A $n$ by $n$ hermitian matrix $A$ is such that

a) $A^{*}=A$ where $*$ represents complex conjugate transpose.

Now what does it mean for $B$ to be complex symmetric? Correct me if I am wrong on this but I think that

1) $B^{T}=B$ where $T$ represents transpose.

Both $A$ and $B$ are normal matrices so they both have an orthogonal set of eigenvectors that form a basis. Say for matrix $A$, these eigenvectors form a unitary matrix $U$ such that $UAU^{*}=D$ where $D$ is a diagonal matrix with eigenvalues along the diagonal.

  • 1
    $\begingroup$ Did you mean complex symmetric in the title? $\endgroup$ Nov 18, 2014 at 5:14
  • $\begingroup$ yes sorry for the typo $\endgroup$
    – Clannad
    Nov 23, 2014 at 12:33

1 Answer 1



$$B = \begin{bmatrix} 1 & x \\ x & 0 \end{bmatrix}, \quad x \in \mathbb{C}.$$

Obviously, $B = B^T$. However,

$$BB^* - B^*B = \begin{bmatrix} 0 & \overline{x} - x \\ x - \overline{x} & 0 \end{bmatrix} = \begin{bmatrix} 0 & -2 \operatorname{Im} x \\ 2 \operatorname{Im} x & 0 \end{bmatrix} \ne 0,$$

for $x \not\in \mathbb{R}$. In other words, complex symmetric matrices need not be normal.

An appropriate decomposition for them is the Takagi's decomposition,

$$B = V D V^T,$$

where $D$ is a real nonnegative diagonal matrix, and $V$ is unitary. Note, however, that this is not a similarity relation, like the eigenvalue decomposition, because $V^T$ is not the inverse of $V$ (unless $V$ happens to be real, of course).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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