⚛️ quantum physics

Power and Limits of Collective Local Measurements in Multicopy State Discrimination

This paper resolves the long-standing question of whether perfect local discrimination of orthogonal quantum states can require more than two copies by demonstrating that while collective local measurements can drastically reduce the copy complexity for maximal-stabilizer eigenbases in odd-prime dimensions to at most three copies, they fail to eliminate multicopy hardness entirely, as there exist explicit bases where the required sample size remains unbounded even under collective processing.

Original authors: Mao-Sheng Li, Yan-Ling Wang, Zhu-Jun Zheng

Published 2026-09-09

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Mao-Sheng Li, Yan-Ling Wang, Zhu-Jun Zheng

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the quantum world, information is often hidden in the delicate arrangement of particles. To read this information, scientists must measure the particles, but the rules of the game change depending on who is doing the measuring and how they are allowed to interact. Imagine a group of scientists, each in their own isolated laboratory, holding a piece of a larger quantum puzzle. They want to figure out exactly which state their piece is in, but they cannot send their particles to a central hub to be measured together; they are restricted to acting only on what is in their own hands. This limitation, known as local measurement, creates a strange gap: states that are perfectly distinct when viewed as a whole can become impossible to tell apart when viewed from the outside. For decades, researchers have wondered if simply having more copies of the same quantum state could bridge this gap. If a scientist has two, three, or more identical copies of a mysterious state, does that extra material give them enough power to solve the puzzle, even without talking to their neighbors?

A team of researchers has now answered this question with surprising precision, revealing that the answer depends entirely on how those extra copies are handled. They discovered that simply having more copies is not enough; the key lies in whether the scientist can process all their copies at the same time, treating them as a single, unified system. In their study, they examined two different ways of using multiple copies. In the first method, the scientist measures each copy one by one, or in separate groups, never letting the copies interact with each other. In the second method, the scientist is allowed to bring all their copies together into one large quantum system and measure them all at once. This second approach, known as collective processing, turns out to be a powerful new resource that changes the rules of the game entirely.

The researchers focused on a specific type of quantum state called a stabilizer state, which are highly structured and useful for error correction in quantum computers. They found that for these states, the difference between the two methods is dramatic. If the scientists are forced to measure their copies individually, the number of copies they need to perfectly identify the state grows without bound as the system gets larger. In some cases, no matter how many copies they gather, they might never be able to tell the states apart if they cannot process them together. However, when the scientists are allowed to process their copies collectively, the situation flips. The researchers proved that for these same states, a very small, fixed number of copies—specifically two or three, depending on the size of the system—is always enough to perfectly identify the state, regardless of how large the system is. This means that the ability to keep quantum coherence across multiple copies and measure them jointly is a distinct and powerful tool, separate from simply having more samples or being able to talk to other labs.

But this advantage is not universal. The team also constructed a different set of quantum states where even the most powerful collective processing fails to provide a simple solution. For these specific states, the number of copies required to identify them perfectly still grows as the system gets larger, even when the scientist is allowed to process all copies together. In this case, the number of copies needed grows with the square root of the number of particles in the system. This finding is crucial because it shows that the power of collective measurement has limits. It is not a magic wand that solves every discrimination problem; rather, its effectiveness depends on the specific structure of the information being hidden. Some quantum codes can be unlocked with just a few copies if measured together, while others remain stubbornly difficult, requiring a vast amount of data even with the best local tools.

The implications of this work reach beyond just identifying states; they redefine what resources are available in distributed quantum networks. The study demonstrates that the ability to store quantum information coherently and process it locally is a resource just as important as the number of copies available or the ability to share entanglement between distant labs. In a future where quantum information is processed across a network of separated nodes, this research suggests that the local memory and processing power of each node will determine what information can be accessed. If a node can only measure one copy at a time, it may be blind to vast amounts of data that would be visible if it could hold and process a few copies simultaneously. Conversely, even with this advanced capability, some information will remain hidden unless the underlying structure of the data allows for it.

The researchers arrived at these conclusions by rigorously proving mathematical bounds on how well different measurement strategies can perform. They did not rely on simulations or guesses; they provided exact proofs showing that for one family of states, the difficulty of individual measurement grows infinitely, while collective measurement stays constant. For another family, they proved that collective measurement still faces an unbounded challenge, with a specific lower limit on the number of copies needed. These results settle a long-standing question about whether a fixed number of copies could ever suffice for all quantum states. The answer is a definitive no. The power of local measurements is not just a matter of quantity; it is a matter of quality and structure. The ability to process multiple inputs together is a distinct resource that can unlock information in some cases but cannot overcome the fundamental hardness of others. This nuanced understanding helps scientists design better quantum networks, knowing exactly where local processing power will pay off and where it will hit a wall.

Technical Summary: Power and Limits of Collective Local Measurements in Multicopy State Discrimination

Problem Statement
This work addresses a fundamental question in quantum information theory regarding the resources required for the perfect local discrimination of orthogonal quantum states. Specifically, it investigates whether the ability to process multiple copies of an unknown state collectively within a single laboratory (as opposed to processing them individually) fundamentally alters the scaling of the required number of copies (copy complexity).

