arxiv: 2510.08467 · v2 · submitted 2025-10-09 · 🪐 quant-ph

Stability of digital and analog quantum simulations under noise

Jayant Rao , Jens Eisert , Tommaso Guaita This is my paper

Pith reviewed 2026-05-18 08:55 UTC · model grok-4.3

classification 🪐 quant-ph

keywords quantum simulationnoise stabilitydigital quantum simulationanalog quantum simulationerror boundsperturbative noiseconcentration bounds

0 comments

The pith

Digital and analog quantum simulators exhibit comparable worst-case error scaling under noise but different average-case cancellations for local observables.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper compares the stability of digital and analog quantum simulators under a variety of perturbative noise models. It derives rigorous worst-case and average-case error bounds specifically for the simulation of local observables. The central finding is that both paradigms scale similarly in the worst case yet display distinct mechanisms of enhanced error cancellation when noise is averaged over. The work also supplies concentration bounds for Gaussian and Brownian noise processes that describe typical rather than extreme deviations. A sympathetic reader cares because these distinctions can guide the choice of architecture for practical noisy quantum devices.

Core claim

The paper establishes that under perturbative noise models, digital and analog quantum simulators show comparable scaling of errors in the worst case when simulating local observables, while exhibiting different forms of enhanced error cancellation on average. Concentration bounds are further derived for Gaussian and Brownian noise processes to capture typical deviations beyond worst-case guarantees.

What carries the argument

Rigorous worst-case and average-case error bounds for noisy quantum simulation of local observables under perturbative noise models, which quantify stability and identify cancellation effects in both digital and analog paradigms.

If this is right

Both simulation paradigms have similar worst-case robustness scaling for local observables under the considered noise.
Distinct average-case cancellation mechanisms imply that specific noise statistics can favor one architecture over the other.
The unified error-bound framework allows direct comparison of robustness across digital and analog devices.
Regimes where digital methods hold intrinsic advantages over analog ones can be identified from the noise type and strength.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The results could inform noise-aware switching protocols in hybrid quantum simulators that combine digital and analog elements.
Extending the bounds beyond local observables might expose different scaling behaviors for nonlocal quantities.
Small-scale experiments on current hardware could directly test the predicted average-case cancellations under engineered noise.
The concentration bounds may connect to error-mitigation techniques already used in variational quantum algorithms.

Load-bearing premise

The analysis assumes perturbative noise models acting on the quantum simulator and focuses on the simulation of local observables.

What would settle it

An experiment that adds controlled perturbative noise to a small quantum device, simulates a local observable such as a two-point correlation function in both digital and analog modes, and checks whether the measured error growth matches the predicted worst-case scaling and average-case cancellation.

Figures

Figures reproduced from arXiv: 2510.08467 by Jayant Rao, Jens Eisert, Tommaso Guaita.

**Figure 1.** Figure 1: FIG. 1: In this work, (a) analog quantum simulation provided by precisely controlled quantum systems naturally [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗

read the original abstract

Quantum simulation is a central application of near-term quantum devices, pursued in both analog and digital architectures. A key challenge for both paradigms is the effect of imperfections and noise on predictive power. In this work, we present a rigorous and physically transparent comparison of the stability of digital and analog quantum simulators under a variety of perturbative noise models. We provide rigorous worst- and average-case error bounds for noisy quantum simulation of local observables. We find that the two paradigms show comparable scaling in the worst case, while exhibiting different forms of enhanced error cancellation on average. We further analyze Gaussian and Brownian noise processes, deriving concentration bounds that capture typical deviations beyond worst-case guarantees. These results provide a unified framework for quantifying the robustness of noisy quantum simulations and identify regimes where digital methods have intrinsic advantages and when we can see similar behavior.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper shows comparable worst-case error scaling for digital and analog simulators under perturbative noise on local observables, with distinct average-case cancellations and concentration bounds for Gaussian and Brownian noise.

read the letter

The paper's core result is that digital and analog quantum simulators show comparable worst-case error scaling for local observables under perturbative noise, but they exhibit different patterns of error cancellation when you look at average-case performance. They also give concentration bounds for Gaussian and Brownian noise that go beyond the worst-case picture. This gives a direct apples-to-apples comparison rather than leaving readers to combine separate papers on each paradigm.

Referee Report

2 major / 3 minor

Summary. The manuscript derives rigorous worst- and average-case error bounds for the simulation of local observables in both digital and analog quantum simulators under a variety of perturbative noise models. It concludes that the two paradigms exhibit comparable scaling in the worst case while displaying distinct forms of enhanced error cancellation on average, and further derives concentration bounds for Gaussian and Brownian noise processes that quantify typical deviations.

