How to find the zero crossing in x and time data sets?

How can I find the zero crossing in the data sets?

figure()

plot(x,t)

[Scott MacKenzie]

Nice plot. Just look at it -- you can easily find where the line crosses 0. Is there something else you might be looking for?

[vimal kumar chawda]

Yes we dont have 0.000000000000 so we have range is 10^-3 so all the value fall under 10^-3 are accepted. Now how can we do it as we have two sets x& t?

[Rena Berman]

(Answers Dev) Restored edit

Answer 1

[Scott MacKenzie]

Here's what I put together. The variable fCross is what you are looking for.

% data from posted matlab.mat files
f = readmatrix('testdata1.txt'); 
t = readmatrix('testdata2.txt');

tiledlayout(3,1);

nexttile;
plot(t,f);
hold on;
axis([1 10 -1.2 1.2]);

nexttile;
fAbove = f .* (f >= 0);
fBelow = f .* (f <= 0);
area(t, fAbove, 'FaceColor', 'r');
hold on;
area(t, fBelow, 'FaceColor', 'g');
axis([1 10 -1.2 1.2]);

nexttile;
fCrossRaw = find(diff(fAbove>0));
fCross = fCrossRaw ./ length(t)*10; % as per axes
plot(fCross, zeros(1,length(fCross)), '*r');
hold on;
axis([1 10 -1.2 1.2]);
xline(fCross, 'color', [.7 .7 .7]);
yline(0, 'color', [.7 .7 .7]);

[vimal kumar chawda]

thank you sir.

[Scott MacKenzie]

@vimal kumar chawda You're welcome. I just updated my answer to make the 3rd plot look a bit better.

Answer 2

[Stefan Schuberth]

If you have (x,y) data and want to do it without using loops try that:

i=find(y(1:end-1).*y(2:end)<0); % index of zero crossings

m=(y(i+1)-y(i))./(x(i+1)-x(i)); % slope

x0=-y(i)./m+x(i); % x coordinates of zero crossings linear interpolated

Answer 3

[Joel Lynch]

idx = find( f(2:end).*f(1:end-1)<0 )

Will return the left-hand indicies of cross-over points.

To get the exact X-values where the cross-over occurs, you would have to do some linear intepolation (inverted)

t_zero = zeros(size(idx));
for i=1:numel(idx)
    j = idx(i); % Need both the index in the idx/t_zero vector, and the f/t vector
    t_zero(i) = interp1( f(j:j+1), t(j:j+1), 0.0, 'linear' );
end

Note: this will fail if the cross-over happens on the last i value (i+1 would extend outside the range of the dataset)

[vimal kumar chawda]

yea, true it fails so how to fix it?

[Joel Lynch]

Simplest solution is to put the interpolation line in an if statement

if j< numel(f)
   % do j:j+1 interpret
else
   t_zero(i) = interp1( f(j-1:j), t(j-1;j), 0.0, 'linear' );
end

or something similar

[vimal kumar chawda]

It worked thanks.

[Scott MacKenzie]

Yes, Joel's code gives the exact cross-over point. Bear in mind, however, that this is exact for the linearly interpolated data. The actual data are empirical, so it is not possible to know the exact cross-over point.

It probably doesn't matter much in this case, since the data appear to be gathered at a high sampling rate.