Question

I have a dataset (stellar spectrum) with around 4000 points. I want to interpolate the x and y data (same length) with this new list, x_new which has nearly 17000 points.

I have found numpy.interp thas does this job, except it is a piecewise linear interpolation. I want it to be smooth.

Any suggestions?

Answer 1

The scipy.interpolate.interp1d has a kind parameter that can be set to e.g. "cubic".

Instad of supplying the x-values you wish to interpolate on directly to this function, you do it like this:

>>> x = np.arange(0, 10)
>>> y = np.exp(-x / 3.0)
>>> f = interpolate.interp1d(x, y, "cubic")
>>> i = np.arange(0.5, 9.5, 1)
>>> f(i)
array([ 0.85826566,  0.61497421,  0.44064828,  0.31573829,  0.22623637,
        0.16210544,  0.11615363,  0.08322771,  0.05963526])

This is a scipy interpolation tutorial that might be helpful.