Significance. If the bounds and cancellation mechanisms hold under the stated assumptions, the work supplies a unified, physically transparent framework for quantifying robustness in near-term quantum simulation. The explicit comparison of scaling behaviors and the derivation of concentration inequalities for specific noise processes constitute a useful contribution for identifying regimes of relative advantage between digital and analog approaches. No machine-checked proofs or open reproducible code are reported, but the focus on falsifiable scaling predictions and perturbative expansions is a clear strength.

major comments (2)

[§3, Eq. (12)] §3, Eq. (12): The worst-case bound for analog simulation is obtained via a first-order perturbative expansion in the noise strength; the manuscript does not quantify the radius of convergence or the size of higher-order corrections, which directly affects whether the claimed comparable scaling with digital simulation remains valid beyond the weak-noise regime.
[§4.3] §4.3: The average-case cancellation for digital simulators is shown under the assumption of independent local noise channels; the derivation does not address spatially correlated noise, which could eliminate the reported advantage and is therefore load-bearing for the headline distinction between the two paradigms.

minor comments (3)

[Notation] The notation for the local observable support size is introduced inconsistently between the main text and the appendices; a single definition should be used throughout.
[Figure 2] Figure 2 caption does not specify the number of disorder realizations used to compute the plotted averages, making it difficult to assess statistical significance of the observed cancellation.
[Discussion] A brief remark on the computational cost of evaluating the derived bounds would help readers assess practical utility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading of our manuscript and the constructive comments. We address each major point below and have revised the manuscript to incorporate clarifications where appropriate, while preserving the core results on perturbative noise bounds.

read point-by-point responses

Referee: [§3, Eq. (12)] §3, Eq. (12): The worst-case bound for analog simulation is obtained via a first-order perturbative expansion in the noise strength; the manuscript does not quantify the radius of convergence or the size of higher-order corrections, which directly affects whether the claimed comparable scaling with digital simulation remains valid beyond the weak-noise regime.

Authors: We agree that the worst-case bound in Eq. (12) is derived from a first-order perturbative expansion. This is the leading contribution in the weak-noise regime that is the focus of the work, where the noise strength ε is small compared to the inverse simulation time. Higher-order terms enter at O(ε²) and are negligible under the stated perturbative assumptions, preserving the linear scaling in ε that matches the digital case. The radius of convergence is governed by the operator norm of the noise terms, which for local models is typically O(1) and thus larger than the ε values of experimental interest. In the revised manuscript we will add a short paragraph after Eq. (12) explicitly stating these assumptions and the suppression of higher-order corrections, thereby clarifying the regime of validity without changing the reported scaling. revision: yes
Referee: [§4.3] §4.3: The average-case cancellation for digital simulators is shown under the assumption of independent local noise channels; the derivation does not address spatially correlated noise, which could eliminate the reported advantage and is therefore load-bearing for the headline distinction between the two paradigms.

Authors: The derivation in §4.3 indeed assumes independent local noise channels, as stated in the text. This assumption is physically motivated for many near-term devices where noise sources are local. While strong spatial correlations could reduce the observed cancellation, the worst-case bounds remain comparable between paradigms regardless of correlations. For the average-case distinction, we will add a clarifying paragraph noting that the reported advantage holds under the independent-noise model and briefly indicating how finite-range correlations would enter the averaging argument. This keeps the headline comparison intact for the noise models analyzed while acknowledging the referee’s point on broader applicability. revision: partial

Circularity Check

0 steps flagged

Derivations rely on standard operator norms and perturbative expansions with no self-referential fitting or definitional loops.

full rationale

The paper derives worst- and average-case error bounds for noisy quantum simulation of local observables under perturbative noise models using established quantum-mechanical techniques such as operator norms and expansions. These bounds directly yield the comparisons of scaling and error cancellation between digital and analog paradigms. No parameters are fitted inside the work and then relabeled as predictions, no self-citations form the load-bearing justification, and no ansatz or uniqueness result is imported from the authors' prior work to close the argument. The central claims follow from the mathematical analysis under explicitly stated assumptions, rendering the derivation self-contained against external benchmarks in quantum information theory.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claims rest on standard mathematical tools of quantum information theory and the modeling choice that noise remains perturbative.

