邵 殿 國(guó)
(1.吉林大學(xué) 數(shù)學(xué)研究所,長(zhǎng)春 130012;2.東北電力大學(xué) 理學(xué)院,吉林 吉林 132012)
?
研究快報(bào)
正倒向隨機(jī)比例系統(tǒng)的隨機(jī)最大值原理
邵 殿 國(guó)1,2
(1.吉林大學(xué) 數(shù)學(xué)研究所,長(zhǎng)春 130012;2.東北電力大學(xué) 理學(xué)院,吉林 吉林 132012)
利用經(jīng)典變分方法、對(duì)偶方法和可料倒向隨機(jī)微分方程,考慮狀態(tài)方程為正倒向隨機(jī)比例方程的隨機(jī)最優(yōu)控制問(wèn)題,得到了該問(wèn)題的隨機(jī)最大值原理.
隨機(jī)控制;隨機(jī)最大值原理;正倒向隨機(jī)比例方程;變分法
目前,關(guān)于隨機(jī)控制系統(tǒng)最大值原理的研究已取得許多結(jié)果[1-10].隨機(jī)延遲微分方程(隨機(jī)時(shí)滯微分方程)在科學(xué)與工業(yè)的很多領(lǐng)域內(nèi)應(yīng)用廣泛.近年來(lái),隨機(jī)比例方程:
(1)
這種特殊的隨機(jī)延遲微分方程也受到廣泛關(guān)注[5].
Bismut[1]通過(guò)引入線(xiàn)性倒向隨機(jī)微分方程,給出了顯式解.Pardoux等[6]給出了更一般形式的倒向隨機(jī)微分方程,并證明了其解的存在唯一性.Peng等[7]得到了一類(lèi)新型倒向隨機(jī)微分方程:可料倒向隨機(jī)微分方程.該隨機(jī)微分方程與隨機(jī)延遲微分方程之間存在對(duì)偶關(guān)系,可利用這種對(duì)偶關(guān)系得到隨機(jī)延遲系統(tǒng)的隨機(jī)最大值原理.當(dāng)考慮金融領(lǐng)域大客戶(hù)投資問(wèn)題時(shí),即涉及利用正倒向隨機(jī)微分方程模擬投資行為.本文研究正倒向隨機(jī)比例系統(tǒng)的隨機(jī)最大值原理.
令(Ω,F,P,{Ft}t≥0)是一個(gè)完備的概率空間,其中Ft是d-維布朗運(yùn)動(dòng){Bt}t≥0生成的信息流.給定T>0,|·|表示m中的范數(shù),〈·,·〉表示內(nèi)積.考慮正倒向隨機(jī)比例系統(tǒng):
(2)
其中:f:[0,T]×n×n×U→n;σ:[0,T]×n×n×U→n×d;g:[0,T]×n×n×m×m×d×U→m;h:n→m.令U是k的非空凸子集.記Uab={v(·)∈U|v(t)}是Ft-適應(yīng)的,且∞.Uab的元素稱(chēng)為容許控制.ξ(·)和η(·)分別是x(·)和v(·)的初始路徑,且<+∞.為方便,記x(qt)為xq.
假設(shè):
(H1)f,σ,g,h關(guān)于(x,xq,y,z,v)線(xiàn)性增長(zhǎng),f,σ,g,h在(x,xq,y,z)內(nèi)連續(xù)可微、偏導(dǎo)數(shù)有界.f,g在v上連續(xù),偏導(dǎo)數(shù)在(x,xq,y,z,v)內(nèi)連續(xù).
(H2)l在(x,xq,y,v)內(nèi)連續(xù)可微、偏導(dǎo)數(shù)有界;h1在x上連續(xù)可微、導(dǎo)數(shù)有界;h2在y上連續(xù)可微、導(dǎo)數(shù)有界.
由假設(shè)(H1),對(duì)于任意的v(·)∈Uab,系統(tǒng)(2)有唯一解(x(·),y(·),z(·)).
定義效用泛函:
最優(yōu)控制問(wèn)題就是找最優(yōu)控制v(·),使效用泛函J(v(·))取最小值.對(duì)系統(tǒng)(2)求變分,得到變分方程:
其中:fx=fx(t,x(t),x(qt),v(t));fxq=fxq(t,x(t),x(qt),v(t));σx=σx(t,x(t),x(qt),v(t));σxq=σxq(t,x(t),x(qt),v(t));fv=fv(t,x(t),x(qt),v(t)).對(duì)g取相同記號(hào).
