parameterize test, single file

Author: janskaar

testsuite/pytests/sli2py_neurons/test_iaf_psc_exp_ps.py

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-# test_iaf_psc_exp_ps_lossless.py
+# test_iaf_psc_exp_ps_neurons.py
 #
 # This file is part of NEST.
 #
@@ -20,19 +20,13 @@
 # along with NEST.  If not, see <http://www.gnu.org/licenses/>.
 
 """
-Synopsis: (test_iaf_psc_exp_ps_lossless) run -> compares response to current step with analytical
-solution and tests lossless spike detection
+Synopsis: (test_iaf_psc_exp_ps_neurons) run -> compares response to current step with analytical
+solution for both iaf_psc_exp_ps and iaf_psc_exp_ps_lossless and tests lossless spike
+detection for iaf_psc_exp_ps_lossless.
 
-Description:
-test_iaf_psc_exp_ps_lossless.py is an overall test of the iaf_psc_exp_ps_lossless model connected
-to some useful devices.
-
-The first part of this test is exactly the same as test_iaf_psc_exp_ps,
-demonstrating the numerical equivalency of both models in usual conditions.
-The only difference between the models, which is tested in the second part,
-is the detection of double threshold crossings during a simulation step
-(so the membrane potential is again below V_th at the end of the step)
-by the lossless model.
+In the first part of the test, a DC current is injected into the neuron using a current generator
+device. The membrane potential is recorder, and compared to the expected analytical solution,
+with a spike happening off-grid.
 
 The second part tests whether the lossless spike detection algorithm [1] is
 working correctly.
@@ -53,15 +47,14 @@
 regions_algorithm.py which they used to generate Fig. 6. Here you can adjust the
 parameters as wished and obtain the respective regions.
 
-
 References:
 [1] Krishnan J, Porta Mana P, Helias M, Diesmann M and Di Napoli E
     (2018) Perfect Detection of Spikes in the Linear Sub-threshold
     Dynamics of Point Neurons. Front. Neuroinform. 11:75.
     doi: 10.3389/fninf.2017.00075
 
 Author:  Jeyashree Krishnan, 2017, and Christian Keup, 2018
-SeeAlso: test_iaf_psc_exp, test_iaf_psc_exp_ps
+SeeAlso: test_iaf_psc_exp
 """
 
 import nest
@@ -70,34 +63,9 @@
 import pytest
 
 
-def test_precise_spiking_dc_input():
-    dt = 0.1
-    dc_amp = 1000.0
-
-    nest.ResetKernel()
-    nest.set(resolution=dt, local_num_threads=1)
-
-    dc_gen = nest.Create("dc_generator", {"amplitude": dc_amp})
-    nrn = nest.Create("iaf_psc_exp_ps_lossless", 1)
-    vm = nest.Create("voltmeter", {"interval": 0.1})
-
-    syn_spec = {"synapse_model": "static_synapse", "weight": 1.0, "delay": dt}
-    nest.Connect(dc_gen, nrn, syn_spec=syn_spec)
-    nest.Connect(vm, nrn, syn_spec=syn_spec)
-
-    nest.Simulate(8.0)
-
-    times = vm.get("events", "times")
-    times -= dt  # account for delay to multimeter
-
-    tau_m = nrn.get("tau_m")
-    R = tau_m / nrn.get("C_m")
-    theta = nrn.get("V_th")
-    E_L = nrn.get("E_L")
-
-    # array for analytical solution
+def compute_analytical_solution(times, theta, V_reset, t_ref, E_L, tau_m, R, dc_amp, dt):
     V_m_analytical = np.empty_like(times)
-    V_m_analytical[:] = nrn.get("E_L")
+    V_m_analytical[:] = E_L
 
     # first index for which the DC current is received by neuron.
     # DC current will be integrated from this time step
@@ -121,13 +89,44 @@ def test_precise_spiking_dc_input():
     # set refractory potential
     # first index after spike, offset by time at which DC current arrives
     crossing_ind = int(exact_spiketime // dt + 1) + start_index
-    num_ref = int(nrn.get("t_ref") / dt)
-    V_m_analytical[crossing_ind : crossing_ind + num_ref] = nrn.get("V_reset")
+    num_ref = int(t_ref / dt)
+    V_m_analytical[crossing_ind : crossing_ind + num_ref] = V_reset
 
     # rise after refractory period
     num_inds = len(times) - crossing_ind - num_ref
     V_m_analytical[crossing_ind + num_ref :] = vm_soln_offset[:num_inds]
+    return V_m_analytical
+
+
+@pytest.mark.parametrize("model", ["iaf_psc_exp_ps", "iaf_psc_exp_ps_lossless"])
+def test_iaf_psc_exp_ps_dc_input(model):
+    dt = 0.1
+    dc_amp = 1000.0
+    nest.ResetKernel()
+    nest.set(resolution=dt, local_num_threads=1)
 
+    dc_gen = nest.Create("dc_generator", {"amplitude": dc_amp})
+    nrn = nest.Create(model, 1)
+    vm = nest.Create("voltmeter", {"interval": 0.1})
+
+    syn_spec = {"synapse_model": "static_synapse", "weight": 1.0, "delay": dt}
+    nest.Connect(dc_gen, nrn, syn_spec=syn_spec)
+    nest.Connect(vm, nrn, syn_spec=syn_spec)
+
+    nest.Simulate(8.0)
+
+    times = vm.get("events", "times")
+    times -= dt  # account for delay to multimeter
+
+    tau_m = nrn.get("tau_m")
+    R = tau_m / nrn.get("C_m")
+    theta = nrn.get("V_th")
+    E_L = nrn.get("E_L")
+    V_reset = nrn.get("V_reset")
+    t_ref = nrn.get("t_ref")
+
+    V_m_analytical = compute_analytical_solution(times, theta, V_reset, t_ref, E_L, tau_m, R, dc_amp, dt)
+    # array for analytical solution
     nptest.assert_array_almost_equal(V_m_analytical, vm.get("events", "V_m"))
 
 
@@ -190,7 +189,6 @@ def test_lossless_spike_detection():
     time_nospike = sr_nospike.events["times"]
     time_missingspike = sr_missingspike.events["times"]
     time_spike = sr_spike.events["times"]
-    print(time_nospike)
     assert len(time_nospike) == 0
     assert time_missingspike == pytest.approx(4.01442)
     assert time_spike == pytest.approx(4.00659)