← all systems · see it in The System
Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies)
Chaos theory is a branch of mathematics and an interdisciplinary area of scientific study. It focuses on underlying patterns and deterministic laws of dynamical systems that are highly sensitive to initial conditions.
Properties published scienceour proposal
Sourced fields, and our own readings.
| aliases | Chaos and nonlinear dynamics (the phenomena of deterministic chaos and the practice that studies them) |
| born at | Poincare's three-body work (1880s); the practice proper from Lorenz and Ueda (1961) and Yorke's naming (1975) |
| born precision | decades (C-1275) |
| boundary | Inside: the practice of studying deterministic dynamical systems that are highly sensitive to initial conditions (its people, models, computer runs, measures and uses), together with the chaotic behaviour it names. Outside: the systems that behave chaotically (weather, hearts, planets, populations), which are its subjects and partners, and stochastic systems, which 'are not dynamical Systems' in its sense (src:oestreicher-2007). |
| nested level | A practice inside science, built on differential equations and maps and on computers; its subjects run at every rung from planetary orbits to neurons. |
| scale | social |
Measures (50) published science
| Lyapunov time, chaotic electrical circuits | about 1 ms |
| Lyapunov time, weather systems | a few days (but unproven) |
| Lyapunov time, inner solar system | 4 to 5 million years |
| useful prediction horizon of a chaotic system | two or three Lyapunov times |
| skilful forecast lead time of midlatitude instantaneous weather today (practical predictability limit) | around 10 days |
| lead time gained by cutting initial-condition uncertainty tenfold | up to 5 days |
| error doubling time from a general circulation model (Charney committee 1966) and the predictability limit it implied | 5 days doubling; about two weeks limit |
| Feigenbaum constant delta (ratio of successive period-doubling intervals) | 4.669201609... dimensionless |
| Feigenbaum delta of the period-3 window's period-tripling route | 55.26 dimensionless |
| onset of chaos in the logistic map (end of the period-doubling cascade) | about 3.56995 growth parameter r |
| fractal dimension of the Feigenbaum attractor | approximately 0.54 dimensionless (Hausdorff or capacity) |
| variables in Lorenz's 1961 weather model; machine precision vs printout | 12 variables; 6 digits vs 3 digits count; digits |
| natural population time series with evidence of chaos | >30% of 172 (one-dimensional models: <10%) percent of series |
| scaling of Lyapunov exponents with body mass among chaotic populations | -1/6 power of body mass |
| extra spikes in postsynaptic targets from one extra spike (rat barrel cortex, in vivo) | about 28 spikes |
| intrinsic membrane-potential noise implied by the amplification | 2.2 to 4.5 (plus or minus) mV |
| duration of the network effect of one extra cortical spike | more than 50 ms |
| slope of the action-potential-duration restitution curve at which periodic beating gives way to alternans | 1 dimensionless slope |
| share of cardiac arrests initially found in ventricular fibrillation | about 10 percent |
| share of three-body scatterings that are resonant (undergo chaotic evolution) | about 50 percent |
| supermassive black hole triples (zero angular momentum) needing sub-Planck-length accuracy for a time-reversible solution | up to 5 percent |
| unpredictable supermassive black hole triples, initially virialised ensemble (about 50% on smaller astrophysical scales) | about 30 percent |
| energy conservation needed for an ensemble of approximate three-body solutions to stand for the true ensemble | better than 1/10 relative energy error |
| probability that Mercury's eccentricity reaches 0.6 within 5 Gyr (secular equations, 1001 integrations) | about 1 to 2 percent |
| observers showing the typical peak preference for intermediate fractal scaling (Experiment 1) | approx. 50 (smooth approx. 20, sharp approx. 20, none approx. 10) percent of observers |
| time Feigenbaum's universality paper spent in referee rejection | 3 years |
| Ueda's 'randomly transitional phenomena' on an analog computer, and how long his report was held back | 27 November 1961; reported 1970 date; year |
