Sliding Boundaries Archive Heatmap Boundaries [BUG]

[LennartPurucker]

• pyribs version: 0.4.0
• Python version: 3.9
• Operating System: Windows

Description

Heyho,

I am trying to plot a heatmap of a sliding boundaries archive that shows the boundaries. Therefore, I set the parameter boundary_lw to a value large than 0.

The following picture shows the result of the function call (Fig. 1):

But what I would like to have the following (Fig. 2):

Steps to Reproduce

The function call I have used is:

from ribs.visualize import sliding_boundaries_archive_heatmap
import matplotlib.pyplot as plt
fig = plt.figure(figsize=(8, 6))
sliding_boundaries_archive_heatmap(archive, boundary_lw=0.5, ax=ax1)
plt.show()

I can not share the archive I am using.

Potential Fix

I found the following solution while looking at the code.

The current implementation of plotting the boundaries is the following:

if boundary_lw > 0.0:
    ax.vlines(x_boundary,
              lower_bounds[0],
              upper_bounds[0],
              color='k',
              linewidth=boundary_lw)
    ax.hlines(y_boundary,
              lower_bounds[1],
              upper_bounds[1],
              color='k',
              linewidth=boundary_lw)

However, to my understanding of the code, we would need to switch the indices. That is, to plot the boundaries correctly, we need the lower and upper bounds of the other dimension. Hence, the code should be:

if boundary_lw > 0.0:
    ax.vlines(x_boundary,
              lower_bounds[1],
              upper_bounds[1],
              color='k',
              linewidth=boundary_lw)
    ax.hlines(y_boundary,
              lower_bounds[0],
              upper_bounds[0],
              color='k',
              linewidth=boundary_lw)

I have used this code to produce Figure 2.

I am not sure if I am just using the code incorrectly or if this is a bug. I would be happy to hear feedback.

[btjanaka]

Hi @LennartPurucker, thank you for reporting this bug! It indeed looks like the boundary lines are being drawn incorrectly. We are addressing this in #280.

[LennartPurucker]

Hi @btjanaka, the pull request looks great, thank you! This solves the bug and my problem. Hence, I am closing the issue.