Module ode

ODE solver and prediction engine.

RK45 (Dormand-Prince) Adaptive Solver

A 4th/5th order adaptive Runge-Kutta solver for systems of ODEs. Used to integrate learned dynamics $\frac{dy}{dt} = f(t, y)$.

Linear Extrapolation Fallback

When no Neural ODE model is available, uses linear extrapolation from the last two observations as a simple baseline.

Structs

OdeResult

Result of an ODE integration.

Rk45Config

RK45 solver configuration.

Traits

OdeSystem

ODE system: dy/dt = f(t, y).

Functions

linear_extrapolate

Predict a future vector using linear extrapolation.

rk45_integrate

Integrate an ODE system from (t0, y0) to t_end using Dormand-Prince RK45.