def get_spike_update_expressions(self, neuron: ASTNeuron, kernel_buffers, solver_dicts, delta_factors) -> List[ASTAssignment]:
"""
Generate the equations that update the dynamical variables when incoming spikes arrive. To be invoked after ode-toolbox.
For example, a resulting `assignment_str` could be "I_kernel_in += (in_spikes/nS) * 1". The values are taken from the initial values for each corresponding dynamical variable, either from ode-toolbox or directly from user specification in the model.
Note that for kernels, `initial_values` actually contains the increment upon spike arrival, rather than the initial value of the corresponding ODE dimension.
"""
spike_updates = []
initial_values = neuron.get_initial_values_blocks()
for kernel, spike_input_port in kernel_buffers:
if neuron.get_scope().resolve_to_symbol(str(spike_input_port), SymbolKind.VARIABLE) is None:
continue
buffer_type = neuron.get_scope().resolve_to_symbol(str(spike_input_port), SymbolKind.VARIABLE).get_type_symbol()
if is_delta_kernel(kernel):
continue
for kernel_var in kernel.get_variables():
for var_order in range(get_kernel_var_order_from_ode_toolbox_result(kernel_var.get_name(), solver_dicts)):
kernel_spike_buf_name = construct_kernel_X_spike_buf_name(
kernel_var.get_name(), spike_input_port, var_order)
expr = get_initial_value_from_ode_toolbox_result(kernel_spike_buf_name, solver_dicts)
assert expr is not None, "Initial value not found for kernel " + kernel_var
expr = str(expr)
if expr in ["0", "0.", "0.0"]:
continue # skip adding the statement if we're only adding zero
assignment_str = kernel_spike_buf_name + " += "
assignment_str += "(" + str(spike_input_port) + ")"
if not expr in ["1.", "1.0", "1"]:
assignment_str += " * (" + \
self._printer.print_expression(ModelParser.parse_expression(expr)) + ")"
if not buffer_type.print_nestml_type() in ["1.", "1.0", "1"]:
assignment_str += " / (" + buffer_type.print_nestml_type() + ")"
ast_assignment = ModelParser.parse_assignment(assignment_str)
ast_assignment.update_scope(neuron.get_scope())
ast_assignment.accept(ASTSymbolTableVisitor())
spike_updates.append(ast_assignment)
for k, factor in delta_factors.items():
var = k[0]
inport = k[1]
assignment_str = var.get_name() + "'" * (var.get_differential_order() - 1) + " += "
if not factor in ["1.", "1.0", "1"]:
assignment_str += "(" + self._printer.print_expression(ModelParser.parse_expression(factor)) + ") * "
assignment_str += str(inport)
ast_assignment = ModelParser.parse_assignment(assignment_str)
ast_assignment.update_scope(neuron.get_scope())
ast_assignment.accept(ASTSymbolTableVisitor())
spike_updates.append(ast_assignment)
return spike_updates
def apply_incoming_spikes(neuron):
"""
Adds a set of update instructions to the handed over neuron.
:param neuron: a single neuron instance
:type neuron: ASTNeuron
:return: the modified neuron
:rtype: ASTNeuron
"""
assert (neuron is not None and isinstance(neuron, ASTNeuron)), \
'(PyNestML.Solver.BaseTransformer) No or wrong type of neuron provided (%s)!' % type(neuron)
conv_calls = OdeTransformer.get_sum_function_calls(neuron)
printer = ExpressionsPrettyPrinter()
spikes_updates = list()
for convCall in conv_calls:
shape = convCall.get_args()[0].get_variable().get_complete_name()
buffer = convCall.get_args()[1].get_variable().get_complete_name()
initial_values = (
neuron.get_initial_values_blocks().get_declarations()
if neuron.get_initial_values_blocks() is not None else list())
for astDeclaration in initial_values:
for variable in astDeclaration.get_variables():
if re.match(shape + "[\']*",
variable.get_complete_name()) or re.match(
shape + '__[\\d]+$',
variable.get_complete_name()):
spikes_updates.append(
ModelParser.parse_assignment(
variable.get_complete_name() + " += " + buffer +
" * " + printer.print_expression(
astDeclaration.get_expression())))
for update in spikes_updates:
add_assignment_to_update_block(update, neuron)
return neuron
def apply_spikes_from_buffers(self, neuron, shape_to_buffers):
"""generate the equations that update the dynamical variables when incoming spikes arrive.
