Plotters

Vectorized drawings of force density datastructures for figures and papers. The 2D plotter is built on the standalone compas_plotter. A second plotter, built with matplotlib, charts the loss histories recorded during optimization.

Note

The plotter needs the standalone compas_plotter for use. Without it, the plotter degrades to a null object that warns on use.

Plotter

Plotter(*args: Any, **kwargs: Any)

A thin wrapper on the compas_plotter.plotter.Plotter.

This object exists for API consistency.

LossPlotter

LossPlotter(loss: Loss, datastructure: FDNetwork | FDMesh, **kwargs: Any)

Chart a loss and its component errors over an optimization history.

Parameters:

• loss (Loss) –

  The loss function whose terms are re-evaluated and plotted.

• datastructure (FDNetwork | FDMesh) –

  The network or mesh the history parameters belong to.

• kwargs (Any, default: {} ) –

  Extra keyword arguments forwarded to the matplotlib figure.

plot

plot(history: EquilibriumParametersState | list[Any], report_breakdown: bool = True, error_names: Iterable[str] | None = None, plot_legend: bool = True, yscale: str = 'log', **eq_kwargs: Any) -> Float[Array, iterations]

Plot the loss and its component errors over a parameter history.

Parameters:

• history (EquilibriumParametersState | list[Any]) –

  The per-iteration parameter states to replay through the loss.

• report_breakdown (bool, default: True ) –

  If True, plot each error and regularization term separately and print its statistics.

• error_names (Iterable[str] | None, default: None ) –

  The subset of term names to break down. If None, every term is shown.

• plot_legend (bool, default: True ) –

  Whether to draw the plot legend.

• yscale (str, default: 'log' ) –

  The matplotlib y-axis scale, e.g. "log" or "linear".

• eq_kwargs (Any, default: {} ) –

  Extra equilibrium model options. Defaults to a single FDM step.

Returns:

• losses ( Float[Array, iterations] ) –

  The total loss at each iteration.

Notes

Equilibrium is recomputed with a dense model vmapped over the history, since the sparse model does not support vmap.

show

show() -> None

Display the plot.

print_error_stats staticmethod

print_error_stats(errors: Float[Array, iterations], error_name: str) -> None

Print first, last, min, and max of an error series.

Parameters:

• errors (Float[Array, iterations]) –

  The error value at each iteration.

• error_name (str) –

  The label to print the statistics under.