shamrock.utils.plot.benchmark Namespace Reference

Functions
	add_end_labels (ax_main, ax_annot, entries, x_pad=0.05, min_gap_px=25, fontsize=9)
	make_std_bench_plot (plot_data, xlabel, ylabel, title, end_label_fmt=lambda y:f"{y:.2f}", before_plot_func=None, dpi=250, figsize=(8, 6), min_gap_px=75, legend_ncol=2)

Detailed Description

Helpers for building standardized benchmark plots: several series on a
log-log scale, each ending in a non-overlapping value callout, with a
legend row underneath. Used across Shamrock's benchmark examples.

Function Documentation

add_end_labels()

shamrock.utils.plot.benchmark.add_end_labels	(		ax_main,
			ax_annot,
			entries,
			x_pad = 0.05,
			min_gap_px = 25,
			fontsize = 9 )

Place non-overlapping value callouts in a dedicated annotation axes.

`ax_annot` must share its y-axis with `ax_main` (i.e. have been created
with ``sharey=ax_main``), so a "data" y-coordinate means the same thing
in either axes. Each callout is linked back to its line's last data
point in `ax_main` with a leader line.

Parameters
----------
ax_main : matplotlib.axes.Axes
    The axes holding the plotted lines.
ax_annot : matplotlib.axes.Axes
    The (typically narrow, spine-less) axes the callout text is drawn
    into. Must share its y-axis with `ax_main`.
entries : list of (float, float, str, color)
    One ``(x, y, text, color)`` tuple per callout, where ``(x, y)`` is
    the data point the leader line points to.
x_pad : float, optional
    Horizontal position of the callout text, in `ax_annot` axes-fraction
    coordinates.
min_gap_px : float, optional
    Minimum vertical spacing between callouts, in display pixels.
fontsize : float, optional
    Font size of the callout text.

Definition at line 12 of file benchmark.py.

make_std_bench_plot()

shamrock.utils.plot.benchmark.make_std_bench_plot	(		plot_data,
			xlabel,
			ylabel,
			title,
			end_label_fmt = lambda y: f"{y:.2f}",
			before_plot_func = None,
			dpi = 250,
			figsize = (8, 6),
			min_gap_px = 75,
			legend_ncol = 2 )

Build a standardized benchmark plot.

Layout: a log-log plot on the left (5/6 of the width), a value-callout
panel on the right (1/6), and a legend row spanning the full width
underneath.

Parameters
----------
plot_data : dict
    Maps a series key to a dict with keys ``x``, ``y``, ``color``,
    ``label``, ``linestyle`` and ``marker``, one entry per line to plot.
xlabel, ylabel, title : str
    Axis labels and title for the main plot.
end_label_fmt : callable, optional
    Formats a series' last y-value into its callout text.
before_plot_func : callable, optional
    Called as ``before_plot_func(ax_main)`` before the series in
    `plot_data` are plotted, e.g. to draw reference lines underneath
    them.
dpi, figsize : optional
    Passed to `matplotlib.pyplot.figure`.
min_gap_px : float, optional
    Minimum vertical spacing between callouts, in display pixels; see
    `add_end_labels`.
legend_ncol : int, optional
    Number of columns in the legend row.

Returns
-------
(matplotlib.figure.Figure, matplotlib.axes.Axes)
    The created figure and its main (plot) axes.

Definition at line 104 of file benchmark.py.

Here is the call graph for this function: