篮球竞彩分析* :CIMSՓĵ

򾺲 www.fasfz.com  

Ǿ}һNϽⷨ

ע⣺ՓW 2001,18(3)61-66l
ʹՈעՓij̎

(EI䛱)

DŽ1,2 ӭ1,2 ؾ1,2

1.BWWϵ| B1160232.IbYc|B116023

ժ ҪBFGSc烞Yڻ׃һN׃߅sǾ}Ļσ?w㷨˻iIȫǿBFGSՔٶȿăcɞһN͹}ȫЧĴΔ_һrσԱC㷨ՔȫӋЧʱȻ烞кܴ
PI~ϷBFGS烞ȫ
ЈD̖O39 īIRaA

1 

    ϵyƹyӋW}Iܶ}Ǿз͹Եͨăgֲֻ@Щ_㷨ǽһOֵc[1]ֻ܉oܺõijʼcпܵóҪȫHͨ^ڶʼcʹÂyֵȡȫֽķȻ˂@N̎ȫֽĸʲɿԵMܴʵȫֽ㷨ȻһҪ}ݶȷȫѽо[2]SCgz㷨ģM˻㷨ȫփ}еđҲõԽԽҕ[3-4]Ątڻ烞BFGSһNкνs}(1)Ļ㷨                   min                         (1)

    ǴڷǾϵyеһN^ձĬF?㿝T?ЃSCԺ;ֲ΢^ȫSC\ПoFǶ׵Ǝ׺νYmҎsyoµû׃SCvԺҎcMЃ[5]һ烞ֲcijЩBҪLrg_ֵ@ЩBrӋrgݱغL[5]?\\iIȫ?ֲ@[5]ödg[6]]usС׃gǞˏa@һcĄtûcBFGSMЃһûBFGSֲһBFGS⸽ijՔٶȺЧ

2  磭BFGSσ

2.1 BFGS

os}ĔMţDдԵ㷨֮һBFGS̎͹ǾҎ}ƵĔWՓAò_rijՔԺ̎팍H}Чܵ˂ҕ[7-9]MţDʹ˶AϢDzֱӋ㺯HesseDzһAݶϢ 혋һϵе ƽHesse BFGSos}min ( )ҪE£

  (1)  o׃xֵx0׃SnBFGSՔȦB0=Iλꇣk=0Ӌ cx0ݶg0

  (2)  ȡsk=-Bk-1gkskһS_Lk cxk+1=xk+kskӋxk+1cݶgk+1

  (3)  ||gk+1||ܦt  BFGSYDE3t(4)

  (4)  ӋBk+1

            (2)

                         (3)

  (5)  k=k+1D(2)

2.2 烞

û↖}(1)rȽ׃ һһPϵIJ Ȼ,Logisticӳʽ(4)a ܉EͬĻ׃ MЃʽ(4) =4ѽC =4ǡƬ [0,1]֮gv

                (4)

  (1)o׃\ӴΔMo Ӌ   

  (2)  

  (3)  

  (4)  k<M

              

             ֲ׃

            Ȼk=k+1 D(2)

          k>Mt  Y

2.3 σ

    緽BFGSBmȫ֘OСֵ}(1)IJE£

step1  OûΔMiter=0׃xֵx0 

step2   ʼcBFGSîǰBFGS = 

step3 ȡ =  ȡ  ȡ  , ^СĔ

step 4  ʼcMлMû = 

step5  < iter=iter+1 Dstep2tstep6

step6  ׃܉EٴMлo ΢С_ step 4Y  < iter=iter+1 Dstep2tӋYݔ  

       ȫ֘Oֵ}max Dȫ֘OС}min 

㷨лǴ󷶇^ʹ㷨ȫ֌BFGSǾֲµMЃ̎С}ن}

3 


        D 1   - ʾD                                        D 2   ʾD

ƒɂdzsڜyԇz㷨ܵĺyԇ㷨

     

                   

QCamel 6ֲOСc(1.607105, 0.568651)(-1.607105, -0.568651)(1.703607, -0.796084)(-1.703607, 0.796084)(-0.0898,0.7126)(0.0898,-0.7126)(-0.0898,0.7126)(0.0898,-0.7126)ɂȫСcСֵ-1.031628 Q Schaffer'sԓПoOֵֻУ00ȫcֵ1˺ֵ܇һȦȡֵ0.990283˺ͣڴ˾ֲOcīI[10]ԓԓĻƄӺ˹xĸMz㷨GAMASM˿\5040rԓΨһȫc܉ҵñĻ㷨ӋCȲSCԄSCa100ͬijʼc@Щʼclһ㷨24μ܉ՔMȡֵͬr  ӋYքe1ͱ2ʾӋrgָڱv133΢CӋrg

ɱ2ҊM=1500rķ ĸʼ_40˕rӋīI[10]Сͬɻ㷨100ʼcīI[5] Ӌ100 Ք˜50000ӋY67ʞ67%ƽӋrg1.2369sʹC㷨100ȫՔ MƽӋrgҲֻ0.2142s1Ҋ㷨īI[5]ķ

