Main

The main loop of the MMC method.

main.compose_structure_tdf(group_tdfs: list[Callable], component_list: list[ComponentGroup]) → Callable[[NDArray[np.float64]], NDArray[np.float64]]

Return a function that unravels the design variables and calculates the combined TDF for the structure.

main.compose_tdfs(tdf: Callable[[NDArray[np.float64]], jnp.ndarray]) → Callable[[NDArray[np.float64]], jnp.ndarray]

Compose a group-level topology description function.

Group tdf takes in an num_components by num_design_variables array

main.connected_components(singed_distance_functions: NDArray[np.float64]) → NDArray[np.bool]

Find connected components in the list of components.

Parameters:

singed_distance_functions – NDArray[np.float64] - The stack of TDFs

Returns:

NDArray[bool] - The connectivity matrix

main.is_converged(iteration, objective_tolerance, objective_history, constraint_value, window_size) → bool

Check if the optimisation has converged.

main.main(max_iterations: int, domain: Domain2D, boundary_conditions: dict, volume_fraction_limit: float, component_list: list[ComponentGroup]) → None

Do MMC method.

main.make_heaviside(tdf: Callable[[NDArray[np.float64]], NDArray[np.float64]], transition_width: float, minimum_value: float = 0.0) → Callable[[NDArray[np.float64]], NDArray[np.float64]]

Smoothed Heaviside function.

https://en.wikipedia.org/wiki/Heaviside_step_function.

An element-wise function that 1 when x > 0, and minimum_value when x < 0. The step is smoothed over the transition_width.

Parameters:

• tdf – Callable - The topology description function to make crisp

• minimum_value – float - The lower value for areas where x<0

• transition_width – float - The size of the transition region

Returns:

NDArray - The smoothed Heaviside of the input array

main.moving_average(values, n)

Calculate a rolling average for values with window size n.

main.plot_values(values: NDArray, domain_shape: tuple[int, int]) → go.Figure

Plot values for debugging.