Incorrect Prior Color for diagonal subplots in diagnotics.pairs_posterior() on dev branch

[thegialeo]

Description

The function diagnostics.pairs_posterior() does not correctly distinguish the prior distribution color from the posterior color in the diagonal subplots. The expected behavior is that the prior and posterior should have distinct colors, as seen in the off-diagonal subplots.

Is this intended behavior? If not, I can make a PR for a temporary fix.

Branch & Commit

• Branch: dev
• Commit Hash: 97c381da61402cfc1a064355a60ae1309ff2d702

Expected vs. Actual Behavior

• Expected: Prior and posterior distributions should have distinct colors on all subplots.
• Actual: Prior color matches posterior color in diagonal subplots, making them indistinguishable.

Minimal Code for Reproduction

import numpy as np
import bayesflow as bf

dummy_prior = np.random.multivariate_normal([0, 0], np.eye(2), size=100)
dummy_posterior = np.random.multivariate_normal([0, 0], 0.1 * np.eye(2), size=100)

fig = bf.diagnostics.plots.pairs_posterior(
    estimates=dummy_posterior,
    priors=dummy_prior,
)

fig.savefig("test_plots.png")

Output: test_plots.png

System Info

• Python version: 3.11.11
• BayesFlow version: latest dev branch
• OS: Ubuntu 24.04

[paul-buerkner]

thanks! will investigate tomorrow. The-Gia Leo Nguyen ***@***.***> schrieb am Mo., 24. März 2025, 16:56:

…

*Description* The function diagnostics.pairs_posterior() does not correctly distinguish the prior distribution color from the posterior color in the *diagonal subplots*. The expected behavior is that the prior and posterior should have distinct colors, as seen in the *off-diagonal subplots*. Is this intended behavior? If not, I can make a PR for a temporary fix. *Branch & Commit* - *Branch:* dev - *Commit Hash:* 97c381d *Expected vs. Actual Behavior* - *Expected:* Prior and posterior distributions should have distinct colors on *all* subplots. - *Actual:* Prior color matches posterior color in *diagonal subplots*, making them indistinguishable. *Minimal Code for Reproduction* import numpy as npimport bayesflow as bf dummy_prior = np.random.multivariate_normal([0, 0], np.eye(2), size=100)dummy_posterior = np.random.multivariate_normal([0, 0], 0.1 * np.eye(2), size=100) fig = bf.diagnostics.plots.pairs_posterior( estimates=dummy_posterior, priors=dummy_prior, ) fig.savefig("test_plots.png") Output: test_plots.png test_plots.png (view on web) <https://github.com/user-attachments/assets/bcff3a46-ed99-43bc-bbf6-7c027232450c> System Info - *Python version:* *3.11.11* - *BayesFlow version:* *latest dev branch* - *OS:* *Ubuntu 24.04* — Reply to this email directly, view it on GitHub <#368>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADCW2ADKUHHPTZFTJUE3WGL2WATK5AVCNFSM6AAAAABZVJNISGVHI2DSMVQWIX3LMV43ASLTON2WKOZSHE2DGNZSGA4TGMI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***> [image: thegialeo]*thegialeo* created an issue (bayesflow-org/bayesflow#368) <#368> *Description* The function diagnostics.pairs_posterior() does not correctly distinguish the prior distribution color from the posterior color in the *diagonal subplots*. The expected behavior is that the prior and posterior should have distinct colors, as seen in the *off-diagonal subplots*. Is this intended behavior? If not, I can make a PR for a temporary fix. *Branch & Commit* - *Branch:* dev - *Commit Hash:* 97c381d *Expected vs. Actual Behavior* - *Expected:* Prior and posterior distributions should have distinct colors on *all* subplots. - *Actual:* Prior color matches posterior color in *diagonal subplots*, making them indistinguishable. *Minimal Code for Reproduction* import numpy as npimport bayesflow as bf dummy_prior = np.random.multivariate_normal([0, 0], np.eye(2), size=100)dummy_posterior = np.random.multivariate_normal([0, 0], 0.1 * np.eye(2), size=100) fig = bf.diagnostics.plots.pairs_posterior( estimates=dummy_posterior, priors=dummy_prior, ) fig.savefig("test_plots.png") Output: test_plots.png test_plots.png (view on web) <https://github.com/user-attachments/assets/bcff3a46-ed99-43bc-bbf6-7c027232450c> System Info - *Python version:* *3.11.11* - *BayesFlow version:* *latest dev branch* - *OS:* *Ubuntu 24.04* — Reply to this email directly, view it on GitHub <#368>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADCW2ADKUHHPTZFTJUE3WGL2WATK5AVCNFSM6AAAAABZVJNISGVHI2DSMVQWIX3LMV43ASLTON2WKOZSHE2DGNZSGA4TGMI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[paul-buerkner]

Should now be fixed :-) Thank you for reporting this issue.