| Lyapunov time, terrestrial planets | about 5 million years |
| Lyapunov time (x10 definition), solar system | about 10,000,000 years |
| probability Mercury's eccentricity exceeds 0.6 within 5 Gyr | about 1 to 2 % |
| horizon of a precise inner-planet solution | precise to about 40 Myr; hopeless beyond 100 Myr |
| Mercury-Venus collision in one allowed run | about 862 Myr from now |
| weather predictability limit | two weeks (from a five-day doubling time) days |
| predictability horizon, plankton food web | 15-30 days |
| plankton mesocosm record | more than 2,300 days, sampled twice a week |
| share of natural population series showing chaos | >30% (multiple methods) vs <10% (one-dimensional models) % of 172 series |
| Lyapunov exponent scaling with body mass, chaotic populations | -1/6 power of body mass |
| Feigenbaum constant delta | 4.669201609... ratio of successive bifurcation intervals |
| onset of chaos in the logistic map (Feigenbaum point) | about 3.56995 r |
| Lorenz's parameters (sigma, rho, beta) | 10, 28, 8/3 dimensionless |
| fractal dimension, Lorenz attractor | around 2.06 (Koch snowflake 1.26; Feigenbaum's bifurcations about 0.45) dimension |
| Lorenz's rounding | 6 digits in the machine, 3 on the printout (0.506127 printed as 0.506) |
| three-body problem: equations and integrals | 18 first-order equations, 12 integrals of motion, reduced to 6 count |
| respiration's share of heart-rate fluctuation | 96 % (coefficient of determination) |
| PhysioNet chaos challenge series | 15 series (5 normal, 5 heart failure, 5 atrial fibrillation), each about 24 h, roughly 100,000 intervals |
| cortical amplification of one extra spike | about 28 additional spikes in postsynaptic targets |
| correlation dimension, brain | approx. 6 (smooth muscle approx. 3) D2 |
| forecast storm-position difference from minor perturbations, day 7 | 1006 km |
| the practice's dated start points | Poincare 1880s; Lorenz 1961-1963; 'chaos' named by Yorke 1975 years |
| fractal preference sub-groups | intermediate ~50%, smooth ~20%, sharp ~20%, none ~10% % of observers |
Open contentions (12) open contention
Where sources or readings disagree. Both stay on the map.
- C-413 Where do thoughts come from?
booked: No theory is proven, so the map lists the main ones side by side and doesn't pick one. It shows where thinking is handled in the brain, not where thoughts start.
the other reading: the positions listed in C-413, none ruled - C-1272 Is the economy chaotic?
booked: The map lists the evidence, which is very mixed, and doesn't rule.
the other reading: economic systems are inherently stochastic; recurrence quantification finds hidden transitions - C-1273 How common is chaos in animal populations?
booked: The map follows the 2022 finding of chaos in more than 30% of 172 population records, and keeps the older view that it is rare alongside.
the other reading: chaos thought rare; one-dimensional models find under 10%; real chaos hard to tell from chaos only in the model - C-1274 Was chaos theory a scientific revolution?
booked: The map lists the claim that it was a paradigm shift and keeps the view that it continued earlier mathematics alongside, without ruling.
the other reading: a continuation of nonlinear dynamics made practical by computers - C-1278 Does the brain code information by spike rates or by precise timing?
booked: The map follows the finding that single spikes are amplified into noise, so the cortex likely uses a rate code, and keeps the precise-timing view alongside.
the other reading: precise timing codes (synfire chains) found very rare in their recordings - C-1282 Does everyone find the same kind of fractal beautiful?
booked: The map keeps both: an average preference for medium-complexity fractals, and four stable groups of people with different tastes.
the other reading: universal aesthetic of fractals / individual taste in stable sub-groups - C-1286 Is the weather truly chaotic?
booked: The map treats weather as chaotic, with a predictability time of a few days that is unproven, and keeps the models showing both chaotic and orderly behaviour alongside.
the other reading: coexisting chaotic and non-chaotic attractors in Lorenz models: 'a dual nature of chaos and order with distinct predictability' - C-1306 Do AI weather models capture the butterfly effect?
