Analog-Digital Quantum Computing with Quantum Annealing Processors

This paper demonstrates that commercially available quantum annealing processors can be extended beyond uniform annealing to perform analog-digital quantum computing by combining fixed many-body Hamiltonian evolution with arbitrary-basis initialization and measurement via auxiliary qubits, thereby enabling foundational applications like coherent oscillations, quantum walks, and Anderson localization.

The Gist

Imagine you have a giant, super-fast library of books (the quantum computer). For years, the only way to use this library was to follow a strict, pre-written script: you walk in, pick a book, read it, and walk out. This is how standard Quantum Annealing works. It's incredibly powerful for finding the "best" solution to a problem (like finding the shortest route for a delivery truck), but it's a bit rigid. You can't easily stop in the middle to ask a specific question about the book's plot or change the ending.

This paper describes a breakthrough where the researchers taught this rigid library to be flexible and conversational. They turned a "one-way street" into a "two-way street," allowing them to not just find answers, but to simulate how complex quantum systems move and change over time.

Here is the breakdown of their new method, using simple analogies:

1. The Old Way: The "Slide"

Think of a standard quantum annealer like a giant slide.

• You put a marble (the problem) at the top.
• You let gravity (the annealing process) pull it down.
• It naturally settles at the bottom (the solution).
• The Problem: Once the marble hits the bottom, the game is over. You can't stop it halfway to see how fast it was going or what path it took. You can only see the final result.

2. The New Way: The "Train with Stops"

The researchers figured out how to turn that slide into a train track with stations.

• The Train (The Analog Part): The train still runs on a fixed track (the quantum processor's natural physics). This is the "analog" part. It's fast and can carry thousands of passengers (qubits) at once.
• The Stations (The Digital Part): The magic is that they built special "stations" at the beginning and the end of the track.
  • Station A (Initialization): Before the train leaves, they can load the passengers into specific seats or make them dance in a specific pattern. This is like applying a "gate" to set the starting state.
  • Station B (Measurement): When the train arrives, they can check the passengers' state in any way they want, not just the default way.

3. The "Multicolor" Trick

The processor has thousands of qubits, but usually, they all move in unison (like a marching band). The researchers introduced "Multicolor Annealing."

• Imagine the marching band is divided into three groups: Red, Blue, and Green.
• In the old days, the conductor told everyone to march forward at the same time.
• In this new experiment, the conductor tells the Red group to march forward, then stops the Blue group, then tells the Green group to march backward.
• By controlling these groups independently, they can use the Red and Green groups as "helpers" to set up and read the Blue group (the main data), without disturbing the Blue group while it's doing its work.

4. What Did They Actually Do?

They used this new "Train with Stations" method to do three cool things:

• The Quantum Dance (Coherent Oscillations): They made a single qubit (a marble) spin and wobble like a top. By changing the "stations," they could watch it spin in any direction they wanted, proving they could control the quantum state precisely.
• The Quantum Wave (Quantum Walk): They sent a "wave" of energy through a chain of 56 qubits. It was like dropping a pebble in a pond and watching the ripples spread. They proved the ripples moved exactly as physics predicted, even with the "helper" qubits nearby.
• The Quantum Traffic Jam (Anderson Localization): They created a "disordered" road (a chain with random bumps). In a normal world, a car would drive through. But in this quantum world, the car got stuck in one spot because the bumps interfered with each other. They successfully simulated this "traffic jam" of quantum particles, which is very hard to do on regular computers.

Why Does This Matter?

Previously, if you wanted to simulate a complex quantum system (like a new drug molecule or a superconductor), you had to break the simulation into tiny, choppy steps (digital gates), which is slow and error-prone. Or, you had to use a rigid annealer that couldn't show you the process, only the result.

This paper shows that commercial quantum annealers (which are already huge and available) can now do Analog-Digital Quantum Computing.

• Analogy: It's like taking a massive, specialized supercomputer built for one thing (solving puzzles) and giving it a universal remote control. Now, it can also act as a high-speed movie camera to record the "movie" of quantum physics happening in real-time.

The Bottom Line

The researchers didn't build a new machine; they found a clever new way to drive the existing one. By using "helper" qubits to set the stage and read the results, they unlocked the ability to simulate complex, moving quantum systems on a massive scale. This opens the door to using these machines for things they were never designed for, like simulating new materials, understanding how diseases spread at a molecular level, or exploring the weird rules of quantum mechanics in ways we couldn't before.

Technical Summary

1. Problem Statement

Quantum annealing (QA) processors, such as those developed by D-Wave, are designed to solve optimization problems by evolving a system from a transverse-field Hamiltonian to a problem Hamiltonian. While they offer massive scale (thousands of qubits) and complex connectivity, they traditionally suffer from limited flexibility:

• Uniform Control: Qubits are typically controlled in unison, attenuating quantum fluctuations uniformly.
• Restricted Operations: The standard protocol limits operations to finding the ground state of a specific Hamiltonian, preventing the study of coherent excited-state dynamics or arbitrary quantum state preparation.
• Lack of Flexibility: Unlike gate-based systems, standard QA lacks the ability to perform arbitrary single-qubit rotations (state preparation) and measurements in arbitrary bases, which are essential for simulating general quantum dynamics and performing quantum tomography.

The central question addressed is: Can the flexibility of gate-based state preparation and measurement be realized within an annealing-based system without sacrificing its scale?