For example, a resulting `assignment_string` could be "I_shape_in += (in_spikes/nS) * 1".
The definition of the spike kernel shape is then set to 0.
"""
spike_updates = []
initial_values = neuron.get_initial_values_blocks()
for declaration in initial_values.get_declarations():
variable = declaration.get_variables()[0]
for shape in shape_to_buffers:
matcher_computed_shape_odes = re.compile(shape + r"(__\d+)?")
if re.match(matcher_computed_shape_odes, str(variable)):
buffer_type = neuron.get_scope(). \
resolve_to_symbol(shape_to_buffers[shape], SymbolKind.VARIABLE).get_type_symbol()
assignment_string = variable.get_complete_name() + " += (" + shape_to_buffers[
shape] + '/' + buffer_type.print_nestml_type() + ") * " + \
self._printer.print_expression(declaration.get_expression())
spike_updates.append(
ModelParser.parse_assignment(assignment_string))
# the IV is applied. can be reset
declaration.set_expression(
ModelParser.parse_expression("0"))
for assignment in spike_updates:
add_assignment_to_update_block(assignment, neuron)
def apply_incoming_spikes(neuron):
"""
Adds a set of update instructions to the handed over neuron.
:param neuron: a single neuron instance
:type neuron: ASTNeuron
:return: the modified neuron
:rtype: ASTNeuron
"""
assert (neuron is not None and isinstance(neuron, ASTNeuron)), \
'(PyNestML.Solver.BaseTransformer) No or wrong type of neuron provided (%s)!' % type(neuron)
conv_calls = OdeTransformer.get_sum_function_calls(neuron)
printer = ExpressionsPrettyPrinter()
spikes_updates = list()
for convCall in conv_calls:
shape = convCall.get_args()[0].get_variable().get_complete_name()
buffer = convCall.get_args()[1].get_variable().get_complete_name()
initial_values = (
neuron.get_initial_values_blocks().get_declarations() if neuron.get_initial_values_blocks() is not None else list())
for astDeclaration in initial_values:
for variable in astDeclaration.get_variables():
if re.match(shape + "[\']*", variable.get_complete_name()) or re.match(shape + '__[\\d]+$',
variable.get_complete_name()):
spikes_updates.append(ModelParser.parse_assignment(
variable.get_complete_name() + " += " + buffer + " * " + printer.print_expression(
astDeclaration.get_expression())))
for update in spikes_updates:
add_assignment_to_update_block(update, neuron)
return neuron
def test_assignment_with_comments(self):
assignment = '\n' \
'/* pre */\n' \
'a = b # in\n' \
'/* post */\n' \
'\n'
model = ModelParser.parse_assignment(assignment)
model_printer = ASTNestMLPrinter()
self.assertEqual(assignment, model_printer.print_node(model))
def test_assignment_with_comments(self):
assignment = '\n' \
'/* pre */\n' \
'a = b # in\n' \
'/* post */\n' \
'\n'
model = ModelParser.parse_assignment(assignment)
model_printer = ASTNestMLPrinter()
self.assertEqual(assignment, model_printer.print_node(model))
def add_state_updates(neuron: ASTNeuron, update_expressions: Mapping[str, str]) -> ASTNeuron:
"""
Adds all update instructions as contained in the solver output to the update block of the neuron.
:param neuron: a single neuron
:param update_expressions: map of variables to corresponding updates during the update step.
:return: a modified version of the neuron
"""
for variable, update_expression in update_expressions.items():
declaration_statement = variable + '__tmp real = ' + update_expression
add_declaration_to_update_block(ModelParser.parse_declaration(declaration_statement), neuron)
for variable, update_expression in update_expressions.items():
add_assignment_to_update_block(ModelParser.parse_assignment(variable + ' = ' + variable + '__tmp'), neuron)
return neuron
def add_state_updates(state_shape_variables_updates, neuron):
# type: (map[str, str], ASTNeuron) -> ASTNeuron
"""
Adds all update instructions as contained in the solver output to the update block of the neuron.
