Lagrange multipliers method in robust control: the l/sup 1/-setting