⚛️ quantum physics

Dynamical Decoupling using Universal Optimal Tracking

This paper introduces a universal optimal tracking approach for designing dynamical decoupling sequences that dynamically compensates for control imperfections and suppresses error accumulation, as experimentally validated on a superconducting-qubit platform.

Original authors: Amit Devra, Emanuel Malvetti, Niklas J. Glaser, Abhishek Agarwal, Ivan Rungger, Santana Lujan, Max Werninghaus, Stefan Filipp, Leo Van Damme, Steffen J. Glaser

Published 2026-06-23

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Amit Devra, Emanuel Malvetti, Niklas J. Glaser, Abhishek Agarwal, Ivan Rungger, Santana Lujan, Max Werninghaus, Stefan Filipp, Leo Van Damme, Steffen J. Glaser

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Problem: The "Drifting" Qubit

Imagine you are trying to keep a spinning top perfectly balanced on a table. In the real world, the table isn't perfectly flat, and there are tiny bumps and vibrations (noise). If you just let the top spin, it will eventually wobble and fall over.

In quantum computers, the "spinning top" is a qubit (the basic unit of information). When the computer isn't actively using a qubit (an "idle" period), environmental noise causes it to drift away from its intended state. This is called decoherence, and it destroys the calculation.

The Old Solution: The "Broken Record"

To stop the top from falling, scientists use a technique called Dynamical Decoupling (DD). Think of this as giving the top a series of precise taps to keep it upright.

For years, the standard method was like playing a broken record:

You play a short, perfect sequence of taps (a "pulse block").
You repeat that exact same sequence over and over again for as long as the qubit needs to sit idle.

The Flaw: In the real world, your hand isn't perfect. Every time you tap the top, you might miss by a tiny fraction. If you repeat the same "imperfect" tap 40 times, those tiny misses add up. By the end of the long idle time, the top has drifted significantly, even though you kept tapping it. The old method assumes the noise is static and doesn't account for the fact that your "taps" might get slightly worse with every repetition.

The New Solution: "Optimal Tracking" (U-TRACK)

The authors of this paper introduced a new method called U-TRACK. Instead of just playing a broken record, imagine a smart coach watching the spinning top in real-time.

Here is how it works:

The Checkpoints: Instead of just checking if the top is upright at the very end, the coach checks its position after every single tap (or small group of taps).
The Correction: If the coach sees the top drifting slightly to the left after the first tap, the next tap isn't just a repeat of the first one. It is slightly adjusted to push the top back to the center.
The Result: The sequence of taps is no longer a repeating pattern. It is a unique, custom choreography where every move compensates for the mistakes of the previous move.

The paper calls this "Universal Optimal Tracking." It ensures that the qubit stays close to its starting point not just at the finish line, but at every single step along the way.

The Experiments: What They Found

The researchers tested this on a real superconducting quantum computer (a type of hardware that uses tiny electrical circuits to act as qubits). They compared three methods:

UR40: One giant, complex sequence designed to work perfectly only at the very end.
XY4: The old "broken record" method (repeating a short block).
U-TRACK4: The new "smart coach" method.

The Results:

Against Static Errors (The "Bumpy Table"): When the noise was constant (like a table that is always slightly tilted), U-TRACK4 kept the qubit stable much better than the others. The "broken record" (XY4) slowly drifted away, and the giant sequence (UR40) was great at the end but let the qubit wander wildly in the middle.
Against Changing Noise (The "Shaking Room"): They also simulated a scenario where the noise changed randomly, like the room shaking unpredictably. Here, U-TRACK4 shined. Because it corrected errors constantly, it could handle sudden changes in the environment much better than the other methods. If the noise suddenly jumped, the "broken record" failed, but the "smart coach" adapted.

The Hidden Pattern: The "XUR" Discovery

One of the most interesting findings was what the computer "invented" when it designed these new sequences.

The researchers expected the computer to create a random, messy pattern of taps. Instead, the computer naturally converged on a specific, elegant structure they call XUR.

Think of it like a dance routine. The computer didn't just pick random moves; it discovered a specific mix of two known dance styles (called MLEV and XY types) and blended them together perfectly.
This is significant because this specific blend had never been used before. It suggests that nature (or math) has a "sweet spot" for stability that the computer found on its own.

Why This Matters

This paper doesn't claim to fix quantum computers entirely, but it offers a better way to protect them while they wait.

Hardware Agnostic: This method works on any type of quantum hardware, not just the one they tested.
Short and Robust: It allows for short, efficient sequences that are very hard to break, which is crucial for real-world quantum computers that are currently very noisy.
Open Source: The authors released the software they used, so other scientists can use this "smart coach" to design their own protection sequences.