axioms (2)

standard math Standard quantum mechanics, operator norms, and perturbative expansions suffice to bound simulation errors
Invoked to obtain rigorous worst- and average-case bounds for local observables.
domain assumption Noise processes are perturbative and can be modeled as Gaussian or Brownian
Required for the concentration bounds and the distinction between worst-case and average-case behavior.

pith-pipeline@v0.9.0 · 5660 in / 1269 out tokens · 34607 ms · 2026-05-18T08:55:05.515044+00:00 · methodology

discussion (0)

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

Analogue quantum simulation with polylogarithmic interaction strengths by extrapolating within phases of matter
quant-ph 2026-05 unverdicted novelty 7.0

For non-critical systems, analogue quantum simulation via perturbative gadgets requires only polylogarithmic interaction strengths through extrapolation within phases of matter.
Mind the gaps: The fraught road to quantum advantage
quant-ph 2025-10 unverdicted novelty 4.0

The authors identify four transitions needed to reach fault-tolerant application-scale quantum computing from current NISQ devices.

Reference graph

Works this paper leans on

79 extracted references · 79 canonical work pages · cited by 2 Pith papers · 8 internal anchors

[1]

J. I. Cirac and P. Zoller, Goals and opportunities in quantum simulation, Nature Phys.8, 264 (2012)

work page 2012
[2]

Fauseweh, Quantum many-body simulations on digital quantum computers: State-of-the-art and future chal- lenges, Nature Comm.15, 2123 (2024)

B. Fauseweh, Quantum many-body simulations on digital quantum computers: State-of-the-art and future chal- lenges, Nature Comm.15, 2123 (2024)

work page 2024
[3]

I. M. Georgescu, S. Ashhab, and F. Nori, Quantum simu- lation, Rev. Mod. Phys.86, 153 (2014)

work page 2014
[4]

Probing the relaxation towards equilibrium in an isolated strongly correlated 1D Bose gas

S. Trotzky, Y.-A. Chen, A. Flesch, I. P. McCulloch, U. Schollwöck, J. Eisert, and I. Bloch, Probing the re- laxation towards equilibrium in an isolated strongly cor- related one-dimensional Bose gas, Nature Phys.8, 325 (2012), arXiv:1101.2659

work page internal anchor Pith review Pith/arXiv arXiv 2012
[5]

Bloch, J

I. Bloch, J. Dalibard, and S. Nascimbene, Quantum sim- ulations with ultracold quantum gases, Nature Phys.8, 267 (2012)

work page 2012
[6]

Bernien, S

H. Bernien, S. Schwartz, A. Keesling, H. Levine, A. Om- ran, H. Pichler, S. Choi, A. S. Zibrov, M. Endres, M. Greiner, V. Vuletic, and M. Lukin, Probing many- body dynamics on a 51-atom quantum simulator, Nature 551, 579 (2017)

work page 2017
[7]

A. A. Houck, H. E. Tuereci, and J. Koch, On-chip quan- tum simulation with superconducting circuits, Nature Phys. 8, 292 (2012)

work page 2012
[8]

Roushan, C

P. Roushan, C. Neill, J. Tangpanitanon, V. Bastidas, A. Megrant, R. Barends, Y. Chen, Z. Chen, B. Chiaro, A. Dunsworth, A. Fowler, B. Foxen, M. Giustina, E. Jef- frey, J. Kelly, E. Lucero, J. Mutus, M. Neeley, C. Quin- tana, Sank, A. Vainsencher, J. Wenner, T. White, H. Neven, G. Angelakis, and J. Martinis, Spectral signa- tures of many-body localization ...

work page 2017
[9]

J.-Y. Choi, S. Hild, J. Zeier, P. Schauß, A. Rubio-Abadal, T.Yefsah, V.Khemani, D.A.Huse, I.Bloch,andC.Gross, Exploring the many-body localization transition in two dimensions, Science 352, 1547 (2016)

work page 2016
[10]

Smith, A

J. Smith, A. Lee, P. Richerme, B. Neyenhuis, P. W. Hess, P. Hauke, M. Heyl, D. Huse, and C. Monroe, Many-body localization in a quantum simulator with programmable random disorder, Nature Phys.12, 907 (2016)

work page 2016
[11]

Clinton, T

L. Clinton, T. Cubitt, B. Flynn, F. M. Gambetta, J. Klassen, A. Montanaro, S. Piddock, R. A. Santos, and E. Sheridan, Towards near-term quantum simulation of materials, Nature Comm.15, 211 (2024)

work page 2024
[12]