booked: The map keeps both findings: some studies say AI models miss the early growth of small errors, another finds the effect in about 12% of tropical cyclones.
the other reading: the butterfly effect does exist in AI weather predictions: double-attractor transitions in ~12% of tropical cyclones (Leung 2026) - C-1309 What exactly counts as chaos?
booked: There is no agreed mathematical definition. The map uses the commonly cited three-property definition and keeps the looser and stricter ones alongside.
the other reading: sensitivity alone (popular accounts); combinatorial chaos without sensitive dependence (src:wp:chaos-theory) - C-1310 Is the brain chaotic?
booked: The map records chaos as shown in single nerve cells and small circuits, and keeps open whether whole-brain activity is chaos or amplified noise.
the other reading: deterministic chaos in the brain with correlation dimension about 6 (Başar and Güntekin 2007); cortical variability as amplified noise favouring a rate code (London 2010, C-1278) - C-1325 Is a healthy heartbeat chaotic?
booked: The map lists the readings and rules none: the healthy beat is complex and loses that in disease, but whether it is chaos, or mostly breathing's rhythm, is open.
the other reading: the question 'is the normal heart rate chaotic?' left open by the journal Chaos (Glass 2009); the fluctuations are mostly respiratory modulation (96%); ask instead whether the heart rate is 'chaotic' due to respiration (Wessel 2009) - C-1339 How far ahead can weather ever be forecast?
booked: The map records the classic two-week limit and notes that the formulas behind it are still under discussion.
the other reading: the validity of Lorenz's and Lilly's formulas for the limit is under discussion (src:wp:butterfly-effect); Lyapunov time of weather 'a few days (but unproven)'
Cross-scale patterns seen here (21) our proposal
Shapes that recur when the same kind of problem is solved at different scales, and how each shows in Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies). Each links to every other system that shows it.
- causal ceiling (structural)
A limit on what any forecaster can know, set by the system's own dynamics: uncertainty grows exponentially, so no meaningful prediction beyond two or three Lyapunov times; beyond 100 Myr a precise inner-planet solution is 'hopeless'; plankton predictability ends at 15-30 days. The exact limit for we - namespace completeness (the map predicts its gaps) (structural)
The map predicts its gaps: Feigenbaum's universality says every one-hump quadratic map bifurcates at the same rate, so the cascade's spacing in a system not yet measured is predicted; Libchaber then observed the cascade in Rayleigh-Benard convection (1979). Sharkovskii's ordering predicts which peri - nested ladder (structural)
Sharkovskii's ordering of periods: a cycle of period three implies cycles of every other length and chaotic orbits (Li and Yorke 1975), each period implying those after it in the order; and Devaney's three properties, where transitivity plus dense periodic orbits imply sensitivity (and on intervals - runaway from an unstable balance (structural)
The balance point is one extra spike per extra spike: 'If it produces more than one, on average, then perturbations would be amplified'; cortex sits past it (about 28), and collisions cap the runaway at a steady state. In the heart, a restitution slope of 1 is the balance: past it, periodic beating - same role, different builds (structural)
One job, 'going from steady to chaotic by repeated period doubling at the same geometric rate', is done by unlike builds: an iterated quadratic map, a convecting fluid (Libchaber), flour-beetle populations (equilibrium to cycles to chaos), and the Lorenz equations found again in lasers, dynamos and - same toolkit, independent builds (structural)
The same result built twice from the same parts (nonlinear equations plus a computer): Lorenz (1961, digital) and Ueda (1961, analog) found chaos the same year; Feigenbaum (1975) and Coullet and Tresser (1978) found universality independently. - smooth to sharp in finite time (structural)
The period-doubling cascade: periods 2, 4, 8, 16... come at ever shorter parameter intervals shrinking by delta about 4.669, so infinitely many doublings fit before a finite parameter, r about 3.56995, where chaos begins and an orbit of infinite period appears (the Feigenbaum attractor). - The Payment to Entropy (structural)
