- Atlas
- Mathematics
- Dynamical Systems
Dynamical Systems
Dynamical systems is the branch of analysis that studies how things change over time when the rule of change is fixed.
11 frameworks, from 1881 to today · Analysis · Mathematics
- Qualitative theory of differential equations 1881-Present
- Stability theory 1892-Present
- Complex dynamics 1918-Present
- Bifurcation theory 1930-Present
- Ergodic theory 1931-Present
- Kolmogorov–Arnold–Moser theorem 1954-Present
- Hyperbolic dynamics 1960-Present
- Chaos theory 1963-Present
- Thermodynamic formalism 1968-Present
- One-dimensional dynamics 1975-Present
- Smooth ergodic theory 1977-Present
Overview
Dynamical systems is the branch of analysis that studies how things change over time when the rule of change is fixed. Given a state—a position, a population, a temperature—and a law that says what happens next, what happens in the long run1? Does the motion settle down, repeat, wander, or blow up? The subject's distinctive move is to stop asking for a formula that solves the equation and instead ask qualitative questions about the whole family of possible motions: the orbit traced by a single starting point, the phase portrait of all orbits at once, and how both respond when the rule is nudged. This reframing turned differential equations from a computational art into a geometric and measure-theoretic science, and it is why the field now reaches from planetary astronomy to weather prediction to the statistics of gases.
Newton, the three-body problem, and the failure of formulas
The subject begins with celestial mechanics. Newton's Principia (1687) gave the law of motion, and for two bodies gravitating alone the solution is complete: ellipses, parabolas, or hyperbolas, exactly as Kepler had described. Add a third body and the equations no longer integrate in closed form. Eighteenth-century mathematicians—Euler, Lagrange, Laplace—pursued perturbation theory, treating the extra pull as a small correction to the two-body answer, and produced remarkably accurate tables of planetary motion. But the method assumed the corrections stayed small forever, and nobody could prove that.
The crisis came with the problem of small divisors12. When two frequencies in a perturbative expansion nearly resonate, the correction terms acquire denominators that shrink toward zero, and the series threatens to diverge. This was not a technical nuisance but a genuine obstacle: it hinted that some motions might be far more complicated than ellipses. Poincaré, working on a prize problem on the stability of the solar system in the late 1880s, discovered that the perturbative series could behave wildly, and in doing so opened the door to the modern subject.
Poincaré and the qualitative theory
Henri Poincaré's Mémoires sur les courbes définies par une équation différentielle (1881–1886) founded the qualitative theory of differential equations. Instead of solving an equation, he drew its phase portrait: in the plane, the curves traced by solutions, the equilibrium points where they can rest, and the special closed orbits (periodic solutions) around which nearby motion may wind. He classified equilibria—nodes, saddles, foci, centers—by the local behavior of solutions near them, and introduced the Poincaré map, which records where a solution returns to a cross-section and converts a continuous flow into a discrete iteration. That device is the bridge between differential equations and the theory of maps.
Poincaré also saw that the three-body problem's phase portrait could contain homoclinic orbits—paths that leave a saddle and return to it—and that near such an orbit the dynamics is tangled in a way he called a "labyrinth." He did not have the tools to describe it fully. The Russian school carried the qualitative program forward: Lyapunov and, later, Andronov and Pontryagin developed the theory of structural stability—the question of whether a phase portrait survives small changes in the equation—and the bifurcation theory of how portraits change as a parameter is tuned.
Lyapunov stability
Aleksandr Lyapunov's doctoral thesis of 1892 gave the field its central notion of stability13. An equilibrium is stable if solutions starting close stay close forever, and asymptotically stable if they also converge to it3. Lyapunov's insight was that one need not solve the equation to decide this. If a scalar function—now called a Lyapunov function—decreases along every solution and has a minimum at the equilibrium, the equilibrium is stable; the function acts like an energy that the motion can only dissipate. The method, later extended by LaSalle and others, became the standard tool of control theory and remains one of the most portable ideas in the subject: it converts a question about long-run behavior into an inequality to be checked.
Measure, recurrence, and ergodic theory
A different question was posed by statistical physics. Boltzmann's ergodic hypothesis proposed that a gas, left to itself, eventually visits every configuration consistent with its energy, so that time averages equal averages over the space of states. Whether that is true, and what it even means, became the founding problem of ergodic theory.