Lloyd, Universal quantum simulators, Science 273, 1073 (1996)

S. Lloyd, Universal quantum simulators, Science 273, 1073 (1996)

work page 1996
[13]

Structure learning of Hamiltonians from real-time evolution

A. Bakshi, A. Liu, A. Moitra, and E. Tang, Struc- ture learning of Hamiltonians from real-time evolution, arxiv:2405.00082

work page internal anchor Pith review Pith/arXiv arXiv
[14]

Anshu, S

A. Anshu, S. Arunachalam, T. Kuwahara, and M. Soleimanifar, Sample-efficient learning of interacting quantum systems, Nature Phys.17, 931 (2021)

work page 2021
[15]

J. H. Cole, S. G. Schirmer, A. D. Greentree, C. J. Wellard, D. K. L. Oi, and L. C. L. Hollenberg, Identifying an experimental two-state Hamiltonian to arbitrary accuracy, Phys. Rev. A71, 062312 (2005)

work page 2005
[16]

S. G. Schirmer and D. K. L. Oi, Two-qubit Hamiltonian tomography by Bayesian analysis of noisy data, Phys. Rev. A 80, 022333 (2009)

work page 2009
[17]

Hangleiter, I

D. Hangleiter, I. Roth, J. Fuksa, J. Eisert, and P. Roushan, Robustly learning the Hamiltonian dynamics of a super- conducting quantum processor, Nature Communications 15, 9595 (2024)

work page 2024
[18]

Kashyap, G

V. Kashyap, G. Styliaris, S. Mouradian, J. I. Cirac, and R. Trivedi, Accuracy guarantees and quantum advantage in analog open quantum simulation with and without noise, Phys. Rev. X15, 021017 (2025)

work page 2025
[19]

Trivedi, A

R. Trivedi, A. F. Rubio, and J. I. Cirac, Quantum advan- tage and stability to errors in analogue quantum simula- tors, Nature Comm.15, 6507 (2024)

work page 2024
[20]

Möbus, J

T. Möbus, J. Sánchez-Segovia, Álvaro M. Alhambra, and Ángela Capel, Stability of thermal equilibrium in long- range quantum systems (2025), arXiv:2506.16451

work page arXiv 2025
[21]

G. C. Knee and W. J. Munro, Optimal Trotterization in universal quantum simulators under faulty control, Phys. Rev. A 91, 052327 (2015)

work page 2015
[22]

J. Xu, C. Zhao, J. Fan, and Q. Zhao, Exponentially decaying quantum simulation error with noisy devices (2025), arXiv:2504.10247

work page arXiv 2025
[23]

Stilck Franca and R

D. Stilck Franca and R. García-Patrón, Limitations of optimizationalgorithmsonnoisyquantumdevices,Nature Phys. 17, 1221 (2020)

work page 2020
[24]

A. A. Mele, A. Angrisani, S. Ghosh, S. Khatri, J. Eisert, D. S. França, and Y. Quek, Noise-induced shallow circuits and absence of barren plateaus, (2024), arXiv:2403.13927

work page arXiv 2024
[25]

Fefferman, S

B. Fefferman, S. Ghosh, M. Gullans, K. Kuroiwa, and K. Sharma, Effect of non-unital noise on random circuit sampling, (2023), arXiv:2306.16659

work page arXiv 2023
[26]

Y. Cai, Y. Tong, and J. Preskill, Stochastic error cancel- lation in analog quantum simulation, in19th Conference on the Theory of Quantum Computation, Communica- tion and Cryptography (TQC 2024), Leibniz International Proceedings in Informatics (LIPIcs), Vol. 310, edited by F. Magniez and A. B. Grilo (Schloss Dagstuhl – Leibniz- Zentrum für Informatik, D...

work page 2024
[27]

Suzuki, General theory of fractal path integrals with applications to many-body theories and statistical physics, J

M. Suzuki, General theory of fractal path integrals with applications to many-body theories and statistical physics, J. Math. Phys.32, 400 (1991)

work page 1991
[28]

P. M. Poggi, N. K. Lysne, K. W. Kuper, I. H. Deutsch, and P. S. Jessen, Quantifying the sensitivity to errors in analog quantum simulation, PRX Quantum1, 020308 (2020)

work page 2020
[29]

Chen and F

C.-F. Chen and F. G. S. L. Brandão, Average-case speedup for product formulas, Comm. Math. Phys.405, 32 (2024)

work page 2024
[30]

