Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Given I have a matrix A of rank 3. I want to create a matrix of Rank 2 which is closest to A in the $ {l}_{2} $ / Frobenius norm. Let's call this matrix F.

Is easy to achieve by the SVD, namely, if $ A = U S {V}^{H} $ by the SVD decomposition $ F = U \hat{S} {V}^{H} $. Where $ \hat{S} $ is the same as $ S $ with the last Singular Value zeroed.

The question is, is there a less computationally intensive method to create F but using the SVD decomposition?


share|improve this question
it's not necessary to compute the full SVD decomposition in order to obtain a low rank approximation. Instead, just compute the leading singular values and their corresponding left and right vectors. –  Victor May Feb 7 '12 at 8:37
There are ways, such as an adaptation of the power method, for computing partial singular value decompositions. You'll want to look into the literature for algorithms. –  J. M. Feb 7 '12 at 10:12
@J.M., Could you refer me to some sources? –  Drazick Feb 7 '12 at 13:23
Nash/Shlien, off the top of my head. There are probably more modern methods, but I don't have my notes with me at the moment... –  J. M. Feb 7 '12 at 13:30
Look for nipals PCA –  Donbeo Aug 14 at 14:39

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.