 ms (pre before post) or -10 ms (post\nbefore pre). The change of the synaptic weight is measured after five of such\npairs. This experiment is repeated five times with different rates of the\nsequence of the spike pairs: 10Hz, 20Hz, 30Hz, 40Hz, and 50Hz.\n\n## References\n\n.. [1] Clopath C, B\u00fcsing L, Vasilaki E, Gerstner W (2010). Connectivity reflects coding:\n a model of voltage-based STDP with homeostasis.\n Nature Neuroscience 13:3, 344--352\n\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "collapsed": false }, "outputs": [], "source": [ "import numpy as np\nimport matplotlib.pyplot as plt\nimport nest" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "First we specify the neuron parameters. To enable voltage dependent\nprefactor ``A_LTD(u_bar_bar)`` add ``A_LTD_const: False`` to the dictionary.\n\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "collapsed": false }, "outputs": [], "source": [ "nrn_params = {'V_m': -70.6,\n 'E_L': -70.6,\n 'C_m': 281.0,\n 'theta_minus': -70.6,\n 'theta_plus': -45.3,\n 'A_LTD': 14.0e-5,\n 'A_LTP': 8.0e-5,\n 'tau_minus': 10.0,\n 'tau_plus': 7.0,\n 'delay_u_bars': 4.0,\n 'a': 4.0,\n 'b': 0.0805,\n 'V_reset': -70.6 + 21.0,\n 'V_clamp': 33.0,\n 't_clamp': 2.0,\n 't_ref': 0.0,\n }" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Hardcoded spike times of presynaptic spike generator\n\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "collapsed": false }, "outputs": [], "source": [ "spike_times_pre = [\n # Presynaptic spike before the postsynaptic\n [20., 120., 220., 320., 420.],\n [20., 70., 120., 170., 220.],\n [20., 53.3, 86.7, 120., 153.3],\n [20., 45., 70., 95., 120.],\n [20., 40., 60., 80., 100.],\n # Presynaptic spike after the postsynaptic\n [120., 220., 320., 420., 520., 620.],\n [70., 120., 170., 220., 270., 320.],\n [53.3, 86.6, 120., 153.3, 186.6, 220.],\n [45., 70., 95., 120., 145., 170.],\n [40., 60., 80., 100., 120., 140.]]" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Hardcoded spike times of postsynaptic spike generator\n\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "collapsed": false }, "outputs": [], "source": [ "spike_times_post = [\n [10., 110., 210., 310., 410.],\n [10., 60., 110., 160., 210.],\n [10., 43.3, 76.7, 110., 143.3],\n [10., 35., 60., 85., 110.],\n [10., 30., 50., 70., 90.],\n [130., 230., 330., 430., 530., 630.],\n [80., 130., 180., 230., 280., 330.],\n [63.3, 96.6, 130., 163.3, 196.6, 230.],\n [55., 80., 105., 130., 155., 180.],\n [50., 70., 90., 110., 130., 150.]]\ninit_w = 0.5\nsyn_weights = []\nresolution = 0.1" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Loop over pairs of spike trains\n\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "collapsed": false }, "outputs": [], "source": [ "for (s_t_pre, s_t_post) in zip(spike_times_pre, spike_times_post):\n nest.ResetKernel()\n nest.SetKernelStatus({\"resolution\": resolution})\n\n # Create one neuron\n nrn = nest.Create(\"aeif_psc_delta_clopath\", 1, nrn_params)\n\n # We need a parrot neuron since spike generators can only\n # be connected with static connections\n prrt_nrn = nest.Create(\"parrot_neuron\", 1)\n\n # Create and connect spike generators\n spike_gen_pre = nest.Create(\"spike_generator\", 1, {\n \"spike_times\": s_t_pre})\n\n nest.Connect(spike_gen_pre, prrt_nrn,\n syn_spec={\"delay\": resolution})\n\n spike_gen_post = nest.Create(\"spike_generator\", 1, {\n \"spike_times\": s_t_post})\n\n nest.Connect(spike_gen_post, nrn, syn_spec={\n \"delay\": resolution, \"weight\": 80.0})\n\n # Create weight recorder\n wr = nest.Create('weight_recorder', 1)\n\n # Create Clopath connection with weight recorder\n nest.CopyModel(\"clopath_synapse\", \"clopath_synapse_rec\",\n {\"weight_recorder\": wr})\n syn_dict = {\"synapse_model\": \"clopath_synapse_rec\",\n \"weight\": init_w, \"delay\": resolution}\n nest.Connect(prrt_nrn, nrn, syn_spec=syn_dict)\n\n # Simulation\n simulation_time = (10.0 + max(s_t_pre[-1], s_t_post[-1]))\n nest.Simulate(simulation_time)\n\n # Extract and save synaptic weights\n weights = wr.get(\"events\", \"weights\")\n syn_weights.append(weights[-1])\n\nsyn_weights = np.array(syn_weights)\n# scaling of the weights so that they are comparable to [1]\nsyn_weights = 100.0*15.0*(syn_weights - init_w)/init_w + 100.0\n\n# Plot results\nfig1, axA = plt.subplots(1, sharex=False)\naxA.plot([10., 20., 30., 40., 50.], syn_weights[5:], color='b', lw=2.5, ls='-',\n label=\"pre-post pairing\")\naxA.plot([10., 20., 30., 40., 50.], syn_weights[:5], color='g', lw=2.5, ls='-',\n label=\"post-pre pairing\")\naxA.set_ylabel(\"normalized weight change\")\naxA.set_xlabel(\"rho (Hz)\")\naxA.legend()\naxA.set_title(\"synaptic weight\")\n\nplt.show()" ] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.6.12" } }, "nbformat": 4, "nbformat_minor": 0 }