def teardown_function(function):
"""Executed by py.test after each test in this module
Since examples might muck with the RC settings, we reset them here.
"""
rc.reload_rc([])
rc.set('decoder_cache', 'enabled', 'False')
def pytest_runtest_setup(item): # noqa: C901
rc.reload_rc([])
rc.set('decoder_cache', 'enabled', 'False')
rc.set('exceptions', 'simplified', 'False')
if not hasattr(item, 'obj'):
return
for mark, option, message in [
('example', 'noexamples', "examples not requested"),
('slow', 'slow', "slow tests not requested")]:
if getattr(item.obj, mark, None) and not item.config.getvalue(option):
pytest.skip(message)
if getattr(item.obj, 'noassertions', None):
skipreasons = []
for fixture_name, option, message in [
('analytics', 'analytics', "analytics not requested"),
('plt', 'plots', "plots not requested"),
('logger', 'logs', "logs not requested")]:
if fixture_name in item.fixturenames:
if item.config.getvalue(option):
break
else:
skipreasons.append(message)
else:
pytest.skip(" and ".join(skipreasons))
if 'Simulator' in item.fixturenames:
for test, reason in TestConfig.Simulator.unsupported:
# We add a '*' before test to eliminate the surprise of needing
# a '*' before the name of a test function.
if fnmatch(item.nodeid, '*' + test):
pytest.xfail(reason)
def test_cache_benchmark(self, varying_param, analytics, plt):
varying = {
'D': np.asarray(np.linspace(1, 512, 10), dtype=int),
'N': np.asarray(np.linspace(10, 500, 8), dtype=int),
'M': np.asarray(np.linspace(750, 2500, 8), dtype=int)
}[varying_param]
axis_label = self.param_to_axis_label[varying_param]
times = [self.time_all(self.get_args(varying_param, v))
for v in varying]
# Restore RC to original settings
default = RC_DEFAULTS['decoder_cache', 'enabled']
rc.set("decoder_cache", "enabled", str(default))
default = RC_DEFAULTS['decoder_cache', 'readonly']
rc.set("decoder_cache", "readonly", str(default))
for i, data in enumerate(zip(*times)):
plt.plot(varying, np.median(data, axis=1), label=self.labels[i])
analytics.add_data(varying_param, varying, axis_label)
analytics.add_data(self.keys[i], data)
plt.xlabel("Number of %s" % axis_label)
plt.ylabel("Build time (s)")
plt.legend(loc='best')
def test_dtype(Simulator, request, seed, bits):
# Ensure dtype is set back to default after the test, even if it fails
request.addfinalizer(lambda: rc.set("precision", "bits", "64"))
float_dtype = np.dtype(getattr(np, "float%s" % bits))
int_dtype = np.dtype(getattr(np, "int%s" % bits))
with nengo.Network() as model:
u = nengo.Node([0.5, -0.4])
a = nengo.Ensemble(10, 2)
nengo.Connection(u, a)
p = nengo.Probe(a)
rc.set("precision", "bits", bits)
with Simulator(model) as sim:
sim.step()
for k, v in sim.signals.items():
assert v.dtype in (float_dtype,
int_dtype), "Signal '%s' wrong dtype" % k
objs = (obj for obj in model.all_objects if sim.data[obj] is not None)
for obj in objs:
for x in (x for x in sim.data[obj] if isinstance(x, np.ndarray)):
assert x.dtype == float_dtype, obj
assert sim.data[p].dtype == float_dtype
def pytest_runtest_setup(item):
rc.reload_rc([])
rc.set('decoder_cache', 'enabled', 'False')
rc.set('exceptions', 'simplified', 'False')
item_name = get_item_name(item)
# join all the lines and then split (preserving quoted strings)
unsupported = shlex.split(" ".join(
item.config.getini("nengo_test_unsupported")))
# group pairs (representing testname + reason)
unsupported = [unsupported[i:i + 2] for i in range(0, len(unsupported), 2)]
for test, reason in unsupported:
# wrap square brackets to interpret them literally
# (see https://docs.python.org/3/library/fnmatch.html)
test = "".join("[%s]" % c if c in ('[', ']') else c for c in test)
