def test_solvers(Simulator, nl_nodirect):
N = 100
decoder_solvers = [
Lstsq(), LstsqNoise(),
LstsqL2(), LstsqL2nz(),
LstsqL1()
]
weight_solvers = [LstsqL1(weights=True), LstsqDrop(weights=True)]
dt = 1e-3
tfinal = 4
def input_function(t):
return np.interp(t, [1, 3], [-1, 1], left=-1, right=1)
model = nengo.Network('test_solvers', seed=290)
with model:
u = nengo.Node(output=input_function)
a = nengo.Ensemble(nl_nodirect(N), dimensions=1)
nengo.Connection(u, a)
ap = nengo.Probe(a)
probes = []
names = []
for solver in decoder_solvers + weight_solvers:
b = nengo.Ensemble(nl_nodirect(N), dimensions=1, seed=99)
nengo.Connection(a, b, solver=solver)
probes.append(nengo.Probe(b))
names.append(
"%s(%s)" %
(solver.__class__.__name__, 'w' if solver.weights else 'd'))
sim = Simulator(model, dt=dt)
sim.run(tfinal)
t = sim.trange()
# ref = sim.data[up]
ref = filtfilt(sim.data[ap], 20)
outputs = np.array([sim.data[probe] for probe in probes])
outputs_f = filtfilt(outputs, 20, axis=1)
close = allclose(t,
ref,
outputs_f,
plotter=Plotter(Simulator, nl_nodirect),
filename='test_decoders.test_solvers.pdf',
labels=names,
atol=0.05,
rtol=0,
buf=100,
delay=7)
for name, c in zip(names, close):
assert c, "Solver '%s' does not meet tolerances" % name
def test_weights(Simulator, nl):
name = 'test_weights'
n1, n2 = 100, 50
def func(t):
return np.array([np.sin(4 * t), np.cos(12 * t)])
transform = np.array([[0.6, -0.4]])
m = nengo.Network(label=name, seed=3902)
with m:
m.config[nengo.Ensemble].neuron_type = nl()
u = nengo.Node(output=func)
a = nengo.Ensemble(n1, dimensions=2, radius=1.5)
b = nengo.Ensemble(n2, dimensions=1)
bp = nengo.Probe(b)
nengo.Connection(u, a)
nengo.Connection(a, b, transform=transform,
solver=LstsqL2(weights=True))
sim = Simulator(m)
sim.run(2.)
t = sim.trange()
x = func(t).T
y = np.dot(x, transform.T)
z = filtfilt(sim.data[bp], 10, axis=0)
assert allclose(t, y.flatten(), z.flatten(),
plotter=Plotter(Simulator, nl),
filename='test_connection.' + name + '.pdf',
atol=0.1, rtol=0, buf=100, delay=10)
def test_weights(Simulator, nl):
name = 'test_weights'
n1, n2 = 100, 50
def func(t):
return np.array([np.sin(4 * t), np.cos(12 * t)])
transform = np.array([[0.6, -0.4]])
m = nengo.Model(name, seed=3902)
u = nengo.Node(output=func)
a = nengo.Ensemble(nl(n1), dimensions=2, radius=1.5)
b = nengo.Ensemble(nl(n2), dimensions=1)
bp = nengo.Probe(b)
nengo.Connection(u, a)
nengo.Connection(a, b, transform=transform,
weight_solver=nengo.decoders.lstsq_L2nz)
sim = Simulator(m)
sim.run(2.)
