The fastest way to find local minimum

What is the most efficient/fastest way to find the circled local minimum below, the lowest minima between two highest peaks? The graph was created by plot(x,y).

[Walter Roberson]

To find it given what starting conditions?

[Hg]

To find the lowest between two highest peaks. I use 'MinPeakProminence' but I'm looking for a better way.

[peaks, locs] = findpeaks(x, y,'MinPeakProminence',100);
min = 100000;
for i = locs(1): locs(2)
   if( x(i) < min )
       min = x(i);
       min_loc = i;
   end
end

[Walter Roberson]

The lowest between the two highest peaks is about about x = 308. The two highest peaks happen to be close together.

[Walter Roberson]

Never name a variable "min"; that interferes with using the MATLAB min() function. For example,

[peaks, locs] = findpeaks(x, y,'MinPeakProminence',100);
[minval, minidx] = min( x(locs(1):locs(2)) );
%minidx is relative to the subset of values, so move it back to full range
minidx = minidx + locs(1) - 1;

[Hg]

Thanks for the advice. I think using 'MinPeakProminence' to get 2 highest peaks is not robust enough. It might get more than 2 peaks like below where the first and second peak have same value.

Answer 1

[Image Analyst]

Invert your signal

invertedSignal = max(signal(:)) - signal;

Now valleys will be peaks. Then call find peaks.

[Hg]

ok, this is genius!