Software-based compensation of AC-line-induced control errors in qubits and qudits

This paper demonstrates a software-based compensation protocol that measures and corrects reproducible AC mains-induced control errors in trapped-ion qubits and qudits, significantly improving gate fidelity and algorithm success rates without requiring additional hardware.

The Gist

Imagine you are trying to conduct a very delicate orchestra, where every musician (a quantum particle) must play a note at the exact right moment and with the exact right pitch. If even a tiny fraction of a second passes, or if the pitch wavers slightly, the music turns into noise, and the performance fails.

This paper describes a clever "software fix" for a specific problem that plagues these quantum experiments: the hum of the building's electricity.

The Problem: The Unwanted Hum

In many laboratories, the power lines that run through the walls (the AC mains) create a tiny, rhythmic vibration in the magnetic field. It's like a faint, invisible drumbeat happening 60 times a second (in North America) or 50 times (in Europe).

For a quantum computer, this is a nightmare. It causes the "notes" (energy levels) of the particles to wobble up and down in perfect time with the power grid.

• The Analogy: Imagine trying to tune a guitar string, but every time you pluck it, the air pressure in the room changes slightly in a rhythmic pattern. The string's pitch keeps shifting just as you are trying to play it. If you don't account for this, your music will sound out of tune, no matter how skilled you are.

Usually, scientists try to fix this with heavy hardware: thick metal shields, special coils to cancel the magnetic field, or better power supplies. This is expensive, bulky, and hard to install.

The Solution: The "Smart Conductor"

The researchers at the University of Waterloo realized that this "electricity hum" isn't random chaos. It is predictable. Because it is tied to the power grid, it happens at the exact same time every single time the experiment starts.

Instead of building a wall to stop the noise, they built a software shield.

1. Listening: First, they measured exactly how the magnetic field wiggles in sync with the power line. They created a "map" of the noise.
2. Predicting: Because the noise is repeatable, they know exactly what the magnetic field will be doing at any given millisecond after they start the experiment.
3. Counter-Acting: They programmed their control system to do the opposite.
   • The Pitch Fix: If the magnetic field tries to make the particle's pitch go up, the software instantly tells the laser to tune the pitch down by the exact same amount.
   • The Timing Fix: If the noise causes the particle to get "out of step" (accumulate extra phase) between notes, the software adjusts the timing of the next note to cancel that drift.

It's like a conductor who hears the room's echo and instantly adjusts the orchestra's tempo so that, to the audience, the music sounds perfectly in sync, even though the room is noisy.

The Results: From Chaos to Clarity

The team tested this "software conductor" on a trapped ion (a single atom held in place by lasers).

• The Before: Without the fix, the quantum gates (the basic operations) were jittery. When they tried to measure how well the computer worked, the results were messy and unreliable. It was like trying to measure the speed of a car while it was driving over a bumpy road; the data looked like random noise.
• The After: With the software compensation turned on, the jitter vanished. The "bumpy road" was smoothed out.
  • They improved the accuracy of their basic gates to 99.93%.
  • They tested a more complex task (a 16-level "qudit" version of a famous algorithm called Bernstein-Vazirani). Without the fix, the computer only got the right answer 10% of the time (basically guessing). With the fix, it got the right answer 70% of the time.

Why This Matters (According to the Paper)

The paper emphasizes that this is a cheap and easy solution.

• No New Hardware: You don't need to buy new shields or coils.
• Software Only: It just requires updating the code that tells the lasers when to fire and what frequency to use.
• Universal: It works for different types of quantum particles and different sizes of quantum systems (from simple two-level bits to complex multi-level "qudits").

In short, the researchers found that instead of fighting the noisy power grid with heavy hardware, they could simply "dance" with it. By predicting the noise and adjusting their steps in real-time, they turned a source of error into a manageable part of the routine, making their quantum computer much more reliable.

Technical Summary

Technical Summary: Software-based compensation of AC-line-induced control errors in qubits and qudits

Problem Statement
Precision quantum-control experiments are frequently degraded by structured, coherent disturbances originating from AC mains power lines (50/60 Hz and harmonics). Unlike stochastic noise sources such as broadband dephasing or relaxation, these line-synchronous perturbations are phase-coherent with the laboratory power grid. They manifest as reproducible, time-dependent errors in time-domain measurements and benchmarking data. In trapped-ion systems, these disturbances typically couple via magnetic pickup, ground loops, or power-supply ripple, causing magnetic-field-induced shifts in energy level structures. This results in time-dependent detunings between ion transitions and local oscillators, as well as uncontrolled phase accumulation on superpositions. These errors violate the assumptions of time-independent, gate-independent noise required for standard characterization techniques like randomized benchmarking, leading to non-exponential decay, sequence-dependent artifacts, and unreliable fidelity estimates. Existing mitigation strategies often rely on hardware modifications (passive shielding, active cancellation coils) or closed-loop drift control, which can be limited by bandwidth, latency, and the complexity of implementing additional sensors or shielding.

