Pith. sign in

REVIEW 5 minor 40 references

Simulating a Hamiltonian's time evolution costs at least Ω(min_K(Kt + t²λ_K²/ε)) two-qubit gates or queries — exactly what composite qDRIFT achieves.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-01 11:35 UTC pith:6HZE7F54

load-bearing objection Composite qDRIFT is tight for a local classical family; the proofs look right, with only minor presentation slips.

arxiv 2607.19852 v1 pith:6HZE7F54 submitted 2026-07-22 quant-ph

Optimal Lower Bounds for Hamiltonian Simulation

Alexander Zlokapa , Richard R. Allen , Aram W. Harrow This is my paper
classification quant-ph MSC 81P6868Q12
keywords Hamiltonian simulationquantum lower boundsqDRIFTcomposite qDRIFTTrotter-Suzukiquery complexitytrace distanceno-fast-forwarding
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper proves that any simulation of a local Hamiltonian H = Σ a_j h_j with non-increasing coefficients a_j must use at least Ω(min_{0≤K≤L}(Kt + t²λ_K²/ε)) two-qubit gates, or classical oracle queries, to reach trace-distance error ε, where λ_K is the total weight of terms past the K largest. The bound is tight: a hybrid algorithm called composite qDRIFT—deterministically Trotterizing the largest terms and randomly sampling the tail—attains this cost. Consequently, no coefficient-only algorithm can beat this hybrid; in particular, proposed 'higher-order qDRIFT' schemes cannot improve the t and ε scaling while keeping L-independence. The paper constructs deliberately simple hard instances—time-dependent sums of local projectors on disjoint qubit pairs—and proves the bound without encoding Boolean functions. This matters for physical systems with many small terms (power-law interactions, electronic structure), where the result implies gate count must grow polynomially in 1/ε, unlike the logarithmic scaling suggested by coherent block-encoding oracle counts.

Core claim

The paper's central claim is that the simulation complexity of a Hamiltonian H(τ)=Σ_j a_j h_j(τ) is fixed by the coefficient vector alone: any channel E_H built from G two-qubit gates that is within ε (in trace distance) of ideal evolution on every input must satisfy G ≥ Ω(min_K(Kt + t²λ_K²/ε)), with λ_K = Σ_{j>K} a_j. The same bound holds for Q classical queries to an oracle that returns term descriptions. The hard instances are piecewise-constant in time: for the gate bound, H acts as Σ a_j |11⟩⟨11| on fresh disjoint qubit pairs each unit interval; for the query bound, single-qubit projectors with hidden random signs. The proof reasons that the ideal cat state across many pairs acquires a

What carries the argument

The argument is carried by two constructions. (1) The pointwise-control identity (eq. C5): writing the off-diagonal |0,0⟩⟨x,y| as a quarter-sum of four entangled projectors |ψ_c⟩⟨ψ_c| converts a worst-case trace-distance guarantee into |f(x,y)−g(x,y)| ≤ 4ε for every basis pair (x,y), and then into a second-moment bound over random branches of the channel. (2) The subset correlation ratio R_r(S), defined as the product of four matrix elements of a branch's Stinespring dilation normalized to the unit disk, which factorizes over pairs touched by two-qubit gates: R_r(S) depends only on S∩τ_r. The tail lower bound uses this factorization to pit two biased sign distributions P_± against each other

Load-bearing premise

The lower bound assumes the simulator must achieve small error in worst-case trace distance on every possible input state, including specially constructed entangled superpositions; if accuracy were required only on simpler input families or on expectation values, the t²λ_K²/ε tail bound would not follow from this proof.

What would settle it

For the commuting Hamiltonian H = (1/L)Σ_{j=1}^L |11⟩⟨11|_j at time t = 1, Theorem 1 predicts that achieving trace-distance error ε = 1/L² on every input requires Ω(L) two-qubit gates (since min_K(K + λ_K²/ε) ≈ L). A reader could search for an algorithm whose gate count grows sublinearly in L while meeting this error; any such algorithm would contradict the tail lower bound.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Theorem 1 sets a concrete gate floor: any channel approximating H = Σ a_j h_j to ε trace distance needs Ω(min_K(Kt + t²λ_K²/ε)) two-qubit gates.
  • Theorem 2 gives the same floor for algorithms accessing the Hamiltonian only through a classical term oracle, matching classically-computed chemistry Hamiltonians.
  • No coefficient-only algorithm can outperform composite qDRIFT; in particular, higher-order randomized compilers cannot simultaneously improve t and ε dependence while staying L-independent.
  • For Hamiltonians with many small terms (power-law interactions, second-quantized chemistry), the gate count must scale polynomially in 1/ε, not logarithmically.
  • Beating the bound requires exploiting structure beyond the coefficient vector—input state, spatial locality, or a specific Hamiltonian.

