The Impact of Markovian Errors on Random Circuits

arXiv:2610.31500 · quant-ph · Submitted 2026-10-22 · Read on arXiv

Listen

Transcript

Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.

Kai: We're looking at the paper: "The Impact of Markovian Errors on Random Circuits".

Mira: Here's the gist of the paper. The gist The paper develops a perturbative, gauge-invariant interpretation of Randomized Benchmarking (RB) error rates for small gate-local Markovian noise by factorizing noisy gates and showing that Pauli-stochastic errors contribute linearly while Hamiltonian errors contribute through a gauge-invariant quadratic form.

Kai: How it works The authors study RB and related random-circuit benchmarks under small gate-local Markovian noise by factorizing each noisy gate into its ideal action followed by the exponential of a small error generator expressed in the Lindbladian basis of Hamiltonian, Pauli-stochastic, Paulicorrelation, and active sectors The goal is to supply such an interpretation in the perturbative regime of small gate-local Markovian noise by factoring each noisy gate into its ideal action followed by the exponential of a small error generator expressed in the Lindbladian basis of Hamiltonian, Pauli-stochastic, Paulicorrelation, and active sectors.

Mira: They use the FOGI framework to identify the gauge-invariant coordinate layer and study the RB L matrix for sequence-asymptotic unitary 2-designs.

Kai: Linear and Quadratic Contributions The main results show that for every fixed sampling distribution with the required second-moment gap, Pauli-stochastic error generators are the only generators that contribute linearly to the RB decay parameter Hamiltonian generators contribute through a gauge-invariant quadratic form.

Paper summary: Kai: The main idea here is they factorize each noisy gate into its ideal action plus an exponential term from a small error generator. This lets them perturb the RB L matrix, which models that decay.

Mira: And they find something specific: for any low-error gate set that fits the standard requirements for randomized benchmarking, it turns out that Pauli-stochastic error rates contribute linearly to that decay parameter.

Lev: But Hamiltonian errors don't just add up simply; they show up through a gauge-invariant quadratic form.

Kai: So, while one type of error is linear and easy to track, the other kind comes in as a quadratic term, which is more complex mathematically.

Mira: They are saying that Pauli-correlation and active generators don't show up in that linear part; they only appear at higher orders.

Lev: For the single-qubit native gate set, specifically Xpi/two Ypi/two they even derive this quadratic form explicitly using three first-order gauge-invariant coordinates <ref:2610.31500#pg1,For the single-qubit native gate set>.

Kai: That means we can translate the error into something more physically meaningful, like these intrinsic over-rotations and an axis tilt error.

Mira: And this whole analysis explains how the direct RB decay rate depends on the sampling distribution and how compilation affects Clifford RB because of these new terms.

Lev: It’s a way to turn those abstract decay rates into functions of elementary errors we can actually measure in the lab.

Conclusion: Kai: So, looking at "The Impact of Markovian Errors on Random Circuits," this paper is really giving us a better way to read the results we get from randomized benchmarking.

Mira: They’re showing that when noise is gate-dependent, you don't just get one single number for error; you get two distinct parts: a linear part from the Pauli stochastic errors and a quadratic part from the Hamiltonian errors.

Lev: That quadratic form simplifies nicely for native gates to three specific geometric parameters, which helps us understand how much we need to worry about compilation differences in Clifford circuits versus direct benchmarking.

Kai: It means that instead of just saying "our gate fidelity is X," we can start saying "the decay rate is determined by the linear Pauli errors plus this quadratic correction based on these three coordinates."

Mira: This moves us away from just looking at a single averaged number toward understanding the underlying structure of the noise itself.

Lev: It’s a way to map out exactly where hardware imperfections are hurting our performance, telling us whether we should focus on controlling those specific intrinsic over-rotations or something else.

JILA · Quantum Performance Laboratory, Sandia National Laboratories

quant-ph

Submitted: 2026-10-22

Updated: 2026-10-22

Comments: 15 pages, 3 figures

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 89/100

The gist: The gist The paper develops a perturbative, gauge-invariant interpretation of Randomized Benchmarking (RB) error rates for small gate-local Markovian noise by factorizing noisy gates and showing that

Key concepts

Randomized Benchmarking (RB): RB measures the quality of quantum gates by running random circuits and observing how quickly the signal decays. The paper focuses on interpreting this decay rate when the noise affecting each gate is small and local, which is common in real quantum hardware.
Pauli-stochastic errors: These are types of noise that affect a gate in a way that can be described by Pauli operators (like X, Y, Z). The paper finds that these specific errors have a direct, linear impact on the overall RB decay parameter.
Hamiltonian error rates: These errors relate to inaccuracies in the actual evolution of the quantum system's Hamiltonian. Unlike stochastic errors, these do not contribute linearly to the RB decay but instead influence it through a quadratic relationship that is independent of certain local gauge shifts.