# We add a '*' before test to eliminate the surprise of needing
# a '*' before the name of a test function.
test = "*" + test
if fnmatch(item_name, test):
if TestConfig.run_unsupported:
item.add_marker(pytest.mark.xfail(reason=reason))
else:
pytest.skip(reason)
def pytest_runtest_setup(item):
rc.reload_rc([])
rc.set('decoder_cache', 'enabled', 'False')
rc.set('exceptions', 'simplified', 'False')
item_name = get_item_name(item)
# join all the lines and then split (preserving quoted strings)
unsupported = shlex.split(
" ".join(item.config.getini("nengo_test_unsupported")))
# group pairs (representing testname + reason)
unsupported = [
unsupported[i:i + 2] for i in range(0, len(unsupported), 2)]
for test, reason in unsupported:
# wrap square brackets to interpret them literally
# (see https://docs.python.org/3/library/fnmatch.html)
test = "".join("[%s]" % c if c in ('[', ']') else c for c in test)
# We add a '*' before test to eliminate the surprise of needing
# a '*' before the name of a test function.
test = "*" + test
if fnmatch(item_name, test):
if TestConfig.run_unsupported:
item.add_marker(pytest.mark.xfail(reason=reason))
else:
pytest.skip(reason)
def load_ipython_extension(ipython):
if IPython.version_info[0] >= 5:
warnings.warn(
"Loading the nengo.ipynb notebook extension is no longer "
"required. Progress bars are automatically activated for IPython "
"version 5 and later.")
elif has_ipynb_widgets() and rc.get('progress', 'progress_bar') == 'auto':
warnings.warn(
"The nengo.ipynb notebook extension is deprecated. Please upgrade "
"to IPython version 5 or later.")
IPythonProgressWidget.load_frontend(ipython)
rc.set('progress', 'progress_bar', '.'.join(
(__name__, IPython2ProgressBar.__name__)))
def load_ipython_extension(ipython):
if IPython.version_info[0] >= 5:
warnings.warn(
"Loading the nengo.ipynb notebook extension is no longer "
"required. Progress bars are automatically activated for IPython "
"version 5 and later.")
elif has_ipynb_widgets() and rc.get('progress', 'progress_bar') == 'auto':
warnings.warn(
"The nengo.ipynb notebook extension is deprecated. Please upgrade "
"to IPython version 5 or later.")
IPythonProgressWidget.load_frontend(ipython)
rc.set('progress', 'progress_bar', '.'.join((
__name__, IPython2ProgressBar.__name__)))
def pytest_runtest_setup(item): # noqa: C901
rc.reload_rc([])
rc.set('decoder_cache', 'enabled', 'False')
rc.set('exceptions', 'simplified', 'False')
if not hasattr(item, 'obj'):
return # Occurs for doctests, possibly other weird tests
conf = item.config
test_uses_compare = getattr(item.obj, 'compare', None) is not None
test_uses_sim = 'Simulator' in item.fixturenames
test_uses_refsim = 'RefSimulator' in item.fixturenames
tests_frontend = not (test_uses_sim or test_uses_refsim)
if getattr(item.obj, 'example', None) and not conf.getvalue('noexamples'):
pytest.skip("examples not requested")
elif getattr(item.obj, 'slow', None) and not conf.getvalue('slow'):
pytest.skip("slow tests not requested")
elif not TestConfig.compare_requested and test_uses_compare:
pytest.skip("compare tests not requested")
elif TestConfig.is_skipping_frontend_tests() and tests_frontend:
pytest.skip("frontend tests not run for alternate backends")
elif (TestConfig.is_skipping_frontend_tests()
and test_uses_refsim
and not TestConfig.is_refsim_overridden()):
pytest.skip("RefSimulator not overridden")
elif (TestConfig.is_skipping_frontend_tests()
and test_uses_sim
and not TestConfig.is_sim_overridden()):
pytest.skip("Simulator not overridden")
elif getattr(item.obj, 'noassertions', None):
options = []
for fixture, option in [('analytics', 'analytics'),
('plt', 'plots'),
('logger', 'logs')]:
if fixture in item.fixturenames and not conf.getvalue(option):
options.append(option)
if len(options) > 0:
pytest.skip("%s not requested" % " and ".join(options))
if not tests_frontend:
item_name = get_item_name(item)
for test, reason in TestConfig.Simulator.unsupported:
# We add a '*' before test to eliminate the surprise of needing
# a '*' before the name of a test function.
if fnmatch(item_name, '*' + test):
pytest.xfail(reason)
def pytest_runtest_setup(item): # noqa: C901
rc.reload_rc([])
rc.set('decoder_cache', 'enabled', 'False')
rc.set('exceptions', 'simplified', 'False')
if not hasattr(item, 'obj'):
return # Occurs for doctests, possibly other weird tests
conf = item.config
test_uses_compare = getattr(item.obj, 'compare', None) is not None
test_uses_sim = 'Simulator' in item.fixturenames
test_uses_refsim = 'RefSimulator' in item.fixturenames
tests_frontend = not (test_uses_sim or test_uses_refsim)
if getattr(item.obj, 'example', None) and not conf.getvalue('noexamples'):
pytest.skip("examples not requested")
elif getattr(item.obj, 'slow', None) and not conf.getvalue('slow'):
pytest.skip("slow tests not requested")
elif not TestConfig.compare_requested and test_uses_compare:
pytest.skip("compare tests not requested")
elif TestConfig.is_skipping_frontend_tests() and tests_frontend:
pytest.skip("frontend tests not run for alternate backends")
elif (TestConfig.is_skipping_frontend_tests()
and test_uses_refsim
and not TestConfig.is_refsim_overridden()):
pytest.skip("RefSimulator not overridden")
elif (TestConfig.is_skipping_frontend_tests()
and test_uses_sim
and not TestConfig.is_sim_overridden()):
pytest.skip("Simulator not overridden")
elif getattr(item.obj, 'noassertions', None):
options = []
for fixture, option in [('analytics', 'analytics'),
('plt', 'plots'),
('logger', 'logs')]:
if fixture in item.fixturenames and not conf.getvalue(option):
options.append(option)
if len(options) > 0:
pytest.skip("%s not requested" % " and ".join(options))
if not tests_frontend:
item_name = get_item_name(item)
for test, reason in TestConfig.Simulator.unsupported:
# We add a '*' before test to eliminate the surprise of needing
# a '*' before the name of a test function.
if fnmatch(item_name, '*' + test):
pytest.xfail(reason)
def pytest_runtest_setup(item): # noqa: C901
rc.reload_rc([])
rc.set('decoder_cache', 'enabled', 'False')
rc.set('exceptions', 'simplified', 'False')
if not hasattr(item, 'obj'):
return # Occurs for doctests, possibly other weird tests
test_uses_sim = 'Simulator' in item.fixturenames
test_uses_refsim = 'RefSimulator' in item.fixturenames
tests_frontend = not (test_uses_sim or test_uses_refsim)
if not tests_frontend:
item_name = get_item_name(item)
for test, reason in TestConfig.Simulator.unsupported:
# We add a '*' before test to eliminate the surprise of needing
# a '*' before the name of a test function.
if fnmatch(item_name, '*' + test):
pytest.xfail(reason)
def pytest_runtest_setup(item): # noqa: C901
rc.reload_rc([])
rc.set('decoder_cache', 'enabled', 'False')
rc.set('exceptions', 'simplified', 'False')
if not hasattr(item, 'obj'):
return
for mark, option, message in [
('example', 'noexamples', "examples not requested"),
('slow', 'slow', "slow tests not requested")
]:
if getattr(item.obj, mark, None) and not item.config.getvalue(option):
pytest.skip(message)
if getattr(item.obj, 'noassertions', None):
skipreasons = []
for fixture_name, option, message in [
('analytics', 'analytics', "analytics not requested"),
('plt', 'plots', "plots not requested"),
('logger', 'logs', "logs not requested")
]:
if fixture_name in item.fixturenames:
if item.config.getvalue(option):
break
else:
skipreasons.append(message)
else:
pytest.skip(" and ".join(skipreasons))
if 'Simulator' in item.fixturenames:
for test, reason in TestConfig.Simulator.unsupported:
# We add a '*' before test to eliminate the surprise of needing
# a '*' before the name of a test function.
if fnmatch(item.nodeid, '*' + test):
pytest.xfail(reason)
def test_validation_error(request):
# Ensure settings are set back to default after the test, even if it fails
request.addfinalizer(
lambda: rc.set("exceptions", "simplified",
str(RC_DEFAULTS["exceptions"]["simplified"])))
nengo.rc["exceptions"]["simplified"] = "False"
with nengo.Network():
with pytest.raises(ValidationError) as excinfo:
nengo.Ensemble(n_neurons=0, dimensions=1)
assert str(excinfo.value) == (
"Ensemble.n_neurons: Value must be greater than or equal to 1 (got 0)")
check_tb_entries(
excinfo.traceback,
[
("test_validation_error",
"nengo.Ensemble(n_neurons=0, dimensions=1)"),
("__call__", "inst.__init__(*args, **kwargs)"),
("__init__", "self.n_neurons = n_neurons"),
("__setattr__", "super().__setattr__(name, val)"),
("__setattr__", "super().__setattr__(name, val)"),
("__set__", "self.data[instance] = self.coerce(instance, value)"),
("coerce", "return super().coerce(instance, num)"),
("coerce", "raise ValidationError(..."),
],
)
nengo.rc["exceptions"]["simplified"] = "True"
with pytest.raises(ValidationError) as excinfo:
nengo.dists.PDF(x=[1, 1], p=[0.1, 0.2])
assert str(
excinfo.value) == "PDF.p: PDF must sum to one (sums to 0.300000)"
check_tb_entries(
excinfo.traceback,
[
("test_validation_error",
"nengo.dists.PDF(x=[1, 1], p=[0.1, 0.2])"),
("__init__", "raise ValidationError(..."),
],
)
def run(self, params):
self.decision_type = params[0]
self.drug_type = params[1]
self.drug = params[2]
self.trial = params[3]
self.seed = params[4]
self.P = params[5]
self.dt = self.P['dt']
self.dt_sample = self.P['dt_sample']
self.t_cue = self.P['t_cue']
self.t_delay = self.P['t_delay']
self.drug_effect_neural = self.P['drug_effect_neural']
self.drug_effect_functional = self.P['drug_effect_functional']
self.drug_effect_biophysical = self.P['drug_effect_biophysical']
self.enc_min_cutoff = self.P['enc_min_cutoff']
self.enc_max_cutoff = self.P['enc_max_cutoff']
self.sigma_smoothing = self.P['sigma_smoothing']
self.frac = self.P['frac']
self.neurons_inputs = self.P['neurons_inputs']
self.neurons_wm = self.P['neurons_wm']
self.neurons_decide = self.P['neurons_decide']
self.time_scale = self.P['time_scale']
self.cue_scale = self.P['cue_scale']
self.tau = self.P['tau']
self.tau_wm = self.P['tau_wm']
self.noise_wm = self.P['noise_wm']
self.noise_decision = self.P['noise_decision']
self.perceived = self.P['perceived']
self.cues = self.P['cues']
if self.drug_type == 'biophysical':
rc.set("decoder_cache", "enabled", "False") #don't try to remember old decoders
else:
rc.set("decoder_cache", "enabled", "True")
with nengo.Network(seed = self.net_seed) as model:
wm = nengo.Ensemble(self.neurons_wm, 2, neuron_type = self.neuron_type,seed = self.ens_seed, label = 'Working Memory')
self.wm_recurrent = nengo.Connection(wm, wm, synapse = self.tau_wm, function = self.wm_recurrent_function, seed = self.con_seed,
solver = nengo.solvers.LstsqL2(weights=True))
with nengo.Simulator(model,dt = self.dt, seed = self.sim_seed) as sim:
pass
weights = self.degrade_synaptic_connection(sim, wm)
with nengo.Network(seed = self.net_seed) as model:
cue = nengo.Node(output = self.cue_function, label = 'Cue')
time = nengo.Node(output = self.time_function, label = 'Time')
inputs = nengo.Ensemble(self.neurons_inputs, 2, seed = self.ens_seed, label = 'Input Neurons')
noise_wm_node = nengo.Node(output = self.noise_bias_function, label = 'Noise injection (WM node)')
noise_decision_node = nengo.Node(output = self.noise_decision_function, label = 'Noise injection (decision node)')
wm = nengo.Ensemble(self.neurons_wm, 2, neuron_type = self.neuron_type,seed = self.ens_seed, label = 'Working Memory')