Where Pith is reading between the lines

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

  • Because the proof hinges on error being small on every input state, a simulation that only guarantees accuracy on product states or low-energy states could in principle bypass the t²λ_K²/ε tail; quantifying such restricted-input costs is a natural extension the paper does not explore.
  • The hard instances are commuting projectors, so the lower bound applies even to the 'easy' case of classical Hamiltonians; this suggests a parallel lower bound for classical randomized simulators that must output bit-string samples, which the paper's channel formalism does not directly address.
  • The paper's unit-time-to-arbitrary-time reduction (fresh blocks per time interval) is a general template: any unit-time lower bound for a commuting Hamiltonian family can be lifted to time t with the same cost structure, a technique that could yield tight t-dependence for Lindbladian or imaginary-time simulations.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

0 major / 5 minor

Summary. The paper proves asymptotically tight lower bounds for Hamiltonian simulation of a natural family of local, bounded-degree classical Hamiltonians with arbitrary term coefficients. Theorem 1 states that any channel approximating the ideal time evolution to trace-distance error ε must use Ω(min_K(Kt + t²λ_K²/ε)) two-qubit gates, where λ_K is the tail mass; Theorem 2 establishes the same bound in a classical (non-superposition) query model. The hard instances are time-dependent, piecewise-constant, and act on fresh qubit blocks in each unit interval. The matching upper bound is composite qDRIFT, re-derived in Appendix A. The proofs use cat-state witnesses, a four-state identity to turn worst-case trace-distance assumptions into pointwise off-diagonal control, and concentration of a subset-correlation ratio.

Significance. If correct, this is a significant result: it rules out 'higher-order qDRIFT' and any other coefficient-only algorithm that beats the hybrid of high-order Trotterization (for large terms) and qDRIFT (for small terms). Unlike prior no-fast-forwarding bounds, it is fine-grained in the coefficient vector a and gives explicit constants (C_t=10⁻⁷, c_t=1/1536, ε₀=10⁻⁵) with complete proofs in Appendices C and D. The hard instances are simple and the proof is elementary. The paper is honest about its scope: it relies on worst-case trace distance over all inputs, on time-dependent Hamiltonians with fresh blocks, and on a classical oracle that cannot be queried in superposition; it does not apply to time-independent Hamiltonians or to coherent-query/block-encoding models. These caveats are limitations, not internal inconsistencies.

minor comments (5)
  1. [Theorem 2 (eq. 14)] The symbol E_H is used both for the output channel and for the expectation over the Hamiltonian distribution. This is confusing; use E_{H∼D} for the expectation and, say, C_H for the output channel.
  2. [Appendix C, eq. (C23)] '8√32ε+128ε≤99√ε' is dimensionally inconsistent; it should read '8√(32ε)+128ε≤99√ε' (or '8√32·√ε+128ε'). The bound is nevertheless correct for ε≤0.1.
  3. [After Lemmas 6 and 7] The sentence 'Combining Lemmas 6 and 7 ... proves Theorem 1' omits the case split needed to handle the negative λ²_K/ε term in Lemma 7. A short derivation (e.g., split on whether λ²_K/ε is larger or smaller than a constant multiple of K) would make the proof self-contained.
  4. [Theorems 1 and 2] The main text says 'sufficiently small ε' without a bound; Appendices C/D require ε≤ε₀=10⁻⁵. The theorem statements should state this bound explicitly or refer to the appendices.
  5. [Abstract and 'In what sense...'] The abstract's final sentence ('contrary to the complexity suggested by counting coherent oracle queries') is a suggestion, not a consequence of the theorems, since the lower bound does not apply to coherent-query models. The 'In what sense' section is honest about this; consider softening the abstract to avoid over-interpretation.

