Normalize distributions by RMS value

[tdbjacobs]

I'm not sure how to address this, but height and curvature distributions are useless when many different scales are included. See screenshot.

This isn't a programming problem - it's a visualization problem. Should we maybe normalize the x axis somehow, like we did in the SDRP paper?

[mcrot]

Would automatically zooming into x make these graphs more useful? Or providing a choice of predefined zoom parameters which put certain details of specific measurements or of the average into focus while neglecting details from other measurements?

[tdbjacobs]

Maybe, but I’m not sure zooming helps because the distributions are on widely different scales.

So inherently, even if you can see one clearly, the others may still look like sticks.

On Thu, Jun 23, 2022 at 5:17 AM Michael Röttger @.***> wrote:

Would automatically zooming into x make these graphs more useful? Or providing a choice of predefined zoom parameters which put certain details of specific measurements or of the average into focus while neglecting details from other measurements?

— Reply to this email directly, view it on GitHub https://github.com/ContactEngineering/TopoBank/issues/844#issuecomment-1164162130, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFOUYWQPTPHFJSLVBIDIFFTVQQTSTANCNFSM5YUGOHMA . You are receiving this because you authored the thread.Message ID: @.***>

-- Tevis Jacobs Mechanical Engineering and Materials Science University of Pittsburgh

[Sent from my phone. Please forgive typos.]

[pastewka]

We could normalize those distribution by the rms value. This way one would be able to see deviations from Gaussianity but of course the absolute scale would be lost.

[tdbjacobs]

I think that is a good solution. At present, there is zero value to these plots for data with multiple scales represented (just looks like a spike). If we normalize, they will at least show Gaussinity/non-Gaussinity.

We should think about somehow saving the RMS value so that a user can restore them to absolute values if they wish.

On Tue, Jul 19, 2022 at 3:21 PM Lars Pastewka @.***> wrote:

We could normalize those distribution by the rms value. This way one would be able to see deviations from Gaussianity but of course the absolute scale would be lost.

— Reply to this email directly, view it on GitHub https://github.com/ContactEngineering/TopoBank/issues/844#issuecomment-1189464472, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFOUYWUOQR3Y67G5AZTNANDVU352RANCNFSM5YUGOHMA . You are receiving this because you authored the thread.Message ID: @.***>