You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I understand the the generalized phase-retrieval problem solves |Ax| = b where both A and x are complex. But can phase-pack be used when A is complex and x is real?
The text was updated successfully, but these errors were encountered:
@tomgoldstein For the case that x is real, the gradient w.r.t. x should be real as well, and I don't think the current code
handle that case, e.g.,
%%Define ingredients for FASTA% m x 1 -> 1 x 1% f(y) = .5 ||y - b||^2
f = @(y) .5*norm(y(:)-b(:))^2;
% m x 1 -> m x 1
gradf = @(y) y(:)-b(:);
% n x n -> 1 x 1% g(z) = mu||X||_nuc, plus characteristic function of the SDP cone
g = @(X) mu*norm(eig(X),1);
% n x n -> n x n% proxg(z,t) = argmin t*mu*nuc(x)+.5||x-z||^2, with x in SDP cone
proxg = @(X,t) projectSemiDefCone(X, mu*t);
I understand the the generalized phase-retrieval problem solves |Ax| = b where both A and x are complex. But can phase-pack be used when A is complex and x is real?
The text was updated successfully, but these errors were encountered: