Random-horizon Markov decision processes with fuzzy rewards on Borel spaces
Open Mathematics · 23.01.2026 · 24 (1)
DOI: 10.1515/math-2025-0267 ↗Kurz gefasst
Fachartikel in der Zeitschrift Open Mathematics, online erschienen am 16.07.2026. Verfasst von Hugo Cruz-Suárez und 2 weiteren Autor:innen. Die Arbeit nennt 26 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
We study a class of Markov decision processes on Borel state spaces motivated by situations where both the horizon and the notion of performance are uncertain. The process may terminate at a random time, generating a survival profile that effectively plays the role of a time-dependent discount factor. To capture imprecise performance specifications, rewards are modeled as trapezoidal fuzzy numbers, and policies are compared through weighted-vertex ranking functionals. Our results show that, under Lyapunov-type drift conditions and standard continuity/compactness assumptions, the ranked control problem admits Bellman optimality equations and optimal Markov policies. Technically, we reduce the original model to an equivalent discounted problem on an augmented state space, which leads to a contraction property, a fixed-point representation of the value function, and a convergent value-iteration scheme. A continuous-state neurostimulation example illustrates the implications in a concrete setting and outlines a practical discretization-based implementation.
Zitieren
APA
Cruz-Suárez, H., Ortega-Gutiérrez, R. I., & Camilo-Garay, C. (2026). Random-horizon Markov decision processes with fuzzy rewards on Borel spaces. Open Mathematics, 24(1). https://doi.org/10.1515/math-2025-0267
Vancouver
Cruz-Suárez H, Ortega-Gutiérrez RI, Camilo-Garay C. Random-horizon Markov decision processes with fuzzy rewards on Borel spaces. Open Mathematics. 2026;24(1). doi:10.1515/math-2025-0267
BibTeX
@article{CruzSurez2026,
author = {Cruz-Suárez, Hugo and Ortega-Gutiérrez, R. Israel and Camilo-Garay, Carlos},
title = {Random-horizon Markov decision processes with fuzzy rewards on Borel spaces},
journal = {Open Mathematics},
year = {2026},
volume = {24},
number = {1},
doi = {10.1515/math-2025-0267}
}
EUVIMED