Chapter 53A: Symplectic and Hamiltonian Matrices¶

Prerequisites: Inner Product Spaces (Ch08) · Matrix Equations (Ch20) · Classical Mechanics Basics

Chapter Outline: From Euclidean to Symplectic Geometry → The Symplectic Form and Standard Skew-symmetric Matrix $J$ → Definition and Group Structure of Symplectic Matrices ($Sp(2n)$) → Spectral Properties of Symplectic Matrices (Eigenvalue Pairing) → Definition and Lie Algebra Structure of Hamiltonian Matrices → The Link between Hamiltonian Matrices and Riccati Equations → Symplectic Decompositions (Symplectic QR and Schur) → Applications: Volume Preservation in Phase Space, State-Costate Equations in Optimal Control, and Quantum Optics

Extension: Symplectic matrices are operators that describe "area-preserving" transformations; they reveal the inherent algebraic symmetries of physical systems under energy conservation, serving as the advanced algebraic language connecting classical mechanics, PDEs, and control theory.

In linear algebra, we are familiar with orthogonal matrices that preserve inner products. In physics, particularly Hamiltonian mechanics, it is more important to preserve a "skew-symmetric inner product." Symplectic Matrices are the algebraic characterization of such operators. they describe the conservation of area in phase space and lead to a class of matrices with perfectly symmetric eigenvalue structures: Hamiltonian Matrices. This chapter explores these special matrices that profoundly influence physical evolution and optimal control.

53A.1 Symplectic Form and Matrices¶

Definition 53A.1 (Standard Symplectic Matrix $J$)

Define the $2n \times 2n$ block matrix $J$ as: $$J = \begin{pmatrix} 0 & I_n \\ -I_n & 0 \end{pmatrix}$$ satisfying $J^2 = -I$ and $J^T = -J$. It defines the symplectic inner product $[x, y] = x^T J y$.

Definition 53A.2 (Symplectic Matrix)

A square matrix $M \in M_{2n}$ is Symplectic if it preserves the symplectic form: $$M^T J M = J$$ The set of all such matrices forms the Symplectic Group $Sp(2n)$.

53A.2 Hamiltonian Matrices¶

Definition 53A.3 (Hamiltonian Matrix)

A square matrix $H \in M_{2n}$ is Hamiltonian if $JH$ is symmetric: $$(JH)^T = JH \iff H^T J + JH = 0$$ Property: If $M(t)$ is a symplectic trajectory with $M(0)=I$, its derivative $\dot{M}(0)$ is necessarily a Hamiltonian matrix.

53A.3 Spectral Properties and Pairing¶

Theorem 53A.1 (Eigenvalue Pairing)

Symplectic: If $\lambda$ is an eigenvalue, then $1/\lambda$ must also be an eigenvalue.
Hamiltonian: If $\lambda$ is an eigenvalue, then $-\lambda$ must also be an eigenvalue. Physical Meaning: This pairing reflects the symmetric balance between stable and unstable modes in conservative physical systems.

Exercises¶

1. [Basics] Calculate the determinant of a $2 \times 2$ symplectic matrix.

Solution

Conclusion: 1. Proof: From $M^T J M = J$, taking the determinant of both sides gives: $\det(M^T) \det(J) \det(M) = \det(J)$. Since $\det(J) = 1$ for $2n \times 2n$, we have $(\det M)^2 = 1 \implies \det M = \pm 1$. More advanced theory (Pfaffians) proves the determinant must be strictly +1, corresponding to volume preservation in phase space.

2. [Hamiltonian] Determine if $H = \begin{pmatrix} A & B \\ C & -A^T \end{pmatrix}$ is Hamiltonian (with $B, C$ symmetric).

Solution

Verification Steps: 1. $JH = \begin{pmatrix} 0 & I \\ -I & 0 \end{pmatrix} \begin{pmatrix} A & B \\ C & -A^T \end{pmatrix} = \begin{pmatrix} C & -A^T \\ -A & -B \end{pmatrix}$. 2. Transpose: $(JH)^T = \begin{pmatrix} C^T & -A^T \\ -A & -B^T \end{pmatrix}$. 3. Since $B, C$ are symmetric, $C^T=C$ and $B^T=B$. Conclusion: $(JH)^T = JH$, so this block form is the standard construction for a Hamiltonian matrix.