Form: information (and through it, thermodynamic). Every hand-off of a chaotic state forward in time pays a toll in predictive worth: an initial error grows by a factor e every Lyapunov time, so a forecast is worthless after two or three of them (src:wp:chaos-theory); the toll is a rate, the Kolmogo - trade one property for another (structural)
AI weather models trade forecast accuracy for physical fidelity: reducing coarse-graining makes predictions 'more physics-like (arrow of time and butterfly-like effects emerge), but forecast accuracy declines'; models sit at different points of the trade. OGY control buys stability with a kick kept - Umwelt: each system carves its own slice of one shared medium (structural)
The practice reads one dynamics through very different slices: a census, a printout, a convection cell, a reservoir's climate, the eye on a pendulum. Delay embedding adds a twist: one well-chosen slice is enough to rebuild the whole attractor's invariants, though not its shape. - conservation law plus closure (partial)
The conserved integrals of motion leave the three-body outcome open; the ergodic hypothesis (outcomes uniform over the accessible phase volume) closes it into closed-form distributions. - coordination-before-mapping (partial)
OGY control steers a chaotic system 'without a detailed model', from a Poincare section alone; reservoir computing forecasts and finds exponents 'model-free'. Partial: a slice of the state stands in for a map. - lightning (partial)
A stored difference released: Mercury's orbit diffuses slowly for hundreds of Myr, then enters a secular resonance with Jupiter and fails (falls onto the Sun at ~1.261 Gyr, or hits Venus at ~862 Myr, Mars ejected at ~822 Myr in one run). Also the field's own discharge: Lorenz's rounded rerun. - same shape, opposite sign (partial)
One process, the linearised growth of the difference between two trajectories: alone it grows as e^(lambda t) (chaos); with diffusive coupling it shrinks as e^((lambda - 2 alpha) t), the sign flipping at a critical coupling (synchronisation). - seed (partial)
A one-line rule unpacks unbounded combinations: the logistic map x -> r x (1 - x) carries cycles of every period and chaos (period three implies all periods); Feigenbaum's universality let the theory unpack across 'many different phenomena'. The butterfly effect alone is the dark twin (a small cause - symmetry breaking (name the cut) (partial)
In the Lorenz system the no-convection state is symmetric; at rho = 1 a pitchfork gives two mirror-image steady-convection states, so the flow must turn one way or the other (the source names the cut, not which way a given flow picks). - templating (partial)
Universality: the same period-doubling cascade and constants in dripping faucets, populations, convection and paced heart cells; one module (the shape of a map with a single hump) reused across unrelated systems. Ubiquity booked as the finding; its forms vary (period doubling, period tripling with d - The carry (partial)
All three faces. (1) Overflow passed on: in the multi-scale atmosphere, uncertainty fills the smallest scales and is transferred upscale, scale by scale, giving a finite-time horizon (src:palmer-2004-meteorological-butterfly; Lorenz 1969 via src:wp:butterfly-effect); in the sandpile, kin of chaos in - who opens the door (partial)
Chaos control opens its door ahead of time: the Poincare section and the target orbit are found first, then the controller 'waits until it comes near a desired periodic orbit in the section' and kicks. The receiver (controller) prepared the port. - holding / entrainment (surface)
Chaos synchronisation and the Kuramoto lockstep: coupled chaotic or oscillating parts fall into a common rhythm (pendulums, fireflies, neurons, a footbridge, Josephson arrays). Surface: the sources show coupling between peers, not a whole setting its parts' clock. - red-queen (surface)
Gene-for-gene co-evolution of crop and pathogen sometimes shows chaotic allele frequencies. Surface: the stalemate is not shown in the source.
Valence: what it trades, and with whom (19) published science
Each exchange with a neighbour, read from both sides: who gains, who pays, and what carries it.