cor = nengo.Ensemble(1, 1, neuron_type = nengo.Direct(),seed = self.ens_seed, label = 'Accuracy sensor')
dbs = nengo.Node(output = self.DBS_function,label = 'Deep Brain Stimulation Node', size_out=1)
if self.decision_type == 'default':
decision = nengo.Ensemble(self.neurons_decide, 2, seed = self.ens_seed, label = 'Decision Maker')
elif self.decision_type == 'basal_ganglia':
utilities = nengo.networks.EnsembleArray(self.neurons_inputs, n_ensembles = 2, seed = self.ens_seed, label = 'Utility network')
BasalGanglia = nengo.networks.BasalGanglia(2, self.neurons_decide)
decision = nengo.networks.EnsembleArray(self.neurons_decide, n_ensembles = 2, intercepts = Uniform(0.2, 1), encoders = Uniform(1,1), seed = self.ens_seed, label = 'Decision ensemble (Basal Ganglia')
temp = nengo.Ensemble(self.neurons_decide, 2,neuron_type = self.neuron_type, seed = self.ens_seed)
bias = nengo.Node([1] * 2, label = 'bias node')
output = nengo.Ensemble(self.neurons_decide, 1, neuron_type = self.neuron_type, seed = self.ens_seed, label = 'Output')
nengo.Connection(cue, inputs[0], synapse = None, seed = self.con_seed)
nengo.Connection(time, inputs[1], synapse = None, seed = self.con_seed)
nengo.Connection(inputs, wm, synapse = self.tau_wm, function=self.inputs_function, seed = self.con_seed)
self.wm_recurrent = nengo.Connection(wm.neurons, wm.neurons, synapse = self.tau_wm, seed = self.con_seed, transform = weights)
nengo.Connection(noise_wm_node, wm.neurons, synapse = self.tau_wm, transform = np.ones((self.neurons_wm,1)) * self.tau_wm, seed = self.con_seed)
if self.DBS:
nengo.Connection(dbs, wm.neurons, synapse = 0, seed = self.con_seed, transform = np.ones((self.neurons_wm,1)))
if self.decision_type == 'default':
wm_to_decision = nengo.Connection(wm[0], decision[0], synapse = self.tau, seed = self.con_seed)
nengo.Connection(noise_decision_node, decision[1], synapse = None, seed = self.con_seed)
nengo.Connection(decision, output,function = self.decision_function, seed = self.con_seed)
nengo.Connection(decision, cor, synapse = 0.025,function = self.f_dec, seed = self.con_seed)
elif self.decision_type == 'basal_ganglia':
wm_to_decision = nengo.Connection(wm[0], utilities.input, synapse = self.tau, function = self.BG_rescale, seed = self.con_seed)
nengo.Connection(BasalGanglia.output, decision.input, synapse = self.tau, seed = self.con_seed)
nengo.Connection(noise_decision_node, BasalGanglia.input,synapse = None, seed = self.con_seed) #added external noise?
nengo.Connection(bias, decision.input, synapse = self.tau, seed = self.con_seed)
nengo.Connection(decision.input, decision.output, transform=(np.eye(2)-1), synapse = self.tau/2.0, seed = self.con_seed)
nengo.Connection(decision.output,temp, seed = self.con_seed)
nengo.Connection(temp,output,function = self.decision_function, seed = self.con_seed, synapse = None)
nengo.Connection(temp, cor, synapse = 0.2,seed = self.con_seed, function = self.f_dec)
probe_wm = nengo.Probe(wm[0],synapse = 0.024, sample_every = self.dt_sample)
probe_spikes = nengo.Probe(wm.neurons, 'spikes', sample_every = self.dt_sample)
probe_output = nengo.Probe(output,synapse = None, sample_every = self.dt_sample)
p_cor = nengo.Probe(cor, synapse = None, sample_every = self.dt_sample)
#data_dir = ch_dir()
#CollapsingGexfConverter().convert(model).write('model.gexf')
print('Running trial %s...\n' %(self.trial+1))
with nengo.Simulator(model,dt = self.dt, seed = self.sim_seed) as sim:
if self.drug_type == 'biophysical':
sim = reset_gain_bias(self.P, model, sim, wm, self.wm_recurrent, wm_to_decision, self.drug)
sim.run(self.t_cue + self.t_delay)
xyz = sim.data[probe_spikes]
abc = np.abs(sim.data[p_cor])
print('Constructing Dataframes...')
