Results should be the same as the convex optimization technique in the "well-defined" entries, though numerical stability is a real concern that I haven't looke

Results should be the same as the convex optimization technique in the "well-defined" entries, though numerical stability is a real concern that I haven't looked at yet. Empirically I haven't had any issues, but if you have entries in the singular vectors sufficiently close to zero you can definitely imagine some problems. I don't think that's unique to this approach, either.