Extended Hamilton--Jacobi Characterization of Value Functions in Optimal Control
In the first place the author introduces yet another type of ``generalized viscosity solution [e.g., \textit{M. Bardi} and \textit{I. Capuzzo-Dolcetta}, ``Optimal control and viscosity solutions of Hamilton-Jacobi-Bellman equations (1997; Zbl 0890.49011)] using the ``outer limit \[ \partial f(x):=\{p\in R^n; \;\exists \;x_m \to x, \;p_m\in \widehat \partial f(x_m): \;p_m\to p\} \] of the (Fréchet/contingent) subdifferential: \[ \widehat \partial f(x)=D^-f(x):= \Biggl\{p\in R^n; \liminf_{y\to x} {{f(y)-f(x)-\langle p,y-x\rangle}\over {\|y-x\|}}\geq 0 \Biggr\}. \] According to author's Def.1.2, a real function \(u(.,.)\) is said to be a ``solution (actually a ``strict viscosity supersolution) of the Hamilton-Jacobi (HJ) equation: \[ u_t(t,x)+H(t,x,u_x(t,x))=0 \;\forall \;(t,x)\in (0,\infty)\times R^n, \;u(0,x)\equiv \varphi (x) \;(x \in R^n) \] if it is proper, l.s.c., \(u(0,\xi) \equiv \varphi (\xi)\) and at each point \((t,x)\in \text{dom} (u(.,.))\subset [0,\infty)\times R^n\) satisfies the differential relations: \[ p_0 +H(t,x,p)=0 \;\forall \;(p_0,p)\in \partial u(t,x), \;t>0; \;p_0 + H(0,x, p)\leq 0 \;\forall \;(p_0,p)\in \partial u(0,x), \;x\in R^n. \] The main result of the paper is Theorem 2.2 stating that under certain fairly mild hypotheses on the data \(\varphi (.)\) and \( H(.,.,.)\), if \(L(.,.,.)\) and \(\Gamma(.)\) are the associated ``Lagrangean and, respectively, the ``cost functional: \[ L(t,x,v)= \sup_{p\in R^n}[\langle p,v\rangle- H(t,x,p)], \;\Gamma(x(.)):= \varphi (x(0))+\int_0^\tau L(t,x(t),x'(t)) dt, \] then the value function: \[ V(\tau,\xi):= \begin{cases} \inf\{\Gamma(x(.)); \;x(.)\in AC, \;x(\tau)=\xi \} \quad \text{if}\quad \tau >0 \\ \varphi (\xi) \quad \text{if}\quad \tau =0, \\ +\infty \quad \text{if}\quad \tau <0 \end{cases}. \] is the unique solution (in the sense above) of the (HJ) equation in the class of proper l.s.c. functions \(u(.,.)\) that have the following ULLG (uniform lower linear growth) property: for each \(T>0\) there exists \(k>0\) such that: \[ u(t,x)\geq -k(1+\|x\|) \;\forall \;(t,x)\in [-T,T]\times R^n. \] The proof of the main result is rather complicated and it is based on some 14 auxiliary results concerning a large number of topics such as: viability, invariance, monotonicity, existence of optimal trajectories, necessary optimality conditions, reachable sets, etc.
- On representation formulas for Hamilton Jacobi's equations related to calculus of variations problems
- A generalized Hamilton-Jacobi-Bellman equation for deterministic optimal control problems
- On viscosity solutions of the Hamilton-Jacobi equation
- scientific article; zbMATH DE number 1907948
- Lower Semicontinuous Solutions of Hamilton–Jacobi–Bellman Equations
- On representation formulas for Hamilton Jacobi's equations related to calculus of variations problems
- Viscosity solutions of HJB equations with unbounded data and characteristic points
- Viscosity solutions to an initial value problem for a Hamilton-Jacobi equation with a degenerate Hamiltonian occurring in the dynamics of peakons
- Existence of solutions to a one-dimensional Hamilton-Jacobi equation with a degenerate Hamiltonian
- Value functions for certain class of Hamilton Jacobi equations
- A generalized Hamilton-Jacobi-Bellman equation for deterministic optimal control problems
- Value functions for Bolza problems with discontinuous Lagrangians and Hamilton-Jacobi inequalities
- Duality and uniqueness of convex solutions to stationary Hamilton-Jacobi equations
- Lower Semicontinuous Solutions of Hamilton–Jacobi–Bellman Equations
- Uniqueness results for a class of hamilton-jacobi equations with singular coefficients
- Existence of viscosity solutions with the optimal regularity of a two-peakon Hamilton-Jacobi equation
- Reduction of lower semicontinuous solutions of Hamilton-Jacobi-Bellman equations
- An error estimate for symplectic Euler approximation of optimal control problems
- Time-Dependent Measurable Hamilton–Jacobi Equations
- The value function representing Hamilton-Jacobi equation with Hamiltonian depending on value of solution
- Optimal trajectories and the neoclassical calculus of variations
- Value function for nonautonomous problems in the calculus of variations
- On nonuniqueness of solutions of Hamilton-Jacobi-Bellman equations
This page was built for publication: Extended Hamilton--Jacobi Characterization of Value Functions in Optimal Control
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4507456)
