Most quantum states are too entangled to be useful as computational resources

Quantum entanglement has long been regarded as the cornerstone enabling quantum computational advantage, exemplified by models like measurement-based quantum computing (MBQC) and cluster states. Early works demonstrated that certain highly entangled states support universal computation, fueling the belief that “more entanglement” correlates with greater power. Recent studies explored random states and critical systems, revealing their high entanglement levels. However, the precise relationship between entanglement quantity and computational utility remained unclear. This paper builds on the geometric measure of entanglement, integrating measure concentration phenomena, to analyze the distribution of entanglement in large quantum systems. It aims to clarify whether the abundance of entanglement in generic states translates into computational advantage or if there exists an optimal “dose” beyond which states become ineffective.

The core issue addressed is whether the prevalent high entanglement in generic quantum states actually confers computational benefits. Prior assumptions suggested that “more entanglement” implies “more power,” but this lacked rigorous quantification. The challenge lies in identifying the threshold at which entanglement ceases to be useful, especially in the context of universal quantum computation. The problem is compounded by the difficulty of characterizing the typical entanglement distribution in high-dimensional state spaces and understanding its implications for classical simulability. The authors seek to rigorously establish whether most states are “useful” or “over-entangled,” and how this affects resource theory and practical quantum computing.

The key innovations include: 1) defining and employing the geometric measure of entanglement (Eg) as a quantitative resource indicator; 2) applying measure concentration bounds to analyze the typical entanglement distribution of Haar-random states; 3) constructing ε-nets over product states to statistically bound overlaps and entanglement levels; 4) linking high Eg values to classical simulability within NP complexity, thus proving most states are non-universal; 5) extending the analysis to physically relevant states like MERA, showing they too are “over-entangled.” These advances provide a rigorous, probabilistic foundation for understanding the scarcity of useful quantum states.

• �� Define geometric entanglement Eg as the negative log of maximum squared overlap with product states. • Use measure concentration inequalities (e.g., Levy’s lemma) to bound the probability that a Haar-random state has Eg below a certain threshold. • Construct ε-nets covering the space of product states, with size polynomial in system size, to discretize the problem. • Calculate the overlap distribution of random states with the net points, deriving exponential bounds on the probability of low overlap. • Relate high Eg to classical simulation complexity via NP problem reductions, showing that states with Eg close to n are classically simulable. • Extend the analysis to scale-invariant MERA states, demonstrating their high entanglement and limited utility.

The work relies on rigorous mathematical proofs, employing measure concentration bounds, ε-net constructions, and Haar measure sampling to analyze the distribution of entanglement. No physical experiments are conducted; instead, the authors derive probabilistic bounds and asymptotic behaviors to demonstrate that the overwhelming majority of states are “over-entangled.” The analysis includes explicit calculations of the probability that a Haar-random state has Eg below a certain threshold, confirming the universality of the “over-entangled” phenomenon across high-dimensional state spaces.

The main quantitative result is that the fraction of states with Eg less than n - 2 log2(n) - 3 is less than e^(-n^2), meaning nearly all states are excessively entangled. Consequently, the probability that a random state can serve as a universal resource diminishes exponentially with system size. The analysis confirms that high entanglement correlates with classical simulability, invalidating the assumption that “more entanglement” always yields “more power.” Even physically relevant states like MERA are shown to fall into this “over-entangled” category, indicating a fundamental limit on the usefulness of generic high-entanglement states.

In the short term, these results guide the selection of quantum states for computation, emphasizing the importance of moderate entanglement levels. In the long term, they influence the design of quantum hardware and algorithms, encouraging the engineering of states with “just right” entanglement to optimize computational advantage, and avoiding resource wastage on “over-entangled” states that cannot outperform classical algorithms.

The analysis primarily depends on the geometric measure of entanglement and asymptotic bounds, which may not fully capture finite-size effects or other entanglement measures relevant in physical systems. Practical state preparation imperfections, noise, and decoherence are not explicitly modeled. Further research is needed to verify these theoretical predictions experimentally and to explore the utility of other entanglement metrics.