Recall that the singular value decomposition (SVD) of a matrix is a factorization , where and are unitary and , with , where where . We sometimes write to specify the matrix to which the singular value belongs.

A standard technique for obtaining singular value inequalities for is to apply eigenvalue inequalities to the Hermitian positive semidefinite matrices or , whose eigenvalues are the squares of the singular values of , or to the Hermitian matrix

whose eigenvalues are plus and minus the singular values of together with zero eigenvalues if .

We begin with a variational characterization of singular values.

Theorem 1.For ,

where .

Proof. The result is obtained by applying the Courant–Fischer theorem (a variational characterization of eigenvalues) to .

As a special case of Theorem 1, we have

and, for ,

The expression in the theorem can be rewritten using (the equality case in the Cauchy–Schwarz inequality). For example, (2) is equivalent to

Our first perturbation result bounds the change in a singular value.

Theorem 2.For ,

Proof.The bound is obtained by applying the corresponding result for the Hermitian eigenvalue problem to (1).

The bound (4) says that the singular values of a matrix are well conditioned under additive perturbation. Now we consider multiplicative perturbations.

The next result is an analogue for singular values of Ostrowski’s theorem for eigenvalues.

Theorem 3. For and nonsingular and ,

where .

A corollary of this result is

where . The bounds (5) and (6) are intuitively reasonable, because unitary transformations preserve singular values and the bounds quantify in different ways how close and are to being unitary.

Next, we have an interlacing property.

Theorem 4. Let , , and . Then

where we define if .

Proof.The result is obtained by applying the Cauchy interlace theorem to , noting that is the leading principal submatrix of order of .

An analogous result holds with rows playing the role of columns (just apply Theorem 4 to ).

Theorem 4 encompasses two different cases, which we illustrate with and . The first case is , so that and

The second case is , so and

Therefore Theorem 3 shows that removing a column from does not increase any singular value and that when no singular value decreases below . However, when the smallest singular value of may be less than the smallest singular value of .

Here is a numerical example. Note that transposing does not change its singular values.

>> rng(1), A = rand(5,4); % Case 1. >> B = A(:,1:end-1); sv_A = svd(A)', sv_B = svd(B)' sv_A = 1.7450e+00 6.4492e-01 5.5015e-01 3.2587e-01 sv_B = 1.5500e+00 5.8472e-01 3.6128e-01 > A = A'; B = A(:,1:end-1); sv_B = svd(B)' % Case 2 sv_B = 1.7098e+00 6.0996e-01 4.6017e-01 1.0369e-01

By applying Theorem 4 repeatedly we find that if we partition then for all for which the left-hand side is defined.

## References

- Roger A. Horn and Charles R. Johnson, Matrix Analysis, second edition, Cambridge University Press, 2013. My review of the second edition.
- Ilse Ipsen, Relative Perturbation Results for Matrix Eigenvalues and Singular Values, Acta Numerica 7, 151–201, 1998.