An ε-active barrier-function method for solving minimax problems