- Galaxy via sys:gravitational-field · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Hamiltonian and Lagrangian mechanics (analytical mechanics: classical mechanics built from energy functions and the stationary action, and the practice of using them) via sys:phase-space · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Classical mechanics (Newtonian mechanics: bodies, forces and motion, and the practice of describing them) via sys:gravitational-field · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Fluid dynamics (the flows of liquids and gases, and the practice of describing them) via sys:fluid · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Supermassive black hole via sys:computer · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Neocortex (six-layered isocortex) via sys:phase-space · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Human (Homo sapiens) via sys:sensory-environment · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Entropy via sys:phase-space · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) even
- Statistical mechanics via sys:sensory-environment · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Habitat and ecosystem via sys:ecosystem · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Solar System via sys:gravitational-field · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Human brain via sys:nervous-system · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Heart via sys:circulatory-system · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Weather via sys:atmosphere-of-earth · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Computer via sys:computer · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Cryptography via sys:computer · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Differential equations (the practice of writing and solving laws of change, and the equations it works on) via sys:mental-representation · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Electric circuit via sys:fluid · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
- Computational physics (computer simulation of physical systems: the practice and the simulated systems it stands for) via sys:computer · Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) gain
Partners and parts published science
Built from / builds (10)
- Phase space
- Computer
- Differential equations (the practice of writing and solving laws of change, and the equations it works on)
- Dynamical systems (a state and a deterministic rule of change)
- Lyapunov exponents (the measure of sensitive dependence)
- Strange attractors
- Bifurcations and routes to chaos (period doubling)
- The Lorenz system (the founding model)
- The logistic map (the archetypal discrete model)
- Delay embedding (attractor reconstruction from one measured series)
Sits inside / holds (4)
- Human (Homo sapiens)
- Science (as a human social and cultural system)
- Differential equations (the practice of writing and solving laws of change, and the equations it works on)
- Physics (the field)
expressed by (19)
- Generation/Synthesis
- Recombination/Composition
- Scaffolding/Enablement (Meta-Tooling)
- Templating / Standardization
- Interoperability/Transduction (Bridging)
- Internal Coordination/Regulation
- Sensing/Scanning
- Hypothesis/Goal Selection
- Mapping/Modeling
- Uncertainty Reduction/Learning
- Active Experimentation / Probing (Optimal Experimental Design)
- Multi-Scale Data Fusion / Triangulation
- Trail-Marking / Externalized Memory (Stigmergy)
- Surprise / Change-Point Detection
- Hypothesis Portfolio / Multiple Hypothesis Tracking
- Offline Simulation & Replay-Guided Planning
- Cascade Induction
- Self-Monitoring
- Quality Control & Error Correction
floor filled by (25)
- Classical mechanics (Newtonian mechanics: bodies, forces and motion, and the practice of describing them)
- Fluid dynamics (the flows of liquids and gases, and the practice of describing them)
- Human (Homo sapiens)
- Science (as a human social and cultural system)
- Phase space
- Habitat and ecosystem
- Solar System
- Heart
- Weather
- Weather forecasting
- Measurement (the practice of comparing a quantity with a unit, and the units and standards it uses)
- Computer
- Statistics: the practice of collecting and reasoning from data, the intuitive statistics it grows from, and the structure (samples, distributions, estimators) it works on
- Cryptography
- Differential equations (the practice of writing and solving laws of change, and the equations it works on)
- KAM theorem
- Computational physics (computer simulation of physical systems: the practice and the simulated systems it stands for)
- Lyapunov exponents (the measure of sensitive dependence)
- Strange attractors
- Bifurcations and routes to chaos (period doubling)
- The Lorenz system (the founding model)
- The logistic map (the archetypal discrete model)
- Reservoir computing
- Fractal
- PhysioNet
represents (1)
Kinds (6)
- Hyperchaos
- Conservative (Hamiltonian) chaos: no attractor, chaotic seas between KAM tori
- Dissipative chaos: motion on a strange attractor
- Spatiotemporal chaos (fields: turbulence, fibrillating heart tissue, the Kuramoto-Sivashinsky equation)
- Combinatorial (or complex) chaos
- Quantum chaos
Notes published science
- The same sensitivity that makes chaos impossible to predict makes it easy to steer: in 1992, tiny electrical pulses timed by chaos theory turned a rabbit heart's chaotic arrhythmia back into a regular beat (src:garfinkel-1992).
