The Strange Mathematics of Pursuit Curves

The setup is almost a children's puzzle. A duck swims in a straight line across a pond at a constant speed. A dog enters the water and begins swimming directly toward the duck, also at a constant speed. The dog continually adjusts its heading so that it is always pointed at the duck's current position. What path does the dog trace through the water?

The path is not a straight line — except in the special case where the dog and duck move along the same line. It is not a circle, not a parabola, not an ellipse, not any of the conic sections that the schoolroom inventory of curves makes available. It is its own thing, and it has a name: the pursuit curve. The first person to derive its equation was Pierre Bouguer, a French hydrographer and mathematician, in 1732. The problem looked like a recreational puzzle in the early eighteenth century. Three centuries later, the same equations describe the trajectory of every infrared-guided missile, the hunting strategies of dragonflies and falcons, the contact-tracing dynamics of disease transmission, and the spiral arm structure of certain galaxies.

Bouguer's analysis was one of the first non-trivial differential equation derivations in the engineering literature, and it set up a class of problems — pursuit curves, evasion curves, optimal-pursuit games — that has remained mathematically active ever since.

The basic equation

Set the duck moving along the y-axis, starting at the origin, with constant speed v. Place the dog at position (a, 0), at time zero, with constant speed kv where k is the speed ratio. The dog's instantaneous heading is along the line from its current position to the duck's current position. The duck's position at time t is (0, vt); the dog's heading is along the vector from the dog to that point.

The differential equation that emerges, after some algebraic manipulation that converts the parametric system into a relation between the dog's x and y coordinates, is:

x · y'' = k · sqrt(1 + (y')²)

This is the pursuit equation, and its solution depends on whether k is greater than, less than, or equal to 1.

If k > 1 (the dog is faster than the duck), the dog catches the duck in finite time. The path is a curve that asymptotically aligns with the duck's direction of motion before contact, and the closing time is a function of the initial distance and speed ratio.

If k < 1 (the dog is slower than the duck), the dog never catches up. The dog's path approaches a straight line behind the duck, with the gap growing without bound. The path itself is a well-defined curve, just one that never terminates in capture.

If k = 1 (equal speeds), the boundary case yields a path that asymptotically converges to a fixed lag distance behind the duck — the dog never catches up, but the gap stops growing. This is the case that produced the cleanest closed-form solution and gave the curve its mathematical character.

From recreational mathematics to missile guidance

The pursuit curve sat in the recreational-mathematics literature for two centuries. Bouguer's derivation was rederived by various authors with various complications added — multiple pursuers, evading prey that adjusts speed, prey that turns at known intervals — and the field accumulated a small library of variant solutions. The applications were mostly thought-experiments.

The applied turn came with World War II. The development of guided weapons created an urgent practical need to understand pursuit dynamics, and the Bouguer framework was the natural starting point. The simplest guidance law, called pure pursuit, sets the missile's heading directly at the target's instantaneous position — exactly the dog-and-duck setup. Pure pursuit works against slow targets but is inefficient against fast ones, because the missile spends most of its time turning rather than closing distance.

The improvement that came from the pursuit-curve mathematics was proportional navigation, in which the missile's heading rate is set proportional to the line-of-sight rate to the target. The math says that for any constant target velocity, proportional navigation produces a straight-line interception trajectory that minimizes turning energy. This is the guidance law used in almost every modern infrared and radar-guided missile, and it traces directly back to the pursuit-curve analysis Bouguer started in 1732.

The biological convergence

The strange thing about pursuit dynamics is that biological predators figured them out independently, by evolution rather than analysis. The dragonfly's hunting strategy was an open question in entomology until 2014, when Stacey Combes at Harvard and a team using high-speed video showed that dragonflies use a strategy mathematically equivalent to proportional navigation. The dragonfly maintains a constant line-of-sight angle to its prey while closing, producing the straight-line intercept trajectory that the missile guidance literature had identified as optimal. The neural circuits that implement this in the dragonfly's tiny brain are now an active research area, and the convergent evolution between biology and engineering — both arriving at the same mathematical solution — is the kind of result that makes the field rewarding.

