# Scientific model — Brusselator reaction kinetics

## Model interpretation

The Brusselator is a canonical, dimensionless, well-mixed autocatalytic reaction model. A commonly used reaction-network representation is:

- A → X
- 2X + Y → 3X
- B + X → Y + D
- X → E

After idealized scaling, the concentration-like state variables X and Y obey:

\[
\frac{dX}{dt}=A-(B+1)X+X^2Y
\]
\[
\frac{dY}{dt}=BX-X^2Y
\]

A and B are positive dimensionless feed/control parameters. Time and state variables are dimensionless. This package uses the equations as an educational dynamical system, not as a calibrated model for a specific chemical mixture.

## Equilibrium

At a steady state, both derivatives vanish. The unique positive equilibrium is:

\[
X^*=A, \qquad Y^*=\frac{B}{A}
\]

## Linear stability

The Jacobian is:

\[
J(X,Y)=\begin{bmatrix}-(B+1)+2XY & X^2\\ B-2XY & -X^2\end{bmatrix}
\]

At the equilibrium:

\[
J^*=\begin{bmatrix}B-1 & A^2\\ -B & -A^2\end{bmatrix}
\]

Therefore:

\[
\operatorname{tr}(J^*)=B-1-A^2, \qquad \det(J^*)=A^2>0
\]

For this two-dimensional system, the equilibrium is linearly asymptotically stable when the trace is negative and unstable when the trace is positive. The Hopf boundary occurs at:

\[
B_H=1+A^2
\]

At the boundary, the eigenvalues are purely imaginary with angular frequency A in the dimensionless time units. The modules distinguish the analytical equilibrium classification from finite-duration numerical observations.

## Numerical integration

The shared engine uses classical fourth-order Runge–Kutta (RK4), default internal step `h = 0.0025`, and a default output sampling interval of approximately `0.05`. The step is fixed for reproducibility. Non-finite or excessively large numerical states terminate the run with an explicit error message rather than being silently clipped.

The integration settings are deliberately common across modules. Changing the observation duration changes the endpoint, not the model equations. The CSV exports contain the sampled output points, not every internal RK4 substep.

## Analysis conventions

- Tail statistics use the final 75% of the observation window.
- X maxima are detected from sampled local maxima.
- Period is the median interval between successive detected X maxima, and is reported only when the tail X peak-to-peak excursion exceeds 0.01 dimensionless units.
- If fewer than two maxima are detected or the tail excursion is below the threshold, period is reported as unavailable; this avoids labelling tiny decaying transients as sustained oscillations.
- The phase-plane arrows are normalized for legibility; their lengths are not a quantitative rate scale.
- The vessel particles are qualitative encodings of concentrations, not molecule counts.

## Conservation-like identity

Adding the two differential equations gives:

\[
\frac{d(X+Y)}{dt}=A-X
\]

This identity is useful as a regression check for the shared derivative function.
