Skip to content

plots

Functions:

  • hinton –

    Draw Hinton diagram for visualizing a 2-D matrix.

hinton

hinton(
    A: Num[COO, "a b"],
    ax: Axes | None = None,
    marker: str = "s",
    update_from: PathCollection | None = None,
) -> Axes | PathCollection

Draw Hinton diagram for visualizing a 2-D matrix.

This has a significant speed-up over the looping version from the matplotlib cookbook, since we utilize the Axes.scatter function, along with some clever math to determine marker scaling to fill the canvas correctly.

Parameters:

  • A

    (Num[COO, 'a b']) –

    2-D matrix to visualize

  • ax

    (Axes | None, default: None ) –

    Optional pre-defined axis to plot on

  • marker

    (str, default: 's' ) –

    Any valid matplotlib.markers option.

Source code in src/affinis/plots.py
def hinton(
    A: Num[COO, "a b"],
    ax: Axes | None = None,
    marker: str = "s",
    update_from: PathCollection | None = None,
) -> Axes|PathCollection:
    """Draw Hinton diagram for visualizing a 2-D matrix.

    This has a significant speed-up over the looping version from the
    [matplotlib cookbook](https://matplotlib.org/stable/gallery/specialty_plots/hinton_demo.html),
    since we utilize the `Axes.scatter` function, along with some clever
    math to determine marker scaling to fill the canvas correctly.

    Args:
      A: 2-D matrix to visualize
      ax: Optional pre-defined axis to plot on
      marker: Any valid [`matplotlib.markers`](https://matplotlib.org/stable/api/markers_api.html#module-matplotlib.markers) option.

    """
    # A = COO(A) if isinstance(A, SparseArray) else COO.from_numpy(A)
    A = convert(A, COO)
    i, j = A.coords[1] + 0.5, A.coords[0] + 0.5
    ax = ax if ax is not None else plt.gca()

    # if update_from is None:
    ax.patch.set_facecolor("gray")
    ax.xaxis.set_major_locator(plt.NullLocator())
    ax.yaxis.set_major_locator(plt.NullLocator())
    ax.set_aspect("equal")
    ax.set_xlim(-0.5, A.shape[1] + 0.5)
    ax.set_ylim(-0.5, A.shape[0] + 0.5)
    plt.gcf().canvas.draw()
    ax.autoscale_view()
    ax.invert_yaxis()

    ext = ax.get_window_extent()
    base_size = ((ext.width) * (72.0 / plt.gcf().dpi) / (A.shape[1] + 2)) ** 2
    size = np.abs(A.data)
    size = base_size * size / size.max()
    color = np.where(A.data > 0, "white", "black")

    if update_from is not None:
        update_from.set_offsets(np.stack([i, j]).T)
        update_from.set_sizes(size)
        update_from.set_facecolors(color)
        return update_from
    else:
        scat = ax.scatter(
            i,
            j,
            s=size,
            c=color,
            marker=marker,
            linewidth=0,
        )
        return scat