Poincaré Sections and Return Maps

The Poincaré section is arguably the most heavily industrialized object in this chapter. In accelerator physics it is not an analysis tool but the native description: a stored particle's turn-by-turn coordinates at a fixed azimuth of the ring are literally iterates of a return map, and storage-ring design is the engineering of that map — tune (rotation number) placement away from low-order rationals, dynamic-aperture surveys that are section portraits, island chains and chaotic layers appearing exactly per Derivation C. In magnetic-confinement fusion, field-line Poincaré plots are the daily diagnostic of stellarator and tokamak design: nested invariant curves are good magnetic surfaces, island chains are resonant field errors, and stochastic (dust-filled) regions leak heat — a KAM textbook rendered in plasma. Celestial mechanics maps stability of exoplanetary systems and designs low-energy spacecraft trajectories through the tangle-generated transport channels of the three-body problem — the "interplanetary superhighway" is a homoclinic web used as infrastructure. Within this course, the section is the connective instrument of the whole chaos block: it defined the maps whose multipliers ran the cascade, displayed the driven pendulum's strange attractor and the double pendulum's KAM transition, and carries Hamiltonian chaos (via area-preserving maps like the standard map) to the KAM and Chirikov material that closes the track. One article remains: the chapter's simplest object — a one-dimensional map studied on its own terms — where everything the section technology reveals can be computed to arbitrary depth.

A long-exposure photograph of an orbit is a blur; the section is the fix. Stretch a pane of glass across phase space where the flow punches through it cleanly, and mark a dot at every piercing (counting only one crossing direction, so each lap contributes once). The blur becomes a ledger of returns: the section discards the one boring fact — that the trajectory keeps moving forward — and keeps everything about whether, where, and how it comes back. In three dimensions, the pane is two-dimensional and the flow has become a planar map: the entire arsenal built for planar systems earlier in this chapter applies again, one level up.

The dots organize into a reading chart that the whole chapter has been using informally. A periodic orbit pierces the pane at the same spot every lap: finitely many dots, forever — one for period-$T$, $k$ for period-$kT$, and the doubling cascade is dots splitting in two. A quasiperiodic orbit winds on a torus; the torus meets the pane in a circle, and the dots hop around that circle by a fixed angle each return, never repeating (irrational rotation) but filling the curve densely — the nested rings of the double pendulum's low-energy sections. Chaos scatters dots into structured dust: fractal layering if the system is dissipative (the driven pendulum's attractor cross-section), a fat chaotic sea with island chains if it is Hamiltonian. One glance at a section and the dynamical taxonomy is read off.

Why is the map sharper than the flow it samples? Because of what it silently removes. Perturbations along the orbit direction always carry the marginal multiplier $1$ — they merely shift the clock — and in the flow's monodromy matrix this trivial eigenvalue sits in the way of every stability statement. The section quotients it out by construction: deviations within $\Sigma$ are exactly the deviations that matter, and the eigenvalues of $DP$ are exactly the nontrivial Floquet multipliers. Better still, the return map converts the continuous bifurcation zoo into the discrete dictionary already mastered in the cascade articles: a multiplier through $+1$ collides orbits (saddle-node), through $-1$ doubles the period (flip), and a complex pair exiting the unit circle births an invariant curve — a torus of the flow — in the Neimark–Sacker bifurcation. And the section is where chaos was discovered. On it, a saddle fixed point carries one-dimensional stable and unstable manifolds — curves, drawable. Poincaré, mapping the restricted three-body problem, asked how the unstable manifold returns and found it must cross the stable one transversally; but the image of a crossing point under $P$ is another crossing point, so one intersection begets infinitely many, and the unstable manifold must fold back and forth across the stable one in ever-wilder oscillations without ever self-intersecting. The resulting homoclinic tangle forces stretch-and-fold dynamics — Smale later showed a horseshoe map, with all its chaos, lives inside every transverse tangle. The figure below is the picture Poincaré described and declined to draw.

