Zoom in on a logarithmic axes

Good day everyone,

I aim to have the standard axes response (in linear values) when zooming in on

a logarithmic distributed frequency axes, but, ... with linkaxes functionality

coupling 2 adjacent figures in the y- (Frequency) direction. How can I acheive this?

My tries are shown in the code below.

close all; clear clc;
%% Mathworks question

n_timepnts  = 500                       ; % Number of time points
n_freqs     = 100                       ; % Number of frequency points
Freq        = logspace(0,3,n_freqs)     ; % Logarithmic distribution
DataArray   = rand(n_freqs,n_timepnts)  ; % Some arbitrary data array
Spectrum    = rand(n_freqs,1).*max(DataArray,[],2)       ; % Arbitrary spectrum
time        = linspace(0,365,n_timepnts); % Arbitrary time array in days

%% imagesc approach
figure(1);clf
set(1,'Position',[  28  250 1187  420],'Name','');
ax(1) = subplot(1,9,1:2);
plot(Spectrum,log10(Freq))
xlabel('Spectral max.')
ylabel(' Frequency [Hz]')
title('Arbitrary Spectrum')
grid on;
%---
ax(2) = subplot(1,9,4:9);
imagesc(time,log10(Freq),DataArray)
set(gca,'YDir','normal');
xlabel('Day of Year')
ylabel(' Frequency [Hz]')
title('Spectral data over time')
%---
linkaxes(ax,'y')

Using the contourf approach (not preferred due to slow response)

%% contourf approach
figure(2);clf
set(2,'Position',[  48  250 1187  420],'Name','');
ax(1) = subplot(1,9,1:2);
semilogy(Spectrum,Freq)
xlabel('Spectral max.')
ylabel(' Frequency [Hz]')
title('Arbitrary Spectrum')
grid on;
%---
[X,Y] = meshgrid(time,log10(Freq));
ax(2) = subplot(1,9,4:9);
contourf(X,Y,DataArray,'LineColor','none')
% contourf(time,Freq,DataArray,'LineColor','none')
set(gca,'YDir','normal');
xlabel('Day of Year')
ylabel(' Frequency [Hz]')
title('Spectral data over time')
%---
% linkaxes(ax,'y')

yields.

Help please!

Kind regards,

Eric

Answer 1

[Kevin Holly]

You can add Listeners as shown below.

%% Parameters
n_timepnts  = 500                       ; % Number of time points
n_freqs     = 100                       ; % Number of frequency points
Freq        = logspace(0,3,n_freqs)     ; % Logarithmic distribution
DataArray   = rand(n_freqs,n_timepnts)  ; % Some arbitrary data array
Spectrum    = rand(n_freqs,1).*max(DataArray,[],2)       ; % Arbitrary spectrum
time        = linspace(0,365,n_timepnts); % Arbitrary time array in days

%% contourf approach
figure(2);clf
set(2,'Position',[  48  250 1187  420],'Name','');
ax(1) = subplot(1,9,1:2);
semilogy(Spectrum,Freq)
xlabel('Spectral max.')
ylabel(' Frequency [Hz]')
title('Arbitrary Spectrum')
grid on;
%---
[X,Y] = meshgrid(time,log10(Freq));
ax(2) = subplot(1,9,4:9);
contourf(X,Y,DataArray,'LineColor','none')
% contourf(time,Freq,DataArray,'LineColor','none')
set(gca,'YDir','normal');
xlabel('Day of Year')
ylabel(' Frequency [Hz]')
title('Spectral data over time')
%---
% Add Listeners

addlistener(ax(1),'YLim', 'PostSet',@(~,events)set(ax(2),'YLim',log10(ax(1).YLim)))
addlistener(ax(2),'YLim', 'PostSet',@(~,events)set(ax(1),'YLim',10.^(ax(2).YLim)))

[Eric Kappel]

I certainly would not have come up with such an elegant solution!!! I think I can exploit this feature a lot more!

Thank you!!!

Kind regards,

Eric