The most discriminable quantum states in the multicopy regime

This paper investigates the quantum state ensembles that maximize success probability in minimum-error discrimination with multiple copies, proving that $k$-designs are optimal for pure states, mixed states can outperform pure ones beyond certain sizes, and quantum systems offer a quadratic advantage over classical counterparts in the multi-copy regime.

The Gist

The Big Picture: The "Guess the Card" Game

Imagine you are playing a game with a friend. Your friend has a deck of cards, but instead of standard playing cards, they have quantum states. These are like "magic cards" that hold information.

The rules are simple:

1. Your friend picks one card from a specific set of $N$ cards.
2. They give you that card.
3. Your job is to guess which card it was.

In the quantum world, some cards look so similar that you can't tell them apart perfectly. If you only get one card (one copy), you might have to guess, and you'll get it wrong sometimes.

The Twist: In this paper, the researchers ask: What if your friend gives you not just one card, but $k$ identical copies of that same card? Does having more copies make it easier to guess? And, more importantly, which specific set of cards should your friend use so that you can guess the best possible?

The Three Main Discoveries

The paper explores three different "universes" of cards: Pure Quantum Cards, Mixed Quantum Cards, and Real (Classical) Cards. Here is what they found:

1. The "Perfect Pattern" (Pure Quantum States)

When you are dealing with "pure" quantum cards (the most ideal, crisp versions), the researchers found a special rule.

• The Analogy: Imagine trying to arrange a set of points on a sphere (like a globe) so that they are as far apart from each other as possible. If you have just a few points, you can arrange them nicely. But if you have many points, the best arrangement is a specific, highly symmetrical pattern known as a $k$-design.
• The Finding: If your friend gives you a set of cards that forms this perfect symmetrical pattern (a $k$-design), you will be able to guess the card better than with any other arrangement. It's like the cards are arranged in the most "spread out" way possible, making them easiest to tell apart when you have multiple copies.
• The Catch: These perfect patterns only exist if you have a lot of cards. If you have fewer cards than the pattern requires, the "best" arrangement is a mystery that requires heavy computer calculations to solve.

2. The "Magic Trick" of Mixed Cards (Mixed Quantum States)

Usually, in the quantum world, "pure" cards are considered the best. You might think that "mixed" cards (cards that are a bit blurry or a combination of different states) would be harder to distinguish.

• The Surprise: The paper shows that if you have too many cards (more than needed for the perfect pattern), the mixed cards actually win.
• The Analogy: Imagine you are trying to identify a specific flavor of ice cream. If you have a perfect set of distinct flavors, you can tell them apart. But if you are forced to add a huge number of flavors, the best strategy isn't to keep adding distinct flavors; it's to add one "plain vanilla" (a completely mixed state) to the mix. This "plain" card acts as a safety net that helps you distinguish the others better than if you tried to use only distinct, pure flavors.
• The Result: In the "many copies" regime, a mix of perfect patterns and a little bit of "blur" (mixed states) gives you the highest chance of winning the game.

3. The "Quantum Advantage" vs. Classical Bits

The researchers also compared these quantum cards to classical cards (like standard bits: 0s and 1s).

• The Finding: Quantum cards are much better at this game than classical cards, but the advantage depends on the type of quantum card.
  • Complex Quantum Cards: These offer a quadratic advantage. In plain English: If you double the number of copies ($k$) you get, your ability to guess improves much faster than it would with classical cards. It's like quantum cards get a "super boost" from having more copies.
  • Real Quantum Cards (Rebits): These are quantum cards that don't use complex numbers (they are "real" numbers only). The paper found that these cards lose most of their superpower. Their advantage over classical cards is tiny—just a small constant bump, not a massive jump.
• The Metaphor: Think of complex quantum cards as a high-performance sports car that gets exponentially faster the more fuel (copies) you give it. Real quantum cards are like a regular sedan; giving it more fuel helps, but it doesn't turn into a rocket ship. This proves that the "weirdness" of complex numbers is essential for the biggest quantum advantages.

How They Solved It

Since mathematically solving this for every possible number of cards and copies is incredibly hard (like trying to solve a 100-piece puzzle where the pieces keep changing shape), the authors used two main tools:

1. Mathematical Proofs: For specific cases (like when the number of cards is huge), they used rigorous math to prove exactly which patterns work best.
2. Computer Simulations: For the tricky cases where no simple formula exists, they wrote computer programs to test millions of different card arrangements. They used a method called "gradient descent" (like rolling a ball down a hill to find the lowest point) to find the best arrangements and "Semidefinite Programming" to prove that no other arrangement could possibly be better.

Summary in One Sentence

This paper figures out the best way to arrange quantum "cards" so you can identify them when you have multiple copies, discovering that perfect symmetrical patterns are best for small sets, mixed states are best for large sets, and that the "magic" of quantum mechanics relies heavily on complex numbers to beat classical computers.

Technical Summary

1. Problem Definition

The paper investigates the fundamental limits of quantum state discrimination in the multicopy regime. Specifically, it addresses the following optimization problem:

• Scenario: An ensemble of $N$ quantum states $\{\rho_i\}_{i=1}^N$ in dimension $d$ is chosen with uniform probability ($p_i = 1/N$). The receiver is given $k$ identical copies of the unknown state, i.e., $\rho_i^{\otimes k}$.
• Goal: Determine the set of states $\{\rho_i\}$ that maximizes the average success probability of correctly identifying the state using an optimal Positive Operator-Valued Measure (POVM).
• Metric: The authors define $k$-copy discriminability, denoted as $\text{Discr}(\{\rho_i\}, k)$, as the maximal average success probability. They seek the maximal $k$-copy discriminability, $\Omega(d, N, k)$, which is the supremum of this metric over all possible sets of states (pure, mixed, real, or classical).