t = sim.trange()
x = func(t).T
y = np.dot(x, transform.T)
z = filtfilt(sim.data[bp], 10, axis=0)
assert allclose(t, y.flatten(), z.flatten(),
plotter=Plotter(Simulator, nl),
filename='test_connection.' + name + '.pdf',
atol=0.1, rtol=0, buf=100, delay=10)
def test_compare_solvers(Simulator, nl_nodirect):
N = 70
decoder_solvers = [
Lstsq(), LstsqNoise(), LstsqL2(), LstsqL2nz(), LstsqL1()]
weight_solvers = [LstsqL1(weights=True), LstsqDrop(weights=True)]
dt = 1e-3
tfinal = 4
def input_function(t):
return np.interp(t, [1, 3], [-1, 1], left=-1, right=1)
model = nengo.Network('test_solvers', seed=290)
with model:
model.config[nengo.Ensemble].neuron_type = nl_nodirect()
u = nengo.Node(output=input_function)
a = nengo.Ensemble(N, dimensions=1)
nengo.Connection(u, a)
ap = nengo.Probe(a)
probes = []
names = []
for solver in decoder_solvers + weight_solvers:
b = nengo.Ensemble(N, dimensions=1, seed=99)
nengo.Connection(a, b, solver=solver)
probes.append(nengo.Probe(b))
names.append("%s(%s)" % (
solver.__class__.__name__, 'w' if solver.weights else 'd'))
sim = Simulator(model, dt=dt)
sim.run(tfinal)
t = sim.trange()
# ref = sim.data[up]
ref = filtfilt(sim.data[ap], 20)
outputs = np.array([sim.data[probe][:, 0] for probe in probes])
outputs_f = filtfilt(outputs, 20, axis=1)
close = allclose(t, ref, outputs_f,
plotter=Plotter(Simulator, nl_nodirect),
filename='test_solvers.test_solvers.pdf',
labels=names,
atol=0.05, rtol=0, buf=100, delay=7)
for name, c in zip(names, close):
assert c, "Solver '%s' does not meet tolerances" % name
def test_solvers(Simulator, nl_nodirect):
N = 100
decoder_solvers = [lstsq, lstsq_noise, lstsq_L2, lstsq_L2nz, lstsq_L1]
weight_solvers = [lstsq_L1, lstsq_drop]
dt = 1e-3
tfinal = 4
def input_function(t):
return np.interp(t, [1, 3], [-1, 1], left=-1, right=1)
model = nengo.Network('test_solvers', seed=290)
with model:
u = nengo.Node(output=input_function)
# up = nengo.Probe(u)
a = nengo.Ensemble(nl_nodirect(N), dimensions=1)
ap = nengo.Probe(a)
nengo.Connection(u, a)
probes = []
names = []
for solver, arg in ([(s, 'decoder_solver') for s in decoder_solvers] +
[(s, 'weight_solver') for s in weight_solvers]):
b = nengo.Ensemble(nl_nodirect(N), dimensions=1, seed=99)
nengo.Connection(a, b, **{arg: solver})
probes.append(nengo.Probe(b))
names.append(solver.__name__ + (" (%s)" % arg[0]))
sim = nengo.Simulator(model, dt=dt)
sim.run(tfinal)
t = sim.trange()
# ref = sim.data[up]
ref = filtfilt(sim.data[ap], 20)
outputs = np.array([sim.data[probe] for probe in probes])
outputs_f = filtfilt(outputs, 20, axis=1)
close = allclose(t, ref, outputs_f,
plotter=Plotter(Simulator, nl_nodirect),
filename='test_decoders.test_solvers.pdf',
labels=names,
atol=0.05, rtol=0, buf=100, delay=7)
for name, c in zip(names, close):
assert c, "Solver '%s' does not meet tolerances" % name
def test_filtfilt():
dt = 1e-3
tend = 3.
t = dt * np.arange(tend / dt)
nt = len(t)
tau = 0.03 / dt
u = np.random.normal(size=nt)
x = filt(u, tau)
x = filt(x[::-1], tau, x0=x[-1])[::-1]
y = filtfilt(u, tau)
with Plotter(nengo.Simulator) as plt:
plt.plot(t, x)
plt.plot(t, y, '--')
plt.savefig('utils.test_filtering.test_filtfilt.pdf')
plt.close()
assert np.allclose(x, y)
def test_weights(Simulator, nl):
name = 'test_weights'
n1, n2 = 100, 50
def func(t):
return np.array([np.sin(4 * t), np.cos(12 * t)])
transform = np.array([[0.6, -0.4]])
m = nengo.Network(label=name, seed=3902)
with m:
m.config[nengo.Ensemble].neuron_type = nl()
u = nengo.Node(output=func)
a = nengo.Ensemble(n1, dimensions=2, radius=1.5)
b = nengo.Ensemble(n2, dimensions=1)
bp = nengo.Probe(b)
nengo.Connection(u, a)
nengo.Connection(a,
b,
transform=transform,
solver=LstsqL2(weights=True))
sim = Simulator(m)
sim.run(2.)
