n-fold convolution of a probability density function

Hello everyone,

I have few questions about convolutions.

1. I have a sample data in the form of "x" vs "f(x)".

Now I want to do 3-fold convolution of "f(x)" and plot the 3-fold convoluted functions w.r.t "x". How to do it?

2. suppose length of f(x) is 10. So, length of "x" is also 10. After 2-fold convolution (i.e f2(x)=f(x)*f(x)), length of f2(x) will be 19, but original length of "x" is still 10. Then how to plot f2(x) w.r.t "x"?

3. which convolution technique I should use, circular or linear?

Kindly give some suggestion.