Institutional Repository of Vilnius University
Multidimensional Compound Poisson and signed Compound Poisson approximations /
Abstract
dc:descriptionCompound Poisson (CP) approximations have a broad range of theoretical and practical applications. However, the absolute majority of research was done for one-dimensional cases. The multidimensional CP approximations are comparatively little researched. In this Thesis we estimate the accuracy of multidimensional CP and signed compound Poisson (SCP) approximations to the sum of random vectors. The choice of approximations (accompanying laws, Hipp-type approximation, Skellam distribution) was determined by the needs of Actuarial Mathematics and theoretical problem of the best infinitely divisible approximation for the sums of independent random vectors (the first uniform Kolmogorov theorem). The thesis presents estimates for (shifted) accompanying CP approximations and Hipp-type SCP approximations to the distribution of the sum of random vectors, extending one-dimensional results to the multivariate case, which is comparatively little researched. Key contributions include improved estimates in local, Kolmogorov, total variation, and l_r (r>1) norms. In addition, non-uniform, local, marginal non-uniform local estimates are derived for the first time. Estimates in total variation are derived under weaker assumptions. These advances fill critical gaps in actuarial models of multivariate aggregate claims distributions and contribute toward solving the multidimensional uniform Kolmogorov problem for the best infinitely divisible approximation of sums of independent random vectors. All results are new. Some of them are new even in one-dimensional case.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Jokubauskienė, Simona,
Subjects
dc:subject × 1Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- eng
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD329822146&prefLang=en_US