3. [Basics] Verify if $J = \begin{pmatrix} 0 & 1 \\ -1 & 0 \end{pmatrix}$ itself is a symplectic matrix.

Solution

Calculation: $J^T J J = (-J) J J = (-J)(-I) = J$. Conclusion: Yes. In 2D, $J$ is both symplectic and Hamiltonian.

4. [Spectral] If a Hamiltonian matrix has an eigenvalue $2+3i$, find the other three eigenvalues that must exist.

Solution

Using Pairing Properties: 1. Being real, it must have the conjugate: $2-3i$. 2. Being Hamiltonian, it must have the negative: $-(2+3i) = -2-3i$. 3. The conjugate of the negative: $-2+3i$. Conclusion: Eigenvalues appear in quartets $\{ \pm \lambda, \pm \bar{\lambda} \}$.

5. [Area Preservation] Prove that a symplectic operator in $\mathbb{R}^2$ preserves the area of a parallelogram.

Solution

Proof: 1. Area is given by the symplectic form $x^T J y$. 2. After transformation: $(Mx)^T J (My) = x^T (M^T J M) y$. 3. By definition of a symplectic matrix, this equals $x^T J y$. Conclusion: The signed area is invariant under symplectic transforms.

6. [Riccati] How are Hamiltonian matrices used to solve Algebraic Riccati Equations (ARE)?

Solution

Algebraic Link: The solution $X$ to an ARE corresponds to the stable invariant subspace of its associated Hamiltonian matrix $H$. By finding the eigenvectors of $H$ and performing block combinations, the positive definite solution can be constructed directly, bypassing non-linear iterations.

7. [Inverse] Prove: If $M$ is symplectic, then its inverse $M^{-1}$ is also symplectic.

Solution

Proof: 1. Given $M^T J M = J$. 2. Left-multiply by $(M^T)^{-1}$ and right-multiply by $M^{-1}$: 3. $J = (M^T)^{-1} J M^{-1} = (M^{-1})^T J M^{-1}$. Conclusion: This satisfies the definition of a symplectic matrix.

8. [Lie Algebra] Prove: The commutator $[H_1, H_2]$ of two Hamiltonian matrices is Hamiltonian.

Solution

Reasoning: This is because Hamiltonian matrices form the Lie Algebra $\mathfrak{sp}(2n)$ of the symplectic group $Sp(2n)$. A Lie algebra is by definition closed under the commutator bracket.

9. [Basics] Write the inverse of the $4 \times 4$ standard symplectic matrix $J$.

Solution

Conclusion: $-J$. Reasoning: Since $J^2 = -I$, $J(-J) = I$. In physics, this corresponds to reversing the orientation of the symplectic rotation.

10. [Application] Briefly state the advantage of "Symplectic Integrators" in orbital calculations.

Solution

Reasoning: 1. Traditional numerical methods (like Runge-Kutta) accumulate energy errors over time, causing planetary orbits to drift. 2. Symplectic integrators strictly enforce $M^T J M = J$ at every step. 3. This ensures the Hamiltonian (total energy) of the system remains almost constant, preserving the closure and stability of orbits over vast time scales.

Chapter Summary¶

Symplectic and Hamiltonian matrices reveal the deep geometric order of dynamical systems:

Guardians of Area: Symplectic matrices ensure the incompressibility of flow in phase space, serving as the algebraic root of Liouville's theorem in statistical mechanics.
Symmetry of Spectra: The $\pm \lambda$ pairing of Hamiltonian eigenvalues perfectly characterizes local stability near equilibrium in conservative systems, providing a natural basis for identifying stable manifolds.
Bridge to Control: Through the link with Riccati equations, symplectic algebra enables the leap from abstract physical symmetry to engineering optimal control solutions, serving as a powerful pillar of modern systems science.