""" test_MRTS.py Tests the MRTS logic in all distances Also, test automatic generation of the threshold Copyright 2023, Thomas Kreuz Distributed under the BSD License """ import numpy as np import pyspike as spk from pyspike import SpikeTrain from pyspike.isi_lengths import default_thresh def test_MRTS(): """ single testcase for all 4 algorithms changing MRTS: """ v1 = [12.0000, 16.0000, 28.0000, 32.0000, 44.0000, 48.0000, 60.0000, 64.0000, 76.0000, 80.0000, ]; v2 = [7.5376, 19.9131, 24.2137, 35.7255, 40.0961, 51.7076, 55.9124, 68.1017, 71.9863, 83.5994, ]; edges=[0, 300] max_tau=1000 sp1 = spk.SpikeTrain(v1, edges) sp2 = spk.SpikeTrain(v2, edges) ## SPIKE-SYNC Results1 = {14:0., 15:.3, 16:.6, 17:.9, 18:1.} for r in Results1: c = spk.spike_sync(sp1, sp2, MRTS=r) np.testing.assert_almost_equal(c, Results1[r]) ## SPIKE Results2 = { 0 : 0.12095, 1 : 0.12095, 2 : 0.12095, 3 : 0.12095, 4 : 0.12095, 5 : 0.12095, 6 : 0.12095, 7 : 0.12095, 8 : 0.12039, 9 : 0.11434, 10 : 0.10900, 11 : 0.10464, 12 : 0.10064, 13 : 0.09418, 14 : 0.08833, 15 : 0.08326, 16 : 0.07882, 17 : 0.07491, 18 : 0.07143, 19 : 0.06832, 20 : 0.06551, 21 : 0.06298, 22 : 0.06067, 23 : 0.05857, 24 : 0.05664, 25 : 0.05487, 26 : 0.05323, 27 : 0.05171, 28 : 0.05030, 29 : 0.04899, 30 : 0.04777, 31 : 0.04662, 32 : 0.04555, 33 : 0.04454, 34 : 0.04359, 35 : 0.04270, 36 : 0.04185, 37 : 0.04105, 38 : 0.04030, 39 : 0.03958, 40 : 0.03890, 41 : 0.03825, 42 : 0.03763, 43 : 0.03704, 44 : 0.03648, 45 : 0.03594, 46 : 0.03542, 47 : 0.03493, 48 : 0.03446, 49 : 0.03401, 50 : 0.03357, } for r in Results2: d = spk.spike_distance(sp1, sp2, MRTS=r) np.testing.assert_almost_equal(d, Results2[r], decimal=5) ## RI Results3 = { 0 : 0.12094, 1 : 0.12094, 2 : 0.12094, 3 : 0.12094, 4 : 0.12094, 5 : 0.12094, 6 : 0.12094, 7 : 0.12094, 8 : 0.12038, 9 : 0.11432, 10 : 0.10899, 11 : 0.10463, 12 : 0.10063, 13 : 0.09417, 14 : 0.08832, 15 : 0.08325, 16 : 0.07882, 17 : 0.07490, 18 : 0.07142, 19 : 0.06831, 20 : 0.06551, 21 : 0.06297, 22 : 0.06067, 23 : 0.05856, 24 : 0.05664, 25 : 0.05486, 26 : 0.05322, 27 : 0.05171, 28 : 0.05030, 29 : 0.04899, 30 : 0.04776, 31 : 0.04662, 32 : 0.04555, 33 : 0.04454, 34 : 0.04359, 35 : 0.04269, 36 : 0.04185, 37 : 0.04105, 38 : 0.04029, 39 : 0.03957, 40 : 0.03889, 41 : 0.03824, 42 : 0.03762, 43 : 0.03703, 44 : 0.03647, 45 : 0.03593, 46 : 0.03542, 47 : 0.03493, 48 : 0.03446, 49 : 0.03400, 50 : 0.03357, } for r in Results3: d = spk.spike_distance(sp1, sp2, MRTS=r, RI=True) #print('%d : %.5f,'%(r, d)) np.testing.assert_almost_equal(d, Results3[r], decimal=5) ## ISI Results4 = { 0 : 0.10796, 1 : 0.10796, 2 : 0.10796, 3 : 0.10796, 4 : 0.10796, 5 : 0.10796, 6 : 0.10796, 7 : 0.10796, 8 : 0.10796, 9 : 0.10796, 10 : 0.10796, 11 : 0.10796, 