Communication complexity of some number theoretic functions (Q2470553)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Communication complexity of some number theoretic functions
scientific article

    Statements

    Communication complexity of some number theoretic functions (English)
    0 references
    14 February 2008
    0 references
    The communication complexity \(C(b)\)\, of a Boolean function \(b(x,y)\),\, where \(x,y\)\, are \(n\)-bits integers, is the smallest possible number of bits exchanged by a two-party communication protocol which allows at least one party to compute the value \(b(x,y)\). In the paper the author obtains lower bounds for the communication complexity of three Boolean functions ``which naturally arise from some decision problems having a clear number theoretic flavour''. Explicitly: the parity \(f(x,y)\)\, of the sum of the binary digits of \(x+y\),\, the parity \(g(x,y)\)\, of the total number of prime divisors of \(x+y\)\, (counted with their multiplicity) and the function \(h(x,y)\)\, defined as 1 if the largest prime divisor of \(x+y\)\, is greater or equal than the threshold \(2^{e^{-1/2}n}\)\, and 0 otherwise. After listing in the Section 2 some auxiliary results the wanted bounds are provided in Section 3 (theorems 1-3): \begin{itemize} \item(1) \(C(f)\geq \beta n+O(1)\),\,\, where \(\beta=\frac{3}{2}(1-\frac{1}{2}\log_2\frac{\sqrt 2}{\sin(\pi/8)})=0.085667\dots\) \item (2) \(C(g)\geq \frac{3}{2}\log_2(n)+O(1)\) \item (3) \(C(h)\geq \frac{3}{2}\log_2(n)+O(1)\) Finally the author shortly discusses (Section 4) some other number theoretic functions for which similar bounds can be obtained.\end{itemize}
    0 references
    0 references
    0 references
    0 references
    0 references
    communication complexity
    0 references
    discrepancy
    0 references
    number theoretic functions
    0 references
    0 references