Circularity Check

0 steps flagged

No significant circularity: the lower-bound derivation is self-contained and its matching upper bound is an external result, not an input to the proof.

full rationale

The central claims of the paper, Theorems 1 and 2, are lower bounds derived from first-principles witness constructions rather than from the upper-bound algorithm being optimized against. The hard Hamiltonians (Definitions B.1–B.4) are explicit local classical Hamiltonians parameterized only by the coefficient vector a; they are not defined in terms of composite qDRIFT or any fitted quantity. The proof chain is internal: the four-state identity (C5) converts the worst-case trace-distance assumption into pointwise control (C7), the second-moment bound (C8) is derived from that control, and the tail and head lower bounds (Lemmas 6–9) follow from subset-correlation ratios, Pinsker/overlap arguments, and Schmidt-coefficient/fidelity estimates. No step equates a fitted parameter with a prediction: the only optimization over K is in the statement of the bound, and the parameters that appear (a_j, lambda_K, epsilon, t) are inputs of the theorem, not fitted to data. The matching upper bound is not used as a premise of the lower-bound proof; it is attributed externally to Hagan–Wiebe [19] and independently re-derived in Appendix A, so the lower bound does not reduce to the upper bound by construction. There is no load-bearing self-citation: the cited prior work includes Haah–Hastings–Kothari–Low, Childs–Su, Chen–Huang–Kueng–Tropp, and Hagan–Wiebe, none of which is authored by the present paper's authors, and the informal sketch explicitly follows [18] only as motivation. The paper honestly states its scope limitations, including the classical, non-superposition oracle model (Definition B.5) and the piecewise-time-dependent fresh-block construction, in the section 'In what sense is composite qDRIFT “optimal”?' These are model choices and stated restrictions, not circular reductions. The apparent numerical slip in the displayed constant of (C23) is non-load-bearing and does not affect the argument's structure. No circular step can be exhibited from the text, so the appropriate finding is a non-finding with score 0.

Axiom & Free-Parameter Ledger

2 free parameters · 6 axioms · 0 invented entities

No fitted parameters in the predictive sense: the two ledger entries above are existential proof devices whose values do not change the resulting theorem. No invented physical or mathematical entities: the hard families (|11⟩⟨11| pair projectors and random-sign |1⟩⟨1| projectors) are explicit Hamiltonian constructions, and Stinespring dilations are standard. All substantive axioms are either standard mathematics or explicitly stated modeling choices.

free parameters (2)
  • η (bias magnitude in hard distributions) = 16√ε/λ_K* (query, Lemma 8); 1000√ε/λ_K* (gate, Lemma 6)
    Hand-chosen to balance two opposing bounds: the separation of the ideal phases (≈ ηλ_K*, giving the √ε goal) against the KL/query or Pinsker/gate cost (≈ η²·(Q or G)). The final rate λ²_K*/ε is independent of η's value; η is an existential proof device, not an empirical fit, but it is a hand-tuned quantity the derivation's constants depend on.
  • p± = (1±η)/2 (biased subset/sign probabilities) = (1±η)/2
    Defines the two hard distributions P± over subsets (gate bound, Lemma 6, eq. after C23) and signs (query bound, eq. D2). Chosen so that the ideal values e^{−iΣa_j} under P+ and P− are ≳ √ε apart while the algorithm's view (S∩τ_r or queried signs) stays close; a proof construction, not fitted to data.
axioms (6)
  • domain assumption Worst-case trace-distance error for all density matrices, including the four entangled superpositions |ψ_c⟩ = (|0,0⟩ + i^c|x,y⟩)/√2 for every (x,y) (eq. C5–C7)
    Load-bearing: converts the approximation guarantee into the pointwise bound |f−g| ≤ 4ε and the second-moment bound (C8) used throughout Lemma 6.
  • domain assumption Classical, non-superposition oracle access in the query model (Definition B.5)
    Defines Theorem 2's model; the lower bound does not apply to coherent block-encoding oracles, and the paper's 'contrary to the block-encoding model' point depends on this modeling choice being the physically relevant one.
  • domain assumption Gate counting: two-qubit gates counted, single-qubit gates free, arbitrary ancillas; approximating channels are mixtures of Stinespring dilations each using ≤ G two-qubit gates (Appendix C)
    Standard in the simulation literature; the connected-component bounds |τ_r| ≤ 2G and |σ_r| ≤ G rely on it.
  • domain assumption Hard instances are piecewise time-dependent with a fresh block of qubits per unit interval (Definitions B.3–B.4; Lemma 4 reduction)
    This is what produces the Kt head term and the t²λ²_K tail term. Static fixed-geometry Hamiltonians are not covered by the theorems; the power-law claim in the abstract is an extrapolation.
  • domain assumption Term normalization: Σ a_j = 1, ‖h_j‖ = 1, a_j sorted, a_1 ≤ π/2 (eq. 1; Definitions B.1–B.2)
    W.l.o.g. by rescaling t; the π/2 constraint is vacuous under normalization but is used in the cos²(x/2) ≤ e^{−x²/4} and Schmidt-coefficient bounds.
  • standard math Standard inequalities: Hölder, Markov, Jensen, Cauchy–Schwarz, Pinsker + KL chain rule, Fuchs–van de Graaf, Stinespring representation, diamond/trace norm relations
    Invoked throughout Appendices C and D (e.g., eqs. C6, C32, D6, C39); standard background.

