Abstract:The paper deals with concentration inequalities for locally small increments of compound empirical processes. In the asymptotic theory of $m$-estimation such inequalities play an essential role in deriving convergence rates for solutions of the sample average approximation method to solve compound stochastic programs, and in particular for $m$-estimators. We develop inequalities dependent on the sample sizes with explicit terms instead of unspecified universal constants. They are applied to study the Sample Average Approximation method for compound stochastic programs. Nonasymptotic upper estimates for the deviation probabilities of the optimal solutions are derived which are dependent on the sample sizes. They allow to conclude immediately convergence rates for the optimal solutions. In the special case of classical risk neutral stochastic programs, we end up with upper estimates of deviation probabilities for $m$-estimators, and their convergence rates. Moreover, we may also demonstrate how to apply the results to sample average approximation of risk averse stochastic programs. In this respect we consider stochastic programs expressed in terms of absolute semideviations and Average Value at Risk. The investigations are based on concentration inequalities from the recent contribution Kratschmer(2024).

Asymptotic results for compound sums in separable Banach spaces

The Law of the Iterated Logarithm and Marcinkiewicz Law of Large Numbers Forb-Valuedu-statistics

Asymptotic analysis of random partitions ∗

Asymptotic results for sums and extremes

Precise Asymptotics in the Baum–Katz and Davis Laws of Large Numbers of P-Mixing Sequences

Precise asymptotics in the law of logarithm under dependence assumptions

LIL and the Approximation of Rectangular Sums of B-valued Random Variables when Extreme Terms Are Excluded

Precise asymptotics of partial sums for multidimensionally indexed random variables.

Higher order asymptotics for large deviations -- Part I

Asymptotics for A Class of Dependent Random Variables

Functional central limit theorem and Marcinkiewicz strong law of large numbers for Hilbert-valued U-statistics of absolutely regular data

Asymptotics for sums of random variables with local subexponential behaviour

Asymptotic smoothness, concentration properties in Banach spaces and applications

Refined distributional limit theorems for compound sums

Concentration inequalities for locally small increments of compound empirical processes with applications to solutions of compound and risk averse stochastical programming

Asymptotics for the sum-ruin probability of a bi-dimensional compound risk model with dependent numbers of claims

Total variation distance and compound poisson approximations for random sums

A note on summability in Banach spaces

Asymptotic Expansions and two-sided Bounds in Randomized Central Limit Theorems

Dependency-dependent Bounds for Sums of Dependent Random Variables