Mathematics > Statistics Theory
[Submitted on 27 Mar 2018 (v1), last revised 8 Feb 2019 (this version, v2)]
Title:Adaptive nonparametric estimation for compound Poisson processes robust to the discrete-observation scheme
View PDFAbstract:A compound Poisson process whose jump measure and intensity are unknown is observed at finitely many equispaced times. We construct a purely data-driven estimator of the Lévy density $\nu$ through the spectral approach using general Calderon--Zygmund integral operators, which include convolution and projection kernels. Assuming minimal tail assumptions, it is shown to estimate $\nu$ at the minimax rate of estimation over Besov balls under the losses $L^p(\mathbb{R})$, $p\in[1,\infty]$, and robustly to the observation regime (high- and low-frequency). To achieve adaptation in a minimax sense, we use Lepskiĭ's method as it is particularly well-suited for our generality. Thus, novel exponential-concentration inequalities are proved including one for the uniform fluctuations of the empirical characteristic function. These are of independent interest, as are the proof-strategies employed to deal with general Calderon--Zygmund operators, to depart from the ubiquitous quadratic structure and to show robustness without polynomial-tail conditions. Part of the motivation for such generality is a new insight we include here too that, furthermore, allows us to unify the main two approaches to construct estimators used in related literature.
Submission history
From: Alberto J. Coca [view email][v1] Tue, 27 Mar 2018 02:27:13 UTC (105 KB)
[v2] Fri, 8 Feb 2019 19:36:30 UTC (95 KB)
Current browse context:
math
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.