In statistical applications, it is often necessary to evaluate multidimensional integrals that cannot be treated in closed form. After describing some application contexts and mentioning some stochastic and deterministic computation methods, we delve into the analytical method of Laplace’s approximation (Gaussian approximation), showing its main ideas, the assumptions for it to be effective, and some notable results. We conclude with the description and discussion of some works that employ, in different ways, the Gaussian approximation in the evaluation of such intractable integrals.
In applicazioni statistiche è spesso necessaria la valutazione di integrali multidimensionali non trattabili in forma chiusa. Dopo aver descritto alcuni contesti di applicazione e accennato ad alcuni metodi di calcolo stocastico e deterministico si approfondisce il metodo analitico dell'approssimazione di Laplace (approssimazione gaussiana) mostrandone le idee principali, le assunzioni affinché sia efficace e alcuni risultati notevoli. Si conclude con la descrizione e la discussione di alcuni lavori che impiegano, secondo modalità diverse, l'approssimazione gaussiana nella valutazione di tali integrali intrattabili.
Metodi basati su approssimazioni gaussiane per integrali multidimensionali in applicazioni statistiche
STANGHERLIN, PIETRO
2022/2023
Abstract
In statistical applications, it is often necessary to evaluate multidimensional integrals that cannot be treated in closed form. After describing some application contexts and mentioning some stochastic and deterministic computation methods, we delve into the analytical method of Laplace’s approximation (Gaussian approximation), showing its main ideas, the assumptions for it to be effective, and some notable results. We conclude with the description and discussion of some works that employ, in different ways, the Gaussian approximation in the evaluation of such intractable integrals.File | Dimensione | Formato | |
---|---|---|---|
Stangherlin_Pietro.pdf
accesso riservato
Dimensione
420.36 kB
Formato
Adobe PDF
|
420.36 kB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/58690