pyquickbench.plot_benchmark#

plot_benchmark(all_vals, all_args, all_funs=None, *, all_fun_names=None, all_out_names=None, plot_intent=None, plot_type=None, mode='timings', setup=None, all_xvalues=None, color_list=['tab:blue', 'tab:orange', 'tab:green', 'tab:red', 'tab:purple', 'tab:brown', 'tab:pink', 'tab:gray', 'tab:olive', 'tab:cyan', (0, 0, 1), (0, 0.5, 0), (1, 0, 0), (0.75, 0, 0.75), (0, 0, 0)], linestyle_list=['solid', 'dotted', 'dashed', 'dashdot', (0, (1, 10)), (0, (1, 1)), (0, (1, 1)), (5, (10, 3)), (0, (5, 10)), (0, (5, 5)), (0, (5, 1)), (0, (3, 10, 1, 10)), (0, (3, 5, 1, 5)), (0, (3, 1, 1, 1)), (0, (3, 5, 1, 5, 1, 5)), (0, (3, 10, 1, 10, 1, 10)), (0, (3, 1, 1, 1, 1, 1))], pointstyle_list=[None, '.', 'o', 'v', '^', '<', '>', '1', '2', '3', '4', '8', 's', 'p', 'P', '*', 'h', 'H', '+', 'x', 'X', 'D', 'd'], alpha=1.0, violin_width_mul=0.5, violin_showmeans=False, violin_showextrema=True, violin_showmedians=False, violin_show_observations=False, violin_side='both', single_values_idx=None, single_values_val=None, logx_plot=None, logy_plot=None, plot_xlim=None, plot_ylim=None, show=False, return_empty_plot=False, fig=None, ax=None, dpi=150, pxl_per_plot_x=1600, pxl_per_plot_y=800, sharex=True, sharey=False, title=None, xlabel=None, ylabel=None, plot_legend=None, ProductLegend=False, legend_location='upper left', plot_grid=True, relative_to_idx=None, relative_to_val=None, transform=None, clip_vals=False, stop_after_first_clip=False, filename=None)[source]#

Plots benchmarks results

Parameters:

  • all_vals (numpy.ndarray) – Benchmark results as returned by pyquickbench.run_benchmark().

  • all_args (dict | typing.Iterable) – Describes the arguments given to the functions in the benchmark.

  • all_funs (dict [str, iterable] | typing.Iterable [str] | None, optional) –

    Benchmarked functions, by default None.

    Only the function.__name__ attribute is used here.

  • all_fun_names (typing.Iterable [str] | None, optional) –

    Names of the benchmarked functions, by default None.

    In case the functions function.__name__ attribute is missing or uninformative.

  • all_out_names (typing.Iterable [str] | None, optional) – Names of the outputs of the functions if mode="vector_output", by default None.

  • plot_intent (typing.Iterable [str] | None, optional) –

    Describes how to handle the axes of the benchmark results array all_vals.

    See Multidimensional benchmarks for usage examples.

    By default None.

  • plot_type (typing.Iterable [str] | None, optional) –

    Any value that is not None overrides the default plot_type.

    Possible values : "bar", "curve", "violin".

    By default None.

  • mode (str, optional) –

    Benchmark mode, i.e. target of the benchmark.

    See Plotting scalar values for usage example.

    Possible values: "timings" or "scalar_output". By default "timings".

  • setup (callable, optional) –

    Function that prepares the inputs for the functions to be benchmarked.

    See Preparing inputs for usage example.

    Only used if all_out_names was not provided and mode="vector_output"

    By default lambda n: {pyquickbench.default_ax_name: n}.

  • all_xvalues (numpy.ndarray | None, optional) –

    Values to be plotted on the x-axis if those differ from argument values.

    See Plotting scalar values for usage example.

    By default None.

  • color_list (list [str], optional) – List of colors for plotted curves, by default default_color_list.

  • linestyle_list (list [str], optional) – List of linestyles for plotted curves, by default default_linestyle_list.

  • pointstyle_list (list [str], optional) – List of point markers for plotted curves, by default default_pointstyle_list.

  • alpha (float, optional) –

    Alpha value for transparency of curves in plot.

    By default, 1. meaning curves are fully opaque.

  • violin_width_mul (float, optional) –

    Factor on violin plot width. Decrease this value for thinner plot, and increase for fatter ones.

