Is it possible to vectorize for only one input?

Suppose there are two variables.

v1 = [1 2 3];
m1 = [1 2 3; 4 5 6];

I hope to create another matrix m2, having the same size with m1, and its (i, j) component is defined by

 m2(i, j) = sum(v1 > m1(i, j))

In this case, clearly one solution is use "for" loop twice.

m2 = zeros(2, 3);

for i = 1:2

    for j = 1:3

      m2(i, j) = sum(v1 > m1(i, j));

    end

end

However I want to know whether we can apply vectorization for the above procedure. To do this, I first thought that I may create an anonymous function

test_opr = @(v, x) sum(v > x);

and vectorize for the only latter input(while fixing v). But I couldn't find a proper way to do this. Is there any useful trick or alternative?

Thanks in advance.

Answer 1

[José-Luis]

 v1 = [1, 2, 3]; m1 = [1, 2, 3; 4, 5, 6];
 your_result = arrayfun(@(x) sum(v1 > x), m1)

Though this is just syntactic sugar around looping. Maybe faster, would need to test:

 v1 = reshape(v1,1,1,[]);
 alt_result = sum(bsxfun(@lt,m1,v1),3)

[sykim14]

Thanks! Your answer helped me a lot. Not just for this specific case, your answer helped me understand arrayfun and bsxfun more deeply.

[José-Luis]

My pleasure.