Q. Zhao, Y. Zhou, A. F. Shaw, T. Li, and A. M. Childs, Hamiltonian simulation with random inputs, Phys. Rev. Lett. 129, 270502 (2022)

work page 2022
[31]

Harley, I

D. Harley, I. Datta, F. R. Klausen, A. Bluhm, D. S. França, A. H. Werner, and M. Christandl, Going beyond gadgets: the importance of scalability for analogue quan- tum simulators, Nature Comm.15, 6527 (2024)

work page 2024
[32]

Holonomic quantum computation: a scalable adiabatic architecture

C. Wassner, T. Guaita, J. Eisert, and J. Carrasco, Holo- nomic quantum computation: a scalable adiabatic archi- tecture, (2025), arXiv:2502.17188

work page internal anchor Pith review Pith/arXiv arXiv 2025
[33]

S.E.S.IoannisKaratzas, Brownian Motion and Stochastic Calculus (Springer New York, NY)

work page
[34]

J.-F. L. Gall,Brownian Motion, Martingales, and Stochas- tic Calculus(Springer Cham)

work page
[35]

Benoist, L

T. Benoist, L. Hänggli, and C. Rouzé, Deviation bounds and concentration inequalities for quantum noises, Quan- tum 6, 772 (2022)

work page 2022
[36]

A. M. Childs, Y. Su, M. C. Tran, N. Wiebe, and S. Zhu, Theory of Trotter error with commutator scaling, Phys. Rev. X 11, 011020 (2021)

work page 2021
[37]

An introduction to quantum filtering

L. Bouten, R. van Handel, and M. James, An introduc- tion to quantum filtering (2006), arXiv:math/0601741 [math.OC]

work page internal anchor Pith review Pith/arXiv arXiv 2006
[38]

Bouten, R

L. Bouten, R. van Handel, and M. R. James, A discrete in- vitation to quantum filtering and feedback control, SIAM Review 51, 239–316 (2009)

work page 2009
[39]

A Trotter-Kato Theorem for Quantum Markov Limits

L. Bouten, R. Gohm, J. Gough, and H. Nurdin, A trotter-kato theorem for quantum markov limits (2015), arXiv:1409.2260 [math-ph]

work page internal anchor Pith review Pith/arXiv arXiv 2015
[40]

Bouten and R

L. Bouten and R. Van Handel, Discrete approximation of quantum stochastic models, Journal of Mathematical Physics 49, 10.1063/1.3001109 (2008)

work page doi:10.1063/1.3001109 2008
[41]

Y. Quek, D. S. França, S. Khatri, J. J. Meyer, and J. Eis- ert, Exponentially tighter bounds on limitations of quan- tum error mitigation, Nature Phys.20, 1648 (2024)

work page 2024
[42]

Exponen- tially tighter bounds on limitations of quantum er- ror mitigation,

R. Takagi, S. Endo, S. Minagawa, and M. Gu, Fundamen- tal limits of quantum error mitigation, npj Quant. Inf.8, 114 (2022), arXiv:2210.11505

work page arXiv 2022
[43]

Eisert, D

J. Eisert, D. Hangleiter, N. Walk, I. Roth, D. Markham, R. Parekh, U. Chabaud, and E. Kashefi, Quantum cer- tification and benchmarking, Nature Rev. Phys.2, 382 (2020)

work page 2020
[44]

H.-Y. Hu, A. M. Gomez, L. Chen, A. Trowbridge, A. J. Goldschmidt, Z. Manchester, F. T. Chong, A. Jaffe, and S. F. Yelin, Universal dynamics with globally controlled analog quantum simulators (2025), arXiv:2508.19075

work page arXiv 2025
[45]

Dutkiewicz, T

A. Dutkiewicz, T. E. O’Brien, and T. Schuster, The ad- vantage of quantum control in many-body Hamiltonian learning, Quantum 8, 1537 (2024)

work page 2024
[46]

Trivedi and J

R. Trivedi and J. I. Cirac, Noise robustness of problem- to-simulator mappings for quantum many-body physics (2025), arXiv:2509.17579 [quant-ph]

work page arXiv 2025
[47]

C.-F. A. Chen, A. Lucas, and C. Yin, Speed limits and locality in many-body quantum dynamics, Rept. Prog. Phys. 86, 116001 (2023)

work page 2023
[48]

Barthel and M

T. Barthel and M. Kliesch, Quasilocality and efficient simulation of Markovian quantum dynamics, Phys. Rev. Lett. 108, 230504 (2012)

work page 2012
[49]

