the Poincare map
/ pwan-kah-RAY map /
A continuous trajectory in three dimensions, looping and weaving, is hard to look at and harder to analyze. Henri Poincare's brilliant simplification was to stop watching the whole flowing curve and instead place a flat screen across its path, recording only the spots where the trajectory pierces the screen. The smooth, tangled flow becomes a sequence of dots on a single surface — and the rule that sends each dot to the next is a map, one dimension lower than the flow. That dimension-reducing device is the Poincare map.
Concretely, choose a surface (the Poincare section) transverse to the flow, so trajectories cross it cleanly rather than running along it. Watch a trajectory and mark each successive point P0, P1, P2, ... where it punches through the section in the same direction. The Poincare map (or first-return map) is the function that takes each piercing point to the next, P_(n+1) = f(P_n) — it answers 'given where the orbit crosses now, where does it cross next?' The crucial trade is dimensional: a flow in three dimensions becomes a map on a two-dimensional surface, and a single continuous variable's worth of complexity is removed. The dynamics translate faithfully: a periodic orbit of the flow becomes a fixed point of the map (it returns to the same spot each loop); a period-2 orbit becomes a 2-cycle; a chaotic flow leaves a fractal scatter of dots — a cross-section of the strange attractor.
The Poincare map is the bridge between the two halves of dynamical systems: it lets the powerful, visualizable theory of maps (fixed points, period-doubling, the logistic-map picture) be applied to the harder world of continuous flows. The stability of a periodic orbit, hard to judge directly, becomes the easy-to-check stability of a fixed point of the map. And it is the practical instrument experimenters use to detect chaos in a real continuous system: sample the state once per cycle, plot the returns, and read off whether you see a point, a loop, or a fractal.
For a forced oscillator driven at period T, take the Poincare section to be 'sample the state once every drive period'. If the response is periodic and locked to the drive, the snapshots land on one point. If the response has doubled its period, they alternate between two points. If the system is chaotic, the snapshots scatter into a fractal cloud — the Poincare map exposes the structure the full trajectory hides.
Sampling a flow once per cycle turns a tangled curve into dots: one point for periodic, a few for a cycle, a fractal for chaos.
The section must be genuinely transverse to the flow — trajectories should cross it, not graze along it — or 'returns' are ill-defined. And the map exists only locally, for trajectories that actually come back to the section; not every flow has a clean global return surface, so a Poincare map is a powerful tool, not a universal one.