Fixes#29615. Instead of explicitly setting the data lims with the min/max values of the edges, pass all the values to allowupdate_datalim to handle them appropriately. I also took the oppurtunity to de-duplicate the limit setting code inpcolor() andpcolormesh(), since I needed to update them to match anyway.

The tests are both very similar, but I couldn't think of a good way to de-duplicate testingpcolor andpcolormesh - suggestions very welcome.

In terms of performance, running the below script locally I see ~10% slowdown with pcolormesh, and no significant slowdown with pcolor.

importnumpyasnpimportmatplotlib.pyplotaspltimporttimex=np.arange(512,dtype=float)y=np.arange(512,dtype=float)z=np.arange(len(x)*len(y)).reshape(len(x),len(y))ts= []t=time.time()foriinrange(25):fig,ax=plt.subplots()# ax.pcolor(x, y, z)ax.pcolormesh(x,y,z)ax.set_yscale("log")fig.canvas.draw()dt=time.time()-tifi>0:# Allow one warmup loopts.append(dt)print(f"t ={dt} s")plt.close("all")t=time.time()print(f"Mean:{np.mean(ts)}")print(f"Stdev:{np.std(ts)}")

pcolormesh

Before:
Mean: 0.08725245793660481
Stdev: 0.005315622438177594

After:
Mean: 0.09809311230977376
Stdev: 0.006966432274025711

pcolor

Before:
Mean: 1.8416528701782227
Stdev: 0.020062453166251135

After:
Mean: 1.8436991373697917
Stdev: 0.021314677717322438

PR checklist

"closes #0000" is in the body of the PR description tolink the related issue
new and changed code istested
Plotting related features are demonstrated in anexample
New Features andAPI Changes are noted with adirective and release note
Documentation complies withgeneral anddocstring guidelines

dstansby added5 commits

March 20, 2025 15:14

Make sure pcolormesh log limits are good

83c3ec8

Fix log scale for pcolor

14c221d

Make pcolor and pcolormesh code similar

736bed8

De-duplicate common limit setting code

9df4fe0

Fix number of missing lines

bcc902e

dstansby marked this pull request as ready for review

March 20, 2025 18:21

timhoffm approved these changes

Mar 24, 2025

View reviewed changes

Copy link

Member

jklymak commentedMar 24, 2025

You changed pcolor and pcolormesh, but are showing an example with scatter. What do the before and after plots look like? I can't really evaluate what has changed.

Copy link

MemberAuthor

dstansby commentedMar 24, 2025

Gah, thanks for spotting that 🤦 . I've updated the script, and added the timings I get.pcolormesh seems to suffer ~10% performance wise, butpcolor doesn't change.

Copy link

Member

jklymak commentedMar 24, 2025

I guess I was more interested in the visual change, which I think this has?

Copy link

MemberAuthor

dstansby commentedMar 25, 2025

Here's before:

And after:

Using:

importnumpyasnpimportmatplotlib.pyplotaspltx=np.arange(512,dtype=float)y=np.arange(512,dtype=float)z=np.arange(len(x)*len(y)).reshape(len(x),len(y))fig,ax=plt.subplots()ax.pcolormesh(x,y,z)ax.set_yscale("log")plt.show()

Copy link

Member

jklymak commentedMar 25, 2025

Okay? I guess I'm not sure how you choose ymin in the case thaty[0]=0. Is this the accepted solution for this for other undefined y/xlimits?

Copy link

MemberAuthor

dstansby commentedMar 25, 2025

If one scatters the same x/y points used for thepcolormesh (that includes a zero value in the ys), there's similar behaviour:

I'm not sure whypcolormesh chooses a slightly lower y-limit, but I'd say that's in the weeds and this is a definitely improvement on past behaviour.

importnumpyasnpimportmatplotlib.pyplotaspltx=np.arange(512,dtype=float)y=np.arange(512,dtype=float)print(y[0])z=np.arange(len(x)*len(y)).reshape(len(x),len(y))fig,axs=plt.subplots(ncols=2)ax=axs[0]ax.pcolormesh(x,y,z)ax.set_yscale("log")ax=axs[1]ax.scatter(x,y)ax.set_yscale("log")plt.show()

Copy link

Member

timhoffm commentedMar 25, 2025

I'm not sure why pcolormesh chooses a slightly lower y-limit,

T.b.c: This may be a center vs. edge topic. Plot only chooses limits based on the points (centers of the mesh). The mesh may choose limits based on its edges.

jklymak approved these changes

Mar 25, 2025

View reviewed changes

Copy link

Member

jklymak left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others.Learn more.

I think I agree this is an improvement.

However, I'm a little concerned that data at zero is being represented at all in this scale. I see this pretty often when folks dopcolormesh(k, t, spectra) andk[0]=0, and the left hand limit makes the left-most bin ends up spread out arbitrarily wide. It would be nice if thisexplicitly indicated what the xmin would be set to, based on what is ink, whereas I still have no idea based on this.

Copy link

Member

timhoffm commentedMar 26, 2025

I believek[0]=0 is not special for us: Actually, the "center" coordinates of the mesh do not matter at all. AFAIK, we calculate the edges of the mesh and the limits are then determined from the positive edge coordinates (0 and negative coordinates are ignored). Note that even for all k>0 the edges can get negative (see#29615 (comment)).

Copy link

MemberAuthor

dstansby commentedMar 26, 2025

Yes, I think that's right. If one scatters the edge coordinates of the mesh, the bottom limit is set similarly:

importnumpyasnpimportmatplotlib.pyplotaspltx=np.arange(512,dtype=float)y=np.arange(512,dtype=float)print(y[0])z=np.arange(len(x)*len(y)).reshape(len(x),len(y))fig,axs=plt.subplots(ncols=2)ax=axs[0]ax.pcolormesh(x,y,z)ax.set_yscale("log")ax=axs[1]edges=np.concat([y-0.5, [y[-1]+0.5]])ax.scatter(edges,edges)ax.set_yscale("log")plt.show()