The universal relation is the communication problem in which Alice and Bob get as inputs two distinct strings, and they are required to find a coordinate on which the strings differ. The study of this problem is motivated by its connection to Karchmer-Wigderson relations, which are communication problems that are tightly related to circuit-depth lower bounds.

In this paper, we prove a direct sum theorem for the universal relation, namely, we prove that solving $m$ independent instances of the universal relation is $m$ times harder than solving a single instance. More specifically, it is known that the deterministic communication complexity of the universal relation is at least $n$. We prove that the deterministic communication complexity of solving $m$ independent instances of the universal relation is at least $m \cdot (n-O(\log m))$.

Added reference to the new simple proof of Alexander Kozachinsky.

The universal relation is the communication problem in which Alice and Bob get as inputs two distinct strings, and they are required to find a coordinate on which the strings differ. The study of this problem is motivated by its connection to Karchmer-Wigderson relations, which are communication problems that are tightly related to circuit-depth lower bounds.

In this paper, we prove a direct sum theorem for the universal relation, namely, we prove that solving $m$ independent instances of the universal relation is $m$ times harder than solving a single instance. More specifically, it is known that the deterministic communication complexity of the universal relation is at least $n$. We prove that the deterministic communication complexity of solving $m$ independent instances of the universal relation is at least $m \cdot (n-O(\log m))$.

Added a few remarks and clarifications.

The universal relation is the communication problem in which Alice and Bob get as inputs two distinct strings, and they are required to find a coordinate on which the strings differ. The study of this problem is motivated by its connection to Karchmer-Wigderson relations, which are communication problems that are tightly related to circuit-depth lower bounds.

In this paper, we prove a direct sum theorem for the universal relation, namely, we prove that solving $m$ independent instances of the universal relation is $m$ times harder than solving a single instance. More specifically, it is known that the deterministic communication complexity of the universal relation is at least $n$. We prove that the deterministic communication complexity of solving $m$ independent instances of the universal relation is at least $m \cdot (n-O(\log m))$.

Comment #1 Authors: Alexander Kozachinsky

Accepted on: 5th September 2017 12:33

Downloads: 274

Keywords:

In \cite{meir2017the} Meir proved that deterministic communication complexity of the $m$-fold direct sum of the universal relation is at least $m(n - O(\log m))$. In this comment we present a log-rank argument which improves Meir's lower bound to $m(n - 1) - 1$.