perform interpolation each row of a matrix

i have a matrix A with one million rows, and each row is monotonic increase. the value in each row is x.

The interpolation is perform to obtain one million yi for each row given a constant xi value (xi is the same for each row) and given an array y (y is also the same for each row). is there an efficient way to get yi other than using loop.

right now i have to use loop like this:

for i = 1:1e6

yi(i) = interpi(A(i,:), y, xi);

end

Answer 1

[Fangjun Jiang]

You can use interp1() directly for this without for-loop.

x=1:10;
A=(1:10)'+(1:10);
xi=5;
y=interp1(x,A',xi)

y = 1×10

6 7 8 9 10 11 12 13 14 15

[wangm]

Thanks for quick response. but Matlab gives error for the second line A=(1:10)' + (1:10):

Error using +

Matrix dimensions must agree.

Are you sure you get the results using your script?

[Fangjun Jiang]

That is fine. That code only works in later versions of MATLAB. The result actually comes from the online execution of the code. You can do that too on this Q&A forum. Make the text as code and click Run.

Hope you could run this in R2011b (that was 10+ years old).

A=repmat(1:10,10,1)+repmat((1:10)',1,10)

[wangm]

Thanks. I understand now. Basically, function interp1 work for column. I just need to transpose the matrix.

[wangm]

wait, you did not answer my question. in my question, matrix A is x, but you matrix A is y.

[Fangjun Jiang]

I was taking some guesses on your question. Now you need to clarify these. "the value in each row is x", "in my question, matrix A is x", "but you matrix A is y".

The best way is to provide a complete numerical example, including the value for x, xi, y.

You don't need to provide one million rows. Three row is sufficient.