February 25, 2018, 5:46pm
How can i solve using CVX, for solving these form of problem;
min(norm(s,1)) ,s is block sparse , and There are two blocks N Dimension. Each block is K spars.
s is (K*N) *1 , Y is (2M) 1 ,A is (2M)(KN) , N is (2M)*1.
st : norm(Y-As,2)< epsilon
The position sparse is the same in each block.
s were a MATLAB variable, rather than a CVX variable, do you know a series of MATLAB commands which will construct
s and properly populate it? If so, you might be well on your way to figuring out how to do it in CVX.
February 25, 2018, 10:14pm
Is this code correct?
for k = 1:K
[vall(k), indexx(k)] = max(s((k-1)*N+1:k*N));
please help me for Correction code.
K= 2 , M=10 ,N=90.
You’ve done nothing to impose (force) your sparsity structure on
s. So the optimal
s returned by CVX might not have any zeros.