t = sim.trange()
x = func(t).T
y = np.dot(x, transform.T)
z = filtfilt(sim.data[bp], 10, axis=0)
assert allclose(t,
y.flatten(),
z.flatten(),
plotter=Plotter(Simulator, nl),
filename='test_connection.' + name + '.pdf',
atol=0.1,
rtol=0,
buf=100,
delay=10)
def test_eval_points(Simulator, nl_nodirect):
rng = np.random.RandomState(0)
n = 100
d = 5
filter = 0.08
dt = 1e-3
eval_points = np.logspace(np.log10(300), np.log10(5000), 11)
eval_points = np.round(eval_points).astype('int')
max_points = eval_points.max()
n_trials = 1
rmses = np.nan * np.zeros((len(eval_points), n_trials))
for j in range(n_trials):
points = rng.normal(size=(max_points, d))
points *= (rng.uniform(size=max_points)
/ norm(points, axis=-1))[:, None]
rng_j = np.random.RandomState(348 + j)
seed = 903824 + j
# generate random input in unit hypersphere
x = rng_j.normal(size=d)
x *= rng_j.uniform() / norm(x)
for i, n_points in enumerate(eval_points):
model = nengo.Network(
'test_eval_points(%d,%d)' % (i, j), seed=seed)
with model:
u = nengo.Node(output=x)
a = nengo.Ensemble(nl_nodirect(n * d), d,
eval_points=points[:n_points])
nengo.Connection(u, a, synapse=0)
up = nengo.Probe(u)
ap = nengo.Probe(a)
with Timer() as timer:
sim = Simulator(model, dt=dt)
sim.run(10 * filter)
t = sim.trange()
xt = filtfilt(sim.data[up], filter / dt)
yt = filtfilt(sim.data[ap], filter / dt)
t0 = 5 * filter
t1 = 7 * filter
tmask = (t > t0) & (t < t1)
rmses[i, j] = rms(yt[tmask] - xt[tmask])
print("done %d (%d) in %0.3f s" % (n_points, j, timer.duration))
# subtract out mean for each model
rmses_norm = rmses - rmses.mean(0, keepdims=True)
with Plotter(Simulator, nl_nodirect) as plt:
mean = rmses_norm.mean(1)
low = rmses_norm.min(1)
high = rmses_norm.max(1)
plt.semilogx(eval_points, mean, 'k-')
plt.semilogx(eval_points, high, 'r-')
plt.semilogx(eval_points, low, 'b-')
plt.xlim([eval_points[0], eval_points[-1]])
plt.xticks(eval_points, eval_points)
plt.savefig('test_decoders.test_eval_points.pdf')
plt.close()
def test_eval_points(Simulator, nl_nodirect):
rng = np.random.RandomState(0)
n = 100
d = 5
filter = 0.08
dt = 1e-3
eval_points = np.logspace(np.log10(300), np.log10(5000), 11)
eval_points = np.round(eval_points).astype('int')
max_points = eval_points.max()
n_trials = 1
rmses = np.nan * np.zeros((len(eval_points), n_trials))
for j in range(n_trials):
points = rng.normal(size=(max_points, d))
points *= (rng.uniform(size=max_points)
/ norm(points, axis=-1))[:, None]
rng_j = np.random.RandomState(348 + j)
seed = 903824 + j
# generate random input in unit hypersphere
x = rng_j.normal(size=d)
x *= rng_j.uniform() / norm(x)
for i, n_points in enumerate(eval_points):
model = nengo.Network(
'test_eval_points(%d,%d)' % (i, j), seed=seed)
with model:
u = nengo.Node(output=x)
a = nengo.Ensemble(nl_nodirect(n * d), d,
eval_points=points[:n_points])
nengo.Connection(u, a, synapse=0)
up = nengo.Probe(u)
ap = nengo.Probe(a)
with Timer() as timer:
sim = Simulator(model, dt=dt)
sim.run(10 * filter)
t = sim.trange()
xt = filtfilt(sim.data[up], filter / dt)
yt = filtfilt(sim.data[ap], filter / dt)
t0 = 5 * filter
t1 = 7 * filter
tmask = (t > t0) & (t < t1)
rmses[i, j] = rms(yt[tmask] - xt[tmask])
print("done %d (%d) in %0.3f s" % (n_points, j, timer.duration))
# subtract out mean for each model
rmses_norm = rmses - rmses.mean(0, keepdims=True)
with Plotter(Simulator, nl_nodirect) as plt:
mean = rmses_norm.mean(1)
low = rmses_norm.min(1)
high = rmses_norm.max(1)
plt.semilogx(eval_points, mean, 'k-')
plt.semilogx(eval_points, high, 'r-')
plt.semilogx(eval_points, low, 'b-')
plt.xlim([eval_points[0], eval_points[-1]])
plt.xticks(eval_points, eval_points)
plt.savefig('test_decoders.test_eval_points.pdf')
plt.close()