Kliesch, C

M. Kliesch, C. Gogolin, and J. Eisert, Lieb-Robinson bounds and the simulation of time-evolution of local ob- servables in lattice systems, inMany-Electron Approaches in Physics, Chemistry and Mathematics, Mathematical Physics Studies, edited by V. Bach and L. Delle Site (Springer, 2014) p. 301

work page 2014
[50]

Pinelis, Optimum Bounds for the Distributions of Mar- tingales in Banach Spaces, The Annals of Probability22, 1679 (1994)

I. Pinelis, Optimum Bounds for the Distributions of Mar- tingales in Banach Spaces, The Annals of Probability22, 1679 (1994)

work page 1994
[51]

J. A. Tropp, User-friendly tail bounds for sums of random matrices, Found. Comput. Math.12, 389 (2012)

work page 2012
[52]

N. T. Dung, Some new concentration inequalities for the itô stochastic integral (2024), arXiv:2310.18699 [math.PR]

work page arXiv 2024
[53]

J.A.Tropp, The expected norm of a sum of indepen- dent random matrices: An elementary approach, (2015), arXiv:1506.04711

work page internal anchor Pith review Pith/arXiv arXiv 2015
[54]

Ikeda and S

N. Ikeda and S. Watanabe,Stochastic Differential Equa- tions and Diffusion Processes, Second Edition(North Holland Publishing Company, 1981)

work page 1981
[55]

(Anthony) Chen, A

C.-F. (Anthony) Chen, A. Lucas, and C. Yin, Speed limits and locality in many-body quantum dynamics, Rep. Prog. Phys. 86, 116001 (2023)

work page 2023
[56]

J. A. Tropp, An introduction to matrix concentration inequalities, (2015), arXiv:1501.01571

work page internal anchor Pith review Pith/arXiv arXiv 2015
[57]

Chen, Concentration of OTOC and Lieb- Robinson velocity in random Hamiltonians, (2021), arXiv:2103.09186

C.-F. Chen, Concentration of OTOC and Lieb- Robinson velocity in random Hamiltonians, (2021), arXiv:2103.09186

work page arXiv 2021
[58]

Mixing properties of stochastic quantum Hamiltonians

E. Onorati, O. Buerschaper, M. Kliesch, W. Brown, A. H. Werner, and J. Eisert, Mixing properties of stochastic quantum Hamiltonians, Commun. Math. Phys.355, 905 (2017), arXiv:1606.01914. 14 A. Notation and preliminary results In this appendix we review and discuss in more detail the precise assumptions that we make on the considered systems and the notati...

work page internal anchor Pith review Pith/arXiv arXiv 2017
[59]

On this latticeZd we will use thel1 distance, which we indicate asd(·,·)

Notation and assumptions We consider a hypercubic latticeZd, ind spatial dimensions. On this latticeZd we will use thel1 distance, which we indicate asd(·,·). With respect to this metric, we denote the ball of radiusR and centerx as BR(x). It contains a number of sites (i.e., a volume) of |BR(x)|= ΛdRd, (A1) where Λd = 2d d!. With this lattice, we associa...

work page
[60]

Firstly, we will often need to estimate the number of terms of a local Hamiltonian satisfying Assumption 1 which are relevant for the truncated HamiltonianHl

Locality in quantum systems Here, we introduce some results concerning what happens when the system introduced above is truncated to a finite system size. Firstly, we will often need to estimate the number of terms of a local Hamiltonian satisfying Assumption 1 which are relevant for the truncated HamiltonianHl. Lemma 1(Truncated Hamiltonian terms). For a...

work page
[61]

We begin by introducing a result that generalizes the well-known Hoeffding inequality to sums of vector valued random variables

Concentration inequalities for random matrices Here, we collect some useful results concerning norm bounds and concentration inequalities for random matrices and stochastic processes. We begin by introducing a result that generalizes the well-known Hoeffding inequality to sums of vector valued random variables. Lemma 3(Pinelis’ lemma [50]). LetX1,...,XT∈C...

work page
[62]

The following expansion is convenient for dealing with expressions written as products of matrices

F urther helpful lemmas We state here a series of helpful technical results, which we will repeatedly use in the following derivations. The following expansion is convenient for dealing with expressions written as products of matrices. 17 Lemma 5(Telescope product). LetA = ∏N i=1Ai, B = ∏N j=1Bj be two products ofk×k matrices. Then A−B = N∑ i=1 i−1∏ j=1 A...

work page
[63]