    By default, 0.5.

  • violin_showmeans (bool, optional) – Whether to show mean values in violin plots, by default False.

  • violin_showextrema (bool, optional) – Whether to show min/max values in violin plots, by default True.

  • violin_showmedians (bool, optional) – Whether to show median values in violin plots, by default False.

  • violin_show_observations (bool, optional) – Whether to show individual observations in violin plot, by default False.

  • violin_side (str | None, optional) –

    Side of the Kernel Density Estimation reconstruction in violin plots. Possible values: "both", "left", "right", None.

    By default "both".

  • single_values_idx (dict | None, optional) –

    Indices of benchmarked values to be fixed by a plot_intent of "single_value".

    See Multidimensional benchmarks for usage example.

    By default None.

  • single_values_val (dict | None, optional) –

    Values of benchmark to be fixed by a plot_intent of "single_value".

    See Multidimensional benchmarks for usage example.

    By default None.

  • logx_plot (bool | None, optional) – How to override log scaling on the x-axis of the plots, by default None.

  • logy_plot (bool | None, optional) – How to override log scaling on the y-axis of the plots, by default None.

  • plot_xlim (tuple | None, optional) – How to override limits on the x-axis of the plots, by default None.

  • plot_ylim (tuple | None, optional) – How to override limits on the y-axis of the plots, by default None.

  • show (bool, optional) – Whether to issue a call to matplotlib.pyplot.show() instead of returning a (fig, ax) tuple, by default False.

  • return_empty_plot (bool, optional) – Whether to prematurely return an empty (fig, ax) tuple. Useful to create an empty plot and reuse it, see Multidimensional benchmarks for instance. By defaut False.

  • fig (matplotlib.figure.Figure | None, optional) –

    User provided matplotlib.figure.Figure object. If None, a new one is generated.

    Typically, this argument is the result of a former call to pyquickbench.plot_benchmark(), potentially using argument return_empty_plot = True.

    By default None.

  • ax (numpy.ndarray [matplotlib.axes.Axes] | None, optional) –

    User provided array of matplotlib.axes.Axes objects as returned by matplotlib.pyplot.subplots().

    Typically, this argument is the result of a former call to pyquickbench.plot_benchmark(), potentially using argument return_empty_plot = True.

    By default None.

  • dpi (int, optional) – Output image resolution, by default 150.

  • pxl_per_plot_x (int, optional) – Output plot width, by default 1600.

  • pxl_per_plot_y (int, optional) – Output plot height, by default 800.

  • sharex (bool, optional) – Whether to share plot x-axis, by default True.

  • sharey (bool, optional) – Whether to share plot y-axis, by default False.

  • title (str | None, optional) – Image title, by default None.

  • xlabel (str | None, optional) – Override argument value as a default for plot x label, by default None.

  • ylabel (str | None, optional) – Override default for plot y label, by default None.

  • plot_legend (bool | dict | None, optional) – Whether to record each axis of the benchmark in a legend, by default None.

  • legend_location (str, optional) – Location of plot legend as given to matplotlib.axes.Axes.legend(), by default "upper left".

  • plot_grid (bool, optional) – Whether to plot a background grid to each plot, by default True.

  • ProductLegend (bool, optional) –

    Whether to detail every curve in the legend, or aggregate benchmark axes, leading to more concise legends.

    By default False.

  • relative_to_idx (dict | None, optional) –

    Indices of benchmarked values against which curves will be plotted.

    See Plotting transformed values for usage example.

    By default None.

  • relative_to_val (dict | None, optional) –

    Values of benchmark against which curves will be plotted.

    See Plotting transformed values for usage example.

    By default None.

  • transform (str | None, optional) –

    Data transformation before plotting, by default None.

    See Plotting transformed values for usage example.

  • clip_vals (bool, optional) –

    Whether to clip values that are out of bounds. Requires the argument plot_ylim tu be set explicitely. By default None.

    See Plotting transformed values for usage example.

  • stop_after_first_clip (bool, optional) – Whether to stop plotting after the first clipped value if clip_vals == True, by default False.

  • filename (str | None, optional) –

    If not None, saves resulting figure in filename.

    By default None.

Returns:

Matplotlib figure and axes with the benchmark results, only if show == False.

Return type:

tuple [ matplotlib.figure.Figure, numpy.ndarray [matplotlib.axes.Axes]] | None