x,y are two variables which are two vectors by N*1. [x y] is the coordinates of an moving object. Dmax is the longest distance that the object can travel within each time slot. So how can I write the following constrain?

if not in CVX, I think the followling expression maybe right.

for i=1:N-1
(x(i+1)-x(i))^2+(y(i+1)-y(i))^2<=Dmax^2
end

Problem solved, thanks! But I have another question. The output of square_pos(x) is 1/-1, for example: the output of square_pos(2) is 1, and the output of square_pos([1;2;3]) is [1;1;1]. I don’t know how the problem appears.

I checked the source code of square_pos(), and in fact square_pos() calls another function square. y = square( max( x, 0 ) );
But the fuction square() doesn’t return the square of the input. It returns 1 or -1. So I guess this is where the problem arises. But how can I fix the problem?

Does the output of which -all square
show anything other than two entries in cvx directories? If so, there is another square on your computer which perhaps is getting in ahead of CVX’s version. However, perhaps that is only a problem for numerical arguments and not for CVX arguments. The version inside the @cvx directory is what is used for CVX arguments, and that is more complicated than the version for numerical arguments.

I changed my MATLAB path, and this time the CVX version square is called succesively. But here is another error: The function 'square' corresponding to the input parameter of type 'double' is not defined