2. Methodology: Analog-Digital Quantum Computing (ADQC)

The authors propose and implement a hybrid protocol called Analog-Digital Quantum Computing (ADQC) on a superconducting quantum annealing processor. This approach combines large-scale analog simulation with digital (gate-based) initialization and readout.

Key Technical Innovations:

• Multicolor Annealing: The team introduces a control scheme where different subsets of qubits follow independent annealing schedules ($s_\alpha(t)$). The qubits are partitioned into three sets:

  • Source Qubits: Used for state initialization.
  • Target Qubits: The main information qubits undergoing analog evolution.
  • Detector Qubits: Used for state measurement.

• Protocol Flow:

  1. Initialization (Digital Layer): Source qubits are forward-annealed to $s=1$ in the presence of a polarizing bias to prepare a known state. They are then strongly coupled to target qubits. The target qubits are annealed to a fixed point $s^*$, and the source qubits are rapidly quenched back to $s=0$. This effectively implements a programmable single-qubit rotation gate on the target qubits.
  2. Analog Evolution: The target qubits evolve under a fixed, time-independent many-body Hamiltonian ($H_{target}$) for a duration $t$. In the weak-coupling regime, this Hamiltonian is well-described by an effective XY model:
     $$H_{eff} = \sum_{\langle i,j \rangle} \frac{J_{ij}}{4}(\sigma^x_i \sigma^x_j + \sigma^y_i \sigma^y_j) - \sum_i \frac{\delta\Delta_i}{2}\sigma^z_i$$
  3. Measurement (Digital Layer): Detector qubits are rapidly quenched to $s=1$, coupling them to the target qubits. By tuning static flux biases on the detectors, the measurement basis can be rotated to any arbitrary axis on the Bloch sphere before reading out in the computational basis.

• Hardware Implementation: The experiments were conducted on a D-Wave Advantage2 processor (1178 qubits). The system utilizes per-qubit programmable static flux biases and "anneal offsets" to control the effective fields and coupling strengths independently.

3. Key Contributions

• Demonstration of ADQC in QA: First successful implementation of analog-digital quantum computing on a commercially available quantum annealer with thousands of qubits.
• Arbitrary Basis Control: Proved that arbitrary single-qubit state preparation and measurement in any basis are possible in an annealing architecture, effectively expanding the accessible state space beyond the ground state.
• Effective XY Model Realization: Showed that the native hardware, when operated at a specific annealing point ($s^*$) in the weak-coupling regime, naturally realizes an effective XY Hamiltonian suitable for simulating fermionic and spin systems.
• High Coherence: Achieved a normalized coherence time $J T_\phi \gtrsim 20$, demonstrating that coherent dynamics can be preserved despite the "always-on" couplings inherent to the annealing architecture.

4. Experimental Results

The authors validated the protocol through several foundational applications:

• Single-Qubit Coherent Oscillations:

  • Demonstrated Larmor precession with arbitrary initialization and measurement angles ($\theta, \phi$) on the Bloch sphere.
  • Extracted relaxation times ($T_1 \approx 32$ ns) and dephasing times ($T_\phi \approx 12$ ns for single qubits, limited by inhomogeneous broadening).
  • Showed that flux biases on source/detector qubits map directly to rotation angles.

• Two-Qubit Spin Exchange:

  • Observed coherent spin exchange between two coupled target qubits under the XY Hamiltonian.
  • Measured in both $\sigma^x$ and $\sigma^z$ bases, showing excellent agreement with analytical solutions for open quantum systems.
  • Extracted a multi-qubit dephasing time $T_\phi \approx 37$ ns, significantly longer than single-qubit measurements due to motional narrowing (suppression of inhomogeneous broadening).

• Multi-Qubit Quantum Walk (Fermionic Dispersion):

  • Simulated a 56-qubit and 124-qubit periodic 1D chain.
  • Prepared a single excitation and observed ballistic propagation and interference patterns.
  • Performed 2D Fourier transforms of space-time data to extract the dispersion relation $E(k)$, which matched the theoretical prediction for non-interacting spinless fermions (via Jordan-Wigner transformation) with no free fitting parameters.

• Anderson Localization:

  • Introduced disorder by programming random detunings ($\delta\Delta_i$) on target qubits in a 124-qubit chain.
  • Initialized a staggered state ($|1010...\rangle$) and measured the "imbalance" (difference in excitation probability between odd and even sites).
  • Results showed that as disorder increased, the imbalance persisted at long times, confirming Anderson localization. The scaling of imbalance with disorder strength ($W$) matched theoretical predictions ($\tilde{I} \propto W^2$).

5. Significance and Impact

• Expanding the Utility of QA: This work fundamentally changes the perception of quantum annealers from mere optimization solvers to versatile quantum simulators capable of studying non-equilibrium dynamics, many-body localization, and complex quantum phases.
• Scalability: By leveraging the existing massive connectivity and scale of commercial QA processors, this approach offers a pathway to simulate large quantum systems that are currently intractable for small-scale gate-based devices.
• Bridge to Universal Computing: The protocol serves as a stepping stone toward universal quantum computing. Extended versions of this "digital-analog-digital" approach have been proposed as pathways to solve classically intractable problems and achieve quantum supremacy.
• Hardware Efficiency: The method utilizes the native hardware capabilities (flux biases, anneal offsets) without requiring a complete redesign of the processor, making it immediately applicable to current and near-future commercial devices.

In conclusion, the paper demonstrates that quantum annealing processors can transcend their traditional limitations to perform high-fidelity, coherent, many-body quantum simulations with arbitrary control, opening new avenues for quantum simulation and computation.