Rise and fall of nonstabilizerness via random measurements

This paper investigates the dynamics of nonstabilizerness (magic) in monitored quantum circuits, revealing that computational basis measurements exponentially suppress magic through Clifford scrambling, whereas rotated non-Clifford measurements can both generate and sustain magic, leading to distinct relaxation behaviors and diagnostic sensitivities between stabilizer nullity and Stabilizer Rényi Entropies.

The Gist

The Big Picture: What is "Magic"?

Imagine you are trying to build a super-computer that can solve problems impossible for today's machines. To do this, you need a special ingredient called "Magic" (scientists call it nonstabilizerness).

Think of Magic like spice in a giant pot of soup.

• Stabilizer states are like plain, bland water. You can make a lot of it easily, and classical computers can simulate it perfectly. It's safe, but boring.
• Magic states are the hot sauce, the saffron, the exotic spices. They make the soup (the quantum computer) powerful enough to do amazing things. Without them, you just have plain water.

The problem? Magic is fragile. If you look at it too closely (measure it), it often disappears. This paper asks: Can we keep the soup spicy even while we keep tasting it?

The Experiment: The "Taste Test" Game

The researchers set up a game with a quantum system (a pot of soup) and two types of "Taste Tests" (measurements).

Scenario 1: The "Blindfolded" Taste Test (Computational Basis)

Imagine you are tasting the soup, but you are blindfolded and only allowed to check if the spoon is "up" or "down" (a standard, boring measurement).

• The Process: You stir the soup wildly (random quantum operations) and then take a blind taste test.
• The Result: Every time you take a taste test, there is a tiny, tiny chance you accidentally find a "bland" spot and the soup loses a little bit of its spice.
• The Analogy: Imagine trying to empty a swimming pool by scooping out one drop at a time. Because the pool is so huge (exponentially large), it takes forever to drain it completely.
• The Discovery: Even though you are constantly "tasting" (measuring) the soup, the Magic is incredibly hard to kill. It takes an exponential number of measurements to turn the spicy soup back into plain water. The random stirring (Clifford unitaries) protects the spice, hiding it from the taste test.

Scenario 2: The "Rotated" Taste Test (Magic Basis)

Now, imagine you change the rules. Before you taste the soup, you tilt your head at a weird angle (a "rotated" measurement).

• The Process: You stir, tilt your head, and taste.
• The Result: This time, the taste test does something surprising. Sometimes it removes spice, but sometimes it accidentally adds spice!
• The Analogy: It's like shaking a jar of glitter. Sometimes the glitter falls out, but if you shake it at just the right angle, the static electricity makes more glitter stick to the sides.
• The Discovery: The system finds a steady state. It doesn't become plain water, nor does it become infinitely spicy. It settles into a "Goldilocks" zone where it maintains a constant, non-zero amount of Magic forever, regardless of what you started with.

The Two Different "Spice Meters"

The paper uses two different ways to measure how much "Magic" is in the soup. They act very differently:

1. The "Nullity" Meter (The Binary Switch):

   • This meter asks: "Is there any spice left?"
   • Behavior: It is very stubborn. It doesn't care how you tilt your head (the angle of measurement). As long as you are measuring, it eventually settles at a specific high level of spice (about $N-1$ units). It's like a light switch that stays "ON" no matter how you jiggle the wire.

2. The "Rényi Entropy" Meter (The Fine-Tuned Scale):

   • This meter asks: "How rich and complex is the spice?"
   • Behavior: This meter is sensitive. If you tilt your head just a tiny bit, the amount of spice it detects changes.
   • The Twist: If you tilt your head slightly, the spice level grows quadratically (like a square). If you tilt it all the way, it grows linearly. This meter reveals a much richer, more detailed picture of the soup's flavor than the simple switch.

The "Relaxation" Time (How fast does it settle?)

The paper also looked at how long it takes for the soup to settle into its final flavor, depending on what you started with:

• Starting with a "Spicy" Soup (Haar-random state): If you start with a pot full of spice, it takes a constant amount of time to settle. It's like a cup of hot coffee cooling down; it cools fast regardless of the room size.
• Starting with "Plain Water" (Stabilizer state): If you start with plain water and try to make it spicy, it takes a long time (linear with the size of the pot). It's like trying to fill a giant swimming pool with a teaspoon; the bigger the pool, the longer it takes.

Why Does This Matter?

This research is a breakthrough because it challenges the old idea that "measuring destroys quantum power."

1. Protection: It shows that quantum computers are naturally very good at hiding their "Magic" from errors (measurements) if they are scrambled correctly.
2. Creation: It shows that measurements aren't just destroyers; under the right conditions, they can actually create and sustain the very resource (Magic) needed for quantum advantage.
3. New Phases of Matter: It suggests we can engineer quantum systems that stay in a "spicy" state forever, even while being constantly observed. This could lead to new types of quantum memory or error-correcting codes that are naturally robust.

Summary in One Sentence

The paper discovers that while measuring a quantum system usually kills its "magic" (computational power), random scrambling protects it, and cleverly angled measurements can actually generate and sustain that magic, turning the act of observation from a destroyer into a creator.

Technical Summary

1. Problem Statement

The paper investigates the dynamics of nonstabilizerness (also known as "magic"), a crucial resource for universal quantum computation, in monitored quantum circuits. While entanglement is a well-studied resource, it is insufficient for quantum advantage (as seen in stabilizer states). Nonstabilizerness quantifies the distance of a quantum state from the set of stabilizer states.

The central problem addressed is how random projective measurements interact with random Clifford unitaries to create or destroy magic. Previous works focused on regimes where the number of measurements scales extensively with system size. This paper specifically analyzes the role of individual measurements and the dynamics of magic decay (in computational bases) and generation (in rotated non-Clifford bases).

