def coerce(self, conn, transform):
if transform is None:
transform = NoTransform(conn.size_mid)
elif is_array_like(transform) or isinstance(transform, Distribution):
transform = Dense((conn.size_out, conn.size_mid), transform)
if transform.size_in != conn.size_mid:
if isinstance(transform, Dense) and (
transform.shape[0] == transform.shape[1]
):
# we provide a different error message in this case;
# the transform is not changing the dimensionality of the
# signal, so the blame most likely lies with the function
raise ValidationError(
"Function output size is incorrect; should return a "
"vector of size %d" % conn.size_mid,
attr=self.name,
obj=conn,
)
else:
raise ValidationError(
"Transform input size (%d) not equal to %s output size "
"(%d)"
% (transform.size_in, type(conn.pre_obj).__name__, conn.size_mid),
attr=self.name,
obj=conn,
)
if transform.size_out != conn.size_out:
raise ValidationError(
"Transform output size (%d) not equal to connection "
"output size (%d)" % (transform.size_out, conn.size_out),
attr=self.name,
obj=conn,
)
# we don't support repeated indices on 2D transforms because it makes
# the matrix multiplication more complicated (we'd need to expand
# the weight matrix for the duplicated rows/columns). it could be done
# if there were a demand at some point.
if isinstance(transform, Dense) and len(transform.init_shape) == 2:
def repeated_inds(x):
return not isinstance(x, slice) and np.unique(x).size != len(x)
if repeated_inds(conn.pre_slice):
raise ValidationError(
"Input object selection has repeated indices",
attr=self.name,
obj=conn,
)
if repeated_inds(conn.post_slice):
raise ValidationError(
"Output object selection has repeated indices",
attr=self.name,
obj=conn,
)
return super().coerce(conn, transform)
def build_no_transform(
model, transform, sig_in, decoders=None, encoders=None, rng=np.random
):
"""Build a `.NoTransform` transform object."""
if decoders is not None or encoders is not None:
return build_dense(
model,
Dense(shape=(transform.size_out, transform.size_in), init=1.0),
sig_in,
decoders=decoders,
encoders=encoders,
rng=rng,
)
return sig_in, None
def build_connection(model, conn):
"""Builds a `.Connection` object into a model.
A brief summary of what happens in the connection build process,
in order:
1. Solve for decoders.
2. Combine transform matrix with decoders to get weights.
3. Add operators for computing the function
or multiplying neural activity by weights.
4. Call build function for the synapse.
5. Call build function for the learning rule.
6. Add operator for applying learning rule delta to weights.
Some of these steps may be altered or omitted depending on the parameters
of the connection, in particular the pre and post types.
Parameters
----------
model : Model
The model to build into.
conn : Connection
The connection to build.
Notes
-----
Sets ``model.params[conn]`` to a `.BuiltConnection` instance.
"""
# Create random number generator
rng = np.random.RandomState(model.seeds[conn])
# Get input and output connections from pre and post
def get_prepost_signal(is_pre):
target = conn.pre_obj if is_pre else conn.post_obj
key = "out" if is_pre else "in"
if target not in model.sig:
raise BuildError("Building %s: the %r object %s is not in the "
"model, or has a size of zero." %
(conn, "pre" if is_pre else "post", target))
signal = model.sig[target].get(key, None)
if signal is None or signal.size == 0:
raise BuildError(
"Building %s: the %r object %s has a %r size of zero." %
(conn, "pre" if is_pre else "post", target, key))
return signal
model.sig[conn]["in"] = get_prepost_signal(is_pre=True)
model.sig[conn]["out"] = get_prepost_signal(is_pre=False)
decoders = None
encoders = None
eval_points = None
solver_info = None
post_slice = conn.post_slice
# Figure out the signal going across this connection
in_signal = model.sig[conn]["in"]
if isinstance(conn.pre_obj,
Node) or (isinstance(conn.pre_obj, Ensemble)
and isinstance(conn.pre_obj.neuron_type, Direct)):
# Node or Decoded connection in directmode
sliced_in = slice_signal(model, in_signal, conn.pre_slice)
if conn.function is None:
in_signal = sliced_in
elif isinstance(conn.function, np.ndarray):
raise BuildError("Cannot use function points in direct connection")
else:
in_signal = Signal(shape=conn.size_mid, name="%s.func" % conn)
model.add_op(SimPyFunc(in_signal, conn.function, None, sliced_in))
elif isinstance(conn.pre_obj, Ensemble): # Normal decoded connection
eval_points, decoders, solver_info = model.build(
conn.solver, conn, rng)
if isinstance(conn.post_obj, Ensemble) and conn.solver.weights:
model.sig[conn]["out"] = model.sig[conn.post_obj.neurons]["in"]
encoders = model.params[conn.post_obj].scaled_encoders.T
encoders = encoders[conn.post_slice]
# post slice already applied to encoders (either here or in
# `build_decoders`), so don't apply later
post_slice = None
else:
in_signal = slice_signal(model, in_signal, conn.pre_slice)
# Build transform
if conn.solver.weights and not conn.solver.compositional:
# special case for non-compositional weight solvers, where
# the solver is solving for the full weight matrix. so we don't
# need to combine decoders/transform/encoders.
weighted, weights = model.build(Dense(decoders.shape, init=decoders),
in_signal,
rng=rng)
else:
weighted, weights = model.build(conn.transform,
in_signal,
decoders=decoders,
encoders=encoders,
rng=rng)
model.sig[conn]["weights"] = weights
# Build synapse
if conn.synapse is not None:
weighted = model.build(conn.synapse, weighted, mode="update")
# Store the weighted-filtered output in case we want to probe it
model.sig[conn]["weighted"] = weighted
if isinstance(conn.post_obj, Neurons):
# Apply neuron gains (we don't need to do this if we're connecting to
# an Ensemble, because the gains are rolled into the encoders)
gains = Signal(
model.params[conn.post_obj.ensemble].gain[post_slice],
name="%s.gains" % conn,
)
if is_integer(post_slice) or isinstance(post_slice, slice):
sliced_out = model.sig[conn]["out"][post_slice]
else:
# advanced indexing not supported on Signals, so we need to set up an
# intermediate signal and use a Copy op to perform the indexing
sliced_out = Signal(shape=gains.shape, name="%s.sliced_out" % conn)
model.add_op(Reset(sliced_out))
model.add_op(
Copy(sliced_out,
model.sig[conn]["out"],
dst_slice=post_slice,
inc=True))
model.add_op(
ElementwiseInc(gains,
weighted,
sliced_out,
tag="%s.gains_elementwiseinc" % conn))
else:
# Copy to the proper slice
model.add_op(
Copy(
weighted,
model.sig[conn]["out"],
dst_slice=post_slice,
inc=True,
tag="%s" % conn,
))
# Build learning rules
if conn.learning_rule is not None:
# TODO: provide a general way for transforms to expose learnable params
if not isinstance(conn.transform, (Dense, NoTransform)):
raise NotImplementedError(
"Learning on connections with %s transforms is not supported" %
(type(conn.transform).__name__, ))
rule = conn.learning_rule
rule = [rule] if not is_iterable(rule) else rule
targets = []
for r in rule.values() if isinstance(rule, dict) else rule:
model.build(r)
targets.append(r.modifies)
if "encoders" in targets:
encoder_sig = model.sig[conn.post_obj]["encoders"]
encoder_sig.readonly = False
if "decoders" in targets or "weights" in targets:
if weights.ndim < 2:
raise BuildError(
"'transform' must be a 2-dimensional array for learning")
model.sig[conn]["weights"].readonly = False
model.params[conn] = BuiltConnection(
eval_points=eval_points,
solver_info=solver_info,
transform=conn.transform,
weights=getattr(weights, "initial_value", None),
)
def test_dimensionality_errors(nl_nodirect, seed, rng):
N = 10
with nengo.Network(seed=seed) as m:
m.config[nengo.Ensemble].neuron_type = nl_nodirect()
n01 = nengo.Node(output=[1])
n02 = nengo.Node(output=[1, 1])
n21 = nengo.Node(output=lambda t, x: [1], size_in=2)
e1 = nengo.Ensemble(N, 1)
e2 = nengo.Ensemble(N, 2)
# these should work
nengo.Connection(n01, e1)
nengo.Connection(n02, e2)
nengo.Connection(e2, n21)
nengo.Connection(n21, e1)
nengo.Connection(e1.neurons, n21, transform=rng.randn(2, N))
nengo.Connection(e2, e1, function=lambda x: x[0])
nengo.Connection(e2, e2, transform=np.ones(2))
# these should not work
with pytest.raises(ValidationError, match="Shape of initial value"):
nengo.Connection(n02, e1)
with pytest.raises(ValidationError, match="Shape of initial value"):
nengo.Connection(e1, e2)
with pytest.raises(ValidationError, match="Transform input size"):
nengo.Connection(e2.neurons,
e1,
transform=Dense((1, N + 1), init=Choice([1.0])))
with pytest.raises(ValidationError, match="Transform output size"):
nengo.Connection(e2.neurons,
e1,
transform=Dense((2, N), init=Choice([1.0])))
with pytest.raises(ValidationError, match="Function output size"):
nengo.Connection(e2,
e1,
function=lambda x: x,
transform=Dense((1, 1)))
with pytest.raises(ValidationError,
match="function.*must accept a single"):
nengo.Connection(e2,
e1,
function=lambda: 0,
transform=Dense((1, 1)))
with pytest.raises(ValidationError, match="Function output size"):
nengo.Connection(n21, e2, transform=Dense((2, 2)))
with pytest.raises(ValidationError, match="Shape of initial value"):
nengo.Connection(e2, e2, transform=np.ones((2, 2, 2)))
with pytest.raises(ValidationError, match="Function output size"):
nengo.Connection(e1, e2, transform=Dense((3, 3), init=np.ones(3)))
# these should not work because of indexing mismatches
with pytest.raises(ValidationError, match="Function output size"):
nengo.Connection(n02[0], e2, transform=Dense((2, 2)))
with pytest.raises(ValidationError, match="Transform output size"):
nengo.Connection(n02, e2[0], transform=Dense((2, 2)))
with pytest.raises(ValidationError, match="Function output size"):
nengo.Connection(n02[1], e2[0], transform=Dense((2, 2)))
with pytest.raises(ValidationError, match="Transform input size"):
nengo.Connection(n02,
e2[0],
transform=Dense((2, 1), init=Choice([1.0])))
with pytest.raises(ValidationError, match="Transform input size"):
nengo.Connection(e2[0],
e2,
transform=Dense((1, 2), init=Choice([1.0])))
# these should not work because of repeated indices
dense22 = Dense((2, 2), init=np.ones((2, 2)))
with pytest.raises(ValidationError, match="Input.*repeated indices"):
nengo.Connection(n02[[0, 0]], e2, transform=dense22)
with pytest.raises(ValidationError, match="Output.*repeated indices"):
nengo.Connection(e2, e2[[1, 1]], transform=dense22)