Pith. sign in

REVIEW 3 major objections 5 minor 55 references

A qubit-oscillator quantum algorithm encodes a nuclear wavepacket as a superposition of frozen Gaussians and evolves it variationally, converging to exact dynamics on harmonic and Morse potentials and reproducing double-well bifurcation and

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 ↗

A variational hybrid qubit-oscillator algorithm evolves frozen Gaussian wavepackets for nuclear dynamics, converging to exact results on harmonic and Morse potentials and showing partial success on double-well bifurcation.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection New CV-DV variational nuclear dynamics ansatz with solid classical benchmarks, but the NISQ pathway claim is contradicted by the paper's own shot-count analysis. the 3 major comments →

arxiv 2608.03907 v1 pith:GW5YKOIX submitted 2026-08-04 physics.chem-ph

A variational hybrid continuous-variable discrete-variable quantum algorithm for adiabatic nuclear dynamics

Rami Gherib , Roya Radgohar , Anastasiia Pusenkova , Seyyed Mehdi Hosseini Jenab , Sara Turcotte , Dany Lachance-Quirion , Nicholas E. Frattini , Scott N. Genin This is my paper
classification physics.chem-ph
keywords variational quantum algorithmcontinuous-variable quantum computingGaussian wavepacket dynamicsfrozen Gaussianstime-dependent variational principlenuclear dynamicsvibronic spectroscopyqubit-oscillator hardware
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.

The reading

This paper tries to establish that a hybrid qubit-oscillator processor can run a variational quantum algorithm for adiabatic nuclear dynamics, with the nuclear wavepacket represented as a superposition of frozen Gaussians. The algorithm encodes 2^N Gaussians on one bosonic mode plus N ancilla qubits, so each additional qubit doubles the size of the basis, and evolves the circuit parameters with the time-dependent variational principle. In numerical tests on three one-dimensional model potentials, the superposition variant converges to the numerically exact autocorrelation function and wavepacket moments as N grows for harmonic and Morse potentials, and it captures the bifurcation and recurrence of a double-well wavepacket, with only moderate accuracy in the later irregular regime. The paper also reports a physical-hardware measurement of the SO2 vibronic autocorrelation function from classically computed variational parameters, and the resulting spectrum matches experimental peak spacings and intensities. If the claims hold, this is a concrete pathway toward simulating molecular vibrational dynamics on near-term quantum devices.

Core claim

On the paper's own terms, the central discovery is that a CV-DV variational ansatz built from frozen Gaussians can reproduce exact solutions of the nuclear time-dependent Schrödinger equation in one dimension. The wave function is loaded into a register of one qumode, one system qubit, and N ancilla qubits so that each ancilla computational basis state labels a frozen Gaussian; the circuit parameters—qubit rotation angles, phase-space displacements, and phases—are then propagated by the McLachlan variational principle for a single Gaussian and the Kramer–Saraceno variational principle for a superposition. Benchmarked against split-operator calculations, the superposition variant converges to

What carries the argument

The central object is the encoded wavepacket ansatz |Ψ(λ)⟩ of Eq. (7): a superposition of 2^N frozen Gaussians, where each Gaussian keeps a fixed width and carries its own phase-space center, amplitude, and phase, all controlled by N ancilla qubits and one system qubit via rotation gates and conditional displacement gates. The evolution is carried by the equations of motion of the time-dependent variational principle—Re(M) λ̇ = (1/ℏ) Im(V) for the single-Gaussian variant and Im(M) λ̇ = −(1/ℏ) Re(V) for the superposition variant—where M is the overlap matrix of parameter derivatives and V is the Hamiltonian-gradient vector. These matrix elements are evaluated by quantum circuits that embed pa

Load-bearing premise

The dynamics are driven by overlap and Hamiltonian-gradient matrix elements estimated by comparing states shifted by a tiny parameter step of 0.001, and the paper does not test this finite-difference approximation against exact derivatives; some of those matrix elements are as small as 10^-7, so errors at this step could materially change the propagated wavepacket.

What would settle it

Recompute the single- and multi-Gaussian propagations on the quadratic, Morse, and double-well potentials using the same variational equations but with exact derivative evaluations (for example, automatic differentiation of the circuit, or a derived parameter-shift rule) and compare the resulting autocorrelation functions to the finite-difference ε=0.001 results; if they diverge significantly, the reported convergence to exact dynamics is an artifact of the derivative approximation rather than a property of the variational ansatz.

