18 lines
648 B
Matlab
18 lines
648 B
Matlab
function gradient = lsqGradient(parameter, x, y)
|
|
% The gradient of the least square error
|
|
%
|
|
% Arguments: parameter, vector containing slope and intercept
|
|
% as the 1st and 2nd element
|
|
% x, vector of the input values
|
|
% y, vector of the corresponding measured output values
|
|
%
|
|
% Returns: the gradient as a vector with two elements
|
|
|
|
h = 1e-6; % stepsize for derivatives
|
|
|
|
partial_m = (lsqError([parameter(1)+h, parameter(2)], x, y) - lsqError(parameter, x, y))/ h;
|
|
partial_n = (lsqError([parameter(1), parameter(2)+h], x, y) - lsqError(parameter, x, y))/ h;
|
|
|
|
gradient = [partial_m, partial_n];
|
|
end
|