Separating NOF communication complexity classes RP and NP

Matei David,Toniann Pitassi

DOI: https://doi.org/10.48550/arXiv.0802.3860

2008-02-27

Abstract:We provide a non-explicit separation of the number-on-forehead communication complexity classes RP and NP when the number of players is up to \delta log(n) for any \delta<1. Recent lower bounds on Set-Disjointness [LS08,CA08] provide an explicit separation between these classes when the number of players is only up to o(loglog(n)).

Computational Complexity

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to distinguish between the randomized communication complexity class (RP) and the non - deterministic communication complexity class (NP) in the multi - player Number - On - Forehead (NOF) communication complexity model. Specifically, the authors provide an implicit method to separate the RP and NP classes in the NOF model when the number of players is at most \( \delta \cdot \log n \) (where \( \delta < 1 \)). Previous research has achieved such separation with a smaller number of players through specific functions (such as Set - Disjointness), but the goal of this paper is to extend this result to a larger number of players.

### Main Contributions
1. **Main Theorem**:
- Theorem 1.1 shows that for all \( \delta < 1 \) and all \( k \leq \delta \cdot \log n \), there exists a function \( F \) such that \( \text{NP}_{cc}^k \not\subseteq \text{BPP}_{cc}^k \), thus \( \text{NP}_{cc}^k \neq \text{RP}_{cc}^k \).

2. **Technical Methods**:
- The authors use a generalized discrepancy method. By constructing specific functions \( F \) and \( G \) and making \( F \) and \( G \) highly correlated under a certain distribution, while \( G \) has low discrepancy, they prove that the randomized communication complexity of \( F \) is high.
- By choosing a random mask function \( \varphi \), the authors construct the function \( \text{Lift}(OR, \varphi) \) and prove that its randomized communication complexity is at least \( n^{\Omega(1)} \).

### Key Concepts
- **NOF Model**: In this model, \( k \) players try to compute a function \( F \). The input is divided into \( k \) parts, and each player can see the inputs of the other \( k - 1 \) parts.
- **Communication Complexity**: It measures how many bits of information need to be exchanged between players in order to correctly compute the function \( F \).
- **Randomized Communication Complexity**: It allows players to use randomized strategies to reduce the amount of communication.
- **Non - deterministic Communication Complexity**: It allows players to use guessing to verify the output of the function.

### Conclusion
Through the above methods, the authors successfully distinguish between the RP and NP classes in the case of more players, providing new insights and tools for the NOF communication complexity theory. This result not only extends the existing research but also lays the foundation for further exploration of multi - player communication complexity problems.