:param state_shape_variables_updates: map of variables to corresponding updates during the update step.
:param neuron: a single neuron
:return: a modified version of the neuron
"""
for variable in state_shape_variables_updates:
declaration_statement = variable + '__tmp real = ' + state_shape_variables_updates[variable]
add_declaration_to_update_block(ModelParser.parse_declaration(declaration_statement), neuron)
for variable in state_shape_variables_updates:
add_assignment_to_update_block(ModelParser.parse_assignment(variable + ' = ' + variable + '__tmp'), neuron)
return neuron
def add_state_updates(state_shape_variables_updates, neuron):
# type: (map[str, str], ASTNeuron) -> ASTNeuron
"""
Adds all update instructions as contained in the solver output to the update block of the neuron.
:param state_shape_variables_updates: map of variables to corresponding updates during the update step.
:param neuron: a single neuron
:return: a modified version of the neuron
"""
for variable in state_shape_variables_updates:
declaration_statement = variable + '__tmp real = ' + state_shape_variables_updates[
variable]
add_declaration_to_update_block(
ModelParser.parse_declaration(declaration_statement), neuron)
for variable in state_shape_variables_updates:
add_assignment_to_update_block(
ModelParser.parse_assignment(variable + ' = ' + variable +
'__tmp'), neuron)
return neuron
def apply_spikes_from_buffers(neuron, shape_to_buffers):
spike_updates = []
initial_values = neuron.get_initial_values_blocks()
for declaration in initial_values.get_declarations():
variable = declaration.get_variables()[0]
for shape in shape_to_buffers:
matcher_computed_shape_odes = re.compile(shape + r"(__\d+)?")
buffer_type = neuron.get_scope(). \
resolve_to_symbol(shape_to_buffers[shape], SymbolKind.VARIABLE).get_type_symbol()
if re.match(matcher_computed_shape_odes, str(variable)):
assignment_string = variable.get_complete_name() + " += (" + shape_to_buffers[
shape] + '/' + buffer_type.print_nestml_type() + ") * " + \
_printer.print_expression(declaration.get_expression())
spike_updates.append(ModelParser.parse_assignment(assignment_string))
# the IV is applied. can be reset
declaration.set_expression(ModelParser.parse_expression("0"))
for assignment in spike_updates:
add_assignment_to_update_block(assignment, neuron)
def apply_spikes_from_buffers(self, neuron, shape_to_buffers):
"""generate the equations that update the dynamical variables when incoming spikes arrive.
For example, a resulting `assignment_string` could be "I_shape_in += (in_spikes/nS) * 1".
The definition of the spike kernel shape is then set to 0.
"""
spike_updates = []
initial_values = neuron.get_initial_values_blocks()
for declaration in initial_values.get_declarations():
variable = declaration.get_variables()[0]
for shape in shape_to_buffers:
matcher_computed_shape_odes = re.compile(shape + r"(__\d+)?")
if re.match(matcher_computed_shape_odes, str(variable)):
buffer_type = neuron.get_scope(). \
resolve_to_symbol(shape_to_buffers[shape], SymbolKind.VARIABLE).get_type_symbol()
assignment_string = variable.get_complete_name() + " += (" + shape_to_buffers[
shape] + '/' + buffer_type.print_nestml_type() + ") * " + \
self._printer.print_expression(declaration.get_expression())
spike_updates.append(ModelParser.parse_assignment(assignment_string))
# the IV is applied. can be reset
declaration.set_expression(ModelParser.parse_expression("0"))
for assignment in spike_updates:
add_assignment_to_update_block(assignment, neuron)
def test_assignment_without_comments(self):
assignment = 'a = b\n'
model = ModelParser.parse_assignment(assignment)
model_printer = ASTNestMLPrinter()
self.assertEqual(assignment, model_printer.print_node(model))
def test_assignment_without_comments(self):
assignment = 'a = b\n'
model = ModelParser.parse_assignment(assignment)
model_printer = ASTNestMLPrinter()
self.assertEqual(assignment, model_printer.print_node(model))
