Measurement-induced state transitions in multi-qubit transmon processors

This paper investigates how the presence of other circuit elements, such as spectator qubits and couplers, in multi-qubit transmon processors modifies the measurement-induced state transition (MIST) thresholds and dynamics of a readout qubit, revealing that these components can both lower the transition threshold and be impacted by the measurement process.

The Gist

The Big Picture: Listening to a Whisper in a Noisy Room

Imagine you are trying to listen to a single person whispering in a quiet room. This is like a quantum computer trying to "read" the state of a single qubit (a tiny bit of quantum information). To do this, scientists use a "readout resonator," which acts like a microphone that sends a signal to check if the qubit is a 0 or a 1.

Usually, this process is gentle. You check the qubit, and it stays exactly as it was. However, the paper explains that if you turn the "microphone" volume up too high (using a strong drive signal), something weird happens: the act of listening actually changes the qubit's state. It's like shouting at a whispering person so loudly that they get startled and start shouting back, changing their answer.

In the scientific world, this is called a Measurement-Induced State Transition (MIST). It happens because the loud signal accidentally hits a "resonance," causing the qubit to jump up energy levels it shouldn't be in. This ruins the computer's calculation.

The Problem: The "Spectator" Effect

Until now, scientists mostly studied this problem with just one qubit in isolation. But real quantum computers have many qubits packed together, like a crowded party.

The authors of this paper asked: What happens when you try to listen to one person (the "Target") while other people (the "Spectators") are standing right next to them?

They found that the presence of these neighbors changes the rules.

• The Surprise: The neighbors can actually make the "listening" process more dangerous. Even if the target qubit is safe on its own, the presence of a neighbor might lower the volume threshold where the target gets startled and jumps.
• The Mechanism: Think of the qubits as tuning forks. If you strike one (the target), the sound waves can travel through the air and vibrate a neighbor (the spectator). Sometimes, the neighbor vibrates in a way that creates a "shortcut" for the target to jump to a wrong energy level.

The Solution: A New Way to Map the Danger Zones

To figure out exactly when and why this happens, the authors invented a new mathematical tool. They call it a "branch analysis," but let's call it "The Two-Path Test."

Imagine you are trying to walk through a forest (the quantum system) to get to a destination (the measurement result).

1. Path A (Coupling-First): You first tie all the trees together with vines (turn on the connections between qubits), and then you start walking.
2. Path B (Drive-First): You start walking first, and then you tie the trees together.

In a perfect world, both paths should lead to the same result. However, the authors discovered that in these quantum forests, the two paths often lead to different places.

• If the paths are the same, the neighbors aren't causing trouble.
• If the paths are different, it means the neighbors have created a "trap" (an avoided crossing) that only appears when the connections are active. This trap is where the qubit gets startled and jumps.

By comparing these two paths, the team can predict exactly how loud the "microphone" can get before the neighbors cause a disaster.

The Twist: The "Tunable Bridge" (Couplers)

In advanced quantum computers, qubits aren't just stuck next to each other; they are often connected by a special switch called a coupler. This is like a bridge between two islands that can be raised or lowered.

The authors tested what happens if you use this bridge.

• Good News: Sometimes, the bridge acts like a noise-canceling headphone. By adjusting the bridge (the coupler), they found specific settings where the "trap" disappears. The neighbors stop causing the target to jump, even if they are close by.
• Bad News: It's tricky. The bridge only works if it is in the right "state" (like being in a specific position). If the bridge itself gets excited or moves, it can actually make the problem worse. Also, the settings that stop the "jumping" are not necessarily the same settings that stop the qubits from interfering with each other in other ways.

The Takeaway

The paper concludes that you cannot design a quantum computer by looking at one qubit at a time. You have to look at the whole crowd.

• Spectators matter: Neighboring qubits can make your measurements less reliable.
• Context matters: A setup that works for one qubit might fail when that qubit is part of a larger chip.
• Couplers are a double-edged sword: They can help fix these problems, but only if tuned very precisely, and they introduce their own new rules to follow.

Essentially, the authors provided a map to help engineers navigate the crowded, noisy environment of a multi-qubit quantum processor so they can listen to the qubits without accidentally startling them.

