TOPICS
Search

Erdős-Littlewood-Offord Inequality


The Erdős-Littlewood-Offord inequality states that if v_1, ..., v_n are real numbers satisfying |v_i|>=1, then at most the binomial coefficient (n; |_n/2_|) of the 2^n signed sums sum_(i=1)^(n)epsilon_iv_i, where each epsilon_i in {-1,1}, lie in any open interval of length 2 (Erdős 1945).

Equivalently, if epsilon_1, ..., epsilon_n are independent and identically distributed random signs, each taking the values -1 and 1 with equal probability, then for every open interval I of length 2,

 P(sum_(i=1)^nepsilon_iv_i in I)<=1/(2^n)(n; |_n/2_|).

The inequality is sharp, as can be seen by taking v_1=...=v_n=1 and choosing an open interval containing a mode of the signed sum.


See also

Anti-Concentration Inequality

Explore with Wolfram|Alpha

References

Erdős, P. "On a Lemma of Littlewood and Offord." Bull. Amer. Math. Soc. 51, 898-902, 1945. https://doi.org/10.1090/S0002-9904-1945-08454-7.

Cite this as:

Weisstein, Eric W. "Erdős-Littlewood-Offord Inequality." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/Erdos-Littlewood-OffordInequality.html

Subject classifications