Catch dfitpack.error from scipy.interpolate.InterpolatedUnivariateSpline

Question

I am interpolating an array and there is the possibility, that I might not have enough knots for the interpolation. In this case, I will need to adjust the degree of the smoothing spline. In the minimal working example given below, this would be the case if N = 3.

from scipy.interpolate import InterpolatedUnivariateSpline
import numpy as np

N = 10       # is working fine
# N = 3      # produces error
x = np.linspace(-3, 3, N)
y = np.exp(-x**2) + 0.1 * np.random.randn(N)
spline = InterpolatedUnivariateSpline(x, y, bbox=[x[0],x[-1]], k=3, ext=0)

x_full = np.linspace(-3,3,10*N)
yspline = spline(x_full)

The error, if N=3 is error: (m>k) failed for hidden m: fpcurf0:m=3. The same error as described here and here. I know what this error means and know why it appears, this is not the issue.

I would like to write a try/except to catch this case and handle it appropriately. What <ErrorType> would this be?

try:
    spline = InterpolatedUnivariateSpline(x, y, bbox=[x[0],x[-1]], k=3, ext=0)
    yspline = spline(x_full)
except <ErrorType>:
    print('something')

(I know I could instead write an if/else statement checking the length of x and y. I will keep this in mind as an alternative. But for know, it would be better to use the try/except statement.)

Answer 1

except Exception . While generally considered bad form (for a reason!), it'll work here.