Hilbert’s sixteenth problem: limit cycles

Imagine arrows at every point of a plane telling a moving dot which way to go. One rule makes a dot one unit from the centre keep circling, while nearby dots inside spiral outward and those outside spiral inward. That repeating loop has no neighbouring repeating loops: it is an isolated periodic orbit, called a limit cycle. Polynomial rules add and multiply coordinates and their whole-number powers. Once their degree—the largest total power in any term—is fixed, is there one finite cap on the number of limit cycles across all such rules?[1] [2]

Arrows telling a moving dot its direction and speed at each point in a plane define a motion rule. When these arrows come from polynomial formulas, how many isolated repeating trajectories can occur with the degree fixed? Write the motion as $x'=P(x,y)$, $y'=Q(x,y)$, with real polynomials of degree at most $d$; primes denote rates of change with time. A limit cycle is an isolated nonstationary periodic orbit; an equilibrium point is not one. Define $H(d)$ as their largest possible count over all such systems, or infinity if those counts are unbounded; this is their supremum. The uniform question asks whether $H(d)$ is finite for each $d$, then asks for its sharp value and possible arrangements.[1] [2]

A single formula can have one isolated loop, while another of the same degree can have more. Even knowing that each individual system has finitely many does not give one cap for the whole family: a collection of finite counts can still grow without bound. The general cap depending only on degree remains unknown, along with the sharp counts and possible arrangements.[1] [2] [4] [5]

Individual finiteness and finiteness of H(d) are different claims. Classical individual-finiteness results do not supply a common bound across all coefficient choices at a fixed degree. The missing estimate must depend only on d, rather than on the particular polynomial pair or a chosen region of its flow. A bound for a restricted family or a different piecewise-defined model does not answer this unrestricted question; sharp counts and the possible arrangements remain further targets.[1] [2] [4] [5]

A limit cycle describes a persistent oscillation with a particular path and amplitude, rather than a motion that settles to rest. Different isolated loops can represent different repeating regimes in the same mathematical system. A degree-only bound would connect the complexity of the local motion rule to how many such long-term patterns can coexist, without needing to solve every trajectory explicitly.[2]

Near a periodic orbit, choose a short transverse segment—a small crossing line—and follow each nearby starting point until its path returns to the segment. This gives a return map. A periodic orbit is a fixed point of that map, and an isolated one is an isolated zero of the return map minus its input. Loop counting is thus connected to controlling zeros of functions built by following a flow, not merely counting roots of the original polynomials. A uniform degree bound would tie those global return behaviours back to the finite algebraic complexity of the local rule.[4] [5]