Trading Circuit Depth for Pulse Sparsity in Chromatic Dynamical Decoupling

AI summary

Overview

Research area: Quantum error suppression — multi-qubit dynamical decoupling (DD) for superconducting quantum processors, combining graph coloring with sign-matrix scheduling.

Technical level: Advanced. The paper is written for researchers familiar with average-Hamiltonian theory, Pauli noise models, and graph coloring; the theorems and proofs use linear algebra over $\mathbb{F}_2$.

Scope (one sentence): The paper introduces two single-axis chromatic DD sequences, Chromatic-Binary DD (CBDD) and Chromatic-Gray DD (CGDD), that trade an exponentially larger circuit depth ($2^C$) for substantially lower pulse density than Chromatic-Hadamard DD (CHaDD), and tests them on a 127-qubit IBM processor.

What This Paper Is About

As quantum processors grow, suppressing decoherence and crosstalk across many qubits at once becomes a bottleneck. Chromatic-Hadamard DD (CHaDD) solves this by coloring the hardware graph and assigning each color a row of a Hadamard matrix, so circuit depth scales with the number of colors $C$ rather than the number of qubits — but it pays for this with a high pulse density (a pulse repetition rate, PRR, above $1/4$, and equal to $2/3$ at $C=3$). This paper asks whether that depth budget can be spent differently to produce sparser pulse schedules, and whether sparser schedules actually preserve quantum states better on real hardware.

Key Contributions

1. Two new sequence families. CBDD (based on a binary sign matrix) and CGDD (based on a Gray-code column ordering) cancel the same Hamiltonian terms as CHaDD to first order, with $\mathrm{PRR} < 2/C$ and $\mathrm{PRR} = 1/C$ respectively, at the price of circuit depth $N = 2^C$ instead of $O(C)$. For $C>2$ these are special cases of non-minimum-depth CHaDD.

2. A minimal characterization of what a sign matrix must do. The authors show that only two properties matter: each color-assigned row must sum to zero (so that $\overline{H}_1 = H_1^x$), and every pair of assigned rows must be orthogonal (so that off-color crosstalk is suppressed). This yields the control-group structure $\mathcal{G} \cong \mathbb{F}_2^r$ and shows that minimum-depth CHaDD has $|\mathcal{G}| = N$.

3. Chromatic-Walsh DD (CWDD). Using that characterization, they identify the row assignment that minimizes the CHaDD pulse count, with $P = C(C+1)/2 + \lceil C/2 \rceil$ pulses, and the PRR bounds of Theorem 1 follow from it.

4. Selective $k$-body cancellation. Because the CBDD row indices are the standard basis vectors of $\mathbb{F}2^C$, CBDD cancels to first order every $k$-body term whose $Y$- and $Z$-type factors act on an odd number of qubits of at least one color — a property CHaDD row assignments need not have (e.g., a $YYY$ term survives the $C=3$ CHaDD average because the entrywise product of rows 1, 2, and 3 of $W\nu$ is the constant row).

Main Findings

• Formal decoupling guarantees. Theorem 1 (CHaDD): depth $C < N \le 2C$ with $\frac{1}{4} + \frac{1}{2C} \le \mathrm{PRR} < \frac{1}{2} + \frac{7}{6C}$ cancels, to first order in $\tau$, all terms in $H_G$ except $H_1^x$ and $H_2^{xx}$. Theorem 2 (CBDD): depth $N = 2^C$ with $\mathrm{PRR} < 2/C$ achieves the same cancellation. Theorem 3 (CGDD): the same with $\mathrm{PRR} = 1/C$.

• Pulse-count comparison. Per sequence repetition, CWDD uses $P = C(C+1)/2 + \lceil C/2 \rceil$ pulses at depth $N = 2^{\lfloor \log_2 C \rfloor + 1} \in O(C)$; CBDD uses $P = 2^{C+1}-2$ at $N = 2^C$; CGDD uses $P = 2^C$ at $N = 2^C$.

• Sparser is better only without robust pulses. In state-preservation experiments on the 127-qubit ibm_strasbourg QPU with the embedded $C=3$ coloring and six Pauli eigenstates, the less dense non-robust sequences outperform CHaDD at short times, in reverse order of their PRRs. Robust variants built from the universally robust UR4 pulse block largely erase this advantage.

• CGDD wins among robust sequences. Among the robust sequences, CGDD accumulates the fewest pulses over a fixed protection time.

• PRR predicts performance. Across non-robust $C=3$ and $C=5$ schedules, equal PRR on a fixed qubit subset is empirically associated with similar state preservation, even across different values of $C$ — supporting PRR as the practical figure of merit when coherent pulse errors dominate.

• A lower bound on pulse count within the full-group traversal. Any cyclic ordering of the $2^C$ distinct columns incurs at least one sign change per transition, so $P \ge 2^C$; the Gray order attains this. Sequences that traverse a smaller control group with repeated elements can go lower still — the four lowest-sequency nonconstant rows of $W_4$ give $P = 12 < 16$ at $C=4$.

• Hardware graphs considered. The IBM heavy-hex lattice has chromatic number $\chi = 2$; the paper also uses a distance-3 5-coloring and a proper 3-coloring of the active red/green/blue (RGB) subgraph, which is minimum for that subgraph and yields an effective kagome lattice. Spectator gray qubits are assigned the constant row and idle.

