Topological Dynamical Decoupling with Complete Pulse Error Cancellation
Nayden P. Nedev 1 and Nikolay V. Vitanov 1 · 2025
Read Topological Dynamical Decoupling with Complete Pulse Error Cancellation by Nayden P. Nedev 1 and Nikolay V. Vitanov 1 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-17692.txt.
Introduction
Nayden P. Nedev and Nikolay V. Vitanov design pulse sequences that resist control errors in a 2025 arXiv paper. Quantum computers are steered by carefully timed pulses. Small errors in pulse area are systematic, meaning they repeat the same way, and they limit control fidelity across many qubit platforms. Their method is called twinned dynamical decoupling.
Each sequence is paired with a twin shifted by a half turn in phase. That pairing cancels common pulse area errors to all orders at exact resonance. The pulse phases within each twin are also chosen to suppress detuning errors as far as possible. A simple formula gives the phases for any sequence length. Tests on IBM and IQM quantum processors matched theory. The plateaus held up better than standard decoupling protocols.
Catalog record
| CREATOR | Nayden P. Nedev 1 and Nikolay V. Vitanov 1 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 17 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-17692
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Topological Dynamical Decoupling with Complete Pulse Error Cancellation Nayden P. Nedev 1 and Nikolay V. Vitanov 1 Abstract Systematic pulse errors remain a major obstacle to high-fidelity quantum control. We present a new family of dynamical decoupling sequences, denoted T n n , that achieve exact cancellation of pulse area errors to all orders by enforcing a simple topological phase condition. Unlike some conventional composite sequences, T n n requires no numerical optimization and admits closed-form analytic phases for arbitrary sequence length, while providing substantial robustness to detuning as well. We demonstrate these sequences on superconducting transmon qubits from both IBM Quantum processor ibm_torino and IQM Quantum processor Garnet, observing population plateaus in close agreement with theory. These results establish a new paradigm for hardware-efficient error suppression, broadly applicable across quantum computing, sensing, and memory platforms. Dynamical decoupling (DD) is a cornerstone technique for mitigating decoherence in quantum systems, offering a practical and hardware-efficient means of prolonging coherence without the overhead of full quantum error correction Viola and Lloyd 1998 ; Viola et al. 1999 ; Viola and Knill 2005 ; Khodjasteh and Lidar 2005a ; Khodjasteh and Lidar 2007 ; Lidar 2012 ; Yang et al. 2011 ; Suter and Álvarez 2016 . Its utility has been repeatedly demonstrated on noisy intermediate-scale quantum devices (NISQ), where experimental constraints preclude large-scale fault-tolerant schemes Álvarez et al. 2010 ; Peng et al. 2011 ; Ezzell et al. 2023 ; Pokharel and Lidar 2024 ; Barthel et al. 2023 ; Xu et al. 2025 ; Tong et al. 2025 . The essence of the method lies in the application of carefully timed and phased control pulses to average unwanted environmental interactions, thus suppressing errors due to both dephasing and relaxation processes. Although traditional DD protocols, such as CPMG Carr and Purcell 1954 ; Meiboom and Gill 1958 , UDD Uhrig 2007 , KDD Souza et al. 2012 , and XY-based sequences Maudsley 1986 ; Khodjasteh and Lidar 2005b ; West et al. 2010 , compensate for errors up to a given order in system–environment couplings, their accuracy remains inherently finite, especially in the presence of systematic control imperfections such as pulse area errors. Dynamical decoupling techniques used to combat dephasing in NMR are known in the literature as rephasing Bernstein et al. 2004 . Such composite pulse methods from NMR have shown that phase engineering can mitigate these errors, but typically only in a finite error range. Here, we present an exact analytic construction of topological dynamical decoupling sequences with infinite order suppression of systematic pulse area errors, while simultaneously achieving significant robustness to detuning errors too. This class of sequences, denoted T n , is derived from a simple criterion imposed on the phases, and is demonstrated on superconducting transmon qubits using IBM’s ibm_torino and IQM’s Garnet processors. By “topological”, we mean that the cancellation of errors is not achieved by delicate fine-tuning, but is enforced by a global constraint on the control phases. Unlike conventional DD, where robustness is finite-order and depends on perturbative expansions, here this topological condition ensures complete error cancellation, independent of microscopic details. The dynamics of a qubit due to an external pulsed interaction is described by a propagator U , which is conveniently parameterized with the complex Cayley-Klein parameters a and b as U=\begin{bmatrix}a&b\\ -b^{*}&a^{*}\end{bmatrix}. (1) For exact resonance ( \Delta=0 ), the Schrödinger equation is solved exactly for any \Omega(t) . Then the Cayley-Klein parameters depend only on the pulse area A=\int_{t_{i}}^{t_{f}}\Omega(t)dt , a=\cos(A/2),\quad b=-i\sin(A/2), (2) with \Omega(t) assumed real. The transition probability is p=|b|^{2}=\sin^{2}(A/2) , and hence, complete population inversion occurs for A=\pi ( \pi pulses) or odd-integer multiples of \pi . This inversion is sensitive to variations in the pulse area: a small deviation from the value \pi , i.e., A=\pi(1+\xi) , causes an error in the inversion of order \mathcal{O}(\xi^{2}) , p=1-\pi^{2}\xi^{2}/4+\mathcal{O}(\xi^{4}). (3) A constant phase shift \phi in the Rabi frequency, \Omega(t)\to\Omega(t)e^{i\phi} , is imprinted into the propagator ( 1 ) as U(A,\phi)=\begin{bmatrix}a&be^{i\phi}\\ -b^{*}e^{-i\phi}&a^{*}\end{bmatrix}. (4) A sequence of N pulses, each with area A_{k} and phase \phi_{k} , produces the propagator U^{(N)}=U_{\phi_{N}}(A_{N})U_{\phi_{N-1}}(A_{N-1})\cdots U_{\phi_{1}}(A_{1}). (5) The objective of any rephasing sequence is to produce the identity propagator, \mathbf{I}=\left[\begin{array}[]{cc}1&0\\ 0&1\end{array}\right], (6) as closely as possible, i.e., by compensating deviations in the relevant parameters to the highest possible order. We define the rephasing error \epsilon as the distance between the actual propagator \mathbf{U} and the target propagator \mathbf{I} : \epsilon=|U_{11}-1|+|U_{12}|. (7) The propagator ( 4 ) can be written in the form U(A,\phi)=e^{-\frac{iA}{2}M(\phi)}, (8) where M(\phi)=\begin{bmatrix}0&e^{i\phi}\\ e^{-i\phi}&0\end{bmatrix}, (9) and A=\pi(1+\xi) . The objective of the present work is to find a set of phases \phi_{k} for which all systematic errors in the pulse area are compensated. This leaves us with the sequence of the form ( 10 ) that we require to be equal to the identity for any deviation \xi from an ideal \pi pulse. The deviation \xi is consistent for all pulses, U^{(n)}=\prod_{k=1}^{n}e^{-i\tfrac{A}{2}M(\phi_{k})}=I (10) To fulfill Eq. ( 10 ), we need to impose two conditions on the phases \phi_{k} and M(\phi_{k}) matrices, respectively, 1. All pairs of neighboring M(\phi_{j}),M(\phi_{k}) matrices need to commute. 2. The phases must be chosen such that \sum_{k=1}^{n}M(\phi_{k})=0. If the two conditions are fulfilled, then U^{(n)}=\exp\!\left(-i\tfrac{A}{2}\sum_{k=1}^{n}M(\varphi_{k})\right)=I. (11) From condition 1 we find \left[M(\phi_{k}),M(\phi_{j})\right]=2i\sin(\phi_{k}-\phi_{j})\sigma_{z}, (12) where \sigma_{z} is the z Pauli matrix. This result shows us that any two matrices M(\phi_{j}),M(\phi_{k}) commute if and only if \sin(\phi_{k}-\phi_{j})=0 . Therefore, \phi_{k}-\phi_{j} must be an integer multiple of \pi . Condition 2 requires \sum_{k=1}^{n}M(\varphi_{k})=\begin{bmatrix}0&\sum_{k=1}^{n}e^{i\varphi_{k}}\\ \sum_{k=1}^{n}e^{-i\varphi_{k}}&0\end{bmatrix}=0 (13) Therefore, exact cancellation requires \sum_{k=1}^{n}e^{i\varphi_{k}}=0. (14) Now, suppose that the two consecutive phases are chosen such that condition one is fulfilled: \phi_{k+1}=\phi_{k}\pm\pi . Then, using the identity e^{\pm(\phi\pm\pi)}=-e^{\pm\phi} , their contributions to the exponential sum ( 14 ) are e^{i\phi_{k}}+e^{i\phi_{k+1}}=e^{i\phi_{k}}+e^{i\phi_{k}+\pi}=e^{i\phi_{k}}-e^{i\phi_{k}}=0 . Hence, each such pair cancels exactly. This leaves us with the following |\phi_{k}-\phi_{j}|=\pi\quad(mod\;2\pi) (15) as the only condition that we need to impose for any pair of neighboring pulses in the sequence to achieve compensation for any systematic error in pulse area. The presented derivation assumes that the system is in resonance ( \Delta=0 ). Still, we find that the results also hold for models where the Rabi frequency is an even function, while the detuning is an odd function of time, such as the Landau-Zener and Allen-Eberly models. Figure 1: Theoretical predictions (top) and measured results from IQM Garnet (bottom) of the |0\rangle -state population for several T n sequences and CPMG with simultaneous deviations in both detuning and pulse area. The Rabi frequency is \Omega=2\pi\times 25 MHz and the pulse duration is 20 ns. Figure 2: Measured |0\rangle -state population for several T n sequences on ibm_torino. Here, CPMGn denotes the sequence with the same number of pulses, but implemented without the phases for reference. Using the newly derived condition, we present a new class of topological DD sequences which cancel exactly any systematic pulse area errors, while simultaneously providing substantial compensation in detuning \Delta as shown in Fig. 1 . We start by fixing the sequence’s structure to fulfill ( 15 ). Then, to determine the actual phases of the sequences, the elements of the resulting propagator \boldsymbol{U}^{(n)} must satisfy the following conditions: [\partial^{k}_{\Delta}U^{(n)}_{11}]_{\Delta=0}=0 and [\partial^{k}_{\Delta}U^{(n)}_{12}]_{\Delta=0}=0 , to the