2. Methodology

The authors study a monitored circuit protocol consisting of $N$ qubits evolving through discrete time steps:

1. Random Clifford Unitary: A global random Clifford unitary $U_C$ is applied to the system.
2. Rotation: A single-qubit rotation $R_x(\theta_M) = \exp(-i\theta_M \pi/8 \sigma_x)$ is applied to the first qubit.
   • If $\theta_M = 0$, the rotation is trivial (Clifford), and the process is purely Clifford.
   • If $\theta_M > 0$, the rotation introduces non-Clifford operations, potentially generating magic.
3. Measurement: The same qubit is measured in the computational basis ($\sigma_z$).
4. Inverse Unitary: The inverse Clifford $U_C^\dagger$ is applied to restore the basis.

This cycle is repeated $t$ times. The authors analyze two distinct regimes:

• Regime A ($\theta_M = 0$): Measurements in the computational basis (non-magic inducing).
• Regime B ($\theta_M > 0$): Measurements in a rotated basis (magic inducing).

Metrics Used:

• Stabilizer Nullity ($\nu$): Defined as $\nu = N - \log_2 |G_S|$, where $|G_S|$ is the size of the stabilizer group. It is a discrete monotone that counts the number of qubits required to describe the state as a stabilizer state.
• Stabilizer Rényi Entropies (SREs): Specifically the second moment ($M_2$), which quantifies the distribution of Pauli expectation values. Unlike nullity, SRE is continuous and sensitive to small perturbations.

The authors employ analytical modeling using Markov chains to describe the transition probabilities of the nullity and numerical simulations for various initial states (Haar-random, product T-states, GUE-evolved states, and computational basis states).

3. Key Contributions and Results

A. Decay of Magic in Computational Basis ($\theta_M = 0$)

• Quantized Step Decay: The stabilizer nullity $\nu$ decays in discrete steps of 1. The system transitions from a high-magic state to a random stabilizer state ($\nu=0$).
• Exponential Protection: The probability of a step-down in nullity is exponentially small in the current nullity value ($P \sim 2^{\nu-N}$). Consequently, erasing all magic requires an exponential number of measurements ($t \sim 2^N$). This reveals that Clifford scrambling provides strong protection against the destruction of magic by local measurements.
• Analytical Model: The authors derive a differential equation for the non-stabilized dimension $y = 2^\nu$:
  $$\frac{dy}{dt} \approx -A_N \frac{1}{2}(y^2 - 1)$$
  This model predicts that for $t \ll 2^N$, $\nu(t) \approx N - \log_2 t$, and for $t \gg 2^N$, $\nu$ decays exponentially to zero.
• SRE Dynamics: For Haar-random states, the SRE ($M_2$) follows the nullity dynamics closely, with plateaus corresponding to specific nullity values. However, for other initial states (like T-states or GUE states), the SRE dynamics deviate from the nullity mapping, showing that SRE captures finer-grained structural details of the state.

B. Generation and Steady State in Rotated Basis ($\theta_M > 0$)

• Competition of Creation and Destruction: When $\theta_M > 0$, the rotation can increase the nullity (create magic), while the measurement tends to decrease it.
• Steady State: The system converges to a non-trivial steady state with finite magic, independent of the initial state.
  • Nullity Steady State: The average steady-state nullity is $\nu_{ss} \approx N - 1.46$. Crucially, this value is independent of the rotation angle $\theta_M$ (as long as $\theta_M \neq 0$) and depends only on system size $N$.
  • Convergence Timescales:
    • Starting from high-magic states (e.g., Haar-random), the system relaxes to the steady state in constant time (independent of $N$).
    • Starting from low-magic states (e.g., stabilizer states $|0\rangle^{\otimes N}$), the system requires linear time ($t \sim N$) to build up extensive magic.
• SRE Steady State: Unlike nullity, the steady-state SRE ($M_{ss}^2$) depends on the rotation angle $\theta_M$.
  • For small angles ($\theta_M \ll 1$), the steady-state magic density scales quadratically: $m_{ss}^2 \propto \theta_M^2$.
  • The steady state is not a Haar-random state (which would have $\nu \approx N$), but rather a state with high entanglement but "low" magic relative to a Haar state, yet non-zero.

4. Significance and Implications

• Robustness of Magic: The work demonstrates that magic is inherently robust against local projective measurements when protected by Clifford scrambling, requiring exponential resources to destroy.
• Measurement as a Resource: The paper challenges the view that measurements only destroy quantum resources. It shows that measurements in rotated bases can sustain and generate magic, acting as a source of computational power in hybrid quantum protocols.
• Phase Transitions and Phases of Matter: The findings provide a tunable protocol to realize a specific phase of matter characterized by volume-law entanglement but constant (non-zero) nonstabilizerness. This is distinct from the "magic-induced computational separation" discussed in recent literature.
• Diagnostic Distinction: The study highlights a sharp distinction between "coarse-grained" diagnostics (Stabilizer Nullity) and "fine-grained" diagnostics (SRE). While nullity is robust and angle-independent in the steady state, SRE reveals a richer structure dependent on the measurement angle and initial state preparation.
• Practical Application: The results suggest that in noisy intermediate-scale quantum (NISQ) devices or error-corrected systems, one can potentially maintain computational resources (magic) even in the presence of frequent measurements, provided the measurement basis is appropriately rotated.

Conclusion

Scocco et al. provide a comprehensive analytical and numerical framework for understanding the interplay between random unitary dynamics and measurements in the context of quantum computational resources. They establish that while Clifford dynamics protect magic from decay, rotated measurements can actively sustain it, leading to a stable, non-trivial steady state that balances entanglement and nonstabilizerness. This offers new insights into the dynamics of monitored quantum circuits and the resource theories of quantum computation.