Second derivative of SmoothKernelDistribution PDF?

Question

Does anyone know why I can't seem to get a second derivative of this interpolating function f? In this example, h[x] is zero for all x.

data = Join[RandomVariate[NormalDistribution[-3, 1], 100], RandomVariate[NormalDistribution[3, 1], 100]];
f = PDF[SmoothKernelDistribution[data]];
g = f';
h = g';
Plot[{f[x], g[x], h[x]}, {x, -10, 10}, PlotRange -> All]

enter image description here

You might try KernelMixtureDistribution to avoid the piecewise linear limitation — Andy Ross, May 02 '15 at 02:01

user15994 · Answer 1 · 2015-05-01T21:40:46.607

4

If you plot f, f', and f'' over a much smaller range, the answer will be obvious: the interpolation function f is piecewise linear, f' is piecewise constant (i.e., a step function), and f'' is identically zero within each interval.

edited May 01 '15 at 21:40

answered May 01 '15 at 21:34

user15994

509
4
7

Oops, I backspaced in the window and the message got messed up. – user15994 May 01 '15 at 21:35

Second derivative of SmoothKernelDistribution PDF?

1 Answers1

Linked