Optimal Rates of Convergence for Deconvolving a Density
Abstract
Suppose we observe the sum of two independent random variables X and Z, where Z denotes measurement error and has a known distribution, and where the unknown density f of X is to be estimated. It is shown that if Z is normally distributed and if f has k bounded derivatives, then the fastest attainable convergence rate of any nonparametric estimator of f is only (log n)-k/1. Therefore deconvolution with normal errors may not be a practical proposition. Other error distributions are also treated. Stefanski-Carroll (1978b) estimators achieve the optimal rates. Our results have versions for multiplicative errors, where they imply that even optimal rates are exceptionally slow.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1988
- Accession Number
- ADA197748
Entities
People
- Peter Hall
- Raymond J. Carroll
Organizations
- Texas A&M University