Geometry of Almost-Conserved Quantities in Symplectic Maps. Part I: Perturbation Theory

1. Problem Statement

The paper addresses the challenge of characterizing the long-term stability and structure of discrete-time, symplectic, and reversible dynamical systems (symplectic maps), particularly those that are non-integrable or chaotic.

• Context: While Noether's theorem links continuous symmetries to exact conservation laws in continuous-time Hamiltonian systems, discrete symplectic maps (common in accelerator physics and celestial mechanics) often lack exact invariants.
• The Gap: Traditional perturbation methods, such as Lie algebra transformations to Birkhoff normal forms, are effective for local analysis near stable fixed points but fail to capture global structures, resonances, and the geometry of chaotic regions (e.g., island chains, separatrices). Furthermore, standard power series expansions often diverge or lose predictive power near resonances (small denominators).
• Goal: To develop a systematic perturbative framework that constructs approximate invariants (almost-conserved quantities) for symplectic maps by leveraging their inherent discrete symmetries (reversibility), providing a geometric description of near-integrable behavior even in chaotic regimes.

2. Methodology

The authors propose a perturbative construction of an approximate invariant $K^{(n)}[p, q]$ for a map $T$ defined in the McMillan form:
$$q' = p, \quad p' = -q + f(p)$$
where $f(p)$ is a smooth force function. The map is reversible, meaning it can be decomposed into two involutions (reversors) $R_1$ and $R_2$.

Key Steps in the Method:

1. Symmetry-Driven Ansatz: Instead of demanding exact invariance ($K[T(\zeta)] - K[\zeta] = 0$), the authors seek an approximate invariant satisfying $K^{(n)}[T(\zeta)] - K^{(n)}[\zeta] = O(\epsilon^{n+1})$. They construct $K^{(n)}$ as a polynomial series in the phase space variables $(p, q)$, expanded in a small parameter $\epsilon$ representing amplitude.
2. Symmetry Constraints: The construction enforces the map's discrete symmetries. Specifically, the approximate invariant must respect the symmetry lines $p=q$ and $p=f(q)/2$. This reduces the number of free coefficients in the polynomial expansion.
3. Handling Indeterminacy (Averaging Procedure): The perturbation equations yield an underdetermined system where certain coefficients (denoted $C_i$) cannot be fixed by the invariance condition alone. To resolve this, the authors introduce an averaging procedure:
   • Transform variables to Floquet coordinates (action-angle-like variables).
   • Minimize the integral of the square of the residual error ($\int R_n^2 d\phi$) over one period.
   • This procedure uniquely determines the free coefficients, effectively selecting the "best" approximate invariant at each order.
4. Resonant Theory: The averaging naturally leads to a resonant theory. By multiplying the invariant by products of resonant denominators ($r_k$), the authors construct non-singular averaged invariants that remain well-defined even when the rotation number is rational (commensurate with unity), a regime where standard perturbation theory typically diverges.
5. Connection to Action-Angle Variables: Using Danilov's Theorem, the authors derive the rotation number $\nu(J)$ and twist coefficients ($\tau_i = d^i\nu/dJ^i$) directly from the constructed approximate invariants, allowing for a direct comparison with Lie algebra methods.

3. Key Contributions

• Conceptual Parallel to Noether: The paper establishes a conceptual bridge between continuous symmetries (Noether's theorem) and discrete symmetries in reversible maps. It demonstrates that discrete symmetries can generate approximate invariants that organize the phase space, even in the absence of exact conservation laws.
• Systematic Perturbative Framework: A novel algorithm to construct approximate invariants order-by-order using only linear algebra and elementary integrals, avoiding the complexity of full Lie algebra transformations for global structures.
• Resolution of Small-Denominator Problems: The introduction of the averaging procedure and the construction of non-singular invariants allows the method to handle low-order resonances (e.g., 1/3, 1/4, 1/5) where traditional expansions fail.
• Geometric Insight: The method reveals the geometry of "island chains" and the structure of separatrices in chaotic maps (like the Hénon map) by analyzing the level sets of the approximate invariants.

4. Results

The authors benchmarked their method against established techniques (Lie algebra, numerical tracking) using the Hénon map (quadratic and cubic) and McMillan maps.

• Twist Coefficients: The method successfully reproduces the twist coefficients ($\tau_0, \tau_1, \dots$) derived from Lie algebra techniques. Specifically, the $n$-th order approximation converges to the exact twist coefficients up to order $\lfloor n/2 \rfloor - 1$.
• Resonance Handling:
  • Quarter-integer resonance ($\nu_0 = 1/4$): The method correctly predicts the transition from circular to square-like invariant structures and the formation of 4-island chains.
  • Third-integer resonance ($\nu_0 = 1/3$): The method captures the "touch-and-go" bifurcation and the complex tangle of stable/unstable manifolds near the origin, providing accurate approximations of the invariant curves even in the presence of local instability.
• Comparison with Numerical Tracking:
  • The approximate invariants provide significantly better estimates of the rotation number $\nu(J)$ and the stability boundary (dynamic aperture) compared to traditional Taylor series expansions.
  • The "averaged" invariants (using the $C_i$ coefficients) show superior agreement with numerical tracking than non-averaged versions, particularly near resonances.
• Global Structure: For integrable McMillan maps, the perturbation series converges exactly to the known polynomial invariants. For maps with infinite series invariants, the method shows almost-everywhere convergence.

5. Significance and Implications

• Accelerator Physics: The work offers a powerful tool for estimating dynamic apertures in particle accelerators. It explains why accelerators remain stable despite nonlinear imperfections: low-order approximate invariants (hidden near the origin) persist and organize the phase space, preventing immediate chaos.
• Unified Framework: The method unifies the treatment of integrable and chaotic systems. It suggests that even in chaotic regimes, the underlying discrete symmetries impose a "skeleton" of order that can be captured by low-order approximations.
• Practical Utility: The approach requires only linear algebra and integration of elementary functions, making it computationally efficient and analytically transparent compared to heavy numerical simulations or complex Lie algebra computations.
• Future Directions: The paper sets the stage for Part II, which applies these methods to large-amplitude dynamics, global invariant reconstruction, and real-world accelerator lattices (e.g., Mu2e, Main Injector, IOTA), demonstrating that a single low-order approximate invariant can describe dynamics across diverse machine configurations.

In summary, this paper provides a rigorous geometric and algebraic framework for understanding the "almost-conserved" quantities in reversible symplectic maps, bridging the gap between local linear stability and global chaotic behavior through the systematic exploitation of discrete symmetries.