The decisive step was George David Birkhoff's and John von Neumann's work of the early 1930s4. Birkhoff's ergodic theorem (1931) proved that for a measure-preserving transformation, time averages converge almost everywhere—so the ergodic hypothesis holds in a precise averaged sense, whether or not the orbit is literally dense14. Von Neumann had proved a mean-convergence version shortly before. Around the same time, Koopman's operator formulation recast the dynamics of a map as a unitary operator on functions, linking ergodic theory to spectral theory. Poincaré's earlier recurrence theorem—that a measure-preserving system returns arbitrarily close to its starting state—had already shown that recurrence is generic, not exceptional15.
The measure-theoretic classification of systems then became a program in its own right: Bernoulli shifts, mixing, and later Kolmogorov's and Sinai's entropy, a number measuring how fast a system generates information, which distinguished systems that had looked alike.
Birkhoff, Smale, and hyperbolic dynamics
The qualitative program reached its modern form with Stephen Smale's work in the 1960s. Smale asked which systems are structurally stable—robust under small perturbation—and found that the answer involves hyperbolicity: at every point, the tangent space splits into directions along which nearby orbits separate exponentially (unstable) and directions along which they converge (stable). The horseshoe map, a simple stretching-and-folding construction, showed that a structurally stable system could nonetheless contain infinitely many periodic orbits and chaotic behavior. Smale's Axiom A systems and the spectral decomposition organized such dynamics into finitely many basic pieces.
This was the era of the Smale school at Berkeley, and of parallel work by Anosov in the Soviet Union on what are now called Anosov diffeomorphisms. Hyperbolic dynamics gave the first rigorous examples of systems whose orbits are genuinely unpredictable yet whose statistical behavior is well behaved—the two facts that chaos theory would later popularize.
Chaos, strange attractors, and the computer
The word chaos entered the wider culture through Edward Lorenz's numerical experiments at MIT in the early 1960s. Running a simplified model of atmospheric convection, Lorenz found that truncating his initial data at the third decimal produced a completely different forecast—sensitive dependence on initial conditions, the butterfly effect. His 1963 paper described a system whose trajectories are confined to a bounded region yet never repeat, drawn to what became known as the Lorenz attractor, a fractal set of non-integer dimension.
The mathematical groundwork for such objects had been laid earlier. Smale's horseshoe, the solenoid, and the work of Ruelle and Takens in 1971 on strange attractors gave the theory its objects. Benoit Mandelbrot's fractal geometry (1975 onward) supplied the language of self-similarity. The 1970s saw chaos become a recognized subject, with the Feigenbaum universality of period-doubling (1978) showing that entirely different systems could share the same route to chaos.
One-dimensional dynamics and the Mandelbrot set
The simplest chaotic systems are maps of an interval, and they became a laboratory. The logistic map $x \mapsto r x(1-x)$, studied by May and others in the 1970s, exhibits a period-doubling cascade as $r$ increases, then chaos, then windows of order again. The Feigenbaum constants—approximately 4.669 and 2.502—describe the universal ratio in which the cascade's bifurcations crowd together, the same for a whole class of maps.
In the complex plane, the iteration of $z \mapsto z^2 + c$ produced the Mandelbrot set (drawn by Mandelbrot in 1980, with earlier work by Brooks, Matelski, and others)6. The Julia sets—the boundaries between starting points that escape and those that stay bounded—had been studied by Gaston Julia and Pierre Fatou around 1918, but the computer made them visible. The field of complex dynamics was reborn, and by the 1980s Douady and Hubbard had proved the Mandelbrot set connected and developed the theory of external rays and renormalization that organizes the parameter plane.
KAM theory: taming the small divisors
The small-divisor problem that had defeated Poincaré was solved, in a precise sense, in the 1950s and 1960s. Kolmogorov (1954), Arnold (1963), and Moser (1962) proved what is now called KAM theory: for a nearly integrable Hamiltonian system, most invariant tori survive a small perturbation, provided a Diophantine condition rules out the worst resonances. The result is a statement of partial stability—the solar system is not guaranteed to be stable forever, but a large set of initial conditions leads to quasi-periodic motion that persists. KAM theory required a new analytic technique, the Newton method with loss of derivatives, which has since become a standard tool.
Thermodynamic formalism and smooth ergodic theory
In the 1960s and 1970s, Sinai, Ruelle, and Bowen developed thermodynamic formalism, importing the machinery of statistical mechanics—Gibbs measures, pressure, entropy—into the study of hyperbolic systems. The idea is to assign a weight to each orbit according to how much it expands, and to find the invariant measure that maximizes a pressure functional. This gave a unified construction of the natural invariant measures for Axiom A systems and connected dynamics to the theory of phase transitions.