df_primary = primary_dataframe(self.P, sim, self.drug,self.trial, probe_wm, probe_output, self.day)
df_firing = firing_dataframe(self.P,sim,self.drug,self.trial, sim.data[wm], probe_spikes, self.day)
return [df_primary, df_firing, abc, xyz]
def run(params):
import nengo
from nengo.dists import Choice,Exponential,Uniform
from nengo.rc import rc
import numpy as np
import pandas as pd
from helper import MySolver, reset_gain_bias, primary_dataframe, firing_dataframe, get_correct
decision_type=params[0]
drug_type=params[1]
drug = params[2]
trial = params[3]
seed = params[4]
P = params[5]
dt=P['dt']
dt_sample=P['dt_sample']
t_cue=P['t_cue']
t_delay=P['t_delay']
drug_effect_neural=P['drug_effect_neural']
drug_effect_functional=P['drug_effect_functional']
drug_effect_biophysical=P['drug_effect_biophysical']
enc_min_cutoff=P['enc_min_cutoff']
enc_max_cutoff=P['enc_max_cutoff']
sigma_smoothing=P['sigma_smoothing']
frac=P['frac']
neurons_inputs=P['neurons_inputs']
neurons_wm=P['neurons_wm']
neurons_decide=P['neurons_decide']
time_scale=P['time_scale']
cue_scale=P['cue_scale']
tau=P['tau']
tau_wm=P['tau_wm']
noise_wm=P['noise_wm']
noise_decision=P['noise_decision']
perceived=P['perceived']
cues=P['cues']
if drug_type == 'biophysical': rc.set("decoder_cache", "enabled", "False") #don't try to remember old decoders
else: rc.set("decoder_cache", "enabled", "True")
def cue_function(t):
if t < t_cue and perceived[trial]!=0:
return cue_scale * cues[trial]
else: return 0
def time_function(t):
if t > t_cue:
return time_scale
else: return 0
def noise_bias_function(t):
import numpy as np
if drug_type=='neural':
return np.random.normal(drug_effect_neural[drug],noise_wm)
else:
return np.random.normal(0.0,noise_wm)
def noise_decision_function(t):
import numpy as np
if decision_type == 'default':
return np.random.normal(0.0,noise_decision)
elif decision_type == 'basal_ganglia':
return np.random.normal(0.0,noise_decision,size=2)
def inputs_function(x):
return x * tau_wm
def wm_recurrent_function(x):
if drug_type == 'functional':
return x * drug_effect_functional[drug]
else:
return x
def decision_function(x):
output=0.0
if decision_type=='default':
value=x[0]+x[1]
if value > 0.0: output = 1.0
elif value < 0.0: output = -1.0
elif decision_type=='basal_ganglia':
if x[0] > x[1]: output = 1.0
elif x[0] < x[1]: output = -1.0
return output
def BG_rescale(x): #rescales -1 to 1 into 0.3 to 1, makes 2-dimensional
pos_x = 0.5 * (x + 1)
rescaled = 0.4 + 0.6 * pos_x, 0.4 + 0.6 * (1 - pos_x)
return rescaled
'''model definition'''
with nengo.Network(seed=seed+trial) as model:
#Ensembles
cue = nengo.Node(output=cue_function)
time = nengo.Node(output=time_function)
inputs = nengo.Ensemble(neurons_inputs,2)
noise_wm_node = nengo.Node(output=noise_bias_function)
noise_decision_node = nengo.Node(output=noise_decision_function)
wm = nengo.Ensemble(neurons_wm,2)
if decision_type=='default':
decision = nengo.Ensemble(neurons_decide,2)
elif decision_type=='basal_ganglia':
utilities = nengo.networks.EnsembleArray(neurons_inputs,n_ensembles=2)
BG = nengo.networks.BasalGanglia(2,neurons_decide)
decision = nengo.networks.EnsembleArray(neurons_decide,n_ensembles=2,
intercepts=Uniform(0.2,1),encoders=Uniform(1,1))
temp = nengo.Ensemble(neurons_decide,2)
bias = nengo.Node([1]*2)
output = nengo.Ensemble(neurons_decide,1)
#Connections
nengo.Connection(cue,inputs[0],synapse=None)
nengo.Connection(time,inputs[1],synapse=None)
nengo.Connection(inputs,wm,synapse=tau_wm,function=inputs_function)
wm_recurrent=nengo.Connection(wm,wm,synapse=tau_wm,function=wm_recurrent_function)
nengo.Connection(noise_wm_node,wm.neurons,synapse=tau_wm,transform=np.ones((neurons_wm,1))*tau_wm)
if decision_type=='default':
wm_to_decision=nengo.Connection(wm[0],decision[0],synapse=tau)
nengo.Connection(noise_decision_node,decision[1],synapse=None)
nengo.Connection(decision,output,function=decision_function)
elif decision_type=='basal_ganglia':
wm_to_decision=nengo.Connection(wm[0],utilities.input,synapse=tau,function=BG_rescale)
nengo.Connection(utilities.output,BG.input,synapse=None)
nengo.Connection(BG.output,decision.input,synapse=tau)
nengo.Connection(noise_decision_node,BG.input,synapse=None) #added external noise?
