The Reverse Order Law and the Riccati Equation

Oskar Kędzierski

2024-08-12

Abstract:We give a full analytic solution to a particular case of the algebraic Riccati equation $XWW^*WX=W^*$ for any matrix $W$ (possibly non-square or non-symmetric) in using the Schur method, terms of the SVD decomposition of $W$. In particular, $(WX)^3=WX$ and $(XW)^3=XW$ for any solution $X$. We show that for $W=AB$, matrix $X=B^+ A^+$ is a solution of this equation if and only if the reverse order law holds, i.e., ${(AB)}^+=B^+ A^+$. For a Hermitian and invertible $W$ the maximal and stabilizing Hermitian solutions is shown to be equal to $W^+$. Equivalence to the equation $XWX=W^+$ is proven.

Rings and Algebras,Optimization and Control

What problem does this paper attempt to address?

The problem that this paper attempts to solve is about the analytical solution of the algebraic Riccati equation \(XWW^*WX = W^*\) and its relationship with the reverse order law. Specifically, the author studies how to use the Schur method and singular value decomposition (SVD) to solve this particular form of the algebraic Riccati equation for any matrix \(W\) (which may be non - square or non - symmetric).

### Main problems
1. **Finding the analytical solution**: The author provides a complete analytical solution through the Schur method and SVD decomposition.
2. **Application of the reverse order law**: It is explored whether the matrix \(X = B^+A^+\) satisfies the above Riccati equation when \(W = AB\), and whether this is equivalent to the reverse order law \((AB)^+ = B^+A^+\) holding.
3. **Periodic properties**: It is shown that for any solution \(X\), \((WX)^3=WX\) and \((XW)^3 = XW\).

### Specific problem description
- **Algebraic Riccati equation**:
\[
XWW^*WX = W^*
\]
where \(W\) is an arbitrary matrix (possibly non - square or non - symmetric).

- **Reverse order law**:
\[
(AB)^+ = B^+A^+
\]
where \(A^+\) and \(B^+\) are the Moore - Penrose pseudo - inverses of matrices \(A\) and \(B\) respectively.

### Solution
The author solves the above problems through the following steps:

1. **Using the Schur method and SVD decomposition**:
- Perform singular value decomposition on \(W\), \(W = U\Sigma V^*\).
- Obtain the form of the solution by analyzing the eigenvectors of the matrix \(M=\begin{bmatrix}0&WW^*\\W^*&0\end{bmatrix}\).

2. **Give the general form of the solution**:
\[
X = V\Sigma^+DU^*+P_{N(W)}YP_{N(W^*)}
\]
where \(D\) is a block - diagonal matrix and \(Y\) is an arbitrary matrix.

3. **Prove the relationship between the reverse order law and the solution of the Riccati equation**:
- When \(W = AB\), \(X = B^+A^+\) satisfies the Riccati equation if and only if the reverse order law holds.

4. **Show the periodic properties**:
- For any solution \(X\), \((WX)^3 = WX\) and \((XW)^3=XW\).

Through these steps, the author not only provides the analytical solution of the algebraic Riccati equation, but also reveals the profound connection between it and the reverse order law.