Hamilton-Jacobi equations for optimal control on multidimensional junctions with entry costs (Q2173296)

The geometry where the problem lives is: there are $n$ half planes $\mathcal P_j$ in $\mathbb R^3$ that intersect at a line $\Gamma, $ boundary of each half-plane. The state or junction is the union of the $\mathcal P_j$. On each $\mathcal P_j$, the dynamics are determinist (given by a differential system) and the optimal control problem is an infinite horizon problem. The dynamics, the set of controls and the running cost $\ell_j(\cdot, \cdot)$ (the integrand of the cost functional) are different for each $\mathcal P_j$. Finally, the trajectory pays an entry cost $c_j(\cdot)$ each time it enters the half-plane $\mathcal P_j$ from $\Gamma$. Under the usual controllability assumptions and regularity conditions, the authors derive the Hamilton-Jacobi equation for this problem and show that there is a unique viscosity solution of the equation (among other difficulties, the entry cost causes discontinuities of the value function). They also show a comparison principle for sub- and supersolutions of the Hamilton-Jacobi equation. There is a version of the results for a weaker notion of controllability.