How to set the lower bound of extrapolation?

Hi!

Is it possible to set a lower bound for extrapolating using the interp2 function? I am dealing with nitrogen concentrations, and a negative concentration cannot exist. However, currently the extrapolation results have negative values. So I do not want the extrapolation to go less than 0. I do not desire to set an upperbound limit.

Thanks! George

[Star Strider]

Is ‘interp2’ producing below-zero values in its interpolation or in its extrapolation? (You have more control over its extrapolation range than interpolation.) What 'method' are you using with ‘interp2’?

[George McFadden]

The interpolation and extrapolation are both producing below 0 values. Thanks!

[charles alexander]

yes, its is possible to set a lower bound.

you will need to find the value of c from the linear discriminant model.

extrapolating will be (c1+c2)/100*values less than 30,

find the percentage value of extrapolation, add the (percentage value to the value of extrapolation:c3).

(c1+c2)/c3*fill_value.

Answer 1

[D G]

You could set the boundary limit manually after the calculation like this.

concentration(concentration < 0) = 0;

I create a function that does just this.

function value = clip(value, clip, type)
  %CLIP Clip values
  % Adjust values in a vector such that no value extends beyond the
  % specified type.
  %
  %   Inputs:
  %       Clip min and max together:
  %           value (vector - 1 by x)
  %           clip  (vector - 1 by x > 1)
  %       Clip min and max seperately
  %           value (vector - 1 by x)
  %           clip  (scalar)
  %           type  (char - min/max)
  %
  %   Outputs:
  %       value - The clipped values

    if length(clip) > 1
        value(max(clip) < value) = max(clip);
        value(value < min(clip)) = min(clip);
    else
        switch lower(type)
            case 'min'
                value(value < clip) = clip;

            case 'max'
                value(clip < value) = clip;
        end %switch
    end
end %clip

And you can use it like this.

concentration = clip(concentration, 0, 'min');

But it is more useful when you want to clip on both ends of the spectrum.

concentration = clip(concentration, [0, 100]);

If you have more than 2 values in the boundary, it choose the min and max as the clipping boundaries.

[Walter Roberson]

concentration(concentration < 0) = 0;

can also be formulated as

concentration = max(0, concentration);

[D G]

So it can, hilarious! It seems a little backwards. Why wouldn't that be min, and the reverse be max?

http://www.youtube.com/watch?v=JyV96ZpJG98#t=04m23s