1    Mȡֵͬr ӋY

_____________________________________________________________________

        M          ȫcĴΔ      ĸ   ƽӋrg 

                                    (-0.0898,0.7126)  (0.0898,-0.7126)
_____________________________________________________________________________________________

  1000         44                39                83%             0.1214s

         3000         53                45                98%             0.1955s

         5000         53                47               100%             0.2142s
________________________________________________________________________________________________

2    Mȡֵͬr ӋY

___________________________________________________________

            M     ȫcĴΔ  ĸ    ƽӋrg
____________________________________________________________________________________

           1500            40                     40%            0.1406s

           5000            73                     73%            0.2505s

          10000            88                     88%            0.4197s

      50000           100                    100%            1.6856s
____________________________________________________________________________________

4  ӋY

ɱ1ͱ2Ҋ㷨ȫ֌SMӶM_ijһĔֵMuȫĸʿ_100

       ՓfMuڅoFrʹ׃vРBԸ1cĻ\MεǎBFGSֲc_ȮǰֲֵСһֲijһc̎Ҫ׃vРBɻ\ӱvԿ֪ijһw}Mu_ijһwޔֵr׃ıvԿԵõ^ģM@һcǿԝMHҲC@һc

    ںԑBsԲͬڲͬ@Ĝyԇ  ֵMuĴСвeͬһ^gͬ\ӴΔʹʼcͬwԕȫĸ,ҪC㷨ȻԸ1ՔȫȻMu ۙӋgYCMr㷨Ĵ_ֲc^mȫMһ㷨ӋrgҪMڞʹ㷨ȫMл

5  YZ

û׃\cMЃзdzֲԓcBFGSYʹڿԽܵӋ܉Ӌõ}H烞Ժһ½㷨YʹǾڱIJõBFGSõLogisticӳa׃ֻǮa׃Чʽ֮һ

\cSC\Dzͬ?rS퍵ͳ՘uhһָ?òƟoҎ\?_ϖ?Ǜ]ڵcSC\^\ӿڸBvļOýyӋĔ\Ӳ؏͵ؽ^ͬһBû׃MЃȲSC׃MЃЃ

烞c½YʹНһNȫփ;׃s}ĿɿMһЧԼΰѻ烞Чڏss}ֵMһоn}

㷨ȫՔԵć񔵌WCM֮

īI

[1]ɽꐱСso飮ǾҎ}ȫփģM˻[J]AWW37(6)19975-9

[2]C A Floudas, A Aggarwal, A R Ciric Global optimum search for nonconvex NLP and MINLP problems[J]. Comput Chem Engng 1989 13(10) 1117~1132

[3]ɽxʸµȣǔֵ㷨һԣDDģM˻㷨[M]ƌW1998

[4]ɽǔֵ㷨ڶԣDDz㷨[M]ƌW1998

[5]YοO烞䑪[J]Փc14(4)1997613-615

[6]ͮꂥӲţ׃߶Ȼ烞䑪[J]cQ14(3)1999285-287

[7]ϯأǾ[M]ߵȽ1992

[8]ϯwP־Ӌ㷽[M]Ϻ?ѧH\?1983

[9]Press W H, Tenkolsky S A, Vetterling W T, Flannery B PNumerical Recipes in C, The Art of Scientific Computing[M] Second edition Cambridge University Press 1992

[10]J C PortsThe development and evaluation of an improved genetic algorithm based on migration and artificial selection[J]IEEE Trans. Syst. Man and Cybern.1994, 24(1)73-85

A Hybrid Approach for Nonlinear Optimization

WANG Denggang1,2 ,  LIU Yingxi1,2 ,  LI-Shouju1,2

(1. Dalian University of Technology, Dalian, 116023     2.State Key Lab. of Struct. Anal. of Ind. Equip., Dalian, 116023) 

AbstractCombined BFGS method with chaos optimization method, a hybrid approach was proposed to solve nonlinear optimization problems with boundary restraints of variables. The hybrid method is an effective approach to solve nonconvex optimization problems, as it given both attentions to the inherent virtue to locate global optimum of chaos optimization method and the advantage of high convergence speed of BFGS method. Numerical examples illustrate that the present method possesses both good capability to search global optima and far higher convergence speed than that of chaos optimization method.
key wordshybrid approachBFGS methodchaos optimization methodglobal optimum


վ䛵ıߵՓģ

1A relialble approach to compute the forward kinematics of robot with uncertain geometric parameters

2rʼ͏ģą^gݷ

3ӅReо

4ⲻ΢һNz㷨

5YɿԷą^g

PՓՈcվ<<<վȫ>>>

gӭӑՓlՓļоIĿ
(Ոڰlԕrژ}ʹcuՓĵ}Ŀо@ӷҞg[)

| CIMSՓ | й | ̓M | | Փ | Ŀ_l | WgYԴ | վȫ | MՓľWվȫ |

line.gif (4535 ֹ)

˸õĞҷgӭӱվͶƱ{

򾺲 Ոc

վ򾺲 gӭL

ע⣺վδSDd

All rights reserved, all contents copyright 2000-2019
վ20003¿WL