Interp1 for non-monotonic timeseries

Hi all,

Recently I have been trying to fix a interp1-related problem. However, I am not able to solve. I need you guys help. Let me give an easy example:

I have a timeseries, y=[-3 -2 -1 1 2 3 4 5 6 5 4 3 2 1 -1 -2 -3 -2 -1 -0.5], and time axis is simply from 1 to 20. I want to find out all the time when y equals 0. It is easy to find out when we do 'plot (x, y)'. But when I use " interp1 (y,x,0)", a lot of problems happens:

_ __Error using griddedInterpolant The grid vectors are not strictly monotonic increasing.

Error in interp1 (line 183) F = griddedInterpolant(X,V,method);_ _

Looking forward to your answer.

Regards, Guojian

Answer 1

[Star Strider]

This is likely the easiest way, and is reasonably fast:

t = 1:20;
y=[-3 -2 -1 1 2 3 4 5 6 5 4 3 2 1 -1 -2 -3 -2 -1 -0.5];

ycsm = y .* circshift(y, [0 -1]);               % Will Be <0 Near Zero Crossing
yzxidx = find(ycsm < 0);                        % Find Negatives

for k1 = 1:length(yzxidx)
    idx = [yzxidx(k1) yzxidx(k1)+1];            % Index Range
    b = [[1; 1] t(idx)']\y(idx)';               % Regression Parameters
    zx(k1) = -b(1)/b(2);                        % Zero Crossings
end

produces these values of ‘t’ for the zero-crossings:

zx =

       3.5    14.5

[Guojian]

Thank you! It is brilliant!

[Star Strider]

My pleasure! And I very much appreciate your compliment!