Indices shown in plot are off

For large vectors, the indices shown in a figure are a bit off. For example:

test = [zeros(1,1079998), 1, zeros(1,1457106-1079998), 1, zeros(1,2159999-1457108)];

find(test) % gives correct indices: 1079999 1457108

figure; plot(test)

% The labeling of the 1's in the plot using data tips give idices of 1080000 and 1457110

This is probably a numeric precision issue.

Answer 1

[Torsten]

My guess is that the x-axis has a resolution of 10.

[Davis]

This is not generally true, for the same length vector:

test = [zeros(1,107998), 1, zeros(1,145716-107998), 1, zeros(1,2159999-145718)];

find(test) % gives 107999 and 145718

figure; plot(test)

% The labeling of the 1's in the plot using data tips give also idices of 107999 and 145718

[Torsten]

Yes, because your new data vector has 1/10 of the elements of the first. Thus its resolution can be chosen as 1.

Try for a data vector with 10x the number of elements of the first. My guess is that the resolution achieved is 100.

[Davis]

I didn't change the vector size; I just moved the non-zero points.

% first vector

test = [zeros(1,1079998), 1, zeros(1,1457106-1079998), 1, zeros(1,2159999-1457108)];

length(test)

% second vector

test = [zeros(1,107998), 1, zeros(1,145716-107998), 1, zeros(1,2159999-145718)];

length(test)

Furthermore. the resolution is location dependent:

test = [zeros(1,107998), 1, zeros(1,1457106-107998), 1, zeros(1,2159999-1457108)];

find(test) % gives 107999 and 1457108

figure; plot(test); % data tips at 107999 (correct) and 1457110 (incorrect)

[Torsten]

Then I don't know and you should report it as a potential bug to TMW:

https://supportcases.mathworks.com/mwsupport/s/casetypeselection?language=en_US&c__caseParameter=productusage&s_tid=srlp_product

[Davis]

Thanks Torsten, I submitted this issue to Mathworks using the link you provided. I'll update when I hear back.

[Davis]

I got the following response from Mathworks:

After initial investigation, the description matches our record and indicates this is likely a known bug. The potential reason is that the datatip is using "%g" as output format. I want to let you know that our development team is aware of this issue. We appreciate your effort in identifying and reporting issues to help us provide our users with better experience using MATLAB.

Meanwhile you could manually set the datatip output format with the following code:

test = [zeros(1,1079998), 1, zeros(1,1457106-1079998), 1, zeros(1,2159999-1457108)];

find(test) % gives correct indices: 1079999 1457108

figure;

p = plot(test)

p.DataTipTemplate.DataTipRows(1).Format ='%d'

[Torsten]

@Davis

Thank you for coming back and posting TMW's answer here.