Use the previously unused position of a tick label to ensure that only a single zero label is rendered when using SymLogNorm. I store the first time a zero label is encountered and return blanks for all following zeros. The documentation on theticker.SymmetricalLogLocator.tick_values shows how these multiple zeros come to be through the linear approximation near zero.

As per discussion below I have taken a different approach, trying to fix the root cause of the problem in the locator (ticker.SymmetricalLogLocator.tick_values) so that the extra zero ticks are never created in the first place.

Fixes#10122

PR Checklist

Has Pytest style unit tests
Code is PEP 8 compliant
N/A New features are documented, with examples if plot related
N/A Documentation is sphinx and numpydoc compliant
N/A Added an entry to doc/users/next_whats_new/ if major new feature (follow instructions in README.rst there)
N/A Documented in doc/api/api_changes.rst if API changed in a backward-incompatible way

jklymak reviewed

Dec 29, 2017

View reviewed changes

Copy link

Member

jklymak left a comment•
edited
Loading

Choose a reason for hiding this comment

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

This seems OK to me. But isn't the root problem that there are two zeros returned by the Locator? Can't we just remove the duplicate entry from the Locator?

Copy link

Member

jklymak commentedDec 29, 2017

matplotlib/lib/matplotlib/ticker.py

Line 2215 in64874e5

classSymmetricalLogLocator(Locator):

matplotlib/lib/matplotlib/ticker.py

Line 2353 in64874e5

returnself.raise_if_exceeds(np.array(ticklocs))

anntzer previously requested changes

Dec 29, 2017

View reviewed changes

Copy link

Contributor

anntzer left a comment

Choose a reason for hiding this comment

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

The strategy of this PR can't work. If you run e.g. examples/pyplots/pyplot_scales.py with this PR applied, and move or zoom the symlog axes, the label at 0 disappears (because it is never redrawn).

I believe that@jklymak's idea is the correct approach.

Copy link

ContributorAuthor

Raab70 commentedDec 29, 2017 via email

That’s a great point. I’ll spend some more time in the locator and the linear approximation around zero and see if there’s a good way to ensure only a single zero value.

…

On Dec 29, 2017, 2:14 PM -0600, Antony Lee ***@***.***>, wrote:@anntzer requested changes on this pull request. The strategy of this PR can't work. If you run e.g. examples/pyplots/pyplot_scales.py with this PR applied, and move or zoom the symlog axes, the label at 0 disappears (because it is never redrawn). I believe that@jklymak's idea is the correct approach. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

tacaswell added this to thev2.2 milestone

Dec 30, 2017

Raab70 force-pushed thebug-fix/logsymnorm-zero-labels branch fromfd130c1 tob26b598Compare

December 30, 2017 15:22

prevent subticks from being added to zero decade

7fe5e4e

Raab70 force-pushed thebug-fix/logsymnorm-zero-labels branch fromb26b598 to7fe5e4eCompare

December 30, 2017 15:27

Copy link

ContributorAuthor

Raab70 commentedDec 30, 2017

@jklymak @anntzer Thanks for your help and patience, I believe this is the root of the issue. The multiple zeros are stemming from subticks being added. This change just prevents the addition of subticks to the zero label (since they will all be zero).

Copy link

Member

tacaswell commentedJan 2, 2018

👍 That makes sense and looks like a reasonable fix.

Could you add a test? I suspect creating a symlog axis and then asserting that there is only 0 zero label it in is the best approach (no need for an image test). The OP example is maybe a bit complex to trigger this, but as a last resort would be ok to use a test.

tacaswell previously requested changes

Jan 2, 2018

View reviewed changes

Copy link

Member

tacaswell left a comment

Choose a reason for hiding this comment

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

Fix looks good, just needs a test.

Anyone can dismiss this when a test is added.

anntzer dismissed theirstale review

January 2, 2018 00:42

implementation changed

adding test for SymLogNorm single zero fix

e9af134

jklymak approved these changes

Jan 2, 2018

View reviewed changes

jklymak dismissedtacaswell’sstale review

January 2, 2018 17:26

Dismissing@tacaswell review as test now added per his wishes

Copy link

Member

jklymak commentedJan 2, 2018

Just for completeness, the test is:

frommatplotlibimportpyplotaspltimportmatplotlibasmplfig,axes=plt.subplots(1,2,True,True)fori,axinenumerate(axes):im=ax.imshow([[0]],norm=mpl.colors.SymLogNorm(1e-5,vmin=-1,vmax=1))cb=plt.colorbar(im,ax=ax)# Clean up the labelszero_labelled=Falseforlabelincb.ax.yaxis.get_ticklabels():iflabel.get_text()==r'$\mathdefault{0}$':ifzero_labelled:label.set_visible(False)zero_labelled=Trueplt.show()