When Are HJB-Equations in Stochastic Control of Delay Systems Finite Dimensional? (Q4804874)

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 1903065
Language Label Description Also known as
default for all languages
No label defined
    English
    When Are HJB-Equations in Stochastic Control of Delay Systems Finite Dimensional?
    scientific article; zbMATH DE number 1903065

      Statements

      When Are HJB-Equations in Stochastic Control of Delay Systems Finite Dimensional? (English)
      0 references
      0 references
      0 references
      28 April 2003
      0 references
      stochastic differential delay equation
      0 references
      stochastic control problem
      0 references
      Hamilton-Jacobi-Bellman equation
      0 references
      viscosity solution
      0 references
      The authors consider a stochastic control problem, where the system is given by a one-dimen\-sional stochastic differential delay equation of the form: \[ \begin{cases} dX(t)=b\bigl(t,X(t), Y(t), Z(t),u(t)\bigr)dt+ \sigma \bigl(t, X(t), Y(t),Z(t),u(t)\bigr) dB(t), \quad t\in(s,T],\;T<\infty,\\ X(t)=\xi(t-s),\quad t\in[s-\delta,s],\;\xi\in C\bigl([-\delta,0];R\bigr),\end{cases} \] where \[ Y(t)=\int^0_{-\delta}e^{\lambda s}X(t+s)s, \quad Z(t)=Z(t-\delta), \] \(b\) and \(\sigma\) are given functions, \(\delta\geq 0\) is the constant delay, \(\lambda\) is constant, \(B\) is Brownian motion and \(u(t)\) is the control. The problem is to maximize the functional \[ J(s,\xi;u)= E^{s,\xi,u}\left [\int^T_sf\bigl(t,X(t), Y(t), u(t)\bigr)dt+ h\bigl(X(T), Y(T)\bigr)\right] \] over some class \({\mathcal U} [s,T]\) of control strategies. The value function is defined by \[ \begin{cases} V(s,\xi) =\sup_{u\in {\mathcal U}[s,T]}J(s,\xi;u),\quad & (s,\xi)\in[0,T)\times C\bigl([-\delta,0];R\bigr),\\ V(T,\xi)=h\bigl(X(0), Y(0)\bigr),\quad & \xi\in C\bigl([-\delta,0];R\bigr).\end{cases} \] The authors prove that the value function solves the Hamilton-Jacobi-Bellman partial differential equation (HJB-equation) and that under certain conditions the HJB-equation has unique viscosity solution. They also give numerical examples with the linear stochastic differential equation and indicate the difficulties with nonlinear models.
      0 references

      Identifiers

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