Theorem 1(Restated, upper bound for worst case errors in analog simulators)

Proof of Theorem 1 We now prove Theorem 1, which we restate here for convenience. Theorem 1(Restated, upper bound for worst case errors in analog simulators). Consider a perturbed analog time evolution V (t) defined by the time-dependent Hamiltonian H′(s) = ∑ γ H′ γ(s), (B1) where H′ γ(s)−Hγ ≤δfor alls<t and allγ∈Γ. We further assume that each local t...

work page
[64]

(C1) Here, for everyγ∈Θl,{Xγ,a}m a=1 is a set ofm Hermitian operators supported onsupp (Hγ) and with∥Xγ,a∥≤1

Perturbations generated by Gaussian processes As discussed in Section IVA, we consider an analog simulator where the the Hamiltonian is perturbed by a noise process of the form Lγ(t) = m∑ a=1 ξγ,a(t)Xγ,a. (C1) Here, for everyγ∈Θl,{Xγ,a}m a=1 is a set ofm Hermitian operators supported onsupp (Hγ) and with∥Xγ,a∥≤1. These operators could correspond, for inst...

work page
[65]

Theorem 4(Restated, Average case bounds for errors in analog simulators with Gaussian perturbations)

Proof of Theorem 4 We repeat the statement of Theorem 4 here. Theorem 4(Restated, Average case bounds for errors in analog simulators with Gaussian perturbations). Consider a perturbed analog time evolution given by the Hamiltonian H′(t) = ∑ γ∈Γ ( Hγ+δ m∑ a=1 ξγ,a(t)Xγ,a ) , (C15) wheret↦→ξγ,a(t) are uncorrelated Gaussian noise processes with time correla...

work page
[66]

For the evolution generated by this equation we can prove the following theorem

Proof of Theorem 5 The white noise version of Equation (C3) is given by the following Ito stochastic differential equation d|ψt⟩=  −iH−δ2 2 m|Θl|∑ σ=1 X2 σ  |ψt⟩dt−iδ m|Θl|∑ σ=1 Xσ|ψt⟩dWσ(t), (C30) where dWσ(t) are standard Wiener process increments. For the evolution generated by this equation we can prove the following theorem. Theorem 5(Restated, A...

work page
[67]

Assume that the implemented Hamiltonian: H′ γ=Hγ+δLγ, (C48) whereLγis chosen from an ensemble of Hermitian matrices withE [Lγ] = 0 and∥Lγ∥≤1

Proof of Theorem 6 Theorem 6(Restated). Assume that the implemented Hamiltonian: H′ γ=Hγ+δLγ, (C48) whereLγis chosen from an ensemble of Hermitian matrices withE [Lγ] = 0 and∥Lγ∥≤1. Then E [∆] ≤Cδtd+1. (C49) Proof.We apply Lemma 6 to∆ , so that ∆ ≤ ∫ t 0 ds  ∑ γ [δLγ,O (s)] . (C50) Now we take expectation values on both sides to find E [∆] ≤ ∫ t...

work page
[68]

Lemma 9(Lindblad Evolution as Average over White Noise)

Proof of Theorem 9 A reason why white noise is physically interesting is because the averaged density matrixρ(t) = E [|ψt⟩⟨ψt|], follows a Lindblad type time evolution. Lemma 9(Lindblad Evolution as Average over White Noise). Assume that the stochastic state vector|ϕ⟩t evolves under Ht as d|ψ⟩t = (( −iH0−1 2 ∑ a S2 aδ2 ) dt−i ∑ a δSadWa(t) ) |ψt⟩, (C55) w...

work page
[69]

This is the most widely used method to decompose unitaries into products of local unitaries which can then be implemented in a quantum circuit

Quantum simulation by Suzuki-T rotter formulas One common approach, which we focus on here is the product formula decomposition of local Hamiltonians. This is the most widely used method to decompose unitaries into products of local unitaries which can then be implemented in a quantum circuit. We consider a local HamiltonianH = ∑ γHγand choose to implemen...

work page
[70]

We will do so using the geometrical locality of theHγterms as stated in Assumption 1

Locality and Suzuki-T rotter products In the following, we will further evaluate the nested commutators appearing in Lemma 10. We will do so using the geometrical locality of theHγterms as stated in Assumption 1. Lemma 11(Nested commutator scaling). We have that  ∑ γ1,...,γp [Hγp, [Hγp−1,...[Hγ2,Hγ1]]]  ≤2p ( Λd2dRd)p−1 [(p−1)!]d|Θl|. (D11) Pr...

