What is the convolution of two probability distribution estimate (ksdensity in matlab)?

Hello Everyone I am wondering how to perform the convolution of two ksdenity given that their vectors are not related. Here is an example of two sampling height data:

[f1,x1]=ksdenity(data_height_1);
[f2,x2]=ksdenity(data_height_2);

How to convolute f1 with f2? what would be the resulted height vector ?

Answer 1

[Ahmet Cecen]

Use:

[f,xi] = ksdensity(x,pts)

To sample them at the same points, even if you have to sample a long tail of zeros to do so. Then just do:

conv(u,v)

[Imane Imane]

Thank you for your answer. I have a question about the resulting vector x of the convolution. As I know: length(conv(u,v))=length(u)+length(v)-1 But I would have no idea about the points corresponding to the convolution elements.

[Ahmet Cecen]

Convolution will give you a vector in a lag based coordinates system. The center point of the resulting vector will be when both u and v are aligned exactly on top of each other (with v reversed). 1 pixel to the next of the center will be u shifted by 1 pixel, then overlayed on top of v and so forth. Look up a wikipedia article on this it should be helpful.

[Imane Imane]

So you are telling me to build the resulting vector by myself, there is no tool in matlab to build it?

[Ahmet Cecen]

No, the tool is x = conv(u,v) I was addressing your question about how to interpret the resulting vector x.