pith-pipeline@v1.3.0-alltime-deepseek · 22491 in / 44792 out tokens · 449511 ms · 2026-08-01T11:35:30.103555+00:00 · methodology

0 comments
read the original abstract

For Hamiltonian $H = \sum_j h_j$, we prove asymptotically tight lower bounds on the gate and query complexities of simulating time evolution on a quantum computer. Our bounds hold for arbitrary term norms $\|h_j\|$, time $t$, and trace-distance error $\epsilon$. The matching upper bound (known as composite qDRIFT) consists of high-order Trotterization of the large terms and a randomized first-order Trotterization of the small terms. Unlike prior work that chooses worst-case $\|h_j\|$ to encode the computation of parity or other Boolean functions in time evolution, our proof is elementary and based on a local, bounded-degree classical Hamiltonian. Our work suggests that for many physical systems (e.g., power-law interactions), gate count must scale polynomially in $1/\epsilon$, contrary to the complexity suggested by counting coherent oracle queries such as those in the block-encoding model.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

40 extracted references · 35 linked inside Pith

  1. [1]

    J. Haah, M. B. Hastings, R. Kothari, and G. H. Low, Quantum algorithm for simulating real time evolution of lattice hamiltonians, SIAM Journal on Computing52, FOCS18 (2023), arXiv:1801.03922

    Pith/arXiv arXiv 2023

  2. [2]

    A. M. Childs and Y. Su, Nearly optimal lattice simula- tion by product formulas, Phys. Rev. Lett.123, 050503 (2019), arXiv:1901.00564

    Pith/arXiv arXiv 2019

  3. [3]

    A. M. Childs, Y. Su, M. C. Tran, N. Wiebe, and S. Zhu, Theory of trotter error with commutator scaling, Phys. Rev. X11, 011020 (2021), arXiv:1912.08854

    Pith/arXiv arXiv 2021

  4. [4]

    M. C. Tran, Y. Su, D. Carney, and J. M. Taylor, Faster digital quantum simulation by symmetry protec- tion, PRX Quantum2, 010323 (2021), arXiv:2006.16248

    Pith/arXiv arXiv 2021

  5. [5]

    S ¸ahino˘ glu and R

    B. S ¸ahino˘ glu and R. D. Somma, Hamiltonian simulation in the low-energy subspace, npj Quantum Information7, 119 (2021), arXiv:2006.02660

    Pith/arXiv arXiv 2021

  6. [6]

    Q. Zhao, Y. Zhou, A. F. Shaw, T. Li, and A. M. Childs, Hamiltonian simulation with random inputs, Phys. Rev. Lett.129, 270502 (2022), arXiv:2111.04773

    Pith/arXiv arXiv 2022

  7. [7]

    Chen and F

    C.-F. Chen and F. G. S. L. Brand˜ ao, Average- case speedup for product formulas, Communications in Mathematical Physics405, 10.1007/s00220-023-04912-5 (2024), arXiv:2111.05324

    Pith/arXiv arXiv 2024

  8. [8]

    M. C. Tran, A. Y. Guo, Y. Su, J. R. Garrison, Z. Eldredge, M. Foss-Feig, A. M. Childs, and A. V. Gorshkov, Locality and digital quantum simulation of power-law interactions, Phys. Rev. X9, 031006 (2019), arXiv:1808.05225

    Pith/arXiv arXiv 2019

  9. [9]

    Ouyang, D

    Y. Ouyang, D. R. White, and E. T. Campbell, Compila- tion by stochastic Hamiltonian sparsification, Quantum 4, 235 (2020), arXiv:1910.06255

