[Paper Review] Bell Violations through Independent Bases Games

Oded Regev|arXiv (Cornell University)|Jan 3, 2011

Quantum Mechanics and Applications4 citations

TL;DR

This paper presents two novel two-player games that exhibit large Bell violations—demonstrating a significant quantum advantage over classical and maximally entangled strategies—using elementary probabilistic and quantum information techniques. The authors achieve improved quantitative bounds compared to prior work, with a Bell violation of order √n/log n, and provide a simpler, more accessible proof than previous operator space-theoretic methods.

ABSTRACT

In a recent paper, Junge and Palazuelos presented two two-player games exhibiting interesting properties. In their first game, entangled players can perform notably better than classical players. The quantitative gap between the two cases is remarkably large, especially as a function of the number of inputs to the players. In their second game, entangled players can perform notably better than players that are restricted to using a maximally entangled state (of arbitrary dimension). This was the first game exhibiting such a behavior. The analysis of both games is heavily based on non-trivial results from Banach space theory and operator space theory. Here we present two games exhibiting a similar behavior, but with proofs that are arguably simpler, using elementary probabilistic techniques and standard quantum information arguments. Our games also give better quantitative bounds.

Motivation & Objective

• To construct two-player games that exhibit large Bell violations, showing a significant quantum advantage over classical and restricted entangled strategies.
• To provide a simpler, more accessible proof of Bell violations compared to prior work relying on advanced functional analysis.
• To improve quantitative bounds on Bell violations, achieving a violation of order √n/log n with n inputs per player.
• To demonstrate that elementary quantum information techniques and probabilistic methods can replace complex operator space theory in analyzing Bell inequalities.

Proposed method

• The games are constructed using independent bases and Fourier analysis on matrix-valued functions, with winning probability expressed in terms of Fourier coefficients of measurement operators.
• The analysis uses the von Neumann entropy and mutual information to bound the sum of squared Fourier coefficients of the measurement operators.
• A quantum state is defined for each input string, with density matrices scaled by n/D to ensure trace 1 and bounded eigenvalues.
• Conditional entropy and strong subadditivity of von Neumann entropy are used to relate mutual information across qubits to the sum of squared deviations of measurement operators.
• The binary entropy function is used to lower-bound the quantum mutual information, which is then related to the trace of squared deviation operators.
• Cauchy-Schwarz inequality is applied to bound the bias in terms of the L2 norms of the Fourier coefficients, leading to the final bound.

Experimental results

Research questions

• RQ1 Can Bell violations of order √n/log n be achieved using simpler, more intuitive techniques than Banach space theory?
• RQ2 Can the analysis of the Junge-Palazuelos game be simplified using quantum information tools like entropy and Fourier analysis?
• RQ3 What is the tightest possible bound on the bias of entangled strategies in games with n inputs and outputs, using elementary methods?
• RQ4 Can games be constructed that show a quantum advantage over both classical and maximally entangled strategies, with explicit and constructive proofs?
• RQ5 How does the use of von Neumann entropy and mutual information help in bounding the Fourier coefficients of quantum measurement operators?

Key findings

• The paper achieves a Bell violation of order √n/log n, matching the best-known upper bounds up to logarithmic factors.
• The winning probability bias is bounded by O(√(log n)/n), which is tighter than previous results.
• The analysis replaces complex operator space theory with elementary quantum information and probabilistic techniques, making the proof more accessible.
• The mutual information between input qubits and the quantum system is shown to be at most log n, which directly leads to the desired bound on the Fourier coefficients.
• The bound ∑i Tr(σ̂i²) ≤ 2 ln 2 · log n is derived using entropy inequalities and the binary entropy function.
• The result confirms that quantum strategies can outperform both classical and maximally entangled strategies in games with n inputs, with a violation that is nearly optimal in terms of input and output size.