An O(log2N) Fully-Balanced Resampling Algorithm for Particle Filters on Distributed Memory Architectures

Resampling is a well-known statistical algorithm that is commonly applied in the context of Particle Filters (PFs) in order to perform state estimation for non-linear non-Gaussian dynamic models. As the models become more complex and accurate, the run-time of PF applications becomes increasingly slo...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Algorithms Ročník 14; číslo 12; s. 342
Hlavní autoři: Varsi, Alessandro, Maskell, Simon, Spirakis, Paul G.
Médium: Journal Article
Jazyk:angličtina
Vydáno: Basel MDPI AG 01.12.2021
Témata:
Accuracy
Algorithms
Complexity
Computer architecture
Distributed Memory
Dynamic models
high performance computing
Hypotheses
message passing interface
parallel computing
Particle Filters
Resampling
Run time (computers)
State estimation
Supercomputers
Workloads
ISSN:1999-4893, 1999-4893
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Resampling is a well-known statistical algorithm that is commonly applied in the context of Particle Filters (PFs) in order to perform state estimation for non-linear non-Gaussian dynamic models. As the models become more complex and accurate, the run-time of PF applications becomes increasingly slow. Parallel computing can help to address this. However, resampling (and, hence, PFs as well) necessarily involves a bottleneck, the redistribution step, which is notoriously challenging to parallelize if using textbook parallel computing techniques. A state-of-the-art redistribution takes O((log2N)2) computations on Distributed Memory (DM) architectures, which most supercomputers adopt, whereas redistribution can be performed in O(log2N) on Shared Memory (SM) architectures, such as GPU or mainstream CPUs. In this paper, we propose a novel parallel redistribution for DM that achieves an O(log2N) time complexity. We also present empirical results that indicate that our novel approach outperforms the O((log2N)2) approach.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1999-4893
1999-4893
DOI:10.3390/a14120342