Arak's inequalities for concentration functions and the Littlewood-Offord problem (Q2631210)

The concentration function of a \(d\)-dimensional random vector \(Y\) with distribution function \(F\) is defined as: \(Q(F,t)=\sup_{x\in \mathbb R^d}P(Y\in x+ tB)\), \(t\geq 0\), where \(B=\{x\in \mathbb R^d: \|x\| \leq 1/2 \}\) is a centered Euclidean ball of radius \(1/2\). Let \(X_1,\dots, X_n\) be independent and identically distributed \(\mathbb R\)-valued random variables. Let \(a=(a_1,a_2,\dots,a_n)\neq 0\), where \(a_k=(a_{k1},a_{k2},\dots,a_{kd}) \in \mathbb R^d\), \(k=1,2,\dots,n\). Motivated by the classical works of \textit{J. E. Littlewood} and \textit{A. C. Offord} [Mat. Sb., Nov. Ser. 12(54), 277--286 (1943; Zbl 0061.01801)], and \textit{P. Erdős} [Bull. Am. Math. Soc. 51, 898--902 (1945; Zbl 0063.01270)], the authors study the behavior of the concentration function of the weighted sums \(S_a=\sum_{k=1}^n X_k a_k\) with respect to the arithmetic structure of the coefficients, and the vector \(S_a\) has the distribution \(F_a\). They also discuss the relations between a certain inverse principle of \textit{T. Tao} and \textit{V. H. Vu} [Ann. Math. (2) 169, No. 2, 595--632 (2009; Zbl 1250.60023)] and a similar principle formulated by \textit{T. V. Arak} [Theory Probab. Appl. 26, 219--239 (1982; Zbl 0481.60024); ibid. 26, 437--451 (1982); translation from Teor. Veroyatn. Primen. 26, 449--463 (1981; Zbl 0495.60037)] for sums of arbitrarily distributed independent random variables.