How can one perform scipy.signal.fftconvolve on a single axis only?

edit

To perform scipy.signal.fftconvolve on a single axis only, you can specify the axis parameter to be the axis you want to perform the convolution on. For example, if you have a 3D array arr and you want to perform the convolution on the second axis only, you can do:

import scipy.signal

kernel = # define your convolution kernel
arr = # define your input array

convolved_arr = scipy.signal.fftconvolve(arr, kernel, mode='same', axis=1)

Here, axis=1 specifies that the convolution should be performed on the second axis of arr. The resulting convolved_arr will have the same shape as arr, but with the convolution applied on the second axis only.

You can also use numpy.apply_along_axis to apply the convolution along a specific axis:

import numpy as np
import scipy.signal

kernel = # define your convolution kernel
arr = # define your input array

convolved_arr = np.apply_along_axis(lambda x: scipy.signal.fftconvolve(x, kernel, mode='same'), axis=1, arr=arr)

Here, lambda x: scipy.signal.fftconvolve(x, kernel, mode='same') defines a function that applies the convolution on a 1D array, which is then applied along axis 1 of arr using numpy.apply_along_axis. The resulting convolved_arr will again have the same shape as arr, but with the convolution applied on the second axis only.