Methodology in Plain English

The authors treat error suppression as a scheduling problem. First, color the graph of qubits so that no two connected qubits share a color — connected qubits are the ones with unwanted always-on $ZZ$ couplings. Then let every qubit of a given color follow the same pattern of $X$ pulses, where the pattern is read off a row of a sign matrix: a pulse occurs wherever the row's sign flips between consecutive time steps.

The design question becomes: which sign matrix? The paper proves that two properties are enough — each row used must have equally many $+$ and $-$ entries, and any two used rows must be orthogonal. Hadamard matrices satisfy this, but so do larger matrices. Enlarging the matrix to $2^C$ columns gives many more time steps to work with, and the extra room lets the schedule be ordered so that qubits are pulsed far less often. Ordering the columns in Gray-code order, where consecutive columns differ in exactly one bit, gives the minimum possible number of pulses for a full traversal of the control group.

To test whether this matters in practice, the authors run state-preservation experiments on a 127-qubit IBM device, preparing each of the six Pauli eigenstates, applying the different DD schedules, and measuring how well the state survives. They compare both non-robust sequences and robust versions built from the UR4 pulse block, which is designed to cancel coherent pulse errors.

Why This Matters

Impact on research. The paper reframes chromatic DD design around two separable resources — circuit depth and pulse density — and shows experimentally that when pulse imperfections dominate, pulse density is the better predictor of performance. That gives hardware practitioners a simple, sequence-independent number to optimize instead of benchmarking full sequences empirically. The minimal sign-matrix conditions (zero-sum rows, pairwise orthogonality) also open a much larger design space than the Hadamard/orthogonal-array framework that has dominated multi-qubit DD.

Real-world applications (as implications of the approach, not results reported here):

• Suppressing idle-time dephasing and always-on $ZZ$ crosstalk in fixed-coupler superconducting processors.
• Supporting quantum error correction and fault-tolerance demonstrations, where DD is inserted into idle windows and must not interfere with syndrome extraction.
• Pulse-level benchmarking and calibration on cloud-accessible QPUs, where pulse budget and duty cycle are constrained.
• Choosing schedules on architectures whose connectivity graph has small chromatic number, where depth is already independent of qubit count.

Industry relevance. The experiments use a commercially available 127-qubit IBM device and standard single-qubit $X$ pulses only, so the methods require no special hardware or multi-qubit control — a low barrier to adoption for groups running cloud QPUs.

Future Directions

• Multi-axis and higher-order sequences. The authors explicitly restrict attention to single-axis, first-order decoupling and leave multi-axis and higher-order chromatic sequences outside the scope of the work.
• Depth becomes prohibitive as $C$ grows. The $2^C$ depth is exponential in the number of colors, so finding sign matrices that approach the low PRR of CGDD at intermediate depth — the paper notes repeated-element sequences over smaller control groups as one such route — is an open design problem.
• Reducing $k$-body cancellation to a code design problem. The observation that CBDD cancels a broader class of $k$-body terms than CHaDD, together with the recent classical-code construction of Ref. [31], suggests systematic selective cancellation as a target.
• Testing beyond the heavy-hex lattice and beyond $C=3$. The hardware comparison uses the embedded $C=3$ coloring (with $C=5$ schedules appearing in the equal-PRR comparison), so whether the PRR-as-figure-of-merit relation holds on other connectivities and larger $C$ is not established.

Target Audience

Graduate students and researchers in quantum information and quantum control, particularly those working on error mitigation for superconducting hardware, pulse-level DD sequence design, or average-Hamiltonian theory. It is also useful for experimentalists on multi-qubit platforms who need to choose an idle-time DD schedule under a fixed pulse budget. Readers without a background in dynamical decoupling and basic linear algebra over $\mathbb{F}_2$ will find the theorems and proofs demanding.

Authors’ abstract

Suppressing decoherence and crosstalk systematically across large networks of qubits is a pressing concern as qubit counts increase rapidly. General multi-qubit dynamical decoupling (DD) has historically been based on Hadamard matrices and orthogonal arrays, with instantaneous-pulse sequences whose circuit depth scales linearly with the number of qubits. Chromatic-Hadamard DD (CHaDD) improves upon these methods by properly coloring the hardware graph and assigning to each color a row of a Hadamard matrix, resulting in a circuit depth that scales linearly with the number of colors $C$, which can be as low as the chromatic number of the graph. Here, we explore the tradeoff between circuit depth and the pulse repetition rate (PRR), the average fraction of time steps in which a qubit is pulsed, i.e., a measure of pulse density. We introduce two single-axis sequences based on binary and Gray matrices, Chromatic-Binary DD (CBDD) and Chromatic-Gray DD (CGDD), which, for $C&gt;2$, are special cases of non-minimum-depth CHaDD and achieve PRR $&lt;2/C$ and PRR $=1/C$, respectively, compared with a PRR above $1/4$ for CHaDD, at the price of a circuit depth of $2^C$. In state preservation experiments on a $127$-qubit IBM processor with $C=3$, the less dense sequences outperform CHaDD at short times when the pulses are not robust; this advantage is largely eliminated by robust versions designed to compensate for coherent pulse errors, and among the robust sequences, CGDD uses the fewest pulses over a fixed protection time. Across non-robust $C=3$ and $C=5$ schedules, equal PRR on a fixed qubit subset is empirically associated with similar state preservation. Thus, the PRR serves as a practical figure of merit for chromatic DD whenever coherent pulse errors dominate.