# Copyright (c) 2017 The University of Manchester
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import numpy
from spinn_utilities.overrides import overrides
from .abstract_connector import AbstractConnector
from .abstract_generate_connector_on_host import (
AbstractGenerateConnectorOnHost)
class ArrayConnector(AbstractConnector, AbstractGenerateConnectorOnHost):
"""
Make connections using an array of integers based on the IDs
of the neurons in the pre- and post-populations.
"""
__slots__ = [
"__array", "__array_dims", "__n_total_connections"]
def __init__(self, array, safe=True, callback=None, verbose=False):
"""
:param array:
An explicit boolean matrix that specifies the connections
between the pre- and post-populations
(see PyNN documentation). Must be 2D in practice.
:type array: ~numpy.ndarray(2, ~numpy.uint8)
:param bool safe:
Whether to check that weights and delays have valid values.
If False, this check is skipped.
:param callable callback:
if given, a callable that display a progress bar on the terminal.
.. note::
Not supported by sPyNNaker.
:param bool verbose:
Whether to output extra information about the connectivity to a
CSV file
"""
super().__init__(safe, callback, verbose)
self.__array = array
# we can get the total number of connections straight away
# from the boolean matrix
n_total_connections = 0
# array shape
dims = array.shape
for i in range(dims[0]):
for j in range(dims[1]):
if array[i, j] == 1:
n_total_connections += 1
self.__n_total_connections = n_total_connections
self.__array_dims = dims
[docs]
@overrides(AbstractConnector.get_delay_maximum)
def get_delay_maximum(self, synapse_info):
return self._get_delay_maximum(
synapse_info.delays, len(self.__array), synapse_info)
[docs]
@overrides(AbstractConnector.get_delay_minimum)
def get_delay_minimum(self, synapse_info):
return self._get_delay_minimum(
synapse_info.delays, len(self.__array), synapse_info)
[docs]
@overrides(AbstractConnector.get_n_connections_from_pre_vertex_maximum)
def get_n_connections_from_pre_vertex_maximum(
self, n_post_atoms, synapse_info, min_delay=None,
max_delay=None):
# Break the array into n_post_atoms units
split_positions = numpy.arange(
0, synapse_info.n_post_neurons, n_post_atoms)
split_array = numpy.array_split(self.__array, split_positions)
# Sum the 1s in each split row
sum_rows = [numpy.sum(s, axis=1) for s in split_array]
# Find the maximum of the rows
max_connections_row = max([x for y in sum_rows for x in y])
if min_delay is None and max_delay is None:
return max_connections_row
return self._get_n_connections_from_pre_vertex_with_delay_maximum(
synapse_info.delays, self.__n_total_connections,
max_connections_row, min_delay, max_delay, synapse_info)
[docs]
@overrides(AbstractConnector.get_n_connections_to_post_vertex_maximum)
def get_n_connections_to_post_vertex_maximum(self, synapse_info):
# Max number per column is required
max_connections_col = 0
for j in range(self.__array_dims[1]):
n_connections_col = 0
for i in range(self.__array_dims[0]):
if self.__array[i, j] == 1:
n_connections_col = 0
if n_connections_col > max_connections_col:
max_connections_col = n_connections_col
return max_connections_col
[docs]
@overrides(AbstractConnector.get_weight_maximum)
def get_weight_maximum(self, synapse_info):
return self._get_weight_maximum(
synapse_info.weights, self.__n_total_connections, synapse_info)
[docs]
@overrides(AbstractGenerateConnectorOnHost.create_synaptic_block)
def create_synaptic_block(
self, post_slices, post_vertex_slice, synapse_type, synapse_info):
pre_neurons = []
post_neurons = []
n_connections = 0
pre_lo = 0
pre_hi = synapse_info.n_pre_neurons - 1
post_lo = post_vertex_slice.lo_atom
post_hi = post_vertex_slice.hi_atom
for i in range(pre_lo, pre_hi+1):
for j in range(post_lo, post_hi+1):
if self.__array[i, j] == 1:
pre_neurons.append(i)
post_neurons.append(j)
n_connections += 1
# Feed the arrays calculated above into the block structure
block = numpy.zeros(
n_connections, dtype=AbstractConnector.NUMPY_SYNAPSES_DTYPE)
block["source"] = pre_neurons
block["target"] = post_neurons
block["weight"] = self._generate_weights(
block["source"], block["target"], n_connections, post_vertex_slice,
synapse_info)
block["delay"] = self._generate_delays(
block["source"], block["target"], n_connections, post_vertex_slice,
synapse_info)
block["synapse_type"] = synapse_type
return block
def __repr__(self):
return f"ArrayConnector({self.__array})"