- A jar of Baltic Sea plankton, kept under perfectly constant conditions and sampled twice a week for over six years, never settled down; its numbers stayed chaotic and could be forecast only 15-30 days ahead, about as far as the weather (src:beninca-2008).
- The new AI weather models may be accurate partly because they miss the butterfly effect: training on coarse-grained data strips out the fast small scales, and putting them back makes the models more physical but less accurate (src:hassanzadeh-2026, preprint; disputed by src:leung-2026).
Layers
The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.
a chaotic system as nature runs it (a state in phase space, a deterministic law, stretching, folding, the attractor, the regime)
- L1: The state: a point in phase space, each possible state one point
- L2: The deterministic law: nonlinear, or infinite-dimensional; a flow needs three or more dimensions; the three-body problem has 12 integrals, too few to close it
- L3: Stretching: nearby states separate exponentially (positive Lyapunov exponent); periodic orbits repel; the positive exponents sum to the Kolmogorov-Sinai entropy
- L4: Folding and boundedness: amplification has to end; mixing; in dissipative systems phase-space volume shrinks, in conservative ones it is kept
- L5: The attractor: a strange attractor with fractal dimension (Lorenz about 2.06), threaded by infinitely many unstable periodic orbits; strange repellers at basin boundaries
- L6: The regime: a parameter picks fixed point, cycles or chaos (pitchfork at rho = 1, period doubling to r about 3.56995, KAM tori breaking into cantori)
a deterministic nonlinear system watched over time: its state, its rule of change and the path it traces (reader's view)
- L1: The nonlinear world and its marks: fluid flow and convection, weather, a double pendulum, three stars, a beating heart, cortex, plankton and insect populations, circuits, road traffic; and their records (time series, a model's printout)
- L2: Taking it in: measured series, iterated computer runs (digital and analog), bench experiments, and the eyes on swinging pendulums and plotted pictures; what was called noise kept as part of the system
- L3: The state space: a model's phase space (flows need three dimensions, maps any) or one rebuilt from a single measured series by delay embedding; Poincare sections and recurrence plots
- L4: Diagnosis: sensitive dependence measured as Lyapunov exponents (a positive largest one with bounded motion), their sum as the Kolmogorov-Sinai entropy; chaos tests on data, and telling real chaos from chaos only in the model
- L5: The structure: strange attractors with fractal dimension and their skeleton of unstable periodic orbits, mixing, bifurcations and routes to chaos with universal constants, coexisting attractors, synchronisation
- L6: Meaning: determinism without long-term predictability; the Lyapunov time and the predictability horizon; statistics and ensembles in place of single paths
- L7: Use back on the world: forecasting (ensembles, machine learning), control and synchronisation, cryptographic keys, space-mission design, cardiac and ecological risk
chaos and nonlinear dynamics as dynamicists describe it (reader's view: irregular behaviour in the world, measurement, the model, iteration on computers, diagnosis, the structure, use)
- L1: Irregular behaviour in the world: fluid flow and mixing dyes, weather, heartbeat irregularities, population numbers, planetary orbits, firing neurons, a rough voice
- L2: Measurement and time series: Holter RR intervals, plankton sampled twice a week, 172 population series, ephemerides, EEG and patch-clamp records, a computer's own printout
- L3: The model: a state in phase space and a deterministic law, a set of ordinary differential equations (Lorenz) or an iterated map (logistic)
- L4: Iteration and integration on computers: the orbit traced step by step (Runge-Kutta, Bulirsch-Stoer, secular equations), the figure drawn
- L5: Diagnosis: sensitivity measured (Lyapunov exponents and times, dimensions), chaos told from noise (nonlinear forecasting, surrogates), results checked against numerical artifacts
- L6: The structure: attractors (fixed point, limit cycle, torus, strange), bifurcations and routes to chaos (period doubling, Feigenbaum universality, KAM breakdown), Sharkovskii's ordering
- L7: Use: forecasting within a horizon and ensembles, control and synchronization of chaos, cryptography, medicine and ecology
