RUS  ENG
Full version
JOURNALS // Vestnik Udmurtskogo Universiteta. Matematika. Mekhanika. Komp'yuternye Nauki // Archive

Vestn. Udmurtsk. Univ. Mat. Mekh. Komp. Nauki, 2026 Volume 36, Issue 1, Pages 23–41 (Mi vuu948)

MATHEMATICS

Analitycal methods for constructing the domain of differentiability of the minimax solution in a class of boundary value problems for a Hamilton type equation

P. D. Lebedev, A. A. Uspenskii

Institute of Mathematics and Mechanics, Ural Branch of the Russian Academy of Sciences, ul. S. Kovalevskoi, 16, Yekaterinburg, 620077, Russia

Abstract: The differential properties of the minimax solution are investigated in a class of plane Dirichlet problems for the Bellman equation. The class of problems is defined by closed non-convex solid boundary sets whose boundaries contain pseudovertices, which are singular points associated with the singularity of the minimax solution. The differential properties of the solution depend on the order of smoothness of the boundary of the boundary set at the pseudovertices and on the cardinality of the values of the metric projection operator onto this set. The paper distinguishes between situations where the operator has single-point values and when the number of projections is greater than one. Using tools from the theory of alpha sets and Efimov–Stechkin support balls, the features of the characteristic function of a non-convex set are investigated. Formulas for its limit values are found, which in a fairly general case facilitate the construction of a Chebyshev layer of the boundary set, which is a region adjacent to the boundary set in which the minimax solution is differentiable. An example and its meaningful interpretation from the point of view of optimal control are given.

Keywords: Bellman equation, minimax solution, alpha set, supporting ball, metric projection, bisector of a set, support ball, Chebyshev layer, control

UDC: 517.977, 514.172.2

MSC: 52A10, 49Q15

Received: 13.01.2026
Accepted: 02.03.2026

DOI: 10.35634/vm260102



© Steklov Math. Inst. of RAS, 2026