work page
[71]

Theorem 2(Restated, upper bound for worst case errors in digital simulators with gate-dependent perturbations)

Proof of Theorem 2 We now prove Theorem 2, whose formal statement we repeat here for convenience. Theorem 2(Restated, upper bound for worst case errors in digital simulators with gate-dependent perturbations). Consider a perturbed Suzuki-Trotter product unitary of orderp = 2k, which takes the form V (p) l,n (t) = n∏ j=1 Υ∏ υ=1 ∏ γ∈Θl Vγ,j,υ, (D17) where e...

work page
[72]

Theorem 3(Restated, upper bound for worst case errors in digital simulators with constant gate perturbations)

Proof of Theorem 3 We now prove Theorem 3, whose formal statement we repeat here for convenience. Theorem 3(Restated, upper bound for worst case errors in digital simulators with constant gate perturbations). Consider a perturbed Suzuki-Trotter product unitary of orderp = 2k, which takes the form V (p) l,n (t) = n∏ j=1 Υ∏ υ=1 ∏ γ∈Θl Vγ,j,υ, (D29) 32 where...

work page
[73]

This can be studied by applying Pinelis’ Lemma, which we introduced in Appendix A

Sums of random matrices One important tool for this discussion is the analysis of the behaviour of the sum of mean-zero random perturbations. This can be studied by applying Pinelis’ Lemma, which we introduced in Appendix A. Lemma 13(Sum of mean-zero random perturbations). Consider a truncation lengthl> 0 and a sequence of random Hermitian operatorsLJ for...

work page
[74]

The main idea is to find anorm of sumstatement to which Lemma 13 can be applied

Bounds on perturbations of T rotter products We continue by proving the second main tool of this part of this work, namely an improved bound for local perturbations of the Trotter formula, which is better than the telescopic sum bound from Lemma 5. The main idea is to find anorm of sumstatement to which Lemma 13 can be applied. Lemma 15(Perturbation Bound...

work page
[75]

Theorem 7 (Restated, average case errors in digital simulators with gate-dependent perturbations)

Proof of Theorem 7 We are now ready to prove Theorem 7, whose formal statement we repeat here for convenience. Theorem 7 (Restated, average case errors in digital simulators with gate-dependent perturbations). Consider a perturbed Suzuki-Trotter product unitary of orderp = 2k, which takes the form V (p) l,n (t) = n∏ j=1 Υ∏ υ=1 |Θl|∏ γ eit n (Hγaγ,υ+δLγ,υ,...

work page
[76]

Proof of Theorem 8 Theorem 8 (Restated, average case errors in digital simulators with constant gate perturbations). Consider a perturbed Suzuki-Trotter product unitary of orderp = 2k, which takes the form V (p) l,n (t) = n∏ j=1 Υ∏ υ=1 |Θl|∏ γ eit nHγaγ,υ+iδLγ,υ,j, (E40) whereLγ,υ,jare random perturbations, drawn independently from a distribution of Hermi...

work page
[77]

As discussed in the main text (Section IV), this only provides an upper bound on the quantityE [∆(ρ)] for a fixed state

A verage scaling of the state-independent error Analogously to what we did in Theorem 6 for analog simulation, we discuss here the average scaling of the quantity∆ . As discussed in the main text (Section IV), this only provides an upper bound on the quantityE [∆(ρ)] for a fixed state. The following results show that this is in fact a loose bound. We pres...

work page
[78]

Define Sn := n∑ i=1 Xi, (E68) where S is known as the random walk

Brownian random walk Let (Xi)i∈I be a sequence ofidentically and independently distributedrandom variables (i.i.d.) with mean0 and variance 1. Define Sn := n∑ i=1 Xi, (E68) where S is known as the random walk. Define the stochastic process Wn(t) := S⌊nt⌋√n t∈[0, 1]. (E69) Then Donsker’s Theorem states that in then→∞limit it converges in distribution to th...

work page
[79]

Given a noise model as described in Eq.(E71), then the following holds

Proof of Theorem 10 Theorem 10(Restated, discrete-Ito). Given a noise model as described in Eq.(E71), then the following holds. 44 •The expected error behaves asE [∆] ≤Cδtd+ 1 2 +O ( 1√n ) . • For a fixed input state, theE [∆(ψ)]≤Cδt d+1 2 +O ( 1√n ) . Proof. The proof is straightforward and follows the same steps as in Theorem 7 and Theorem 11, we thus o...

work page