Interpolating data in 1D with 2D sample points and a 2D matrix

I have a matrix of Temperatures ( T - [120xN] ) at different heights ( H - [120xN] ). I want to obtain the Temperatures at a set vector of heights ( Hnew - [100x1] ], where N is the total number of lines I want to independently interpolate in 1D.

I can solve this with a loop quite easily

Tnew = zeros(length(Hnew),N);
for i-1:N
   Tnew(:,i) = interp1(H(:,i),T(:,i),Hnew);
end

But since I have many iterations to perform (N ~ 2 million), I am looking for a solution without a loop.

What is the optimum way to do this?

Answer 1

[Vijay]

Hello @Jose Luis Villaescusa Nadal

The use of loop does not create any performance disadvantage. The bottleneck is the interpolation part. If your computer has large number of cores, >4, then you can utilize parfor pragma to parallelize your loop.

Tnew = zeros(length(Hnew),N); 
parfor 
    for i-1:N 
        Tnew(:,i) = interp1(H(:,i),T(:,i),Hnew); 
    end 

Please use the link below for more information

Execute for-loop iterations in parallel on workers - MATLAB parfor - MathWorks India

Hope that helps