nengo.Connection(bias,decision.input,synapse=tau)
nengo.Connection(decision.input,decision.output,transform=(np.eye(2)-1),synapse=tau/2.0)
nengo.Connection(decision.output,temp)
nengo.Connection(temp,output,function=decision_function)
#Probes
probe_wm=nengo.Probe(wm[0],synapse=0.01,sample_every=dt_sample)
probe_spikes=nengo.Probe(wm.neurons, 'spikes', sample_every=dt_sample)
probe_output=nengo.Probe(output,synapse=None,sample_every=dt_sample)
'''SIMULATION'''
print 'Running drug \"%s\", trial %s...' %(drug,trial+1)
with nengo.Simulator(model,dt=dt) as sim:
if drug_type == 'biophysical': sim=reset_gain_bias(
P,model,sim,wm,wm_recurrent,wm_to_decision,drug)
sim.run(t_cue+t_delay)
df_primary=primary_dataframe(P,sim,drug,trial,probe_wm,probe_output)
df_firing=firing_dataframe(P,sim,drug,trial,sim.data[wm],probe_spikes)
return [df_primary, df_firing]
class TestCacheBenchmark(object):
n_trials = 25
setup = '''
import numpy as np
import nengo
import nengo.cache
from nengo.rc import rc
model = nengo.Network(seed=1)
with model:
a = nengo.Ensemble({N}, dimensions={D}, n_eval_points={M})
b = nengo.Ensemble({N}, dimensions={D}, n_eval_points={M})
conn = nengo.Connection(a, b)
'''
without_cache = {
'rc': 'rc.set("decoder_cache", "enabled", "False")',
'stmt': 'sim = nengo.Simulator(model)'
}
with_cache_miss_ro = {
'rc': '''
rc.set("decoder_cache", "enabled", "True")
rc.set("decoder_cache", "readonly", "True")
''',
'stmt': '''
nengo.cache.DecoderCache().invalidate()
sim = nengo.Simulator(model)
'''
}
with_cache_miss = {
'rc': '''
rc.set("decoder_cache", "enabled", "True")
rc.set("decoder_cache", "readonly", "False")
''',
'stmt': '''
nengo.cache.DecoderCache().invalidate()
sim = nengo.Simulator(model)
'''
}
with_cache_hit = {
'rc': '''
rc.set("decoder_cache", "enabled", "True")
rc.set("decoder_cache", "readonly", "False")
sim = nengo.Simulator(model)
''',
'stmt': 'sim = nengo.Simulator(model)'
}
labels = ["no cache", "cache miss", "cache miss ro", "cache hit"]
keys = [l.replace(' ', '_') for l in labels]
param_to_axis_label = {
'D': "dimensions",
'N': "neurons",
'M': "evaluation points"
}
defaults = {'D': 1, 'N': 50, 'M': 1000}
def time_code(self, code, args):
return timeit.repeat(
stmt=code['stmt'], setup=self.setup.format(**args) + code['rc'],
number=1, repeat=self.n_trials)
def time_all(self, args):
return (self.time_code(self.without_cache, args),
self.time_code(self.with_cache_miss, args),
self.time_code(self.with_cache_miss_ro, args),
self.time_code(self.with_cache_hit, args))
def get_args(self, varying_param, value):
args = dict(self.defaults) # make a copy
args[varying_param] = value
return args
@pytest.mark.slow
@pytest.mark.noassertions
@pytest.mark.parametrize('varying_param', ['D', 'N', 'M'])
def test_cache_benchmark(self, varying_param, analytics, plt):
varying = {
'D': np.asarray(np.linspace(1, 512, 10), dtype=int),
'N': np.asarray(np.linspace(10, 500, 8), dtype=int),
'M': np.asarray(np.linspace(750, 2500, 8), dtype=int)