Summary

The gist The paper develops a perturbative, gauge-invariant interpretation of Randomized Benchmarking (RB) error rates for small gate-local Markovian noise by factorizing noisy gates and showing that Pauli-stochastic errors contribute linearly while Hamiltonian errors contribute through a gauge-invariant quadratic form.

How it works

The authors study RB and related random-circuit benchmarks under small gate-local Markovian noise by factorizing each noisy gate into its ideal action followed by the exponential of a small error generator expressed in the Lindbladian basis of Hamiltonian, Pauli-stochastic, Paulicorrelation, and active sectors The goal is to supply such an interpretation in the perturbative regime of small gate-local Markovian noise by factoring each noisy gate into its ideal action followed by the exponential of a small error generator expressed in the Lindbladian basis of Hamiltonian, Pauli-stochastic, Paulicorrelation, and active sectors. They use the FOGI framework to identify the gauge-invariant coordinate layer and study the RB L matrix for sequence-asymptotic unitary 2-designs.

Linear and Quadratic Contributions

The main results show that for every fixed sampling distribution with the required second-moment gap, Pauli-stochastic error generators are the only generators that contribute linearly to the RB decay parameter Hamiltonian generators contribute through a gauge-invariant quadratic form. This quadratic form is unchanged when a linearized Hamiltonian gauge shift is added to h, so it depends only on the equivalence class of h in H/ΓH. The resulting formula explains both the sampling dependence of direct RB and the compilation dependence of Clifford RB.

Error Model Formalism

The authors introduce an error generator formalism based on four kinds of errors in the small Lindbladian error basis: Hamiltonian (H), stochastic (S), correlation (C), and active (A). The implemented gate is expressed as Gek = e Lk Gk, where Lk is the generator of the error process. They parameterize an imperfect gate by a vector⃗ϵ(k) of error rates and split these parameters into four sectors: vectors⃗h, ⃗s, ⃗c, and vec a of Hamiltonian, stochastic, correlation, active rates.

Results for the Native Gate Set

For the single-qubit native gate set of Xπ/2 and Yπ/2, they derive the quadratic form explicitly in terms of three first-order gauge-invariant (FOGI) coordinates: two intrinsic over-rotations and one relational axis-tilt error. The resulting formula explains the dependence of direct RB on the sampling distribution and the compilation dependence of Clifford RB.

Direct RB and Compilation Dependence

The analysis of direct RB shows that for coherent errors, r is a quadratic function of θX, θY, and ϕ. The behavior in direct RB is dominated by over-rotation errors when the sampling probability omegaX is close to 0 or 1, while near p = 1/2, the relational coordinate F3 has maximal sensitivity.

Clifford RB and Compilation

For Clifford RB, the mean-length normalized rate re/mκ varies significantly across compilations because it is sensitive to finer, gauge-invariant structure in the coherent error model filtered through the compilation. The quadratic description of coherent Clifford RB is given by Q(2)CRB = 1/9Πdg + 1/18Π⊥, which shows that to quadratic order, Clifford RB is insensitive to the gauge subspace and weights the three symmetric collective directions by 1/9.

Conclusion

The paper establishes that stochastic Pauli error rates contribute linearly to the RB decay parameter while Hamiltonian error rates contribute through a gauge-invariant quadratic form, which simplifies to three FOGI coordinates for the native gate set. This analytic formula explains both the sampling dependence of direct RB and the compilation dependence of Clifford RB.

How it works

The authors develop a perturbative, gauge-invariant interpretation of RB for small gate-local Markovian noise on random circuits by factorizing noisy gates and showing that Pauli-stochastic errors contribute linearly while Hamiltonian errors contribute through a gauge-invariant quadratic form. They use the FOGI framework to identify the gauge-invariant coordinate layer and study the RB L matrix for sequence-asymptotic unitary 2-designs.