The peregrine falcon's high-speed hunting strategy uses a related but distinct mathematical structure called an attached-eye-fixation pursuit, in which the falcon keeps the prey image fixed at a particular angle on its retina rather than at the center of vision. The mathematical analysis, worked out by Vance Tucker in the 2000s, shows that this slightly off-axis strategy is aerodynamically superior at the high speeds falcons reach in dives, because it allows them to maintain a head position with lower drag than direct-pursuit head-on viewing.

Predator-prey dynamics in ecology more broadly use pursuit-curve mathematics in the analysis of capture probabilities, optimal escape strategies, and the evolutionary stability of various behavioral combinations. The Lotka-Volterra equations of population dynamics are a higher-level abstraction, but the per-encounter capture probability that feeds into them depends on pursuit-curve analysis at the individual level.

The epidemiological adaptation

The contact-tracing problem in epidemiology turns out to have pursuit-curve structure. Disease propagation is a pursuit problem in which the disease is the pursuer and the susceptible population is the evader, with the additional complication that the population is a stochastic distribution rather than a single point. The analysis frames detection-and-isolation as a pursuit-rate problem: how fast does contact tracing have to be, relative to disease propagation, to bring R-effective below 1?

The answer, which is implicit in the pursuit-curve mathematics, is that the catch-up condition is a speed ratio above 1 — the tracing system has to be faster than the disease's transmission rate. The exact threshold depends on the disease's serial interval and the proportion of transmission occurring before symptom onset, but the mathematical framing was useful during COVID-19 modeling and explained why some countries' contact-tracing systems were effective at controlling outbreaks while others, despite identical policy intentions, were structurally too slow to catch up.

The galactic case

The galactic-arm application is more recent and more speculative. Some galaxies — particularly grand-design spirals — have arm structures that fit logarithmic spiral models well, but a subset of barred and irregular galaxies have arm structures that fit pursuit-curve models better than density-wave or material-arm models. The interpretation, due to work by Frank Shu and collaborators, is that the arms are produced by stellar populations chasing density gradients rather than orbiting in coordinated patterns. The mathematics of stellar pursuit in a rotating frame produces structures that match Bouguer-style pursuit curves at galactic scale, three centuries and many orders of magnitude removed from a dog chasing a duck.

The deeper observation

The pursuit curve is one of those mathematical structures that started as a self-contained puzzle and turned out to describe a wide range of natural and engineered phenomena unconnected to the original problem. The list at this point includes: missile guidance, dragonfly hunting, falcon hunting, predator-prey ecological dynamics, contact-tracing in epidemiology, certain galactic arm structures, the path of a ship trying to intercept another ship, optimal-search-and-rescue strategies, the behavior of certain robotic vehicles, and the dynamics of group-formation in animal swarms.

This is the Wigner unreasonable-effectiveness pattern: a mathematical structure developed for one problem turns out to be the right description for problems that nobody anticipated when the structure was developed. Bouguer was thinking about a recreational dog-and-duck problem. He produced a differential equation that, three centuries later, sits at the heart of how missiles intercept aircraft, how dragonflies catch mosquitoes, and how galaxies maintain spiral structure. The world appears to have a small library of these mathematical structures, and the work of physics, biology, and engineering is partly the work of recognizing which library function is the right one for each problem.

The pursuit curve is a good example because it is simple — three lines of differential equation, derivable in a long afternoon — and its applications are wide. Most of the deeper mathematical structures of the modern world have similar genealogies. The Fourier transform was developed for heat conduction; it underlies everything from MP3 encoding to MRI imaging. Group theory was developed for solving polynomial equations; it underlies particle physics and chemistry. The pursuit curve is a small example, but the lesson is the same: pure-mathematical structures developed in one century turn out to describe the practical problems of the next, and the world is more mathematically structured than it has any obvious right to be.