Integrator

class brainpy.dyn.Integrator(size, keep_size=False, sharding=None, name=None, mode=None, tau=10.0, x_initializer=ZeroInit, method='exp_auto', init_var=True)

Integrator Model.

This class implements an integrator model, in which its dynamics is given by:

\[\tau \frac{dx}{dt} = - x(t) + I(t)\]

where \(x\) is the integrator value, and \(\tau\) is the time constant.

Parameters:

• tau (Union[float, TypeVar(ArrayType, Array, Variable, TrainVar, Array, ndarray), Callable]) – float, ArrayType, Initializer, callable. Membrane time constant.

• method (str) – str. The numerical integration method. Default “exp_auto”.

• x_initializer (Union[Callable, TypeVar(ArrayType, Array, Variable, TrainVar, Array, ndarray)]) – ArrayType, Initializer, callable. The initializer of \(x\).

• size (Union[int, Sequence[int]]) – int, or sequence of int. The neuronal population size.

• sharding (Optional[Sequence[str]]) – The sharding strategy.

• keep_size (bool) – bool. Keep the neuron group size.

• mode (Optional[Mode]) – Mode. The computing mode.

• name (Optional[str]) – str. The group name.

reset_state(batch_size=None, **kwargs)

Reset function which resets local states in this model.

Simply speaking, this function should implement the logic of resetting of local variables in this node.

See https://brainpy.readthedocs.io/en/latest/tutorial_toolbox/state_resetting.html for details.

supported_modes: Optional[Sequence[bm.Mode]] = (<class 'brainpy._src.math.modes.TrainingMode'>, <class 'brainpy._src.math.modes.NonBatchingMode'>)

Supported computing modes.

update(x=None)

The function to specify the updating rule.