Dynamical systems

Dynamical systems by writing the equations

The module starts from the thing a mathematician actually writes down: the equations. Type x' = s*(y-x) or x = r*x*(1-x), and Magic Stat detects whether it is an ODE or a map, typesets the equation as you write, and gives you the attractor, the bifurcation diagram, the Lyapunov spectrum and the rest.

The honest part

Equations in, dynamics out — and the second half: real data

There are two ways to study a dynamical system. The first is to know the equations: write them, integrate them, and explore the attractor, the bifurcation structure, the fixed points and their stability, the Poincaré sections and the Lyapunov exponents. Magic Stat takes the equations in the notation a person would use on paper — derivatives x', maps x_{t+1}, sums, roots and fractions — parses them symbolically and integrates them.

The second is to have only a time series and ask whether it is chaotic. The module implements the methods used for that question in ecology — s-map, Jacobian and direct Lyapunov exponents, the 0–1 test, recurrence quantification, permutation entropy and horizontal visibility graphs — as functions over any real series, not only over simulated ones.

It is not a computer algebra system. sympy is used for parsing, the symbolic Jacobian and the LaTeX preview, but the module is numerical. If you need closed-form solutions, stability proofs or symbolic manipulation, use a computer algebra system; Magic Stat is for numerical exploration and for the chaos tests on data.

What it does

What Magic Stat gives you for dynamical systems

How it works

From equations to a result

  1. Write the equations. in the canvas or as text — x' = s*(y-x) for an ODE, x = r*x*(1-x) for a map — or load a classic model from the library.
  2. Set the parameters and initial conditions. the variables you did not write as derivatives or next-state become parameters automatically, each with an editable value.
  3. Integrate or iterate. choose the solver settings (time span, step) and the analysis — attractor, bifurcation, Lyapunov, Poincaré, cobweb or basin.
  4. Or load a real series. and run the chaos-detection methods to see whether the data behave like a chaotic system.
  5. Read and export. the figures go to the gallery and the run is written into the report with its equations and parameters.
Options

The settings, in the dialog

Integration: Euler, RK4 or RK45, with the time span and step set in the dialog; maps are iterated for the number of generations you specify. Bifurcation: iterations per point, transient to discard and number of points.

Real-series methods: s-map lag and embedding by cross-validation, JLE/DLE, the 0–1 test, RQA, permutation entropy and HVG. Symbols: a palette of structures, functions, operators and Greek letters, or typed directly.

Frequently asked

Dynamical-systems questions, answered honestly

Do I have to know the equations?

For the first half of the module, yes — it is an equation-first tool. For the second half, no: if you have only a measured series, the chaos-detection methods (s-map, Lyapunov exponents, 0–1 test, RQA) work on the data directly.

Which solver should I use?

RK4 is the sensible default for smooth ODEs, with RK45 when the system is stiff or the time span is long; Euler is there for teaching and for maps. The step size matters — a chaotic system integrated with too large a step produces numerical chaos, not the system's.

What does a positive Lyapunov exponent mean?

It means nearby trajectories separate exponentially, which is the defining feature of chaos. The module reports the spectrum, not only the leading exponent; but a positive exponent from a short or noisy series should be treated as an indication, not a proof.

Is this a computer algebra system?

No. sympy is used for parsing, the symbolic Jacobian and the LaTeX preview, but the module is numerical. For closed-form solutions and symbolic work, use a CAS such as Mathematica, Maple or sympy directly.

Is my data uploaded somewhere?

No. Everything runs locally; the only automatic signal is an anonymous installation counter.

Time series analysis → Nonlinear regression →