12 : 0.10704, 13 : 0.10103, 14 : 0.09547, 15 : 0.09065, 16 : 0.08643, 17 : 0.08271, 18 : 0.07940, 19 : 0.07644, 20 : 0.07378, 21 : 0.07137, 22 : 0.06918, 23 : 0.06718, 24 : 0.06534, 25 : 0.06366, 26 : 0.06210, 27 : 0.06066, 28 : 0.05932, 29 : 0.05807, 30 : 0.05691, 31 : 0.05582, 32 : 0.05480, 33 : 0.05384, 34 : 0.05294, 35 : 0.05209, 36 : 0.05128, 37 : 0.05052, 38 : 0.04980, 39 : 0.04912, 40 : 0.04847, 41 : 0.04785, 42 : 0.04727, 43 : 0.04671, 44 : 0.04617, 45 : 0.04566, 46 : 0.04517, 47 : 0.04470, 48 : 0.04425, 49 : 0.04382, 50 : 0.04341, } for r in Results4: d = spk.isi_distance(sp1, sp2, MRTS=r) np.testing.assert_almost_equal(d, Results4[r], decimal=5) print('OK1') def test_autoThresh(): """ Automatic determination of MRTS """ edges = [0, 1000] spikes1 = SpikeTrain([64.88600, 305.81000, 696.00000, 800.0000], edges) spikes2 = SpikeTrain([67.88600, 302.81000, 699.00000], edges) spikes3 = SpikeTrain([164.88600, 205.81000, 796.00000, 900.0000], edges) spikes4 = SpikeTrain([263.76400, 418.45000, 997.48000], edges) spike_train_list = [spikes1, spikes2, spikes3, spikes4] Thresh = default_thresh(spike_train_list) print('default_thresh got %.4f'%Thresh) np.testing.assert_almost_equal(Thresh, 325.4342, decimal=4, err_msg="default_thresh") c1 = spk.spike_sync(spikes1, spikes2, MRTS=Thresh) c2 = spk.spike_sync(spikes1, spikes2, MRTS='auto') np.testing.assert_almost_equal(c1, c2, err_msg="spike_sync") # apply it to the first example avove v1 = [12.0000, 16.0000, 28.0000, 32.0000, 44.0000, 48.0000, 60.0000, 64.0000, 76.0000, 80.0000, ]; v2 = [7.5376, 19.9131, 24.2137, 35.7255, 40.0961, 51.7076, 55.9124, 68.1017, 71.9863, 83.5994, ]; edges=[0, 300] sp1 = spk.SpikeTrain(v1, edges) sp2 = spk.SpikeTrain(v2, edges) t = default_thresh([sp1, sp2]) ## Look at all 4 algorithms c1 = spk.spike_sync(sp1, sp2, MRTS=t) c2 = spk.spike_sync(sp1, sp2, MRTS='auto') np.testing.assert_almost_equal(c1, c2, err_msg="spike_sync2") print('SS thresh %.3f, results %.3f'%(t,c1)) # compare with: {14:0., 15:.3, 16:.6, 17:.9, 18:1.} c1 = spk.spike_distance(sp1, sp2, MRTS=t) c2 = spk.spike_distance(sp1, sp2, MRTS='auto') np.testing.assert_almost_equal(c1, c2, err_msg="spike_distance") c1 = spk.spike_distance(sp1, sp2, MRTS=t, RI=True) c2 = spk.spike_distance(sp1, sp2, MRTS='auto', RI=True) np.testing.assert_almost_equal(c1, c2, err_msg="RI") c1 = spk.isi_distance(sp1, sp2, MRTS=t) c2 = spk.isi_distance(sp1, sp2, MRTS='auto') np.testing.assert_almost_equal(c1, c2, err_msg="ISI") c1 = spk.spike_directionality(sp1, sp2, MRTS=t) c2 = spk.spike_directionality(sp1, sp2, MRTS='auto') np.testing.assert_almost_equal(c1, c2, err_msg="directionality") print('OK2') if __name__ == "__main__": test_MRTS() test_autoThresh()