'Interp1 error. Value X should be distinct.'

I have matrix A below. When I try using interp1, I get this error: ??? Error using ==> interp1 at 259. The values of X should be distinct.

A=  [17.4000,0.9949;
    16.7000,0.9964;
    15.9000,0.9978;
    15.3000,0.9993;
    14.5000,1.0000;
    13.8000,1.0000;
    13.1000,0.9985;
    12.4000,0.9971;
    11.7000,0.9942]

Val=interp1(A(:,1),A(:,2),15.0000)

Matlab treats 15.3 and 14.5 as 15? How do I get around this problem?

Answer 1

[Jan]

Finding duplicates is much cheaper, when you sort the data. Therefore I suggest to sort them explicitly, when you want to avoid the implicit sorting in unique():

[A1, index] = sort(A(:,1));
A2          = A(index, 2);
uniq        = [true, diff(A1) ~= 0];
Value       = interp1(A1(uniq), A2(uniq), 15.0000);

Btw. for a single value interp1 is not efficient. A hard-coded linear interpolation would be much cheaper. But this matters only, if this code section takes a significant part of the computing time.

[zohar]

+1 Just....

uniq        = [true; diff(A1) ~= 0];

Answer 2

[Walter Roberson]

MATLAB R2012a outputs a value for me when I copy and paste your code.

However, if you were to alter your code slightly to

Val=interp1(A(:,2),A(:,1),15.0000)

then you would have a problem because A(:,2) contains two copies of 1.0000

Please recheck on your system.

[Yun Inn]

Thank you. I found a duplicate somewhere else in the matrix and eliminated it with UNIQUE. However, UNIQUE sorted the data. Is there any way to eliminate duplicate without sorting?

[Andrei Bobrov]

v = [2 5 6 5 8 3];

[a,b] = unique(v,'first');
ii = sort(b);
out = a(ii);

or

for R2012a and later

out = unique(v,'stable');