summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorMario Mulansky <mario.mulansky@gmx.net>2015-05-13 11:26:38 +0200
committerMario Mulansky <mario.mulansky@gmx.net>2015-05-13 11:26:38 +0200
commit40e6372c0813cb8f34abd0ffee353bde7a96f158 (patch)
tree30c6f906be64e0085681abcc573b70bb31d2f51e
parent6f418a5a837939b132967bcdb3ff0ede6d899bd2 (diff)
new profiles example
-rw-r--r--examples/profiles.py66
1 files changed, 66 insertions, 0 deletions
diff --git a/examples/profiles.py b/examples/profiles.py
new file mode 100644
index 0000000..05494bd
--- /dev/null
+++ b/examples/profiles.py
@@ -0,0 +1,66 @@
+""" profiles.py
+
+Simple example showing some functionality of distance profiles.
+
+Copyright 2015, Mario Mulansky <mario.mulansky@gmx.net>
+
+Distributed under the BSD License
+"""
+
+
+from __future__ import print_function
+
+import pyspike as spk
+
+spike_trains = spk.load_spike_trains_from_txt("PySpike_testdata.txt",
+ edges=(0, 4000))
+
+##### ISI PROFILES #######
+
+# compute the ISI profile of the first two spike trains
+f = spk.isi_profile(spike_trains[0], spike_trains[1])
+
+# ISI values at certain points
+t = 1200
+print("ISI value at t =", t, ":", f(t))
+t = [900, 1100, 2000, 3100]
+print("ISI value at t =", t, ":", f(t))
+print("Average ISI distance:", f.avrg())
+print()
+
+# compute the multivariate ISI profile
+f = spk.isi_profile_multi(spike_trains)
+
+t = 1200
+print("Multivariate ISI value at t =", t, ":", f(t))
+t = [900, 1100, 2000, 3100]
+print("Multivariate ISI value at t =", t, ":", f(t))
+print("Average multivariate ISI distance:", f.avrg())
+print()
+print()
+
+# for plotting, use the get_plottable_data() member function, see plot.py
+
+
+##### SPIKE PROFILES #######
+
+# compute the SPIKE profile of the first two spike trains
+f = spk.spike_profile(spike_trains[0], spike_trains[1])
+
+# SPIKE distance values at certain points
+t = 1200
+print("SPIKE value at t =", t, ":", f(t))
+t = [900, 1100, 2000, 3100]
+print("SPIKE value at t =", t, ":", f(t))
+print("Average SPIKE distance:", f.avrg())
+print()
+
+# compute the multivariate SPIKE profile
+f = spk.spike_profile_multi(spike_trains)
+
+# SPIKE values at certain points
+t = 1200
+print("Multivariate SPIKE value at t =", t, ":", f(t))
+t = [900, 1100, 2000, 3100]
+print("Multivariate SPIKE value at t =", t, ":", f(t))
+print("Average multivariate SPIKE distance:", f.avrg())