removing ticks

on 1 Apr 2011

on 18 Nov 2015

Friedrich (view profile)

If I use
set(gca,'xtick',[])
the ticks vanish as expected, but the exponent, common for all ticks, remains in the plot at the end of the axis.
How can I avoid this

Friedrich (view profile)

on 1 Apr 2011

Hi Hans,
now I get it. In this special case you have to deactivate the xticklabels, too. This should works fine:
A(1,1:21)= 100000:100020;
B(1:100,1)= 1:100;
C(1:100,1:21)=33;
figure;
pcolor(A,B,C)
set(gca,'xtick',[])
set(gca,'xticklabel',[])
Best regards,
Friedrich

Jan

Jan (view profile)

on 1 Apr 2011
Removing the XTickLabel is enough and the XTick can be omitted.

Friedrich (view profile)

on 1 Apr 2011

Hi Hans,
which MATLAB and Operating System are you using? I ask this because the following code works fine for me in R2010b and Win7 64bit:
>> nn2 = 1.0e-156*[ 1 2];
>> plot(nn2)
>> set(gca,'YTick',[])
Best regards,
Friedrich

Friedrich (view profile)

on 1 Apr 2011

Hi Hans,
could you please post a working code? Because I am not able to reproduce your problem. Under Win7 64bit and R2009a the following code work fine:
>>n = 6;
>>r = (0:n)'/n;
>>theta = pi*(-n:n)/n;
>>X = r*cos(theta);
>>Y = r*sin(theta);
>>C = r*cos(2*theta);
>> pcolor(X*1e-4,Y*1e-4,C)
>> set(gca,'XTick',[])
So it must be related to your code.
Friedrich

Jan (view profile)

on 1 Apr 2011

A workaround:
A(1,1:21)= 100000:100020;
B(1:100,1)= 1:100;
C(1:100,1:21)=33;
figure;
ax = axes('XTick', [], 'nextplot', 'add');
pcolor(ax, A,B,C)
set(ax, 'YLim', [min(B), max(B)]);

hans

hans (view profile)

on 1 Apr 2011
Thank You this also works !
The strategy seems to be to avoid the generation of the ticks, so it doesn't hurt that the delete function has a bug..

David (view profile)

on 19 Aug 2011

I found this elsewhere.
1 vote Kelly Kearney answered 17 days ago Change the renderer from OpenGL to zbuffer.
set(gcf, 'renderer', 'zbuffer');Why this fixes the problem, I really have no idea. But I encounter it a lot when I add dateticks to my axes. Not sure if it's intended behavior or a bug, but most renderers eliminate the factor when manual tick labels are added; OpenGL does not (or at least doesn't always).

1 Comment

Aurelien Queffurust

Aurelien Queffurust (view profile)

on 5 Mar 2014
I confirm this solution allows to removes exponent on my Windows 64-bit

Greg (view profile)

on 18 Nov 2015

I realize this is a pretty old thread, but I think I've figured out what might have been causing a lot of the confusion in there after running into this issue myself...
Basically, the "opengl" renderer in older versions of MATLAB seems to have a bug that, when a custom XTickLabel is set, the axis exponent is still displayed. This does bug does not appear to happen with the "zbuffer" or "painters" Renderers.
The size of the "B" matrix matters because when you add a very large graphics object to a plot, MATLAB automatically switches the renderer over to opengl (presumably because of the better performance vs. painters or zbuffer).
So, when you execute this code in R2011b, you get no exponent on the axis:
figure
plot(1e6*(1:10),rand(10))
set(gca,'XTickLabel',{'a','b'})
But then if you flip the renderer to opengl, the exponent will appear:
set(gcf,'Renderer','opengl')
You can get rid of it again by going back to painters or zbuffer:
set(gcf,'Renderer','zbuffer')
set(gcf,'Renderer','painters')
I've tested this in R2011b (where the described issue exists) and in R2015a (where it does not), both on 64-bit Windows.

the cyclist (view profile)

on 1 Apr 2011

Can you give a simple example of code that exhibits this problem, because I am not seeing that behavior. For example, does this code
figure
plot(10.^(1:10),1:10)
set(gca,'XTick',[])
leave the exponent remaining? It does not for me.

hans (view profile)

on 1 Apr 2011

I'm using R2009a and Win7 64bit.
pcolor(Times{Sensor}, Freq{Sensor}, Pow{Sensor});
set(gca,'xtick',[])
It's true, when I just use
plot(a,b);
set(gca,'xtick',[])
the exponent vanishes. But not in the upper code with pcolor.

hans (view profile)

on 1 Apr 2011

A very simple code with data generating my problem is:
clear
A(1,1:21)= 100000:100020;
B(1:100,1)= 1:100;
C(1:100,1:21)=33;
figure;
pcolor(A,B,C)
set(gca,'xtick',[])
a simple code that does not generate the problem is just a slightly smaller matrix
clear
A(1,1:21)= 100000:100020;
B(1:10,1)= 1:10;
C(1:10,1:21)=33;
figure;
pcolor(A,B,C)
set(gca,'xtick',[])

hans

hans (view profile)

on 1 Apr 2011
I have modified this answer slightly in the way I included a code that worked fine and one not working fine for comparison.
Just the size of matrix seems to make a difference.
Jan

Jan (view profile)

on 1 Apr 2011
Bug confirmed for Matlab 2009a32 on WinXP32. Please send a bug report to MathWorks.

hans (view profile)

on 1 Apr 2011

Yes, You are right, that works ! Thank You.
However I'm still confused:
1: That I do not have to remove the xticklabel, if I use a matrix for B with a slightly smaller number of elements.
2: That the extracted exponential part of the ticks is not part of the tick but is a ticklabel
Do You have an explanation which could help me to understand ?

Show 1 older comment
hans

hans (view profile)

on 1 Apr 2011
OK, thank You for the explanation of question 2.
Do You also have an idea regarding question 1 ?
Jan

Jan (view profile)

on 1 Apr 2011
The exponent string looks different for these two methods: For the bigger matrix, the exponent is set above the base number, while for the smaller the exponent is set to the right top. It seems like the strings are created by different methods depending on the number of YTicks! And only one method has the bug.
hans

hans (view profile)

on 1 Apr 2011
That suggests a quick and dirty programming of the Matlab people has remained in the code ...

hans (view profile)

on 1 Apr 2011

OK, thank You Friedrich and Jan Both Your suggestions work.
To summarize the explanation of Jan why the problem occurs: The exponent object is attached to the XTickLabels.
What I still do not understand is:
Why does the size of the matrix B make a difference, if the leaving behind of the exponent occurs or not.

hans (view profile)

on 1 Apr 2011

a continuation of this problem occurs if I apply the command
datetick
This changes the number-labels into dates but leaves behind the 10^5 from the original numbers at the axis.
Any suggestions how to deal with this other than rewriting the datetick.m ?