EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Lévy Langevin Monte Carlo for sampling from heavy-tailed target distributions

Anita Behme, Claudius Lütke Schwienhorst

Extremes · 13.07.2026

DOI: 10.1007/s10687-026-00544-9 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Extremes, online erschienen am 13.07.2026. Verfasst von Anita Behme und 1 weiteren Autor:innen. Die Arbeit nennt 27 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Center for Scalable Data Analytics and Artificial Intelligence Dresden/Leipzig.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

We extend the Lévy Langevin Monte Carlo method studied by Oechsler (2024): Choosing a heavy-tailed target distribution we prove convergence of a solution of a stochastic differential equation to this target. Hereby, the stochastic differential equation is driven by a compound Poisson process - unlike in the case of a classical Langevin diffusion. The method allows one to sample from non-smooth targets and distributions with separated modes with exponential convergence to the invariant distribution, which in general cannot be guaranteed by the classical Langevin diffusion in presence of heavy tails. The method is promising due to the possibility of a simple implementation because of the compound Poisson noise term.

Förderung

  • Center for Scalable Data Analytics and Artificial Intelligence Dresden/Leipzig

Zitieren

APA

Behme, A., & Schwienhorst, C. L. (2026). Lévy Langevin Monte Carlo for sampling from heavy-tailed target distributions. Extremes. https://doi.org/10.1007/s10687-026-00544-9

Vancouver

Behme A, Schwienhorst CL. Lévy Langevin Monte Carlo for sampling from heavy-tailed target distributions. Extremes. 2026. doi:10.1007/s10687-026-00544-9

BibTeX
@article{Behme2026,
  author = {Behme, Anita and Schwienhorst, Claudius Lütke},
  title = {Lévy Langevin Monte Carlo for sampling from heavy-tailed target distributions},
  journal = {Extremes},
  year = {2026},
  doi = {10.1007/s10687-026-00544-9}
}
Kostenlos starten →