nengolib.synapses.BoxFilter

nengolib.synapses.BoxFilter(width, normalized=True)

A discrete box-filter with a given width, and optionally unit area.

This filter is also known as a “box blur”, and has the effect of smoothing out the input signal by taking its rolling mean over a finite number of time-steps. Its properties are qualitatively similar to the continuous-time Lowpass().

Parameters:

width : integer

Width of the box-filter (in time-steps).

normalized : boolean, optional

If True, then the height of the box-filter is 1/width, otherwise 1. Defaults to True.

Returns:

sys : LinearSystem

Digital system implementing the box-filter.

See also

z, Lowpass()

Examples

Simulate a Nengo network using a box filter of 10 ms for a synapse:

>>> from nengolib.synapses import BoxFilter
>>> import nengo
>>> with nengo.Network() as model:
>>>     stim = nengo.Node(output=lambda _: np.random.randn(1))
>>>     p_stim = nengo.Probe(stim)
>>>     p_box = nengo.Probe(stim, synapse=BoxFilter(10))
>>> with nengo.Simulator(model) as sim:
>>>     sim.run(.1)

>>> import matplotlib.pyplot as plt
>>> plt.step(sim.trange(), sim.data[p_stim], label="Noisy Input", alpha=.5)
>>> plt.step(sim.trange(), sim.data[p_box], label="Box-Filtered")
>>> plt.xlabel("Time (s)")
>>> plt.legend()
>>> plt.show()

(Source code)