Why do the non-zero singularvalues of a matrix keep increasing in magnitude as more redundant rows are added to it

157 Views Asked by At

Let us consider a matrix $\mathbf{A}\in\mathbb{R}^{n\times n}$ such that rank of the matrix is $n$. If we do the singular value decomposition(SVD) of this matrix, we'll get $4$ non-zero singular values.

Now let us add more rows to A such that the rank of $\mathbf{A}$ stays constant at $n$. Every time we add a row, we perform SVD on the augmented matrix. We see that as we keep on adding more redundant rows to $\mathbf{A}$, the non-zero singular values keep on shifting to right i.e. their value keeps on increasing. Can anyone explain this phenomena to me? Is it due to eigenvalue repulsion?

1

There are 1 best solutions below

0
On BEST ANSWER

Recall that the squares of the singular values are the eigenvalues of $A^*A$. So $$ B = \begin{pmatrix} A \\ b \end{pmatrix} \implies B^*B = A^*A+ b^*b , $$ and the eigenvalues increase (though perhaps not strictly) by the min-max principle.