2. Methodology

The authors employ a combination of analytical proofs, asymptotic analysis, and numerical optimization techniques:

• Analytical Framework:

  • Utilization of Symmetric Subspaces ($\text{Sym}_d^k$) and Permutation Invariant Subspaces ($\text{Perm}_d^k$) to bound the discrimination probability.
  • Application of Quantum State $k$-designs, which are ensembles that mimic the statistical properties of the Haar measure up to the $k$-th moment.
  • Connection to Classical Information Theory, specifically the multiplicative Bayes capacity of classical channels, to solve the classical analogue of the problem.
  • Use of the Cauchy-Schwarz inequality and trace inequalities to derive upper bounds.

• Numerical Techniques:

  • Lower Bounds: Obtained via Grid Search (discretizing the Bloch sphere) combined with Semidefinite Programming (SDP) to optimize measurements for fixed states. They also utilize Gradient Descent (Adam algorithm) to optimize both states and measurements simultaneously in non-convex spaces.
  • Upper Bounds: Derived using a hierarchy of SDP relaxations based on the Doherty-Parrilo-Spedalieri (DPS) hierarchy and the Positive Partial Transpose (PPT) criterion. This approximates the separability of the state-measurement tensor to bound the maximum achievable probability.

3. Key Contributions and Results

A. Pure Quantum States and $k$-Designs

• Main Result: For pure states, if the number of states $N$ is large enough to support a state $k$-design, the set of states forming that design achieves the maximal $k$-copy discriminability.
• Formula: When a $k$-design exists, the maximal success probability is:
  $$\Omega_q^{\text{pure}}(d, N, k) = \frac{1}{N} \binom{d+k-1}{k}$$
  where $\binom{d+k-1}{k}$ is the dimension of the symmetric subspace.
• Implication: If $N$ is insufficient for a $k$-design, the problem becomes computationally hard, and the optimal states are likely approximate $k$-designs (e.g., equilateral triangles on the Bloch sphere for $d=2, N=3$).

B. Mixed Quantum States

• Counter-Intuitive Finding: Unlike the single-copy regime (where pure states are optimal), in the multicopy regime, mixed states can outperform pure states when $N$ exceeds the size required for a $k$-design.
• Mechanism: Pure states span only the symmetric subspace. Mixed states allow the discrimination strategy to utilize the larger permutation-invariant subspace.
• Optimal Construction: The optimal set often consists of a $k$-design (pure states) combined with the maximally mixed state ($I/d$). The measurement includes an outcome orthogonal to the symmetric subspace ($M_{N} = I - \Pi_{\text{sym}}$), which "clicks" for the mixed state but never for the pure states, thereby increasing the total success probability.

C. Classical States and Quantum Advantage

• Classical Analogue: The problem of discriminating classical states (probability distributions) is mapped to the multiplicative Bayes capacity of independent uses of a classical channel.
• Exact Solution: For $N \ge \binom{d+k-1}{k}$, the maximal classical discriminability is exactly solvable:
  $$\Omega_{\text{cl}}(d, N, k) = \frac{1}{N} \text{cap}_d(k)$$
• Quadratic Advantage: The paper proves a quadratic quantum advantage in the number of copies $k$.
  • Classical scaling: $\Omega_{\text{cl}} \sim \frac{1}{N} k^{(d-1)/2}$.
  • Quantum (pure) scaling: $\Omega_q^{\text{pure}} \sim \frac{1}{N} k^{d-1}$.
  • For $d=2$, classical scales as $\sqrt{k}$, while quantum scales as $k$.

D. Real Quantum States (Rebits)

• Reduced Advantage: When restricting states to real density matrices (defined in $\mathbb{R}^d$), the quantum advantage over classical systems is strongly reduced.
• Scaling: For $d=2$, real qubits (rebits) offer only a constant advantage over classical bits, rather than the quadratic advantage seen in complex qubits.
• Geometry: Numerical evidence suggests that for real qubits, regular polygons (regular $N$-gons on the Bloch circle) are the most discriminable sets, even when $N < k$.

4. Significance and Implications

1. Fundamental Limits: The work establishes the ultimate bounds on how well quantum states can be distinguished when multiple copies are available, revealing that mixedness can be a resource in multicopy scenarios.
2. Connection to Designs: It solidifies the link between state $k$-designs and optimal discrimination, suggesting that finding optimal states is equivalent to finding $k$-designs. This has implications for quantum tomography and cryptography.
3. Quantum vs. Classical: The demonstration of a quadratic advantage for complex quantum systems over classical systems highlights the specific role of complex numbers and the geometry of the symmetric subspace in quantum information processing.
4. Real vs. Complex: The finding that real quantum systems lose the quadratic advantage underscores the necessity of complex Hilbert spaces for maximal quantum performance in discrimination tasks.
5. Methodological Contribution: The introduction of rigorous SDP-based upper and lower bounds provides a toolkit for analyzing state discrimination in regimes where analytical solutions are intractable.

5. Open Questions

The authors identify several open problems, including:

• Proving that regular polygons are optimal for real qubits for all $k$.
• Determining the asymptotic behavior of real states for $d > 2$.
• Investigating the role of non-uniform prior distributions.
• Exploring the application of these bounds to quantum information leakage and cryptographic protocols.

In summary, this paper provides a comprehensive characterization of the most discriminable quantum states in the multicopy regime, revealing that the optimal strategy depends critically on the number of states, the availability of copies, and whether the system is complex, real, or classical.