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.

Consider a complex $N-1 \times 1$ vector $b$ and a complex constant c. Let $I$ denote the $N-1 \times N-1$ identity matrix. Then what can we say about the eigenvalues of the matrix

\begin{align} \begin{bmatrix} I & b \\ b^H & c \end{bmatrix} \end{align}

share|improve this question

2 Answers 2

up vote 3 down vote accepted

Apart from Alexander Gruber's answer, we may also determine the eigenvalues in two other ways.

Call your matrix $A$. As the case $b=0$ is trivial, we assume that $b\not=0$. First, note that if $v\perp b$, then $(v^T, 0)$ will be a left eigenvector of $A$ corresponding to the eigenvalue $1$. Hence the multiplicity of this eigenvalue is at least $n-2$.

Next, if $\lambda\not=1$ is an eigenvalue of $A$, then by a determinant formula for block matrices, we have $0=\det(A-\lambda I)=\det(I-\lambda I)\det[(c-\lambda)-b^T(I-\lambda I)^{-1}b]$ and therefore $(c-\lambda)-\|b\|^2/(1-\lambda)=0$. Hence the remaining two eigenvalues of $A$ are given by the two roots of the quadratic equation $\lambda^2 - (c+1)\lambda + (c-\|b\|^2)=0$.

Alternatively, suppose $w^T=(b^T+v^T, q-1)$ is a left eigenvectors of $A$, where $v\perp b$. Then $w^TA = \left(qb^T + v^T,\ \|b\|^2 + b^Tv + (q-1)c\right)$. In order that $qb^T+v^T$ is a multiple of $b^T+v^T$, we must have $v=0$ and the eigenvalue is $q$. So, by comparing the last elements of $w^T$ and $w^TA$, we must have $\|b\|^2 + (q-1)c = q(q-1)$, i.e. $q^2 - (c+1)q + (c-\|b\|^2)=0$.

share|improve this answer
I suppose that determinant formula you are referring is related to schur complements. The determinant is the product of the block matrix and its schur complement? –  dineshdileep Dec 17 '12 at 9:16
elegant answer!!. –  dineshdileep Dec 17 '12 at 9:18
@dineshdileep Yes, it's the product of the determinants of a block and its Schur complement. –  user1551 Dec 17 '12 at 9:39

Call your matrix $M_b$. Then $$\text{det}M_b-\lambda I_{n}=\left((c-\lambda)(\lambda-1)+\sum_{i=1}^{n-1}b_i^2\right)(\lambda-1)^{n-2},$$ so you're going to have $1$ as an eigenvalue of multiplicity $n-2$ and then $$\frac{1}{2}\left(1+c\pm\sqrt{(c+1)^2+4\left(-c+\sum_{i=1}^{n-1}b_i^2\right)}\right)$$ as your other two eigenvalues.

share|improve this answer
Thanks a lot for the answer!!. –  dineshdileep Dec 17 '12 at 9:19

Your Answer


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

Not the answer you're looking for? Browse other questions tagged or ask your own question.