Merge pull request #8 from SpiNNakerManchester/test_base

Moved Integeration tests out of Spynakker

README.md

@@ -1,14 +1,95 @@
 # Cortical microcircuit simulation: PyNN version
 _Stored for easy access for people within the organisation_
 
-**Contributors:**  
-Sacha van Albada ([email protected])  
-Maximilian Schmidt  
-Jannis Schücker  
+**Contributors:**
+Sacha van Albada ([email protected])
+Maximilian Schmidt
+Jannis Schücker
 Andrew Rowley
 Alan Stokes
 
 This is an implementation of the multi-layer microcircuit model of early
 sensory cortex published by Potjans and Diesmann (2014) _The cell-type specific
 cortical microcircuit: relating structure and activity in a full-scale spiking
 network model_. Cerebral Cortex 24 (3): 785-806, [doi:10.1093/cercor/bhs358](https://doi.org/10.1093/cercor/bhs358)
+
+It has been run on three different back-ends: NEST, SpiNNaker, and the ESS (emulator of HMF)
+
+# Instructions
+
+1. Ensure you have the desired back-end.
+
+   For SpiNNaker see https://spinnakermanchester.github.io/latest/spynnaker.html
+
+   For NEST see http://www.nest-initiative.org/index.php/Software:Download
+   and to enable full-scale simulation, compile it with MPI support
+   (use the --with-mpi option when configuring) according to the instructions on
+   http://www.nest-initiative.org/index.php/Software:Installation
+
+2. Install PyNN according to the instructions on
+   http://neuralensemble.org/docs/PyNN/installation.html
+
+3. Run the simulation by typing ```python run_microcircuit.py <simulator>``` in
+   your terminal in the folder containing this file, where ```<simulator>``` is one
+   of ```nest``` or ```spinnaker``` (by default ```spinnaker is selected```).  There
+   are several potential arguments which can be seen by typing
+   ```python run_microcircuit.py <simulator> -h```.  A few useful ones include:
+
+    - --sim_duration - The simulation duration in milliseconds (default 1000)
+    - --output_path  - Where output files should be written (default results)
+
+6. Output files and basic analysis:
+
+   - Spikes are written to .txt files containing IDs of the recorded neurons
+     and corresponding spike times in ms.
+     Separate files are written out for each population and virtual process.
+     File names are formed as 'spikes'+ layer + population + MPI process + .txt
+   - Voltages are written to .dat files containing GIDs, times in ms, and the
+     corresponding membrane potentials in mV. File names are formed as
+     voltmeter label + layer index + population index + spike detector GID +
+     virtual process + .dat
+
+   - If 'plot_spiking_activity' is set to True, a raster plot and bar plot
+     of the firing rates are created and saved as 'spiking_activity.png'
+
+
+The simulation was successfully tested with NEST revision 10711 and MPI 1.4.3.
+Plotting works with Python 2.6.6 including packages numpy 1.3.0,
+matplotlib 0.99.1.1, and glob.
+
+## Known issues:
+
+- At least with PyNN 0.7.5 and NEST revision 10711, ConnectWithoutMultapses
+  works correctly on a single process, but not with multiple MPI processes.
+
+- When saving connections to file, ensure that pyNN does not create problems
+  with single or nonexistent connections, for instance by adjusting
+  lib/python2.6/site-packages/pyNN/nest/__init__.py from line 365 as follows:
+
+  if numpy.size(lines) != 0:
+      if numpy.shape(numpy.shape(lines))[0] == 1:
+          lines = numpy.array([lines])
+          lines[:,2] *= 0.001
+          if compatible_output:
+              lines[:,0] = self.pre.id_to_index(lines[:,0])
+              lines[:,1] = self.post.id_to_index(lines[:,1])
+      file.write(lines, {'pre' : self.pre.label, 'post' : self.post.label})
+
+- To use saveConnections in parallel simulations, additionally ensure that
+  pyNN does not cause a race condition where the directory is created by one
+  process between the if statement and makedirs on another process: In