Smooth ergodic theory then asked how much of this survives when hyperbolicity is only partial or absent. The Pesin theory (1977) showed that nonuniform hyperbolicity—exponential separation along most, but not all, orbits—still yields invariant measures and entropy formulas. The subject remains active, with the Katok–Hasselblatt synthesis and ongoing work on systems that are neither uniformly hyperbolic nor integrable.
The present
Dynamical systems today is a meeting point rather than a single method. The qualitative and hyperbolic traditions descend through Smale and Anosov into the modern theory of uniformly and nonuniformly hyperbolic systems. Ergodic theory, from Birkhoff and von Neumann, has grown into a vast field touching number theory, probability, and operator algebras. KAM theory and perturbation methods remain the tools for celestial mechanics and Hamiltonian systems. One-dimensional and complex dynamics, energized by the Mandelbrot set, continue to produce deep results on renormalization and parameter spaces. Thermodynamic formalism and smooth ergodic theory connect the subject to statistical physics1.
Open questions are plentiful: the precise measure of stability in the solar system over astronomical timescales, the classification of smooth invariant measures for partially hyperbolic systems, the structure of parameter spaces beyond the quadratic family, and the extent to which the statistical behavior of chaotic systems can be predicted from their geometry. The field's founding insight—that one can understand motion without solving for it—remains its organizing principle.
References
- Dynamical system (introduction, Near periodic orbits, History, Ergodic systems), Wikipedia
- Kolmogorov–Arnold–Moser theorem, Wikipedia
- Lyapunov stability (History, introduction), Wikipedia
- Ergodic theory, Wikipedia
- Poincaré recurrence theorem, Wikipedia
- Mandelbrot set (introduction, History), Wikipedia
The frameworks of Dynamical Systems
Each framework is a way of working the field has used: when it held, and the question it answers.
- 1881-Present
Qualitative theory of differential equations
What do all solutions of an equation do in the long run?
- 1892-Present
- 1918-Present
Complex dynamics
What happens when a complex number is fed back into itself?
- 1930-Present
- 1931-Present
- 1954-Present
Kolmogorov–Arnold–Moser theorem
Which regular motions survive a small disturbance?
- 1960-Present
Hyperbolic dynamics
Which systems keep their structure under small changes?
- 1963-Present
- 1968-Present
Thermodynamic formalism
How much information does a system generate over time?
- 1975-Present
One-dimensional dynamics
How do simple maps on a line become complicated?
- 1977-Present
Smooth ergodic theory
How do smooth systems behave statistically almost everywhere?
How the ideas connect
Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.
Hyperbolic dynamics grew out of Qualitative theory of differential equations
Smale's program of structural stability and hyperbolic sets grew directly out of the qualitative theory's question of which phase portraits survive perturbation, extending it from the plane to higher dimensions.
Carried by Stephen Smale; Henri Poincaré.
Smooth ergodic theory grew out of Ergodic theory
Smooth ergodic theory grew out of ergodic theory by adding differentiability hypotheses, using Lyapunov exponents and Pesin theory to connect measure-theoretic behavior with the geometry of the tangent map.
Carried by Yakov Pesin; Anatole Katok.
Stability theory influenced by Qualitative theory of differential equations
Lyapunov built his stability theory on Poincaré's qualitative analysis of equilibria and their local behavior, giving the nodes, saddles and foci a quantitative criterion for whether nearby solutions stay close.
Carried by Aleksandr Lyapunov; Henri Poincaré.
Complex dynamics influenced by Qualitative theory of differential equations
Fatou and Julia took Poincaré's program of iterating a map and studying orbits and periodic points and applied it to rational maps of the complex plane, founding complex dynamics.
Carried by Pierre Fatou; Gaston Julia.
Bifurcation theory grew out of Qualitative theory of differential equations
Andronov and Pontryagin, working in Poincaré's qualitative tradition, studied how phase portraits change as a parameter varies and developed structural stability and the bifurcation of planar systems.
Carried by Aleksandr Andronov; Lev Pontryagin.
Ergodic theory influenced by Qualitative theory of differential equations
Poincaré's recurrence theorem, proved within his qualitative study of flows, supplied ergodic theory with its first structural result and the notion that recurrence is generic in measure-preserving systems.
Carried by Henri Poincaré; George David Birkhoff.