    Pith/arXiv arXiv 2020

  10. [10]

    Motta, E

    M. Motta, E. Ye, J. R. McClean, Z. Li, A. J. Minnich, R. Babbush, and G. K.-L. Chan, Low rank representa- tions for quantum simulation of electronic structure, npj Quantum Information7, 83 (2021), arXiv:1808.02625

    Pith/arXiv arXiv 2021

  11. [11]

    Lloyd, Universal quantum simulators, Science273, 1073 (1996)

    S. Lloyd, Universal quantum simulators, Science273, 1073 (1996)

    1996

  12. [12]

    A. M. Childs and N. Wiebe, Hamiltonian simulation us- ing linear combinations of unitary operations, Quantum Info. Comput.12, 901–924 (2012), arXiv:1202.5822

    Pith/arXiv arXiv 2012

  13. [13]

    D. W. Berry, A. M. Childs, R. Cleve, R. Kothari, and R. D. Somma, Simulating hamiltonian dynamics with a truncated taylor series, Phys. Rev. Lett.114, 090502 (2015), arXiv:1412.4687

    Pith/arXiv arXiv 2015

  14. [14]

    G. H. Low and I. L. Chuang, Optimal hamiltonian sim- ulation by quantum signal processing, Phys. Rev. Lett. 118, 010501 (2017), arXiv:1606.02685

    Pith/arXiv arXiv 2017

  15. [15]

    G. H. Low and I. L. Chuang, Hamiltonian simulation by qubitization, Quantum3, 163 (2019), arXiv:1610.06546

    Pith/arXiv arXiv 2019

  16. [16]

    Campbell, Random compiler for fast hamiltonian simulation, Phys

    E. Campbell, Random compiler for fast hamiltonian simulation, Phys. Rev. Lett.123, 070503 (2019), arXiv:1811.08017

    Pith/arXiv arXiv 2019

  17. [17]

    A. M. Childs, A. Ostrander, and Y. Su, Faster quantum simulation by randomization, Quantum3, 182 (2019), arXiv:1805.08385

    Pith/arXiv arXiv 2019

  18. [18]

    Chen, H.-Y

    C.-F. Chen, H.-Y. Huang, R. Kueng, and J. A. Tropp, Concentration for random product formulas, PRX Quan- tum2, 040305 (2021), arXiv:2008.11751

    arXiv 2021

  19. [19]

    Hagan and N

    M. Hagan and N. Wiebe, Composite quantum simula- tions, Quantum7, 1181 (2023), arXiv:2206.06409

    Pith/arXiv arXiv 2023

  20. [20]

    Jin and X

    S. Jin and X. Li, A partially random trotter algorithm for quantum Hamiltonian simulations, Communications on Applied Mathematics and Computation7, 442 (2025), arXiv:2109.07987

    Pith/arXiv arXiv 2025

  21. [21]

    Rajput, A

    A. Rajput, A. Roggero, and N. Wiebe, Hybridized meth- ods for quantum simulation in the interaction picture, Quantum6, 780 (2022), arXiv:2109.03308

    Pith/arXiv arXiv 2022

  22. [22]

    O. Kiss, M. Grossi, and A. Roggero, Importance sam- pling for stochastic quantum simulations, Quantum7, 977 (2023), arXiv:2212.05952

    Pith/arXiv arXiv 2023

  23. [23]

    Pocrnic, M

    M. Pocrnic, M. Hagan, J. Carrasquilla, D. Segal, and N. Wiebe, Composite QDrift-product formulas for quantum and classical simulations in real and imagi- nary time, Physical Review Research6, 013224 (2024), arXiv:2306.16572

    Pith/arXiv arXiv 2024

  24. [24]

    D. W. Berry, G. Ahokas, R. Cleve, and B. C. Sanders, Ef- ficient quantum algorithms for simulating sparse Hamil- tonians, Communications in Mathematical Physics270, 359 ((2007)), quant-ph/0508139

    Pith/arXiv arXiv 2007

  25. [25]

    A. M. Childs and R. Kothari, Limitations on the simula- tion of non-sparse hamiltonians, Quantum Info. Comput. 10, 669–684 (2010), arXiv:0908.4398

    Pith/arXiv arXiv 2010

  26. [26]