The setup: an autonomous flow and a well-placed pane of glass,

where: $g$ — any smooth scalar whose zero set is the section surface; the transversality condition $\nabla g \cdot \mathbf{f} \neq 0$ says the flow crosses $\Sigma$ at a nonzero angle, never grazing it. Transversality is the entire licence for what follows: it guarantees (Derivation A) a smooth, locally unique first-return time $\tau(\mathbf{x})$, and with it the Poincaré map

where: $\boldsymbol{\phi}_t$ — the flow map (solution operator); $P$ maps the $(n-1)$-dimensional surface to itself. Note that the return time varies from point to point — the stroboscopic map of driven systems, $P = \boldsymbol{\phi}_T$ with fixed $T = 2\pi/\omega_{\text{drive}}$, is the special case where the drive phase supplies a section with constant return time. The dictionary begins with periodic orbits:

where: a $k$-cycle of $P$ likewise corresponds to a closed orbit crossing the pane $k$ times. Stability of the orbit is stability of the fixed point: linearize the map,

where: $\mu_1, \ldots, \mu_{n-1}$ — the characteristic (Floquet) multipliers of the periodic orbit. Their relation to the flow's own linearization is the monodromy matrix over one period,

where: $\mathcal{T}\exp$ — the time-ordered exponential (the propagator of the variational equation from the Lyapunov article, evaluated over one period). The displayed identity is the key structural fact: the velocity vector itself solves the variational equation, so $M$ always has the eigenvalue $1$ along the flow — useless for stability, and removed by the section: the spectrum of $M$ is $\{1\} \cup \{\mu_i\}$ (Derivation A). Volume bookkeeping crosses over intact. Liouville's formula for the variational determinant gives, per return,

where: the integral runs along the orbit. Hamiltonian flow: $\nabla\cdot\mathbf{f} = 0$, so $\det DP = 1$ — the return map is area-preserving (indeed symplectic), multipliers come in reciprocal pairs, and nothing can attract: the KAM world of curves, islands, and seas. Dissipative flow: $|\det DP| < 1$ — dots are drawn inward, and fixed points, invariant circles, or fractal attractor sections await. The driven pendulum's $\det \mathrm{D}P = e^{-2\beta T} \approx 0.009$ was equation (6) with constant divergence. Finally, the bifurcation dictionary — the classification this chapter has used piecemeal, now in one line:

where: the exits through $\pm 1$ were the working machinery of the cascade articles; the complex exit creates an attracting invariant curve around the fixed point — the section of a newborn torus, the discrete sibling of the Hopf bifurcation that opened this chapter.

Which statement answers which question

1 · Reading a stroboscopic ledger (confidence builder)

A mechanical oscillator driven at $f_d = 50$ Hz is sampled once per drive cycle. The section shows 6 dots, and the fitted return-map Jacobian at the associated cycle has eigenvalues $0.3$ and $-0.8$. Classify the motion, its stability, and predict what happens if the second multiplier drifts to $-1.02$.

Step 1. Six dots = a 6-cycle of the stroboscopic map: the response repeats every $6$ drive periods — a period-6 subharmonic at $f_d/6 = 8.33$ Hz, period $0.12$ s.

Step 2. Stability: both multipliers inside the unit circle ($|0.3| < 1$, $|-0.8| < 1$): the subharmonic is stable. The negative sign of $-0.8$ means transverse deviations alternate sides at each return — harmless, but a warning of which exit is nearest.

Step 3. If the multiplier reaches $-1.02$: the $-1$ exit is the flip — the 6-cycle destabilizes and a stable 12-cycle is born; the section's 6 dots each split in two. By the cascade articles, expect further doublings at parameter intervals shrinking by $\delta = 4.669$, with chaos an accumulation point away.

2 · One number decides a limit cycle (exam level)

Using the planar multiplier formula, evaluate the stability of the van der Pol cycle at $\mu = 0.1$ and at $\mu = 0.5$, and give the number of laps needed for a 10% amplitude perturbation to decay below 1%.

Step 1. From Derivation B, the cycle's single multiplier is $\mu_{\text{cyc}} = \exp\big(\oint\nabla\cdot\mathbf{f}\,\mathrm{d}t\big) = e^{-2\pi\mu}$ in the small-$\mu$ regime (using $x \approx 2\cos t$).

