非线性方程组的一个不使用罚函数和滤技能的算法.pdf

àó’ŒÆa¬Æ Ø© š‚5§|˜‡Ø¦^v¼êÚÈEâŽ{ Á ‡ ©‰Ñ ¦)š‚5§|˜‡#Ž{§Äkòš‚5 §|=z˜‡š‚55y§2¦^˜‡Ø¦^v¼êÚÈE⎠{¦)ù‡š‚55y §3JacobiÝ ˜—÷ ^‡ey ²dŽ {)?ÛS4:Ñ´š‚5§|) ÏL3Ž{ ¥Ú ?Eâ5ŽÑŒUMaratosA§Œ ±y ²ù‡{´ÛÜ ‡‚5Âñ …i: š‚55y ئ^v¼êÚÈEâ{ ‚|¢ ‡‚5Âñ i š‚5§|˜‡Ø¦^v¼êÚÈEâŽ{ AN ALGORITHM WITHOUT A PENALTY FUNCTION OR A FILTER FOR NONLINEAR EQUATIONS ABSTRACT We present a new algorithm for solving a system of nonlinear equations. The system of nonlinear equations is reformulated into a nonlinear programming prob- lem firstly, and then we solve the problem by a method without a penalty function or a filter. Under the standard assumption that Jacobi matrices are uniformly full rank, it is proved that every limit point of the sequence generated by the algorithm is a solution of the system of nonlinear equations. Introducing the second-order correction technique to the algorithm for overcoming the so-called Maratos effect, this algorithm will locally have superlinear convergence. KEY WORDS: nonlinear programming, no a penalty function or a filter tech- nique , line search, superlinear convergence ii š‚5§|˜‡Ø¦^v¼êÚÈEâŽ{ ÎÒ`² 1. .− L«Euclidean‰ê. 2. A − L«åjacobian Ý . 3. g(x)− L«8I¼êFÝ. k 4. x − L«1 k S“ •þx §x −L«•þx 1j ‡©þ. j k 5. A − L«1 k S“ Ý A §A −L«•þA 1j ‡. j 6. αk = O(β k )−L«3Õáuk ~êM §¦αk ≤ Mβ k .