CODEWITHSUNDEEP
pythonconvolution
Kyle
Kyle

Reputation: 33

convolve and deconvolve two arrays

So I am trying to figure out how to get convolve and deconvolve to work properly. Here's my code: 

import numpy as np
from scipy import interpolate
from scipy import signal
import matplotlib.pyplot as plt

a = np.arange(0, 50, 1)
b = np.array([1,2,1])

aconvolved = signal.convolve(np.sin(a), b)
adeconvolved = signal.deconvolve(aconvolved, b)

plt.figure()
plt.plot(a, np.sin(a), 'g')
plt.plot(a, aconvolved[0], 'r')
plt.plot(a, adeconvolved[0], 'b')
plt.draw()

When I convolve "a" about a single point (like b = [2]), instead of an array, it works just fine. But when I attempt to do it about more than 1 point, or an array, it just doesn't work. I receive errors such as:

"ValueError: x and y must have same first dimension"

I get this same error if I instead use:

a = np.arange(1, 50, 1)
b = np.array([1,2,1])

Even though in this case, I know they do both start at 1. If I do the same thing, but with

b = np.array([1,2,1])

I get an error of:

"ValueError: BUG: filter coefficient a[0] == 0 not supported yet"

Anyone have any tips for what I am doing wrong?

Upvotes: 2

Views: 2981

Answers (1)

DrV
DrV

Reputation: 23510

You do not get the error where you'd think you get it. This works:

import numpy as np
from scipy import interpolate
from scipy import signal
import matplotlib.pyplot as plt

a = np.arange(0, 50, 1)
b = np.array([1,2,1])

aconvolved = signal.convolve(np.sin(a), b)
adeconvolved = signal.deconvolve(aconvolved, b)[0]

plt.figure()
plt.plot(a, np.sin(a), 'g')
plt.plot(a, aconvolved[1:-1], 'r') 
plt.plot(a, adeconvolved, 'b')
plt.draw()

The differences:

  • convolution creates some extra samples to the ends, hence the aconvolved[1:-1] (if you do not want to do this, then use keyword argument mode='same' for the convolve to receive a vector which has the same length as the input, but then that'll cause problems with the deconvolution output length).
  • deconvolution returns a tuple, where the actual result is the first part, hence signal.deconvolve(aconvolved, b)[0]

Upvotes: 2

Related Questions