+  lib/python2.6/site-packages/pyNN/recording/files.py for instance replace
+
+  os.makedirs(dir)
+
+  by
+
+  try:
+      os.makedirs(dir)
+  except OSError, e:
+      if e.errno != 17:
+          raise
+      pass
+
+Reinstall pyNN after making these adjustments, so that they take effect
+in your pyNN installation directory.

integration_tests/README.md

@@ -0,0 +1,3 @@
+These tests depend on https://github.com/SpiNNakerManchester/TestBase
+
+This is not added automatically by setup.py

integration_tests/__init__.py

@@ -0,0 +1,14 @@
+# Copyright (c) 2017-2020 The University of Manchester
+#
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.

integration_tests/test_microcircuit.py

@@ -0,0 +1,37 @@
+# Copyright (c) 2017-2021 The University of Manchester
+#
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+from spinnaker_testbase import ScriptChecker
+import os
+import sys
+import stat
+
+
+class TestMicrocircuit(ScriptChecker):
+
+    def test_microcircuit(self):
+        self.runsafe(self.microcircuit)
+
+    def microcircuit(self):
+        self.check_script("run_microcircuit.py")
+        for result_file in [
+                "spikes_L23E.pkl", "spikes_L23I.pkl",
+                "spikes_L4E.pkl", "spikes_L4I.pkl",
+                "spikes_L5E.pkl", "spikes_L5I.pkl",
+                "spikes_L6E.pkl", "spikes_L6I.pkl",
+                "spiking_activity.png"]:
+            result_path = os.path.join("results", result_file)
+            assert(os.path.exists(result_path))
+            assert(os.stat(result_path)[stat.ST_SIZE])

microcircuit/nest_specific_info.py

@@ -1,8 +1,8 @@
 import os
 
 from past.builtins import xrange
-from sim_params import NestParams
-from constants import DC, NEST_NEURON_MODEL, CONN_ROUTINE
+from .sim_params import NestParams
+from .constants import DC, NEST_NEURON_MODEL, CONN_ROUTINE
 import numpy
 
 
@@ -84,15 +84,27 @@ class NestSimulatorInfo(NestParams):
     ]
 
     def __init__(
-            self, parallel_safe=True, n_scaling=1.0, k_scaling=1.0,
-            neuron_model=NEST_NEURON_MODEL, conn_routine=CONN_ROUTINE,
-            save_connections=False, voltage_input_type='random',
-            delay_dist_type='normal', input_type=DC, record_fraction=True,
-            n_record=100, frac_record_spikes=1.0, record_v=False,
-            n_record_v=20, frac_record_v=0.1, record_corr=False,
-            pyseed=2563297, master_seed=124678,
-            tau_syn_name='tau_syn_ex'):
-        super(NestSimulatorInfo, self).__init__()
+            self, timestep=0.1, sim_duration=10000.0, min_delay=0.1,
+            max_delay=100.0, n_nodes=1, outfile='output.txt',
+            errfile='errors.txt', output_path='results', output_format='pkl',
+            conn_dir='connectivity', n_procs_per_node=24, wall_time='8:0:0',
+            memory='4gb',
+            mpi_path=(
+                '/usr/local/mpi/openmpi/1.4.3/gcc64/bin/'
+                'mpivars_openmpi-1.4.3_gcc64.sh'),
+            backend_path='/path/to/backend', pynn_path='/path/to/pyNN',
+            parallel_safe=True, n_scaling=1.0,
+            k_scaling=1.0, neuron_model=NEST_NEURON_MODEL,
+            conn_routine=CONN_ROUTINE, save_connections=False,
+            voltage_input_type='random', delay_dist_type='normal',
+            input_type=DC, record_fraction=True, n_record=100,
+            frac_record_spikes=1.0, record_v=False, n_record_v=20,
+            frac_record_v=0.1, record_corr=False, pyseed=2563297,
+            master_seed=124678, tau_syn_name='tau_syn_ex'):
+        super(NestSimulatorInfo, self).__init__(
+            timestep, sim_duration, min_delay, max_delay, n_nodes, outfile,
+            errfile, output_path, output_format, conn_dir, n_procs_per_node,
+            wall_time, memory, mpi_path, backend_path, pynn_path)
         self.parallel_safe = parallel_safe
         self.n_scaling = n_scaling
         self.k_scaling = k_scaling