Performing calculation across large data set

I have a large matrix of latitude and longitude coordinates (343212 x 2 double). I want to calculate the distance between each set of coordinates in this matrix and a single reference coordinate. I am using the LatLon Distance function.

What I do not know how to do is loop through each pair of coordinates in the data set so that it returns a vector with the distance between each pair of coordinates and the reference point.

Any help is much appreciated, thank you.

Answer 1

[Guillaume]

Assuming you're using R2016b, this modified function will work straight away with your matrix and single reference

function [d1km, d2km] = betterlldistkm(latlon1, latlon2)
   %  latlon1: either a single point (1x2) or a matrix of points (nx2)
   %  latlon2: either a single point (1x2) or a matrix of points (nx2)
   %  if both latlon1 and latlon2 are matrices of point, both must have the same number of points
   radius = 6371;
   latlon1 = latlon1 * pi/180;
   latlon2 = latlon2 * pi/180;
   delta = latlon1 - latlon2;
   a = sin(delta(:, 1)/2).^2 + cos(latlon1(:, 1)).*cos(latlon2(:, 1)).*sin(delta(:, 2)/2).^2;
   c =  2*atan2(sqrt(a), sqrt(1-a));
   d1km = radius*c;
   x = delta(:, 2) .* cos(latlon1(:, 1)+latlon2(:, 1))/2;
   y = delta(:, 1);
   d2km = radius * hypot(x, y);
end

Usage example:

a = rand(34312, 2);
b = rand(1);
[d1, d2] = betterlldistkm(a, b)

That's all.

Answer 2

[Iddo Weiner]

a = rand(10,2); %change this to your data
out = nan(length(a));
for i = 1:(length(a))
    out(i,1) = abs(mean([a(i,1),a(i,2)]) - constant part;
        % change this function to the one you want to use, I just invented
        % something for the example
end

[Guillaume]

Please, do not use length on matrices. Your code will fail if the input matrix only has one row (since length will then return the number of columns). Use size with a explicit dimension, in this case size(a, 1).

I wouldn't recommend length for vectors either. numel is safer.