the boardroom (shared floors)
- L3: The state space: a model's phase space or one rebuilt from a measured series; Poincare sections, recurrence plots; how many dimensions are needed
- L3: The model: each input turned into a state in phase space and a deterministic law (differential equations or an iterated map)
- L4: Iteration and integration on computers: repeated application of simple formulas, the orbit traced and drawn
- L5: Diagnosis: the largest Lyapunov exponent with boundedness taken as the mark of chaos; dimensions; nonlinear forecasting to tell chaos from noise; checks against numerical artifacts
- L6: The structure: attractor types, strange attractors, bifurcations and routes to chaos, Feigenbaum universality, Sharkovskii's ordering, KAM
- L6: Meaning: deterministic yet unpredictable; the Lyapunov time and the horizon of two or three of them; ensembles and ergodic statistics in place of single paths
- L7: Use: forecasting within a horizon, control and synchronization, cryptography, medicine and ecology
way in: Brain signals
- L1: Neurons and networks: chaos proven at axons and single or coupled cells; one extra spike amplified in cortex
- L2: Recordings: EEG with its attractor dimensions; in vivo whole-cell patch clamp
way in: Coupling: another chaotic system's signal
- L1: Two chaotic systems coupled or one driving the other
- L2: Strong enough coupling suppresses the divergence: the two move as one while each stays chaotic
way in: Drive: the energy or forcing in, set by a control parameter (heating, pacing, growth rate)
- L1: Air heated from below and cooled from above (the Lorenz model's setting); a heart paced faster; a population's growth rate
- L2: The control parameter picks the regime: steady, periodic, period-doubled, chaotic (8, 16, 32 values beyond r = 3.54409)
way in: Emotion: surprise at a run gone different; pleasure in fractal pictures
- L1: Surprise: the restarted weather run predicted completely different weather
- L2: Preference for fractal-like scaling: a 'universal' peak at intermediate slopes on average, four stable sub-groups of observers
way in: Experiment: the bench (convection cells, circuits, paced heart cells)
- L1: A fluid layer heated from below: as the bottom gets hotter, the rolls grow complex in space and time and the flow turns chaotic
- L2: Instruments follow the bifurcation cascade (Libchaber's convection experiment, 1979); paced myocytes record irregular action potentials
way in: Hearing: rough voices and irregular sounds
- L1: Vocal fold vibration with nonlinear phenomena (subharmonics, irregular, rough voice)
- L2: The voice heard and recorded, analysed with traditional voice analysis and with nonlinear dynamics, which complement each other
way in: Learned model: a reservoir computer trained on the data
- L1: A limited time series of measurements as input
- L2: A high-dimensional reservoir driven by the data; output weights learned by linear regression
- L3: The autonomous reservoir runs on its own and reproduces the attractor's 'climate'
way in: Learned models: a reservoir computer trained on the data
- L1: A limited time series of measurements
- L2: The reservoir driven by the data, its response recorded
- L3: Output weights learned by linear regression
- L4: The autonomous reservoir run on its own output, reproducing the attractor's climate
way in: Orbits: the planets and three bodies under gravity
- L1: Three or more gravitating bodies; Poincare's first manifestation of chaos
- L2: Initial conditions and ephemerides fed to integrations: 1001 secular solutions, 10 very close initial conditions
way in: Perturbation: the tiny difference a chaotic system takes in (a rounding, a butterfly, one extra spike, a Planck length)
- L1: A tiny change in the state: a rounded digit, a single extra spike in one neuron
- L2: Amplified exponentially along the flow at the rate of the largest Lyapunov exponent
- L3: Grown to the size of the system: errors magnified to macroscopic scales; a different path
way in: Population numbers
- L1: Abundances rising and falling: beetles from equilibrium to cycles to chaos; plankton over orders of magnitude
- L2: Time series: a mesocosm sampled twice a week for more than 2,300 days; a global database of 172 series
way in: Records: measured time series (population censuses, heart beats, weather observations)
- L1: A global database of 172 population time series; one observable sampled at intervals
