# -*- coding: utf-8 -*- # # plot_sequence_EI_networks_inputs_weights.py # # Copyright 2019 Sebastian Spreizer # The MIT License import numpy as np import pylab as pl from mpl_toolkits.axes_grid1 import make_axes_locatable import matplotlib.patches as patches from lib.dbscan_cluster import detect_wrap from lib.panel_label import panel_label from lib.ploscb_formatting import set_fontsize import lib.ax_spines as ax_spines import lib.protocol as protocol set_fontsize() simulation = 'sequence_EI_networks' params = {} landscapes = [ {'mode': 'symmetric'}, {'mode': 'homogeneous', 'specs': {'phi': 3}}, {'mode': 'random'}, {'mode': 'Perlin', 'specs': {'size': 4}}, {'mode': 'Perlin_uniform', 'specs': {'size': 3}}, ] nrow = ncol = 120 npop = nrow * ncol recstart = 500. in_spike_count = [] in_cluster_count = [] in_cluster_mean_lifespan = [] params['landscape'] = landscapes[-1] for mean in np.arange(0., 501., 50.): for std in np.arange(0., 501., 50.): print('Mean: %s, std: %s' % (mean, std)) params['noiseE'] = {'mean': mean, 'std': std} params['noiseI'] = {'mean': mean, 'std': std} gids, ts = protocol.get_or_simulate(simulation, params) if len(ts) == 0: in_spike_count.append(0) in_cluster_count.append(0) in_cluster_mean_lifespan.append(0) continue idx = (gids - 1 < nrow * ncol) * ts > recstart ts, gids = ts[idx], gids[idx] in_spike_count.append(len(ts)) nclusters, clusters = detect_wrap(ts, gids, nrow, ncol, td=4, eps=4) in_cluster_count.append(nclusters - 1) if nclusters == 1: in_cluster_mean_lifespan.append(0) else: lifespan = [] for cidx in range(nclusters - 1): t = ts[clusters == cidx] dt = np.max(t) - np.min(t) lifespan.append(dt) in_cluster_mean_lifespan.append(np.mean(lifespan)) params = {} re_spike_count = [] re_cluster_count = [] re_cluster_mean_lifespan = [] params['landscape'] = landscapes[-1] for Ji in [5., 10., 15., 20., 25.]: for g in [6, 7, 8, 9, 10]: print('Ji: %s, g: %s' % (Ji, g)) params['Ji'] = Ji params['g'] = g gids, ts = protocol.get_or_simulate(simulation, params) if len(ts) == 0: re_cluster_count.append(0) re_spike_count.append(0) re_cluster_mean_lifespan.append(0) continue nrow = ncol = 120 npop = nrow * ncol idx = (gids - 1 < nrow * ncol) * ts > 500. ts, gids = ts[idx], gids[idx] re_spike_count.append(len(ts)) nclusters, clusters = detect_wrap(ts, gids, nrow, ncol, td=4, eps=4) re_cluster_count.append(nclusters) if nclusters == 1: re_cluster_mean_lifespan.append(0) else: lifespan = [] for cidx in range(nclusters - 1): t = ts[clusters == cidx] dt = np.max(t) - np.min(t) lifespan.append(dt) re_cluster_mean_lifespan.append(np.mean(lifespan)) def save_obj(obj, name): with open(name + '.pkl', 'wb+') as f: pickle.dump(obj, f, pickle.HIGHEST_PROTOCOL) def load_obj(name): with open(name + '.pkl', 'rb') as f: return pickle.load(f) colors = pl.rcParams['axes.prop_cycle'].by_key()['color'] fig, ax = pl.subplots(2, 3, dpi=300) fig.subplots_adjust(wspace=.7) im0 = ax[0, 