Kolmogorov–Arnold–Moser theorem influenced by Qualitative theory of differential equations
KAM theory resolved the small-divisor problem that Poincaré had encountered in the three-body problem, proving that most invariant tori survive perturbation and answering a question posed within the qualitative theory.
Carried by Andrey Kolmogorov; Vladimir Arnold; Jürgen Moser.
One-dimensional dynamics influenced by Complex dynamics
One-dimensional dynamics revived and extended the Fatou-Julia study of iterated maps, applying renormalization and the period-doubling picture to interval and circle maps.
Carried by Mitchell Feigenbaum.
One-dimensional dynamics influenced by Bifurcation theory
The period-doubling cascade is a bifurcation sequence, and Feigenbaum's universality was formulated as a statement about how bifurcations accumulate, drawing on the bifurcation theory of parameterized families.
Carried by Mitchell Feigenbaum.
Hyperbolic dynamics influenced by Ergodic theory
Hyperbolic dynamics borrowed the measure-theoretic and operator tools of ergodic theory, and Sinai and Bowen showed that Axiom A systems carry invariant measures with strong statistical properties.
Carried by Yakov Sinai; Rufus Bowen.
Thermodynamic formalism influenced by Ergodic theory
Thermodynamic formalism imported Kolmogorov and Sinai's entropy and the ergodic-theoretic classification of systems, recasting entropy as a pressure and connecting it to Gibbs measures.
Carried by Yakov Sinai; David Ruelle.
Chaos theory influenced by Hyperbolic dynamics
Chaos theory took its rigorous objects from hyperbolic dynamics: Smale's horseshoe and the solenoid supplied the first examples of systems with sensitive dependence and infinitely many periodic orbits.
Carried by Stephen Smale.
Thermodynamic formalism grew out of Hyperbolic dynamics
Thermodynamic formalism grew out of the study of hyperbolic and Axiom A systems, where Sinai, Ruelle and Bowen built Gibbs measures and pressure as tools for their invariant measures.
Carried by Yakov Sinai; David Ruelle; Rufus Bowen.
Smooth ergodic theory influenced by Hyperbolic dynamics
Pesin theory generalized hyperbolicity to nonuniform hyperbolicity, showing that almost every orbit of a smooth system with nonzero Lyapunov exponents behaves hyperbolically, a direct extension of Anosov-Smale ideas.
Carried by Yakov Pesin.
One-dimensional dynamics grew out of Chaos theory
The logistic map and the Feigenbaum cascade became the laboratory of chaos theory, and the universality of period-doubling was established through the one-dimensional maps studied in this period.
Carried by Robert May; Mitchell Feigenbaum.
Bifurcation theory influenced by Stability theory
The loss of stability of an equilibrium as a parameter is tuned is the trigger for a bifurcation, so bifurcation theory took Lyapunov's stability criteria as the signal that a phase portrait is about to change.
Carried by Aleksandr Andronov; Aleksandr Lyapunov.
Smooth ergodic theory influenced by Stability theory
Lyapunov exponents, the quantitative measure of separation rates central to smooth ergodic theory, descend from Lyapunov's stability analysis of linearized systems.
Carried by Aleksandr Lyapunov.
Hyperbolic dynamics influenced by Bifurcation theory
The study of how portraits change under parameter variation fed Smale's question of which systems are robust, since structural stability is precisely the absence of bifurcation under small perturbation.
Carried by Stephen Smale.
Hyperbolic dynamics rival of Kolmogorov–Arnold–Moser theorem
KAM theory and hyperbolic dynamics offered competing pictures of typical Hamiltonian behavior in the same decades, one emphasizing invariant tori and persistence, the other exponential instability and dense periodic orbits.
Carried by Andrey Kolmogorov; Stephen Smale.
Beyond the field
Where Dynamical Systems's ideas came from in other fields, and what they led to there: the strongest 12 of 56 links.
Stability theory influenced Synchronous stability theory in Power Systems
Synchronous stability theory applies the stability theory of dynamical systems, especially the equal-area criterion and Lyapunov-type reasoning, to generator swing equations. The mathematical stability framework was carried into power system analysis.
Complex dynamics influenced by Normal family in Complex Analysis
Montel's theory of normal families supplied the compactness tool that Fatou and Julia used to prove the basic theorems on iteration of rational functions and the structure of Fatou and Julia sets.
Carried by Paul Montel; Pierre Fatou; Gaston Julia.
