CODEWITHSUNDEEP
pythonnumpymatplotlibscipy
StarStrides
StarStrides

Reputation: 121

scipy interpolate gives unbounded value

I have a data set data[xi,yi,zi] I'd like to plot (with interpolation values). With scipy.interpolate, everything looks almost perfect however the interpolation is generating some values beyond the bounds of the input data. for example suppose zi is bound by 0 < zi < 1, the rbf interpolatation seems to be returning interpolated values out of the bounds (e.g. >1); here my simplified attempt:

N=100
data=[xi yi zi]
xis = np.linspace(xi.min(), xi.max(), N)
yis = np.linspace(yi.min(), yi.max(), N)
XI, YI =  np.meshgrid(xis,yis)
rbf = scipy.interpolate.Rbf(xi, yi, zi, function='linear')
ZI=rbf(XI,YI)
print ZI.max()
->1.01357328514

Is there a way to pass limits to Rbf and let it know not to go past zi.max() and zi.min() ?

Upvotes: 1

Views: 913

Answers (1)

user6655984
user6655984

Reputation:

Interpolation with radial basis functions may result in values above the maximum and below the minimum of the given data values. (Illustration). This is a mathematical feature of the method, one cannot pass in an option to disable it. Two possible solutions:

  • use np.clip to clip the interpolant between the min-max values of the data, when plotting it.
  • use piecewise linear interpolation instead (scipy.interpolate.LinearNDInterpolator), which is guaranteed to respect the minimum and maximum of data values.

Upvotes: 2

Related Questions