ButterBandstop

Description

Digital band-stop (notch) Butterworth IIR filter -- suppresses frequencies in [low_cutoff, high_cutoff]. Produces a 2N-order filter from the order-N analog prototype.

\[ \text{ButterBandstop}(N, W_{\text{low}}, W_{\text{high}}) \]

• order (int, \(\ge 1\)): order of the analog prototype.

• low_cutoff, high_cutoff (floats in \((0, 1)\), with low < high): stop band as fractions of the Nyquist frequency.

Bit-exact match to scipy.signal.butter(order, [low, high], btype='bandstop') + scipy.signal.lfilter. Verified to ~1e-9 in tests/test_signal.py.

NaN handling

Policy: ignore. A NaN in any input at index t causes the function to skip that step: output at t is NaN and internal state is unchanged. Subsequent finite samples are processed as if step t had not occurred.

Examples

Description

from screamer import ButterBandstop
# 60 Hz notch on 1kHz-sampled data: Nyquist = 500 Hz, cutoffs ~58/500 .. 62/500
notch = ButterBandstop(order=2, low_cutoff=0.116, high_cutoff=0.124)
out = notch(signal)