The problem builds upon the 1999 inquiry by Bennett et al. [1] concerning whether perfect local discrimination of orthogonal states can require more than two copies. While this was resolved for adaptive protocols processing copies separately [50], the setting where laboratories are permitted to perform joint quantum operations on their local copies (collective local measurements) remained open. The authors seek to determine if collective access across repeated local inputs constitutes a distinct resource that can reduce copy complexity from unbounded to constant, or if multicopy hardness persists even under collective processing.

Methodology
The authors analyze four distinct measurement models defined by the combination of spatial constraints (Local Operations and Classical Communication, LOCC; or Separable measurements, SEP) and copy architecture (Individual-copy vs. Collective-copy):

Individual-copy ($ind$): Laboratories perform operations on single copies, potentially with classical feed-forward between copies.
Collective-copy ($col$): Laboratories act jointly on the tensor product of all $r$ local copies held by that laboratory.

The core analytical tool is the product overlap bound. The authors define:

$\gamma(B)$ : The maximum overlap between a state in basis $B$ and any normalized product state (for individual copies).
$\Gamma_r(B)$ : The maximum overlap between the $r$ -fold tensor product of a state in $B$ and any normalized product state across the grouped laboratory partitions (for collective copies).

Using Proposition 1, they establish that the success probability of perfect discrimination is bounded by $D^{r-1}\gamma(B)^r$ for individual copies and $D^{r-1}\Gamma_r(B)$ for collective copies, where $D$ is the total Hilbert space dimension. This converts the operational problem into a geometric estimation of intrinsic overlaps.

To derive lower bounds on copy complexity, the authors employ Lemma 1 (Moment-to-overlap uniformization). This lemma allows them to convert pointwise moment estimates of ensembles (averages of $|\langle \Phi | \Psi \rangle|^{2m}$ ) into uniform upper bounds on the maximum product overlap. This technique is applied to two specific families of bases:

Maximal-stabilizer eigenbases in odd-prime local dimensions ( $p$ ).
Flat-amplitude local-unitary orbit bases constructed using finite-field moment curves.

Key Contributions and Results

1. Unbounded Separation for Stabilizer Bases (Theorems 1 & 2)
For any fixed odd prime local dimension $p$ , the authors identify a sequence of complete maximal-stabilizer eigenbases $\{B_n\}$ where the copy complexity behaves drastically differently under the two architectures:

Individual-copy limit: The copy complexity under individual-copy SEP measurements diverges with the system size $n$ . Specifically, $N^{SEP}_{ind}(B_n) \gtrsim \frac{1}{\sqrt{p}-1}\sqrt{\frac{n}{\log_p n}}$ . Below this threshold, the identification probability remains close to random guessing.
Collective-copy limit: Every such basis can be perfectly distinguished by one-round collective LOCC using at most three copies (and only two copies if $p \equiv 1 \pmod 4$ ).
Mechanism: The collective protocol utilizes scalar-weighted observables. By choosing weights $c_a$ such that $\sum c_a^2 = 0$ and $\sum c_a \neq 0$ , the protocol cancels local commutation phases of the stabilizer generators while preserving global eigenvalue information, rendering the observables simultaneously measurable locally.
Information Gap: For $p \equiv 1 \pmod 4$ , two collective copies recover the full $\log_2 D$ bits of classical information, whereas individual-copy SEP measurements below the threshold scale recover only $o(\log_2 D)$ bits.

2. Persistence of Multicopy Hardness (Theorem 3)
The authors demonstrate that collective processing does not universally eliminate the need for many copies. For every fixed local dimension $d \ge 2$ , they construct a complete orthonormal basis $\{A_{n,d}\}$ (based on flat-amplitude states with finite-field phases) such that:

The copy complexity under collective SEP measurements remains unbounded as $n \to \infty$ .
The lower bound scales as $N^{SEP}_{col}(A_{n,d}) \gtrsim \sqrt{n}$ .
Strong Converse: For any number of copies $r = \lfloor \alpha \sqrt{n} \rfloor$ with $\alpha < 1$ , the success probability decays exponentially, and the accessible information is bounded by $(\alpha^2 + o(1))\log_2 D$ .
This result proves that no fixed number of collective copies suffices for all complete bases, establishing that bounded collective copy complexity is structure-dependent.

Significance and Claims
The paper claims to resolve the open question regarding the power of collective local measurements in multicopy state discrimination. Its primary significance lies in identifying collective local processing across repeated inputs as a distinct resource, separate from both the number of samples (copy number) and spatial measurement power (LOCC vs. SEP).

Resource Distinction: The work establishes that local quantum memory and the ability to perform joint operations on repeated inputs (inter-copy coherence) can unlock extensive classical information without requiring quantum communication between nodes.
Structural Dependence: The results highlight that the "compressibility" of multicopy discrimination is not a universal property of entanglement or orthogonality but depends on the specific operator structure of the basis. While stabilizer bases admit a finite-copy collective decoder, generic flat-phase bases do not.
Data Hiding: The authors suggest these findings define a form of "measurement-restricted data hiding" where the restriction is temporal (inability to maintain coherence across inputs) rather than spatial.

The paper concludes by noting that the optimal worst-case scaling of collective copy complexity for general bases remains an open question, though their construction establishes an $\sqrt{n}$ lower bound. It also leaves open whether the scalar-weight mechanism is unique to stabilizers or part of a broader compatibility principle.

More like this