In summary: The paper introduces a smarter way to protect quantum bits from noise. Instead of blindly repeating the same imperfect actions, the new method constantly checks and corrects the path, keeping the quantum state stable even when the environment is messy or changing.

Technical Summary: Dynamical Decoupling using Universal Optimal Tracking

Problem Statement
Quantum systems are inherently fragile, suffering from decoherence due to environmental noise. Dynamical Decoupling (DD) is a standard technique to suppress these errors by applying control pulses to average out unwanted system-environment interactions. However, conventional DD sequences face a critical limitation: they typically rely on the periodic repetition of a short pulse block (e.g., XY4). If the pulse block contains imperfections (such as detuning or amplitude errors), the net propagator deviates from the identity operator. Consequently, residual errors accumulate across successive repetitions, degrading state recovery over long idle times. While "Universal Robust" (UR) sequences offer robustness against parameter deviations, they often require long sequences optimized as a single block, which fails to refocus the state at intermediate times, making them sensitive to mid-sequence parameter jumps or time-dependent noise. Existing solutions like supercycles (e.g., XY64) improve robustness but are often restricted to specific block lengths and phase patterns.

Methodology
The authors introduce U-TRACK, a new family of DD sequences based on Universal Optimal Tracking. This approach generalizes the tracking concept—originally developed for NMR to decouple heteronuclear spins—to a fully state-independent setting.

Core Concept: Unlike conventional methods that optimize only the final state or repeat a fixed block, U-TRACK enforces that the qubit's evolution approximates the identity operator not just at the end of the sequence, but also at predefined intermediate "waypoints" (after each pulse block). This dynamic compensation prevents error accumulation across cycles.
Optimization Framework: The pulse phases for each block are optimized using a hybrid algorithm combining:
1. GRAPE (Gradient Ascent Pulse Engineering): To efficiently compute local gradients.
2. Genetic Algorithm: To perform a structured global search across the pulse space, avoiding suboptimal local minima.
Flexibility: The framework supports an arbitrary number of blocks ( $M$ ) and an arbitrary number of pulses ( $N$ ) per block. The study focuses on the $N=4$ case (U-TRACK4) for demonstration.
Implementation: The sequences were numerically simulated and experimentally validated on a superconducting transmon qubit platform (Walther-Meißner-Institut and IQM devices).

Key Contributions

Universal Optimal Tracking: A design principle that creates non-periodic pulse structures by dynamically correcting residual errors at every block, ensuring robustness even as the number of blocks increases.
Discovery of XUR Structure: The optimization naturally converges to a specific phase pattern termed XUR, which is a structured mixture of MLEV-type and XY-type blocks within a single sequence. This combination had not been previously considered in literature, where these types were studied separately.
Optimization Efficiency: The authors demonstrate that initializing the optimization with an XUR ansatz (rather than random phases) significantly reduces convergence time and improves solution quality.
Hardware-Agnostic Toolkit: The optimization framework is released as an open-source Julia package, UTrack.jl, providing ready-to-use sequences for various lengths.

Results

Static Imperfections: Experimental and numerical results show that U-TRACK4 maintains high process fidelity across a broad range of static detuning ( $\delta$ ) and amplitude errors ( $\epsilon$ ). Unlike XY4, which shows fidelity degradation due to error accumulation over time, and UR40, which lacks intermediate refocusing, U-TRACK4 suppresses error accumulation while maintaining regular refocusing.
Time-Dependent Noise (Random Telegraph Noise): Under random telegraph noise (RTN) modeling two-level fluctuators:
- In the high-frequency regime ( $\lambda \gtrsim 10^3$ MHz), all sequences perform similarly as fluctuations average out.
- In the low-frequency and quasi-static regimes, U-TRACK4 and XY4 significantly outperform UR40. UR40 is highly sensitive to mid-sequence parameter jumps, leading to fragmented high-fidelity regions.
- Comparison with XY4: In the low-frequency regime, U-TRACK4 generally matches or slightly outperforms XY4, particularly when detuning amplitudes are high, where XY4 exhibits sharp fidelity degradation.
Robustness to Parameter Jumps: Simulations of a system jumping to new parameters mid-sequence confirm that U-TRACK4 retains a centered, high-fidelity region, whereas UR40's performance degrades severely.

Significance and Claims
The paper claims that U-TRACK establishes a practical and hardware-agnostic approach for designing short, robust DD sequences suitable for noisy quantum devices. By extending the tracking concept to a universal, state-independent setting, the method overcomes the trade-off between periodic refocusing (which accumulates errors) and long-block optimization (which lacks intermediate stability). The authors emphasize that the resulting sequences are not merely repetitions of known supercycles but represent a novel structural combination (XUR) that offers superior robustness against both static imperfections and time-dependent noise. The work suggests that such dynamically optimized sequences could become essential components for enhancing the fidelity of quantum algorithms in scalable quantum hardware.