0].imshow(np.array(in_spike_count).reshape(11, 11) / 5. / 120 / 120, origin='bottom') im1 = ax[0, 1].imshow(np.array(in_cluster_count).reshape(11, 11) / 5., origin='bottom') im2 = ax[0, 2].imshow(np.array(in_cluster_mean_lifespan).reshape(11, 11), origin='bottom') for i in range(3): ax[0, i].set_xticks([0,0,5,10]) ax[0, i].set_yticks([0,0,5,10]) ax[0, i].set_xticklabels([0] + range(0, 501, 250)) ax[0, i].set_yticklabels([0] + range(0, 501, 250)) ax[0, 0].set_xlabel('Input standard deviation (pA)') ax[0, 0].set_ylabel('Input mean (pA)') divider = make_axes_locatable(ax[0, 0]) cax = divider.append_axes("right", size="5%", pad=0.05) cbar0 = pl.colorbar(im0, cax=cax) cbar0.set_ticks([0,7,14]) cbar0.set_label('Firing rate [spikes/s]', fontsize=6) divider = make_axes_locatable(ax[0, 1]) cax = divider.append_axes("right", size="5%", pad=0.05) cbar1 = pl.colorbar(im1, cax=cax) cbar1.set_ticks([0,50,100]) cbar1.set_label('Number of clusters per seconds', fontsize=6) divider = make_axes_locatable(ax[0, 2]) cax = divider.append_axes("right", size="5%", pad=0.05) cbar2 = pl.colorbar(im2, cax=cax) cbar2.set_ticks([0,0,40,80]) cbar2.set_label('Mean lifespan of cluster [ms]', fontsize=6) panel_label(ax[0, 0], 'a', x=-.5) panel_label(ax[0, 1], 'b', x=-.3) panel_label(ax[0, 2], 'c', x=-.3) for i in range(3): circle = patches.Circle((2, 7), .5, linewidth=1, edgecolor=colors[1], facecolor='none') ax[0, i].add_patch(circle) im0 = ax[1, 0].imshow(np.array(re_spike_count).reshape(5, 5) / 5. / 120 / 120, origin='bottom') im1 = ax[1, 1].imshow(np.array(re_cluster_count).reshape(5, 5) / 5., origin='bottom') im2 = ax[1, 2].imshow(np.array(re_cluster_mean_lifespan).reshape(5, 5), origin='bottom') for i in range(3): ax[1, i].set_xticks([0,2,4]) ax[1, i].set_xticklabels([6,8,10]) ax[1, i].set_yticks([0,2,4]) ax[1, i].set_yticklabels([5,15,25]) ax[1, 0].set_xlabel('g') ax[1, 0].set_ylabel('Synaptic weight (pA)') divider = make_axes_locatable(ax[1, 1]) cax = divider.append_axes("right", size="5%", pad=0.05) cbar1 = pl.colorbar(im1, cax=cax) cbar1.set_ticks([0,20,90,160]) cbar1.set_label('Number of clusters per seconds', fontsize=6) divider = make_axes_locatable(ax[1, 0]) cax = divider.append_axes("right", size="5%", pad=0.05) cbar0 = pl.colorbar(im0, cax=cax) cbar0.set_ticks([0,5,20,35]) cbar0.set_label('Firing rate [spikes/sec]', fontsize=6) divider = make_axes_locatable(ax[1, 2]) cax = divider.append_axes("right", size="5%", pad=0.05) cbar2 = pl.colorbar(im0, cax=cax) cbar2.set_ticks([0,5,20,35]) cbar2.set_label('Mean lifespan of cluster [ms]', fontsize=6) panel_label(ax[1, 0], 'd', x=-.5) panel_label(ax[1, 1], 'e', x=-.3) panel_label(ax[1, 2], 'f', x=-.3) for i in range(3): circle = patches.Circle((2, 1), .5, linewidth=1, edgecolor=colors[1], facecolor='none') ax[1, i].add_patch(circle) filename = 'sequence_EI_networks_inputs_weights' fig.savefig(filename + '.png', format='png', dpi=300) fig.savefig(filename + '.pdf', format='pdf')