Equivalent models for finite-fuel stochastic control (Q3775455)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 4035712
Language Label Description Also known as
default for all languages
No label defined
    English
    Equivalent models for finite-fuel stochastic control
    scientific article; zbMATH DE number 4035712

      Statements

      Equivalent models for finite-fuel stochastic control (English)
      0 references
      0 references
      0 references
      0 references
      1986
      0 references
      singular control
      0 references
      local time
      0 references
      Gittins index
      0 references
      filtration
      0 references
      one-dimensional Brownian motion
      0 references
      stochastic control problem with finite fuel
      0 references
      optimal stopping
      0 references
      reflected follower problem
      0 references
      monotone follower problem with absorption
      0 references
      Let \((\Omega,{\mathcal F},P)\) be a probability space endowed with a filtration (\({\mathcal F}_ t)\) satisfying usual conditions, and let \[ {\mathcal B}=\{\xi;\quad \xi =\xi^+-\xi^-,\quad \xi^{\pm}\in {\mathcal A}\},\quad {\mathcal B}(y)=\{\xi \in {\mathcal B};\quad \xi^+_{\infty}+\xi^-_{\infty}\leq y\quad a.s.\},\quad 0\leq y\leq \infty \] where \({\mathcal A}\) is the class of nondecreasing, left-continuous and adapted processes \(\xi =\{\xi_ t\), \({\mathcal F}_ t\); \(0\leq t<\infty\}\) with \(\xi_ 0=0\). Also denote a process X, one component of the state, by \(X_ t=x+W_ t+\xi_ t\), \(0\leq t<\infty\), where \(x\in {\mathfrak R}\) is an initial state, \(\{W_ t,{\mathcal F}_ t;0\leq t<\infty \}\) is a standard, one-dimensional Brownian motion, and \(\xi\in {\mathcal B}\). Then the object of the stochastic control problem with finite fuel y is to minimize the expected total cost E[\(\int^{\infty}_{0}e^{-\alpha t}[h(X_ t)dt+d{\check \xi}_ t]]\) over \(\xi\in {\mathcal B}(y)\). Here \(\alpha >0\), \({\check \xi}_ t=\xi^+_ t+\xi^-_ t\) and h is a nonnegative function of class \(C^ 2\) on \({\mathfrak R}\) with \(h(0)=0\) such that h'' is non-increasing and \(0<k\leq h''(x)\leq K<\infty\), \(x\in {\mathfrak R}\). To solve the control problem explicitly it is reduced to simpler stochastic optimization problems such as optimal stopping, reflected follower problem and monotone follower problem with absorption.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references