Abstract
We present a novel tight bound on the quantum violations of the CGLMP inequality in the case of infinitely many outcomes. Like in the case of Tsirelson's inequality, the proof of our new inequality does not require any assumptions on the dimension of the Hilbert space or kinds of operators involved. However, it is seen that the maximal violation is obtained by the conjectured best measurements and a pure, but not maximally entangled, state. We give an approximate state which, in the limit where the number of outcomes tends to infinity, goes to the optimal state for this setting. This state might be potentially relevant for experimental verifications of Bell inequalities through multi-dimensional entangled photon pairs.
Introduction
In the study of quantum nonlocality, Bell inequalities define the boundary between classical "local realistic" theories and quantum mechanics. The most famous example is the CHSH inequality, which applies to two parties (Alice and Bob) making two possible measurements with two outcomes each. Tsirelson’s inequality provides the absolute "speed limit" for how much quantum mechanics can violate that classical bound.
However, as we move toward high-dimensional quantum systems—such as photons with many possible orbital angular momentum (OAM) states—we require inequalities for more than two outcomes ($d > 2$). The CGLMP inequality (named after Collins, Gisin, Linden, Massar, and Popescu) generalizes Bell's work for $d$ outcomes. While numerical studies have long suggested what the quantum limit might be for large $d$, a rigorous, analytical "tight" bound for the limit where $d$ goes to infinity was previously missing. This research fills that gap by deriving a tight "Tsirelson-like" quantum bound for Bell inequalities involving an infinite number of measurement outcomes.
Methods and Analytical Framework
The researchers focused on a simplified version of the CGLMP inequality that deals with the probabilities of Alice's and Bob's outcomes being related (e.g., $A < B$). To find the tight quantum bound, the study adopted a specific set of projective measurements (Fourier-transform based) previously conjectured to be optimal for $d$-outcome Bell tests.
By transforming the discrete sum representing the quantum violation into a continuous integral as the number of outcomes $d \to \infty$, the authors were able to perform an analytical limit analysis. Instead of assuming a maximally entangled state, they proposed a "regularized" ansatz for the state's Schmidt coefficients. By showing this state achieves the bound of zero for their specific probability sum, they proved the inequality is "tight," meaning it represents a fundamental limit that cannot be improved further.
Key Insights: The "Not-So-Maximally" Entangled State
One of the most counter-intuitive findings in high-dimensional Bell tests is that maximally entangled states are not the most effective at violating local realism. The authors provide an explicit analytical form for the state that achieves the maximum violation in the infinite limit.
As shown in the numerical simulations (Figure 2), the quantum violation of the classical CGLMP bound grows as $d$ increases, eventually hitting the newly defined "floor" at zero. The researchers introduced a simple analytical approximate state: $$|\psi_d\rangle \sim \sum_{k=0}^{d-1} \frac{|kk\rangle}{\sqrt{(k+1)(d-k)}}$$
Numerical results demonstrate that this state is an excellent approximation of the true optimal state even for finite $d$, making it a prime candidate for experimentalists working with high-dimensional entanglement.
Entanglement Entropy Scaling
A critical discovery involves the scaling of entanglement. The study found that for the optimal state, the ratio of entanglement entropy ($E$) to the maximum possible entropy ($\log d$) converges to exactly $1/2$. This suggests that "half-maximal" entanglement is sufficient to reach the ultimate quantum limit in this setting.
Figure 3 illustrates this convergence. While the maximally entangled state would maintain a ratio of 1.0, the optimal and approximate states show a clear downward trend toward the theoretical limit. Much like the original Tsirelson inequality, this new bound is universal; it does not require assumptions about the dimension of the Hilbert space or the specific nature of the operators involved.
Conclusion and Future Work
While this work establishes a fundamental limit for the $2 \times 2 \times \infty$ setting (two parties, two measurements each), several challenges remain. Generalizing this to $N$ measurements per side ($2 \times N \times \infty$) is significantly more complex, as the optimal states in those cases are less understood.
Furthermore, while the paper suggests using high-dimensional OAM of entangled photon pairs, preparing the specific non-maximally entangled state required to reach the tight bound is technically demanding. Finally, the authors note a potential trade-off: while their inequality is geometrically "tight," it may not be the "strongest" in terms of statistical significance (Kullback-Leibler divergence) for all experimental tests, suggesting that mathematical elegance and experimental utility must be carefully balanced in future research.
Summary updated