Ergodic theory influenced by Statistical mechanics in Mathematical Physics
Ergodic theory arose from Boltzmann and Gibbs's statistical mechanics, in particular the ergodic hypothesis that time averages equal phase-space averages, which von Neumann and Birkhoff made rigorous.
Carried by Ludwig Boltzmann; Josiah Willard Gibbs; John von Neumann.
Ergodic theory influenced Ergodic-theoretic methods in Additive Number Theory
Dynamical-systems ergodic theory, including recurrence and mixing, is the machinery Furstenberg adapted into the correspondence principle. The same measure-preserving systems studied in dynamics became the vehicle for proving arithmetic progression theorems.
Carried by Hillel Furstenberg.
Ergodic theory led to Ergodic Theory and Mixing Times in Markov Processes
Mixing times quantify the mixing property defined in ergodic theory and dynamical systems; the later work on cutoff and spectral gaps grew out of that tradition.
Carried by George David Birkhoff.
Hyperbolic dynamics grew out of Symbolic dynamics and hyperbolic systems in Nonlinear Dynamics And Chaos
Hyperbolic dynamics grew directly out of the symbolic-dynamics and geodesic-flow tradition of Hadamard, Birkhoff and Morse, with Anosov and Smale's work of the 1960s generalising uniformly hyperbolic behaviour and coding orbits symbolically.
Carried by Dmitri Anosov; Stephen Smale; Jacques Hadamard.
Chaos theory influenced Nonlinear time-series analysis in Nonlinear Dynamics And Chaos
Takens' embedding theorem and the dynamical-systems account of attractors gave nonlinear time-series analysis its central tools for reconstructing phase space from data. The field grew out of the mathematics of chaotic dynamical systems.
Carried by Floris Takens.
Chaos theory influenced by Dynamical systems theory in Applied Mathematics
Lorenz's discovery of sensitive dependence grew out of the qualitative theory of dynamical systems descending from Poincaré and Birkhoff, which supplied the phase-space framework he used.
Carried by Edward Lorenz; Henri Poincaré.
Chaos theory influenced Strange attractors and fractal geometry in Nonlinear Dynamics And Chaos
Lorenz's discovery of sensitive dependence and his butterfly attractor gave the empirical object that Ruelle and Takens analysed in 1971 as 'strange attractors'. The fractal geometry of attractors was then built on the chaotic flows that chaos theory had identified.
Carried by Edward Lorenz; David Ruelle; Floris Takens.
Chaos theory influenced Chaos control and synchronisation in Nonlinear Dynamics And Chaos
Pecora and Carroll's synchronisation scheme and the OGY control method were built on the dynamical-systems theory of chaotic attractors and unstable periodic orbits. The control and synchronisation field grew directly out of that mathematics.
Carried by Louis Pecora; Thomas Carroll; Edward Ott.
Thermodynamic formalism influenced by Statistical mechanics in Thermodynamics
Statistical mechanics, through the Gibbs ensembles and transfer-operator methods, was the direct source of the formalism Sinai and Ruelle developed for dynamical systems.
Carried by Yakov Sinai; David Ruelle.
Thermodynamic formalism influenced by Statistical Mechanics in Thermodynamics
Thermodynamic formalism was built by Sinai, Ruelle and Bowen by importing the Gibbs formalism of statistical mechanics, treating invariant measures as Gibbs states on symbolic systems.
Carried by Yakov Sinai; David Ruelle; Rufus Bowen.

Dynamical Systems is a planet in Nousfarer
Its city has a tower for each of these 11 frameworks. Land, walk in, and climb one concept at a time.
Land on Dynamical SystemsRelated fields
Fields its ideas are linked to
- Nonlinear Dynamics And Chaos 17 links
- Ordinary Differential Equations 9 links
- Applied Mathematics 6 links
- Complex Analysis 2 links
- Operator Theory 2 links
- Signals And Systems 2 links
- Thermodynamics 2 links
- Classical Mechanics 2 links
- Markov Processes 2 links
- Astrodynamics 1 link
- Flight Dynamics 1 link
- Numerical Analysis 1 link
More in Analysis
- Approximation Theory
- Calculus Of Variations
- Complex Analysis
- Functional Analysis
- Geometric Measure Theory
- Harmonic Analysis
- Mathematical Analysis
- Measure Theory
- Operator Theory
- Ordinary Differential Equations
- Partial Differential Equations
- Real Analysis
See where Dynamical Systems sits on the Map of Human Knowledge