《《1996 Implementation of Interior Point Methods for Large Scale Linear Programming》.pdf

Impleme且tatlo口。f Interior Point Methods for Large Scale Linear Prograrnrnin旷l Erling n. Andersen ! .Ja.rek Conrhio1 Csa.h a. MéR元4rosf Xlao|1e Xu|| Iζchnical Rcport 1996.:.1 Lυ2 ilah. HEC Ceneva. Section of i\lana2elllent Studies University of Ceneva , 102 Bd Carl 飞!ogt ， CH.1211 Cene飞••a. I. S矶.it 冗erland f .Jan uary 2 t 吨 1996 Ahstrad 1n the p ö. si 10 y俨乱m the inl♂rior poinL mdh川1s (11- 1\1) 101 linear progr a.mm .in):!; hav( gaîncd cxtrn,ordînary întcrcst 勘乱n ah(飞rnatrvc tυth(飞 sp旧时 simpkx bíl配] mcthod日 Thîs h a.s initi a.ted a fruitful competition betwcen the two types of algorithms w11ich h a.s le a. cl to very dficicnt implcmcntations 011 t 叫11 sicles. The si只nI丑cant clifference b叽v\e011 intcrior poillt and 臼.i mplex ba.:.,; ed 1盯Lhod吕.ir:; rdle可 icd noL only .in [， h俨 ihcülTi .i ca.1 ba.ck只rüLLnd bLLL a.bo .i n thc pr a. dical implcmentation. 1n this paper v\e givc an overviev\ of the most important cha.racicr.i日L.i cr:; 01 adv旧lC 川.1 .iJnpklneni川101览。r .inte 1101 pO lJli 1盯Lhod吕 lu臼 L ， \V俨I-l1er:; cnL th俨 infca.sibk-primn,l-dllal algυrithm which is widcly consid川叫 t lH飞 most c而cicnt gcncrn,l pllrposc lPlvL Our discussion includes various algorithmic enhanccmcnts of the basi c. algorithm.