scimba_torch.flows.create_solution

Create solutions using various ODE integrators.

Functions

create_solution(x0, mu, m, dt, f[, solver])

Create a solution using the specified solver.

solution_to_training_format(data[, solver])

Convert solution data to training format based on the solver used.
create_solution(x0, mu, m, dt, f, solver='RK4')[source]

Create a solution using the specified solver.

Parameters:

  • x0 (Tensor) – initial condition,

  • mu (Tensor) – parameter of the equation,

  • m (int) – number of time steps,

  • dt (float) – time step,

  • f (list[Callable]) – list of functions defining the ODE,

  • solver (str) – the solver to use (default: “RK4”).

Returns:

  • A tuple of tensors (t, mu, res) where:

    • t: time steps,

    • mu: parameters repeated for each time step,

    • res: solution at each time step.

  • The solution tensor res.

Return type:

A tuple containing

solution_to_training_format(data, solver='RK4')[source]

Convert solution data to training format based on the solver used.

Parameters:

  • data (Tensor) – The solution tensor to be converted.

  • solver (str) – The solver used to generate the solution (default: “RK4”).

Return type:

Tensor

Returns:

The reshaped solution tensor suitable for training.