R help - Apr 2009 - [help]how to estimate kernel density over samples from importance sampling

Dear all:
First we run simulation through normal distribution, like p(x)=1/b, b
is length of sample length, the properties of simulation we got f(x)
has an unknow distribution. Through kernel density estimation we could
get f(x)'s approximate distribution.
Now I have run simulation through importance sampling,sampling
distribution is q(x). The data (g(x))we got have a importance weight(w
(x)=p(x)/q(x), p(x) is our initial sample distribution, q(x) is
sampling distribution). The problems is, how to get same approximate
distribution of f(x) again?
Thanks for your attention