回复 第5楼 的 Ihavenothing：回复 第1楼 的 wenfo：
One more contributing factor is that density() does not return a function, but rather the evaluation of the KDE at a given number of selected points through FFT. This rarely causes problems and is much more computationally efficient, but may be the culprit for the cauchy situation with extremely heavy tails. The following could be solutions to this point:
<br />
plot(density(a, from=-4, to=4),xlim=c(-4,4))<br />
plot(density(a, n=20L*length(a)),xlim=c(-4,4))<br />
</p>
