Abstract:Given a structured matrix $A$ we study the problem of finding the closest normal matrix with the same structure. The structures of our interest are: Hamiltonian, skew-Hamiltonian, per-Hermitian, and perskew-Hermitian. We develop a structure-preserving Jacobi-type algorithm for finding the closest normal structured matrix and show that such algorithm converges to a stationary point of the objective function.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: Given a matrix \(A\) with a specific structure, how to find the closest normal matrix (normal matrix) with the same structure. Here, the structure includes Hamiltonian, skew - Hamiltonian, per - Hermitian and perskew - Hermitian.
Specifically, the paper focuses on the following optimization problem:
\[ \min_{X \in N \cap S} \| X - A \|_F \]
where:
- \(A\) is a matrix with a specific structure \(S\).
- \(N\) represents the set of normal matrices.
- \(\| \cdot \|_F\) represents the Frobenius norm.
The background of this problem can be traced back to the problem of finding the closest normal matrix to an arbitrary unstructured matrix \(A\in\mathbb{C}^{n\times n}\):
\[ \min_{X \in N} \| X - A \|_F^2 \]
This problem has been independently solved by Gabriel and Ruhe, and Higham has summarized its important results in the literature. However, when it comes to maintaining the matrix structure, this problem becomes more complex and challenging.
To solve this problem, the author has developed a structure - preserving Jacobi - type algorithm and proved that this algorithm can converge to the stable point of the objective function. Through this method, the matrix can be made as close as possible to the normal matrix while maintaining the matrix structure.
### Specific steps and formulas
1. **Hamiltonian and skew - Hamiltonian matrices**:
- Define the special forms of these matrices and give their diagonalized forms.
- Use the Schur decomposition theorem to decompose the Hamiltonian or skew - Hamiltonian normal matrix into a specific form.
2. **per - Hermitian and perskew - Hermitian matrices**:
- Similarly, define the special forms of these matrices and give their diagonalized forms.
- Decompose them using a similar method.
3. **Jacobi - type algorithm**:
- Develop an iterative algorithm to maximize the Frobenius norm of the diagonal elements through a series of structure - preserving rotation operations.
- These rotation operations include symplectic and perplectic rotations, with the specific forms as follows:
- Symplectic rotation:
\[
R(i,j,\phi,0)=\begin{bmatrix}
\cos\phi&-\sin\phi\\
\sin\phi&\cos\phi
\end{bmatrix}_{i,n + i}
\]
- Perplectic rotation:
\[
R(i,2n - i+ 1,\phi,-\pi/2)=\begin{bmatrix}
\cos\phi&i\sin\phi\\
i\sin\phi&\cos\phi
\end{bmatrix}_{i,2n - i+ 1}
\]
4. **Selecting the rotation angle**:
- Calculate the rotation angles \(\phi\) and \(\alpha\) in each iteration to maximize the Frobenius norm of the diagonal elements.
- Use the partial derivative method to solve for the optimal angles.
Through these methods, the paper successfully solves the problem of finding the closest structure - preserving normal matrix and provides an effective numerical algorithm implementation.