Interp1 - The grid vectors must contain unique points

I have a text file that has been plotted and I need to find the x value from a known y value. However there isn't a data point at this point so I have interpolated. This has worked for some txt files but not others.

ymax = max(y); 
xmax = find(y == ymax); 
xmax = x(xmax);
yhalfmax = max(y)/2; 
xhalfmax = interp1(y, x, yhalfmax, 'spline'); %interpolate to generate a point at the yhalfmax point

 Error: Error using gridded Interpolant. The grid vectors must contain unique points.

I have tried using 'unique' but this has changed the plotting so can't be used.

[Matt J]

I have tried using 'unique' but this has changed the plotting so can't be used.

How can it "change the plotting", if all you've done is throw away duplicate points?

[D F]

When the following is used to replace the interp1 line above, the curve changes shape. I'm unsure why this would change the plot curve unless it is deleting usefull data points

[x, index] = unique(y); xhalfmax = interp1(y(index), x,yhalfmax, 'linear');

Answer 1

[Star Strider]

My usual approach to the problem of non-unique independent variable values for interp1 is to add a very small, increasing value to each element.

Example —

XData = sort(randi(9, 1, 10))                                   % Create Data
XDataUnique = XData + linspace(0, 1, length(XData))*1E-3        % Add Increments To Each Element

I used ‘1E-3’ here to illustrate the idea. In practice, I use a much smaller multiplier, ‘1E-10’ or so.

[D F]

Thanks but this is unfortunately still giving the same error. I have tried to just straight forwardly multiply the interpolated data by 1.001 to make sure it doesn't duplicate the normal x,y data but this has still failed

[Star Strider]

My pleasure.

You have to add an increasing vector of small values to the data with duplicate entries to eliminate the problem with non-unique values. Adding a single scalar value to each element changes nothing.

Run my code to see how it works.

[D F]

Aaah okay that makes sense! I have run the code and it is still returning the same error.

I then exchanged the x in the interp1 line of my code for XDataUnique however the index exceeded the matrix dimension, any thoughts?

[Star Strider]

Looking at your code, it seems that you need to use it to create your ‘y’ data to do your interpolation, not your ‘x’ data.

Try this:

y = sort(randi(9, 1, 10))                                   % Create Data
YDataUnique = y + linspace(0, 1, length(y))*1E-3            % Add Increments To Each Element

I honestly have no idea what is causing the index error. The vector size should not change with my code.