Technical Summary

Technical Summary: Measurement-Induced State Transitions in Multi-Qubit Transmon Processors

Problem Statement
Dispersive readout of transmon qubits in circuit quantum electrodynamics (cQED) is known to lose its quantum non-demolition (QND) character at small to moderate measurement drive amplitudes. This phenomenon, termed Measurement-Induced State Transitions (MIST) or drive-induced unwanted state transitions (DUST), originates from Landau-Zener transitions at accidental multi-photon resonances. While the mechanisms governing MIST in single isolated transmons are well-characterized using Floquet branch analysis, the impact of these transitions in multi-qubit quantum processor units (QPUs) remains largely unexplored. In a multi-qubit chip, the presence of "spectator" components—such as neighboring qubits and couplers—can alter the MIST threshold of a measured transmon. Understanding how these spectator modes influence the readout fidelity and duration is crucial for scaling superconducting quantum processors.

Methodology
The authors introduce a general framework to characterize measurement-induced transitions when the qubit under readout is coupled to other circuit elements. The core of this methodology is a Multimode Branch Analysis that extends the single-qubit Floquet branch analysis to multiple modes.

The method relies on comparing two distinct "labeling" procedures (or gedanken experiments) to identify when spectator modes induce unwanted transitions:

1. Coupling-First (CF) Approach: The qubit-spectator coupling is adiabatically increased from zero to its target value before the measurement drive is ramped up. This method tracks the system through avoided crossings that occur as photons are added to the coupled system.
2. Drive-First (DF) Approach: The measurement drive is ramped up on the uncoupled system first, and then the qubit-spectator coupling is adiabatically increased. This method follows diabatic branches at crossings that would otherwise be avoided in the coupled system.

The spectator-induced critical photon number ($n_{crit}^{spec}$) is defined as the lowest drive amplitude (photon number) at which the state labels assigned by the CF and DF methods diverge. A divergence indicates that the spectator has induced an avoided crossing in the quasienergy spectrum that would not exist in the absence of the spectator, leading to a potential state transition.

Key Contributions and Results

• Static Coupling (Two Transmons):
  The authors applied their method to a system of two capacitively coupled transmons (one measured, one spectator).

  • They demonstrated that the spectator can significantly lower the MIST threshold of the readout qubit.
  • The critical photon number is drastically reduced in regions where the qubit and spectator are near resonance ($\omega_q \approx \omega_s$).
  • This effect extends over a broad frequency range (hundreds of MHz) around the resonance, driven by the effective drive enabling single-excitation exchange interactions.
  • For two-excitation states, transitions can lead to leakage out of the computational subspace, posing risks for quantum error correction.

• Tunable Coupling (Qubit-Spectator-Coupler):
  The study was extended to a system where two qubits interact via a frequency-tunable transmon coupler.

  • Suppression of MIST: Surprisingly, the presence of a coupler can suppress spectator-induced MIST. In specific parameter regimes, the coupler cancels the matrix elements responsible for mixing the qubit and spectator states, turning avoided crossings into simple crossings. This results in a higher critical photon number (i.e., more robust readout) compared to the static coupling case.
  • State Dependence: This mitigation is highly dependent on the quantum state of the coupler. If the coupler is in its ground state, MIST may be suppressed; if it is excited, the suppression may vanish.
  • Parameter Space Complexity: The authors identified diagonal regions in the $(\omega_s, \omega_c)$ plane where critical photon numbers are unusually large due to the cancellation of charge operator matrix elements (derived via Bogoliubov transformations).
  • Decoupling of Constraints: Crucially, the parameter regions where spectator-induced MIST is suppressed do not necessarily coincide with the regions where the residual static ZZ interaction between qubits is canceled. Optimizing a coupler to eliminate ZZ coupling does not automatically minimize MIST.

Significance and Claims
The paper claims to provide a necessary framework for understanding strong drives and nonlinearities in multi-mode cQED systems. The authors emphasize that optimizing the readout of a single, well-isolated qubit does not guarantee optimal performance when that qubit is embedded in a QPU.

The primary significance lies in the demonstration that:

1. Spectator modes can drastically alter the MIST threshold, creating broad regions of frequency crowding that are difficult to avoid in large-scale processors.
2. While tunable couplers offer a mechanism to mitigate spectator-induced leakage, they introduce new constraints on frequency placement and are sensitive to the coupler's internal state.
3. The conditions for suppressing MIST are distinct from the conditions for canceling static ZZ interactions, implying that readout optimization requires a holistic view of the full circuit rather than just static coupling parameters.

The authors conclude that these insights are relevant not only for readout but for any strongly driven nonlinear circuit, including parametric couplers, reset protocols, and state stabilization.