dynabench.solver

This module contains all different backends for the DynaBench solver.

Classes

`BaseSolver`(equation, grid, initial_generator)	Base class for all solvers.
`PyPDESolver`(equation, grid, initial_generator)	Solver class for solving PDEs using the py-pde library.

class dynabench.solver.BaseSolver(equation: BaseEquation, grid: Grid, initial_generator: InitialCondition, parameters: dict = {}, **kwargs)[source]

Bases: object

Base class for all solvers.

Parameters:

equation (dynabench.equation.BaseEquation) – The equation to solve.
grid (dynabench.grid.Grid) – The grid on which the equation is to be solved.
initial_generator (dynabench.initial.InitialCondition) – The initial condition generator from which the initial condition is to be generated.
parameters (dict, default {}) – Dictionary of parameters for the solver.

generate_filename(t_span: List[float], dt_eval: float, seed: int, hash_truncate: int = 8)[source]

solve(random_state: int = 42, t_span: List[float] = [0, 1], dt_eval: float = 0.1, out_dir: str = 'data/raw')[source]

Solve the equation.

Parameters:

random_state (int, default 42) – The random state to use for the initial condition.
t_span (List[float], default [0, 1]) – The time span for the solution.
t_eval (List[float], default None) – The time points at which the solution is to be evaluated.

Returns:

The solution of the equation.

Return type:

np.ndarray

class dynabench.solver.PyPDESolver(equation: BaseEquation, grid: Grid, initial_generator: InitialCondition, parameters: dict = {}, **kwargs)[source]

Solver class for solving PDEs using the py-pde library.

Parameters:

equation (dynabench.equation.BaseEquation) – The equation to solve.
grid (dynabench.grid.Grid) – The grid on which the equation is to be solved.
initial_generator (dynabench.initial.InitialCondition) – The initial condition generator from which the initial condition is to be generated.
parameters (dict, default {}) – Dictionary of parameters for the solver. See the documentation of py-pde and scipy’s solve_ivp for more information.

solve(random_state: int = 42, t_span: List[float] = [0, 1], dt_eval: float = 0.1, out_dir: str = 'data/raw')[source]

Solve the equation.

Parameters:

random_state (int, default 42) – The random state to use for the initial condition.
t_span (List[float], default [0, 1]) – The time span for the solution.
t_eval (List[float], default None) – The time points at which the solution is to be evaluated.

Returns:

The solution of the equation.

Return type:

np.ndarray