How can I populate the rows of a difference matrix without using a for loop?

I am creating Matlab code to construct the following matrix:

In my data, x and y are 101-dimensional real vectors and N is a vector of six real numbers. My Matlab implementation is

% Real inputs of length 101 for x,y and length 6 for X,Y
x = rand(1,101);
y = rand(1,101);
X = rand(6,1);
Y = rand(6,1);
n = length(x);
N = length(X);
G = zeros((N-1),2,n);

% Distance calculation
d = @(k) sqrt((X(k) - x).^2 + (Y(k) - y).^2);  % L2 Norm / Euclidean distance
xdiff = @(k) (((X(1) - x) ./ d(1)) - ((X(k+1) - x) ./ d(k+1)));
ydiff = @(k) (((Y(1) - y) ./ d(1)) - ((Y(k+1) - y) ./ d(k+1)));
GBlock = @(k) [xdiff(k) ydiff(k)];

% This seems inefficient - is there a better alternative?
Z = all(G>=0,2);
for k = 1:size(G,1) 
    if Z(k)
        G(k,:) = GBlock(k);
    end
end

Creating Gblock followed by a for loop seems inefficient. Is there a way to vectorize this code and avoid writing a for loop?

[Matt J]

Your code doesn't match your equation for G. The equation shows no differences between X and y but such differences are present in your code, e.g., "X(k+1) - y" as well as "Y(1) - x"

[Matthew Kehoe]

Yes, I incorrect indexed two locations of x and y. I have updated my original question.

[Matt J]

Very well, but what about my answer below?

Answer 1

[Matt J]

X=X(:);
Y=Y(:);
x=reshape(x,1,1,[]);
y=reshape(y,1,1,[]);

d=hypot( X - x , Y - y );

Diff=@(U,u) ((U(1)-u)./d(1,1,:)) - ((U(2:end) - u) ./ d(2:end,1,:));

G=[Diff(X,x), Diff(Y,y)];