In the paper, to solve the global optimization problems, we propose a novel
parameter-free filled function. Based on the non-parameter filled function, a new filled
function algorithm is designed. In the algorithm, the selection and adjustment of parameters can be ignored by the characteristic that the filled function is parameter-free. In
addition, in the region lower than the current local minimizer of the objective function, the
filled function is continuously differentiable which enables any gradient descent method
to be used as a local search method in the algorithm. Through numerical experiments by
solving two test problems, the effectiveness of the algorithm is verified.