← all systems · see it in The System
Differential equations (the practice of writing and solving laws of change, and the equations it works on)
In mathematics, a differential equation is an equation that relates one or more unknown functions and their derivatives. In applications, the functions generally represent physical quantities, the derivatives represent their rates of change, and the differential equation defines a relationship between the two.
Properties published scienceour proposal
Sourced fields, and our own readings.
| aliases | Differential equations |
| born at | Newton's three kinds in Methodus fluxionum (about 1671, published 1736); Bernoulli's equation 1695 |
| born precision | decades |
| boundary | Inside: the practice of stating a law as a relation between a quantity and its rates of change, and of finding, approximating or characterizing the functions that satisfy it (by hand with separation of variables and integration, by computer algebra, by numerical stepping, by qualitative and stability analysis, by fitting equations to data); the brain systems that carry it in a person; and the structures it works on (ordinary and partial, linear and nonlinear equations, their order, initial and boundary conditions, existence and uniqueness theorems, solution families and their long-run behaviour). Outside: the calculus it is built in (derivative, integral, limits), the functions and algebra it takes over, the physical, chemical, biological and economic systems whose laws are written in it, the people and machines that carry it, and difference equations (a separate kind it borrows numerical schemes from). |
| nested level | A part of calculus (calculus lists differential equations among its parts) and of mathematics; runs on calculus's derivative and integral (calculus L6), on functions (function-math L6) and algebra's transformations (algebra L6), and on writing's notation; carried by people (rung 11), numerical and computer-algebra software (rung 12) and trained networks (rung 13); the structure as a role at rung 0. |
| scale | social |
Measures (18) published science
| Newton's three kinds of differential equation (Methodus fluxionum) | 1671 (published 1736) year |
| characteristic clock: from Newton's three kinds (1671) to Fourier's heat equation (1822) | about 150 years |
| first mention of stiff equations | 1952 year |
| Lotka's equations to Volterra's publication | 1910; 1925; 1926 year |
| Lorenz system published | 1963 year |
| Navier-Stokes existence and smoothness named a Millennium Prize Problem | 2000 year |
| AI agents in the claimed 2026 Navier-Stokes blow-up construction | around 10,000 agents |
| development of ODE solvers since Runge and Kutta | more than 120 years |
| Euler method error orders | per step proportional to h^2; global proportional to h order in step size h |
| Euler method expected total rounding error | proportional to epsilon / sqrt(h) machine epsilon over root step size |
| first-order ODE accuracy, trained network (beam 50) against Mathematica (30 s) | 97.0 against 77.2 percent of 500 test equations |
| second-order ODE accuracy, trained network (beam 50) against Mathematica (30 s) | 81.0 against 61.6 percent of 500 test equations |
| trained network: time to solution | typically less than a second s per equation |
| SINDy: years experts took on the vortex-shedding dynamics | nearly 30 years |
| Hodgkin-Huxley cable model size | 320 coupled ODEs |
| Monte Carlo ensemble of Hodgkin-Huxley parameter sets | 300,000 parameter sets |
| group problem-solving interlude (the practice's working clock in class) | 2-5 minutes |
| Turing's theory to engineered cellular Turing patterns | over 70 years |
Open contentions (4) open contention
Where sources or readings disagree. Both stay on the map.
- C-1176 Are differential equations the laws of nature, or a way of describing them?
booked: The map keeps both readings and rules neither.
the other reading: laws are DEs because a law links a state to the state immediately following, and change is continuous (Siddiqui 2014); continuity is 'one of the most important assumptions in classical physics'; there are processes that do not change continuously (Siddiqui 2014, same text); equation discovery assumes few terms govern the dynamics, 'this assumption holds for many physical systems' (Brunton et al. 2016): a modelling choice that works, not shown to be nature's form - C-1183 Is there one right set of numbers in the classic model of a nerve impulse?
booked: The map keeps both: the original 1952 parameter set, and the 2026 reading of a range of valid settings in which the original is a minority.
the other reading: the standard one-way pipeline: voltage-clamp data reduced to a single best-fit parameter set, combined into a deterministic spiking model (as described by Korngreen 2026); Korngreen 2026: an experimentally constrained ensemble of behaviours; the canonical set is a regularly firing minority, not a unique or dominant solution - C-1196 Has the Navier-Stokes blow-up problem been solved?
booked: The map keeps the problem open and records the September 2026 claim for the forced case as a claim, with its priority dispute.
the other reading: claim: a solution with a smooth external force that develops a singularity in finite time, formalized in Lean, built with about 10,000 AI agents (src:wp:navier-stokes-existence-and-smoothness); the Clay Mathematics Institute still considers the problem 'active' as of October 2026; the unforced and real-analytic-force cases remain unsolved (same source); priority: Alpöge and Buckmaster derived closely related results on the Euler equations used in the work; the method builds on Córdoba and Martínez-Zoroa 2023 (same source) - C-1207 What makes an equation 'stiff'?
booked: The map uses the working definition (a problem that needs implicit numerical methods), and notes that a precise definition is disputed.
the other reading: a necessary condition based on the trace of the Jacobian over the interval (src:wp:stiff-equation); 'no precise definition of stiffness exists' (as sometimes suggested in the literature, same source)
Cross-scale patterns seen here (11) our proposal
Shapes that recur when the same kind of problem is solved at different scales, and how each shows in Differential equations (the practice of writing and solving laws of change, and the equations it works on). Each links to every other system that shows it.
- conservation law plus closure (structural)
Newton's second law, F = m d2x/dt2, cannot be reduced to a closed first-order equation 'for a general force F(x)': the form of the force must be supplied (src:siddiqui-2014). And every equation leaves its constants open until initial or boundary conditions close them, 'the number of initial or bound - nested ladder (structural)
Existence conditions ladder: continuity gives local existence only (Peano); Lipschitz continuity gives local existence and uniqueness (Picard-Lindelöf); with Grönwall's conditions it extends to a global result (src:wp:ordinary-differential-equation). - smooth to sharp in finite time (structural)
Finite-time blow-up is the open question at the centre of Navier-Stokes existence and smoothness; a 2026 claim (an AI company, about 10,000 agents) is of a forced solution that 'develops a singularity in finite time', contested (priority dispute; the Clay institute still lists the problem 'active'). - templating (structural)
One equation swapped between systems and still working: the wave equation for light, sound and pond waves; one diffusion equation for many diffusion processes; Lotka's 1910 equations for autocatalytic chemistry redeployed in 1925 for predator and prey; Turing's reaction-diffusion equations from 1952 - The Payment to Entropy (structural)
Three forms. (1) Chain: each numerical step hands its error to the next; the Euler method's error per step goes as h^2 and at a given time as h, and rounding errors add up, expected total proportional to epsilon over root h, so shrinking the step to cut truncation raises the rounding toll ('the effe - trade one property for another (structural)
Precision bought with speed: ODE-networks 'can explicitly trade numerical precision for speed' (src:chen-2018); a trained network buys accuracy with beam width (81.2 to 97.0 percent on first-order equations from beam 1 to 50, src:lample-charton-2019); the Euler step trades truncation against roundin - Umwelt: each system carves its own slice of one shared medium (structural)
The practice's own slice of the shared world: nine channels, six human (equation, plot, words, hands, thought, emotion) and three others (measured data turned into equations, solvers fed equations, an engineered cell colony reading diffusing signals). Each picks only the rates of change of its quant - runaway from an unstable balance (partial)
Linear stability of a fixed point decides whether a small push dies or grows ('investigation of its stability leads to a linear differential equation', src:wp:differential-equation); in the Lorenz system's chaotic regime small differences grow exponentially (src:wp:lorenz-system); a numerical runawa - seed (partial)
A differential equation defines 'a class and not a particular member of the class' (src:siddiqui-2014): one short statement unpacks a family of solutions, one per choice of the n constants; and 'Many fundamental laws of physics and chemistry can be formulated as differential equations' (src:wp:diffe - The carry (partial)
Two faces. The wait: a stepping method cannot take step n+1 until step n is done (y(n+1) = y(n) + h f(y(n)), src:wp:stiff-equation), and on a stiff problem the explicit chain runs at the pace of its fastest-decaying part: 'an exceptionally small timestep would be required if the Euler method is used - causal ceiling (surface)
Lorenz: a deterministic system whose 'chaotic nature makes long-term prediction practically impossible' (src:wp:lorenz-system). The source says 'practically', so whether the limit is structural or a limit of means is not settled here; booked surface.
Valence: what it trades, and with whom (23) published science
Each exchange with a neighbour, read from both sides: who gains, who pays, and what carries it.
