stoch_prog (1)

stoch_prog (1)

ID:40961847

大小:85.24 KB

页数:21页

时间:2019-08-12

stoch_prog (1)_第1页
stoch_prog (1)_第2页
stoch_prog (1)_第3页
stoch_prog (1)_第4页
stoch_prog (1)_第5页
资源描述:

《stoch_prog (1)》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库

1、Stochasticprogramming•stochasticprogramming•’certaintyequivalent’problem•violation/shortfallconstraintsandpenalties•MonteCarlosamplingmethods•validationsources:Nemirovsky&ShapiroEE364A—StochasticProgramming1Stochasticprogramming•objectiveandconstraintfunctionsfi(x,ω)d

2、ependonoptimizationvariablexandarandomvariableω•ωmodels–parametervariationanduncertainty–randomvariationinimplementation,manufacture,operation•valueofωisnotknown,butitsdistributionis•goal:choosexsothat–constraintsaresatisfiedonaverage,orwithhighprobability–objectiveiss

3、mallonaverage,orwithhighprobabilityEE364A—StochasticProgramming2Stochasticprogramming•basicstochasticprogrammingproblem:minimizeF0(x)=Ef0(x,ω)subjecttoFi(x)=Efi(x,ω)≤0,i=1,...,m–variableisx–problemdataarefi,distributionofω•iffi(x,ω)areconvexinxforeachω–Fiareconvex–hen

4、cestochasticprogrammingproblemisconvex•Fihaveanalyticalexpressionsinonlyafewcases;inothercaseswewillsolvetheproblemapproximatelyEE364A—StochasticProgramming3ExamplewithanalyticformforFi•f(x)=kAx−bk2,withA,brandom2•F(x)=Ef(x)=xTPx−2qTx+r,whereTT2P=E(AA),q=E(Ab),r=E(kbk

5、2)•onlyneedsecondmomentsof(A,b)•stochasticconstraintEf(x)≤0canbeexpressedasstandardquadraticinequalityEE364A—StochasticProgramming4‘Certainty-equivalent’problem•‘certainty-equivalent’(a.k.a.‘meanfield’)problem:minimizef0(x,Eω)subjecttofi(x,Eω)≤0,i=1,...,m•roughlyspeaki

6、ng:ignoreparametervariation•ifficonvexinωforeachx,then–fi(x,Eω)≤Efi(x,ω)–sooptimalvalueofcertainty-equivalentproblemislowerboundonoptimalvalueofstochasticproblemEE364A—StochasticProgramming5Stochasticprogrammingexample•minimizeEkAx−bk1;AijuniformonA¯ij±γij;biuniformon

7、¯bi±δi•objectivePDFsforstochasticoptimalandcertainty-equivalentsolutions•lowerboundfromCEproblem:5.96stochasticsolution024681012141618certaintyequivalentsolution024681012141618EE364A—StochasticProgramming6Expectedviolation/shortfallconstraints/penalties•replaceEfi(x,ω

8、)≤0with–Efi(x,ω)+≤ǫ(LHSisexpectedviolation)–E(maxifi(x,ω)+)≤ǫ(LHSisexpectedworstviolation)•variation:addviolation/shortfallp

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。