    D. W. Berry, A. M. Childs, R. Cleve, R. Kothari, and R. D. Somma, Exponential improvement in precision for simulating sparse Hamiltonians, inProceedings of the forty-sixth annual ACM symposium on Theory of com- puting(2014) pp. 283–292, arXiv:1312.1414

    Pith/arXiv arXiv 2014

  27. [27]

    D. W. Berry, A. M. Childs, and R. Kothari, Hamilto- nian simulation with nearly optimal dependence on all parameters, inProceedings of the 2015 IEEE 56th An- nual Symposium on Foundations of Computer Science (FOCS), FOCS ’15 (IEEE Computer Society, USA, 2015) p. 792–809, arXiv:1501.01715

    Pith/arXiv arXiv 2015

  28. [28]

    Atia and D

    Y. Atia and D. Aharonov, Fast-forwarding of Hamilto- nians and exponentially precise measurements, Nature communications8, 1572 (2017), arXiv:1610.09619

    Pith/arXiv arXiv 2017

  29. [29]

    R. P. Feynman, Simulating physics with computers, International Journal of Theoretical Physics21, 467 (1982)

    1982

  30. [30]

    Campbell, Shorter gate sequences for quantum com- puting by mixing unitaries, Physical Review A95, 042306 (2017), arXiv:1612.02689

    E. Campbell, Shorter gate sequences for quantum com- puting by mixing unitaries, Physical Review A95, 042306 (2017), arXiv:1612.02689

    Pith/arXiv arXiv 2017

  31. [31]

    M. B. Hastings, Turning gate synthesis errors into inco- herent errors (2016), arXiv:1612.01011

    Pith/arXiv arXiv 2016

  32. [32]

    P. K. Faehrmann, M. Steudtner, R. Kueng, M. Kiefer- ova, and J. Eisert, Randomizing multi-product formu- las for Hamiltonian simulation, Quantum6, 806 (2022), arXiv:2101.07808

    Pith/arXiv arXiv 2022

  33. [33]

    Nakaji, M

    K. Nakaji, M. Bagherimehrab, and A. Aspuru-Guzik, High-order randomized compiler for Hamiltonian simula- tion, PRX Quantum5, 020330 (2024), arXiv:2302.14811

    Pith/arXiv arXiv 2024

  34. [34]

    J. D. Watson and J. Watkins, Exponentially reduced cir- cuit depths using trotter error mitigation, PRX Quantum 6, 030325 (2025), arXiv:2408.14385

    Pith/arXiv arXiv 2025

  35. [35]

    J. D. Watson, Randomly compiled quantum simula- tion with exponentially reduced circuit depths (2025), 6 arXiv:2411.04240

    Pith/arXiv arXiv 2025

  36. [36]

    G. H. Low, Hamiltonian simulation with nearly opti- mal dependence on spectral norm, inProceedings of the 51st Annual ACM SIGACT Symposium on Theory of Computing, STOC 2019 (Association for Comput- ing Machinery, New York, NY, USA, 2019) p. 491–502, arXiv:1807.03967

    Pith/arXiv arXiv 2019

  37. [37]

    R. King, G. H. Low, R. Babbush, R. D. Somma, and N. C. Rubin, Quantum simulation with sum-of-squares spectral amplification, Phys. Rev. Lett.136, 110601 (2026), arXiv:2505.01528

    Pith/arXiv arXiv 2026

  38. [38]

    G. H. Low, R. King, D. W. Berry, Q. Han, A. E. De- Prince, A. F. White, R. Babbush, R. D. Somma, and N. C. Rubin, Fast quantum simulation of electronic struc- ture by spectral amplification, Physical Review X15, 10.1103/pb2g-j9cw (2025), arXiv:2502.15882

    arXiv 2025

  39. [39]

    J. Lee, D. W. Berry, C. Gidney, W. J. Huggins, J. R. McClean, N. Wiebe, and R. Babbush, Even more effi- cient quantum computations of chemistry through ten- sor hypercontraction, PRX Quantum2, 030305 (2021), arXiv:2011.03494

    Pith/arXiv arXiv 2021

  40. [40]

    hard-code

    W. F. B. Jr., L. Kim, and M. Marvian, Limits of advan- tage in randomization in quantum protocols (2026), draft manuscript. 7 Appendix A: Upper bound from Hagan–Wiebe [19] Here we justify the claim that eq. (8) is an achievable cost for composite qDRIFT. This fact is essentially proved in [19] but the notation there requires some translation. To avoid a c...

    2026