- Vector calculus (the calculus of fields: gradient, divergence, curl; the practice and the structures it works on) via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Quantum mechanics via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) 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:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- General relativity via sys:written-record · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Linear algebra (the practice of solving, transforming and taking apart linear systems with vectors and matrices, and the structures it works with) via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Calculus (the practice of handling change and accumulation by limits, and the structures it works on) via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Cell via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Neuron (nerve cell) via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Human (Homo sapiens) via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Wave via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Harmonic oscillator via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Weather via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Measurement (the practice of comparing a quantity with a unit, and the units and standards it uses) via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Computer via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Algebra (the practice of solving for unknowns and transforming expressions, and the structures it works with) via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Artificial neural networks (trained sequence-to-sequence models) via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Classical mechanics (Newtonian mechanics: bodies, forces and motion, and the practice of describing them) via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Complex analysis (the practice of working with functions of a complex variable, and the holomorphic structure it works on) via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Computational physics (computer simulation of physical systems: the practice and the simulated systems it stands for) via sys:written-record · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Fluid dynamics (the flows of liquids and gases, and the practice of describing them) via sys:written-record · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Fourier analysis (the practice of splitting signals into waves and rebuilding them, and the structure it works on) via sys:sensory-environment · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
- Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) via sys:mental-representation · Differential equations (the practice of writing and solving laws of change, and the equations it works on) gain
Partners and parts published science
Built from / builds (8)
- Calculus (the practice of handling change and accumulation by limits, and the structures it works on)
- Derivatives
- Ordinary differential equations
- Partial differential equations
- Initial and boundary conditions
- Numerical methods
- The qualitative theory (dynamical systems)
- Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies)
Sits inside / holds (3)
- Mathematics (the field: the practice and the structures it studies)
- Calculus (the practice of handling change and accumulation by limits, and the structures it works on)
- Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies)
expressed by (11)
- Generation/Synthesis
- Recombination/Composition
- Scaffolding/Enablement (Meta-Tooling)
- Templating / Standardization
- Quality Improvement/Refinement
- Interoperability/Transduction (Bridging)
- Encapsulation & Modularity (Abstraction)
- Mapping/Modeling
- Uncertainty Reduction/Learning
- Offline Simulation & Replay-Guided Planning
- Quality Control & Error Correction
floor filled by (25)
- Hamiltonian and Lagrangian mechanics (analytical mechanics: classical mechanics built from energy functions and the stationary action, and the practice of using them)
- General relativity
- Neuron (nerve cell)
- Human (Homo sapiens)
- Language (human language)
- Writing (the practice of marking language on a surface, and the script plus rules it works with)
- Wave
- Human brain
- Weather
- Measurement (the practice of comparing a quantity with a unit, and the units and standards it uses)
- Computer
- Artificial neural networks (trained sequence-to-sequence models)
- Euler method
- Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it
- Calculus (the practice of handling change and accumulation by limits, and the structures it works on)
- Classical mechanics (Newtonian mechanics: bodies, forces and motion, and the practice of describing them)
- Ordinary differential equations
- Partial differential equations
- The qualitative theory (dynamical systems)
- Stiff equation
- Separation of variables
- Computational physics (computer simulation of physical systems: the practice and the simulated systems it stands for)
- Fluid dynamics (the flows of liquids and gases, and the practice of describing them)
- Fourier analysis (the practice of splitting signals into waves and rebuilding them, and the structure it works on)
- Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies)
represents (1)
Kinds (12)
- Calculus (the practice of handling change and accumulation by limits, and the structures it works on)
- Ordinary differential equations
- Partial differential equations
- Linear differential equation
- Non-linear differential equation
- Stochastic differential equation
- Delay differential equation
- Differential algebraic equation
- Integro-differential equation
- Stochastic partial differential equation
- Difference equation (recurrence relation)
- Integral equation
Notes published science
- The famous 1952 Hodgkin-Huxley numbers are not the typical answer. Refitting their own published data with its scatter and running 300,000 versions of the squid-axon equations, a 2026 study found most versions fire one spike and stop; the canonical 1952 set sits in a regularly firing minority, 'rather than representing a unique or dominant solution' (src:korngreen-2026).
- The predator-prey equations ecologists use began as chemistry: Lotka first wrote them in 1910 for autocatalytic chemical reactions, and only in 1925 turned them on predators and prey; Volterra published the same equations independently in 1926 (src:wp:lotka-volterra-equations).
- A machine found in data an equation that took people decades: sparse regression on measurements recovered the dynamics of vortex shedding behind an obstacle, 'a system that took experts in the community nearly 30 years to resolve' (src:brunton-2016).
- A Millennium Prize problem may have just moved: in September 2026 an AI company claimed a forced Navier-Stokes solution that blows up in finite time, built with about 10,000 AI agents and checked in Lean; the Clay institute still calls the problem 'active', and there is a priority dispute (src:wp:navier-stokes-existence-and-smoothness).