Step 2. At $\mu = 0.1$: $\mu_{\text{cyc}} = e^{-0.628} = \boxed{0.53}$ — perturbations halve each lap. At $\mu = 0.5$: $e^{-3.14} = \boxed{0.043}$ — near-deadbeat return to the cycle in one lap (though at $\mu = 0.5$ the $x = 2\cos t$ approximation is already fraying; treat as an estimate).

Step 3. Decay laps at $\mu = 0.1$: need $0.53^n < 0.1$, i.e. $n > \ln 10/0.628 = 3.7$ — $\boxed{4}$ laps. The section converts "how stable is this oscillation" from a vague adjective into a per-lap contraction factor.

3 · Rotation numbers and dot counts (applied)

A two-frequency system has $\omega_1 = 1.000$ and $\omega_2 = 0.625$ (rad/s). (i) Describe its section at $\theta_1 = 0$. (ii) Repeat for $\omega_2 = 0.618034$. (iii) How would you distinguish the two cases from 200 measured returns?

Step 1. $\rho = 0.625 = 5/8$: rational. The return map is rotation by $2\pi\cdot 5/8$; every orbit closes after 8 returns. Section: exactly $\boxed{8}$ dots, visited in the fixed order $0, 5, 2, 7, 4, 1, 6, 3$ (multiples of 5 mod 8) — a mode-locked $8{:}5$ resonance.

Step 2. $\rho = 0.618034 \approx (\sqrt5 - 1)/2$: the golden mean, maximally irrational. Dots never repeat; after $n$ returns they fill the invariant circle to spacing $\sim 2\pi/n$ with famous uniformity. Section: a smooth closed curve materializing dot by dot.

Step 3. With 200 returns: case (i) shows 8 point-clusters whose widths are set only by noise — additional returns revisit clusters, never fill gaps; case (ii) shows 200 distinct points with maximum gap shrinking like $1/n$. Quantitatively, plot the number of distinct dot positions (at measurement resolution) versus $n$: saturation at 8 versus linear growth is unambiguous by $n \approx 30$. This dot-count census is exactly how mode-locking tongues are mapped in driven-oscillator experiments.

References. Strogatz, Nonlinear Dynamics and Chaos, 2nd ed., §8.7 (Poincaré maps) · Guckenheimer & Holmes, Nonlinear Oscillations, Dynamical Systems, and Bifurcations of Vector Fields, §1.5 · Ott, Chaos in Dynamical Systems, 2nd ed., ch. 3–4 · Goldstein, Poole & Safko, Classical Mechanics, 3rd ed., ch. 11 · Lichtenberg & Lieberman, Regular and Chaotic Dynamics, 2nd ed. · Hénon & Heiles (1964), Astron. J. 69, 73

Provenance. Construction of $P$ via transversality and the implicit function theorem follows Guckenheimer–Holmes §1.5 and Strogatz §8.7. Monodromy structure, trivial multiplier $M\mathbf{f}=\mathbf{f}$, and $\det DP=\exp\int\nabla\cdot\mathbf{f}\,\mathrm{d}t$ are standard Floquet/Liouville results (Ott ch. 3–4); a verifier should re-derive the trivial multiplier by differentiating the flow equation and the determinant formula from Jacobi's identity. The planar cycle multiplier $\exp\oint\nabla\cdot\mathbf{f}\,\mathrm{d}t$ matches Strogatz §7.1; the van der Pol evaluation $\oint\mu(1-4\cos^2t)\,\mathrm{d}t=-2\pi\mu$ is recomputed directly. The bifurcation dictionary at $\mu=+1,-1,e^{\pm i\omega}$ is textbook (Guckenheimer–Holmes ch. 3). Torus section = rigid rotation and the rational/irrational dichotomy follow the action-angle normal form with Weyl equidistribution.

Builds on: Phase Portraits and the Phase Plane · Fixed Points and Equilibria · Linear Stability Analysis and the Jacobian · Classification of Fixed Points: Nodes, Saddles, Spirals, Centers · Conservative Systems and Hamiltonian Phase Flow Leads to: Poincaré–Bendixson Theorem · Limit Cycles and Self-Sustained Oscillations · Hopf Bifurcation and Birth of Limit Cycles · Period-Doubling Cascade and the Route to Chaos · Bifurcations and the Loss of Stability