定理1假設(shè)(H1),(H2)成立,(x*,y*,v*)是控制問(wèn)題的最優(yōu)對(duì),則可得可料正倒向隨機(jī)比例系統(tǒng):
系統(tǒng)(3)的解(p(·),A(·))稱(chēng)為伴隨過(guò)程.進(jìn)一步可得
證明:對(duì)效用泛函求變分,得
(4)
其中v=t/q.
dA(t)=-[gyA(t)+F]dt+gz(t)A(t)dB(t).
(5)
(6)
取C=lx,D=lxq,F=ly,消除式(4)中x1和y1的變分項(xiàng).將式(5),(6)代入式(4)中,有
(7)
由式(7)得Hamilton函數(shù):
H=l(t,x,xq,y,v)+〈p,f(t,x,xq,v)〉+〈A,g(t,x,xq,y,z,v)〉+〈k,σ(t,x,xq,v)〉,
相應(yīng)的伴隨方程為式(3).
[1] Bismut J M.Conjugate Convex Functions in Optimal Stochastic Control [J].J Math Anal Appl,1973,44(2):384-404.
[2] Kushner H J.Necessary Conditions for Continuous Parameter Stochastic Optimization Problems [J].SIAM J Control Optim,1972,10(3):550-556.
[3] PENG Shige.A General Stochastic Maximum Principle for Optimal Control Problems [J].SIAM J Control Optim,1990,28(4):966-979.
[4] PENG Shige,WU Zhen.Fully Coupled Forward-Backward Stochastic Differential Equations and Applications to Optimal Control [J].SIAM J Control Optim,1999,37(3):825-843.
[5] FAN Zhencheng,LIU Mingzhu,CAO Wanrong.Existence and Uniqueness of the Solutions and Convergence of Semi-implicit Euler Methods for Stochastic Pantograph Equations [J].J Math Anal Appl,2007,325(2):1142-1159.
[6] Pardoux é,PENG Shige.Adapted Solution of a Backward Stochastic Differential Equation [J].Systems Control Lett,1990,14(1):55-61.
[7] PENG Shige,YANG Zhe.Anticipated Backward Stochastic Differential Equations [J].Ann Prob,2009,37(3):877-902.
[8] HUANG Jianhui,SHI Jingtao.Maximum Principle for Optimal Control of Fully Coupled Forward-Backward Stochastic Differential Delayed Equations [J].ESAIM Control Optim Calc Var,2012,18(4):1073-1096.
[9] JI Shaolin,WEI Qingmeng.A Maximum Principle for Fully Coupled Forward-Backward Stochastic Control Systems with Terminal State Constraints [J].J Math Anal Appl,2013,407(2):200-210.
[10] WEI Qingmeng.The Optimal Control Problem with State Constraints for Fully Coupled Forward-Backward Stochastic Systems with Jumps [J/OL].Abstr Appl Anal,2014-02-13.http://dx.doi.org/10.1155/2014/216053.
(責(zé)任編輯:趙立芹)
StochasticMaximumPrincipleofForward-BackwardStochasticPantographSystems
SHAO Dianguo1,2
(1.InstituteofMathematics,JilinUniversity,Changchun130012,China;2.CollegeofScience,NortheastDianliUniversity,Jilin132012,JilinProvince,China)
We made an investigation into the stochastic optimal control problem of the stochastic delayed system described by forward-backward stochastic pantograph equations with the aid of classical variational approach,duality method and the anticipated backward stochastic differential equations,obtaining the maximum principle for this problem.
stochastic control;stochastic maximum principle;forward-backward stochastic pantograph equation;variational approach
10.13413/j.cnki.jdxblxb.2015.03.18
2015-03-09.< class="emphasis_bold">網(wǎng)絡(luò)出版時(shí)間
時(shí)間:2015-03-25.
邵殿國(guó)(1976—),男,漢族,博士研究生,講師,從事隨機(jī)動(dòng)力系統(tǒng)的研究,E-mail:shaodgnedu@163.com.
國(guó)家自然科學(xué)基金(批準(zhǔn)號(hào):11401089)和吉林省科技發(fā)展計(jì)劃項(xiàng)目(批準(zhǔn)號(hào):20130101065JC).
http://www.cnki.net/kcms/detail/22.1340.O.20150325.1812.001.html.
O211.63
:A
:1671-5489(2015)03-0451-03