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 law of change: an unknown quantity tied to its own rates of change, solved forward from a starting state or across a boundary (reader's view)
- L1: The changing world and its marks: a quantity that changes (a falling ball, a cooling body, a vibrating string, predator and prey numbers, an axon's voltage) and its records (measured time series, the written equation, a solution curve)
- L2: Taking it in: eyes on whiteboard and page, ears on the lecture and the group's talk, hands in gesture; machines take in data and token strings
- L3: Codes (the ports): in professional mathematicians, analysis statements (the home of differential equations) activate a bilateral frontal, intraparietal and ventrolateral temporal network shared with number, sparing language areas
- L4: Setting up and transforming: a law written as a relation between the present state and the next (Newton's second law, Newton's law of cooling); separation of variables with its inner resources (grouping, moving, 'pulling out'); integration; a step taken by a numerical method
- L5: Choosing and checking: which class the equation is in and so which method (separation, limits, guess-and-check, numerical, implicit for stiff), whether a solution exists and is unique, whether the numbers are stable; and the failure to connect two equivalent methods
- L6: The structure: equations relating functions to their derivatives; ordinary and partial, linear and nonlinear, their order; initial and boundary conditions (as many as the order); existence and uniqueness (Peano, Picard-Lindelöf); solution families and their long-run behaviour (dynamical systems, chaos)
- L7: Use: the laws of physics and chemistry written as differential equations; models in biology and economics; design and prediction from celestial motion to bridges to neurons
the boardroom (shared floors)
- L3: Codes: the bilateral frontal, intraparietal and ventrolateral temporal network takes up equation, plot and words; language and general-knowledge areas are spared
- L4: Setting up and transforming: the law as a link between present and next state; separation of variables and its inner moves ('pulling out', grouping, moving); integration; numerical steps
- L5: Choosing and checking: the method from the equation's class, existence and uniqueness, numerical stability, and how sure one is (the two equivalent methods students fail to connect)
- L6: The structure: ordinary and partial, linear and nonlinear, order and conditions, existence and uniqueness theorems, solution families and long-run behaviour
- L7: Use: physical and chemical laws, models in biology and economics, design and prediction
way in: Emotion: pleasure (and silence) in the lecture
- L1: A lecture written to help students take pleasure in the mathematics of physics; the class falls silent, then laughs
way in: Hands: whiteboard writing and gesture in group problem solving
- L1: Groups of 2-3 students working the time-dependent Schrödinger equation on whiteboards, 2-5 minutes per interlude, filmed for discourse, gesture and writing
way in: Living input: diffusing signals read by an engineered gene circuit (a reaction-diffusion system run by cells)
- L1: Chemicals reacting and diffusing in a growing bacterial population
- L2: A synthetic genetic network in each cell answers the local concentrations; stripes form across the colony
way in: Machine input: an equation given to a solver (computer algebra, a trained network, a numerical integrator)
- L1: The equation as input: a CAS command (dsolve, DSolve), or a prefix token sequence for a trained network
- L2: The solver returns a solution, no solution, or times out; a trained network proposes candidates checked by substitution
way in: Machine input: measured time series turned into an equation (sparse regression)
- L1: Time-series measurements of a system's state (oscillators, the Lorenz system, a fluid's vortex shedding)
- L2: A library of candidate right-hand-side terms
- L3: Sparse regression keeps the fewest terms that fit
way in: Sight: solution curves and phase pictures (a trajectory, the Lorenz butterfly)
- L1: A solution plotted: the looping Lorenz attractor, x(t) and v(t) plotted by students, an Euler polygon against the exact curve
way in: Sight: the written equation (dy/dx, partial derivatives, a law on the board)
- L1: The equation written: Newton's three kinds, a law such as F = m d2x/dt2 on the blackboard, the time-dependent Schrödinger equation on a whiteboard
way in: Thought: a law imagined as the link from one instant to the next
- L1: The law held as a connection between neighbouring moments, a continuous chain of states
way in: Words: the law said and asked about (lecture, inquiry dialogue, group talk)
- L1: A teacher's question and students' answers about why laws are differential equations
way out: Running a solver (a numerical step, a CAS command, a trained network)
- L1: Numbers out: the approximate solution at each time
- L4: One step: the equation replaced by a difference equation and advanced
- L5: The method and its settings chosen (step size, implicit for stiff, tolerance)
way out: Saying it (posing the question, explaining the law, confirming an answer)
- L1: The spoken confirmation
- L6: The reason given in words: a differential equation defines a class, not one member
way out: Writing the lines (the law set up, variables separated, the integral and the constants)
- L1: Marks on the whiteboard, filmed
- L4: The separated equation written, the time part pulled out from the space part