An ``average distance'' inequality for large subsets of the cube (Q757399)

From MaRDI portal
scientific article
Language Label Description Also known as
English
An ``average distance'' inequality for large subsets of the cube
scientific article

    Statements

    An ``average distance'' inequality for large subsets of the cube (English)
    0 references
    0 references
    0 references
    1992
    0 references
    For a subset \(A\subset \{0,1\}^ n\) the average distance in A is defined by \(dist(A)=\frac{1}{| A|^ 2}\sum_{x\in A}\sum_{y\in A}H(x,y)\), where \(H(x,y)=| \{i| x_ i\neq y_ i\}|\) is the Hamming distance between \((x_ 1,...,x_ n)\) and \((y_ 1,...,y_ n)\). We show that \(dist(A)\geq \frac{n+1}{2}-\frac{2^{n-1}}{| A|}\) for every \(A\subset \{0,1\}^ n\). This bound is good for large subsets of the cube. It is tight for \(| A| =2^{n-1}\), where the subcubes of dimension n-1 are the only extremal configurations.
    0 references
    0 references
    average distance
    0 references
    Hamming distance
    0 references