Fast convolution with sdpvar objects

[Paul Gab]

Dear all,

I am fairly new with Yalmip. I am working with Matlab and I am trying to implement the convolution between two vectors, where one is a decision variable, sdpvar, and the other is a known data, double.

At the moment, I am using the matlab function conv. However, the convolution takes very long, about 45 minutes, having vectors of about 8000 elements each.

I was wondering if there is a way for speeding this up, e.g. by using fast convolution functions, such as fconv or convolve2, with sdpvar objects as input.

Thank you for your help!

[Johan Löfberg]

I've checked in a quick fix in develop which brings it down to a couple of seconds. Can probably be done much faster, but the problem is that the built-in conv operator only works on full matrices so for speed I have to write the conv manually (the sdpvar basis has very  sparse structure)

>> a = sdpvar(8000,1);b = randn(8000,1);

>> tic;y = conv(a,b);toc

Elapsed time is 5.714931 seconds.

[Paul Gab]

Pardon, I found it on github!

Thank you very much!