Methodology
The authors propose a software-defined compensation framework that leverages the reproducibility of line-synchronous noise rather than attempting to suppress it at the source. The core premise is that if a perturbation is reproducible with respect to the mains phase, its deterministic contribution can be measured in a line-triggered frame and corrected by updating the control sequences in software.

1. Calibration and Measurement: The experiment is synchronized to a line trigger. The instantaneous detuning $\delta\omega(t)$ and accumulated phase $\Phi_{AC}(t)$ are measured using Ramsey-type sequences at various delays within the line cycle. The dominant perturbation is identified as a magnetic-field fluctuation $\Delta B(t)$, which is calibrated by measuring transitions with known magnetic-field sensitivities ($\kappa$). The waveform is fitted to a harmonic model of the line frequency.
2. Compensation Protocol: The control system updates the frequency and phase of the programmed oscillator tone for each pulse $j$ at time $t_j$:
   • Frequency Update: The oscillator frequency is adjusted to match the instantaneous transition frequency: $\omega^{(prog)}_{L,j} = \omega_0 + \kappa \Delta B(t_j)$. This ensures resonance during the pulse.
   • Phase Update: The phase of the oscillator is updated to cancel the phase accumulated by the qubit states due to the detuning between pulses. The correction is calculated as $\phi^{(comp)}_j = \int_0^{t_j} \delta\omega(t') dt' - t_j \delta\omega(t_j)$.
3. Extension to Qudits: The framework is generalized to multilevel systems (qudits). Since different transitions within a qudit Hilbert space have different magnetic-field sensitivities, the compensation is applied pulse-by-pulse, scaling the measured magnetic-field waveform by the specific sensitivity $\kappa_{mn}$ of each addressed transition $|m\rangle \leftrightarrow |n\rangle$.

Key Contributions and Results
The authors demonstrate this protocol using a single trapped $^{137}\text{Ba}^+$ ion confined in a linear Paul trap, controlling optical transitions between the $6s\,^2S_{1/2}$ and $5d\,^2D_{5/2}$ manifolds.

• Detuning and Phase Suppression:
  • Detuning: The calibrated AC line contribution to the detuning was reduced by a factor of $21(9)$, and the fitted harmonic amplitude was reduced by $8(1)\times$.
  • Phase: The fitted AC phase amplitude was reduced below the measurement uncertainty (a reduction factor of $90\times$ in matched-filter scale), effectively eliminating the deterministic phase error.
• Randomized Benchmarking:
  • On a magnetic-field-sensitive qubit ($\kappa \approx 3.2$ MHz/G), uncompensated line-synchronous errors caused large sequence-dependent fluctuations, rendering the standard randomized-benchmarking decay model unreliable.
  • With compensation enabled, these fluctuations were suppressed, allowing the recovery of a standard exponential decay. The extracted average gate fidelity improved from $99.78(3)\%$ to $99.93(1)\%$, comparable to a magnetic-field-insensitive qubit.
• Qudit Algorithm Performance:
  • The protocol was applied to a single-qudit Bernstein–Vazirani algorithm implemented on a 16-level qudit ($d=16$).
  • Without compensation, the success probability was $10(7)\%$, consistent with random chance due to accumulated detuning and phase errors across the multilevel circuit.
  • With compensation enabled, the success probability increased to $70(9)\%$, demonstrating that the method preserves coherent phase relationships across complex, multilevel operations.

Significance and Claims
The paper claims that reproducible line-synchronous noise can be treated as a calibrated control-frame error and corrected without additional hardware. The approach is characterized as:

• Cost-effective: It requires no new equipment, system redesign, or reconfiguration.
• Accessible: It relies on fundamental prerequisites of quantum control: transition frequency determination, frequency/phase control, and line-phase triggering.
• Generalizable: It is applicable to most qubit and qudit platforms experiencing frequency or phase decoherence from line-synchronous sources.
• Compatible: It does not obstruct other noise suppression techniques (e.g., shielding) and can be combined with them.

The authors note that the technique's effectiveness is limited by the reproducibility of the signal and the precision of its characterization, which is constrained by the system's overall coherence times. They also acknowledge that the current implementation introduces dead time in the processor duty cycle due to fixed line-trigger synchronization, suggesting that continuous phase tracking could recover this time in future iterations. The work establishes that software-defined compensation can significantly enhance algorithmic performance and gate fidelity in the presence of structured environmental noise.