- L2: Delay embedding: the series and its lagged copies rebuild the attractor (Takens 1981)
way in: Sight: a chaotic motion watched (the double pendulum, smoke, convection rolls)
- L1: A double pendulum swinging, its path never repeating
way in: Sight: irregular things seen, and the figures that draw them
- L1: Light from irregular motion and from the computer's figures: mixing dyes, the double pendulum, Lorenz's plotted attractor
- L2: The viewer's visual system reading fractal-like scaling in images: an average preference for intermediate scaling, in four stable sub-groups
way in: Sight: the pictures (attractors, bifurcation diagrams, fractals)
- L1: Figures and images from computer runs: the butterfly-shaped Lorenz attractor, the logistic map's bifurcation diagram
- L2: The visual system detects and discriminates fractal-like scaling; preference follows that ability
way in: Simulation: the model iterated on a computer
- L1: Finite precision: Lorenz's LGP-30 held 6 digits, the printout 3 (0.506127 typed back as 0.506)
- L2: Integration step after step, which magnifies truncation and rounding errors to macroscopic scales; arbitrary-precision codes (Brutus) to control it
way in: The heartbeat (felt as interoception; read as RR intervals)
- L1: The normal heartbeat fluctuating in a complex manner even at rest; modulated by respiration
- L2: Holter ECG RR-interval series (about 24 hours, roughly 100,000 intervals each), shared openly so methods can be compared
way in: The machine's own numbers
- L1: Arithmetic at finite precision: computers truncate numbers
- L2: The printout: six-digit values printed to three digits and typed back in
way in: Theory: the equations (Lorenz system, logistic map)
- L1: A simplified model written down: three ODEs for convection (Lorenz 1963); a one-line demographic map (May 1976)
- L2: Analysis: nonlinearity (or infinite dimension) is needed; Poincare-Bendixson rules out strange attractors in two-dimensional flows
way in: Thought (inner): determinism, models and equations
- L1: The idea to be tested: Laplace's universal determinism (1778), Poincare's chance from small causes
- L2: Equations constructed by the human brain, taken as representations of reality
way in: Thought: the idea that small causes can have large effects
- L1: Contested origin (C-413). One recorded route: Poincare's idea, carried from the three-body problem to meteorology
way in: Weather: the convecting atmosphere
- L1: Atmospheric convection, the system Lorenz simplified into three equations
- L2: Observations, noisy and sparse, turned into an analysis by an ensemble Kalman filter
way out: Chaos-based cryptography
- L1: Primitives running on computers: image encryption, hash functions, stream ciphers
- L7: Use a chaotic map's control parameters and initial condition as the key
way out: Control of chaos (small timed perturbations)
- L1: Electrical stimuli delivered to the heart at irregular times
- L5: Wait at the Poincare section, then compute the small kick
- L7: Choose one of the unstable periodic orbits to stabilise
way out: Design: spacecraft routes on chaotic transport, keys from chaotic maps
- L1: Control parameters and initial conditions of chaotic maps used as keys
- L7: A table-top benchmark for saddle-mediated transport carried to energy-efficient mission design
way out: Forecasting: ensembles and horizons (weather, three bodies, machine-learned forecasts)
- L4: A lead time set by the horizon: around 10 days for midlatitude weather today
- L6: An ensemble of approximate solutions stands for the true ensemble when energy is kept to better than 1/10
way out: Forecasts within the horizon (ensembles, probabilities)
- L1: The forecast in people's hands (storm tracks, ecological forecasts with their limits)
- L5: Ensembles: AI and physics models run from perturbed starts; the spread carries the butterfly effect
- L7: The horizon set: no meaningful prediction beyond two or three Lyapunov times
way out: Publishing and teaching: papers, symposia, popular books
- L7: The first symposium on chaos (New York, December 1977)
way out: Synchronized chaotic circuits and networks
- L1: Coupled chaotic circuit systems in experiment
- L7: Build synchronizing systems and tune them to output chosen chaotic signals
way out: Writing and teaching (papers, reviews, popular books)
- L1: Books and meetings for the public and the fields
- L6: Concepts named and passed on (chaos 1975, strange attractor 1971)