}[varying_param]
axis_label = self.param_to_axis_label[varying_param]
times = [self.time_all(self.get_args(varying_param, v))
for v in varying]
# Restore RC to original settings
default = RC_DEFAULTS['decoder_cache', 'enabled']
rc.set("decoder_cache", "enabled", str(default))
default = RC_DEFAULTS['decoder_cache', 'readonly']
rc.set("decoder_cache", "readonly", str(default))
for i, data in enumerate(zip(*times)):
plt.plot(varying, np.median(data, axis=1), label=self.labels[i])
analytics.add_data(varying_param, varying, axis_label)
analytics.add_data(self.keys[i], data)
plt.xlabel("Number of %s" % axis_label)
plt.ylabel("Build time (s)")
plt.legend(loc='best')
@staticmethod
def reject_outliers(data):
med = np.median(data)
limits = 1.5 * (np.percentile(data, [25, 75]) - med) + med
return data[np.logical_and(data > limits[0], data < limits[1])]
@pytest.mark.compare
@pytest.mark.parametrize('varying_param', ['D', 'N', 'M'])
def test_compare_cache_benchmark(self, varying_param, analytics_data, plt):
stats = pytest.importorskip('scipy.stats')
d1, d2 = analytics_data
assert np.all(d1[varying_param] == d2[varying_param]), (
'Cannot compare different parametrizations')
axis_label = self.param_to_axis_label[varying_param]
print("Cache, varying {0}:".format(axis_label))
for label, key in zip(self.labels, self.keys):
clean_d1 = [self.reject_outliers(d) for d in d1[key]]
clean_d2 = [self.reject_outliers(d) for d in d2[key]]
diff = [np.median(b) - np.median(a)
for a, b in zip(clean_d1, clean_d2)]
p_values = np.array([2. * stats.mannwhitneyu(a, b)[1]
for a, b in zip(clean_d1, clean_d2)])
overall_p = 1. - np.prod(1. - p_values)
if overall_p < .05:
print(" {label}: Significant change (p <= {p:.3f}). See plots"
" for details.".format(
label=label, p=np.ceil(overall_p * 1000.) / 1000.))
else:
print(" {label}: No significant change.".format(label=label))
plt.plot(d1[varying_param], diff, label=label)
plt.xlabel("Number of %s" % axis_label)
plt.ylabel("Difference in build time (s)")
plt.legend(loc='best')
def load_ipython_extension(ipython):
if has_ipynb_widgets() and rc.get('progress', 'progress_bar') == 'auto':
IPythonProgressWidget.load_frontend(ipython)
rc.set('progress', 'progress_bar', '.'.join(
(__name__, IPython2ProgressBar.__name__)))
def load_ipython_extension(ipython):
if has_ipynb_widgets() and rc.get('progress', 'progress_bar') == 'auto':
IPythonProgressWidget.load_frontend(ipython)
rc.set('progress', 'progress_bar', '.'.join((
__name__, IPython2ProgressBar.__name__)))
def pytest_configure(config):
rc.reload_rc([])
rc.set('decoder_cache', 'enabled', 'false')
def pytest_configure(config):
rc.reload_rc([])
rc.set('decoder_cache', 'enabled', 'false')
def pytest_configure(config):
rc.reload_rc([])
rc.set("decoder_cache", "enabled", "false")
def pytest_runtest_setup(item):
rc.reload_rc([])
rc.set("decoder_cache", "enabled", "False")
rc.set("exceptions", "simplified", "False")
rc.set("nengo.Simulator", "fail_fast", "True")