the KAM theorem
/ K-A-M (Kolmogorov, Arnold, Moser) /
A pristine solar system, if each planet felt only the Sun, would trace perfect, eternally repeating orbits. But the planets tug on one another too. Does that small mutual pull, acting over eons, eventually derail the orderly clockwork into chaos, or does the order largely survive? The KAM theorem is the deep answer: most of the orderly motion survives a small enough disturbance, but not all of it.
The setting is a Hamiltonian system that is integrable, meaning its motion, in the right (action-angle) coordinates, is a set of independent, steady rotations that wind forever around nested tori (doughnut-shaped surfaces) in phase space, each labeled by conserved actions. The Kolmogorov-Arnold-Moser theorem states that when such a system is perturbed by a small non-integrable term, most of these invariant tori are not destroyed but merely deformed, and quasi-periodic motion persists on them, provided the perturbation is small enough and the system meets a non-degeneracy condition. The tori that survive are precisely those whose frequency ratios are sufficiently irrational, far from any low-order resonance in the precise sense of a Diophantine condition. Tori with rational or near-rational frequency ratios are the ones that break up first, seeding thin layers of chaos near the resonances.
KAM (announced by Kolmogorov in 1954, with rigorous proofs by Arnold and Moser around 1962-63) is the resolution of a centuries-old worry about the stability of the solar system, and the bridge between integrable order and full chaos: it explains why phase space at small perturbation is a marbled mixture of surviving KAM tori and thin chaotic layers, rather than all-or-nothing. Honest caveats: the theorem is perturbative, guaranteeing survival only for sufficiently small perturbations, and the rigorous threshold is often far smaller than what is observed numerically. In systems with three or more degrees of freedom the surviving tori no longer separate phase space into sealed compartments, so trajectories can slowly leak through the chaotic web over enormous times, a phenomenon called Arnold diffusion.
For the outer solar system, KAM-type results help explain why the giant planets' orbits, though mutually perturbing, remain quasi-periodic and bounded over billions of years rather than promptly dissolving into chaos.
Why the clockwork mostly holds: gently deformed invariant tori survive a small enough disturbance.
KAM does not promise eternal stability for any perturbation strength: it is a small-perturbation theorem, its guaranteed thresholds are far stricter than numerics suggest, and for three or more degrees of freedom the surviving tori no longer wall off phase space, allowing slow Arnold diffusion.