AN AGGREGATE FUNCTION METHOD FOR NONLINEAR PROGRAMMING

摘要: This paper presents a new method, called the "aggregate function" for solvingnonlinear programming problems. At first, we use "maximum" constraint in place of theoriginal set to convert multi-constrained optimization problem non-smoothbut singly constrained problem; then employ surrogate concept and themaximum entropy principle derive smooth function, by which non-smooth maximumconstraint is approximated original converted con-strained furthermore, develop multiplier penalty algorithm. The presentalgorithm has merits stable fast convergence ease computer implementation,and particularly suitable solving nonlinear with large num-ber constraints.