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

If this is right

  • If the convergence claim holds, vibrational and vibronic spectra can be obtained by Fourier transforming the autocorrelation function produced by the variational circuit, with accuracy improving monotonically as ancilla qubits are added.
  • The exponential scaling of basis size with ancilla count means that simulating wavepacket dynamics with many Gaussians requires only logarithmic qubit resources, making the approach a candidate for near-term devices whenever the target state lies within the ansatz's reachable subspace.
  • For double-well systems, the algorithm reliably captures the physics that matters for tunneling and inversion spectroscopy—wavepacket splitting and coherent recurrence—while the later irregular regime remains approximate; users should match the number of Gaussians to the dynamical timescale of interest.
  • The physical autocorrelation measurement on a cavity-transmon device demonstrates that oscillator-qubit platforms can serve as hardware accelerators for computing the spectroscopic quantities that flow from variational dynamics, even when the variational parameters themselves are obtained classically.
  • The variational equations of motion, together with the circuit constructions for M and V, can be transported to other potentials and, as the authors note, extended toward nonadiabatic dynamics and higher dimensions by assigning Gaussians to electronic states.

Where Pith is reading between the lines

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

  • The finite-difference derivative step with ε=0.001 is the most replaceable component: exact parameter-shift rules for conditional-displacement gates would likely eliminate the reported ~10^18-shot sampling bottleneck and could make the full variational loop executable on hardware, not just the final autocorrelation readout.
  • The ansatz's expressivity restriction—Gaussian centers live in an (N+1)-dimensional subspace of the full 2^N-dimensional center space—implies a sharp test: prepare target superpositions whose Gaussian centers are not sum-distinct, and the algorithm should fail no matter how many ancillas are added; such a test would cleanly separate representational limits from propagation error.
  • For the double-well case, the moderate accuracy in the irregular regime is consistent with the need for time-dependent Gaussian widths or non-Gaussian gates; adding squeezing operations to the qumode could improve late-time dynamics without increasing ancilla count.
  • The one-dimensional demonstrations suggest a natural scaling probe: apply the same variational loop to a two-mode system where exact multi-configuration dynamics exhibit mode correlation; whether the CV-DV ansatz captures entanglement between modes would be a decisive next experiment.
Share X Bluesky LinkedIn Reddit HN

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

3 major / 5 minor

Summary. The manuscript proposes a variational hybrid continuous-variable/discrete-variable (CV-DV) quantum algorithm for adiabatic nuclear dynamics. The nuclear wavefunction is encoded as a superposition of frozen Gaussians on a register of one qumode plus qubits, with the variational parameters evolved according to the time-dependent variational principle. Two variants are considered: a single-FG variant using the McLachlan variational principle and a multi-FG variant using the Kramer-Saraceno variational principle. The required matrix elements of the variational equations are estimated with CV-DV circuits and finite-difference derivative approximations. The authors benchmark the multi-FG variant against split-operator exact dynamics for 1D harmonic, Morse, and double-well potentials, reporting convergence with increasing Gaussian number for the first two and moderate agreement for the double-well. They also compute the SO2 photoelectron autocorrelation function on a cavity-transmon device, but using variational parameters obtained classically rather than by running the variational EOM on the hardware. The paper explicitly discusses three limitations: sampling overhead, reduced expressivity of the ansatz, and arbitrariness of the initial Gaussian spacing.

Significance. If the numerical results are taken at face value, the paper provides a useful proof of concept that a CV-DV circuit parameterization of a frozen-Gaussian superposition can reproduce exact 1D dynamics for harmonic and anharmonic models, and can partially capture double-well bifurcation and recurrence. The independent split-operator benchmarks and the candid discussion of limitations in Sec. III F strengthen the credibility of the classical emulation results. The hardware demonstration of the autocorrelation circuit is a modest but real experimental step. However, the advertised 'pathway for simulating molecular dynamics on NISQ devices' is not supported by the paper's own analysis: the core variational EOM requires roughly 10^18 shots per matrix element (Sec. III F 1), and the hardware experiment only evaluates the autocorrelation function with classically prepared parameters. The significance of the work is therefore primarily as a classical emulation study and a hardware-demonstration of a single circuit block, not as a viable NISQ algorithm as currently presented.

major comments (3)
  1. [Abstract, Sec. III B, Sec. III F 1] The abstract and conclusion claim that this work 'establishes a pathway for simulating molecular dynamics on NISQ devices.' Section III F 1 states that N_shots ≈ 10^18 for a single matrix element, making even the single-FG variant 'out of reach for computation on physical quantum hardware.' Section III B does not run the variational EOM on hardware; it uses classically computed variational parameters and only evaluates the autocorrelation circuit. Thus the central NISQ-pathway claim is contradicted by the manuscript's own evidence. Please either reframe the claim as a classical proof of concept with a hardware demonstration of a sub-circuit, or provide a concrete, quantitative route by which the sampling overhead could be reduced within a NISQ setting.
  2. [Sec. II A, Eqs. (3)-(5)] The variational equations are built entirely from finite central differences with a fixed epsilon = 0.001 for all parameters. No convergence study in epsilon, no comparison with analytic derivatives, and no error estimate is provided. Since Sec. III F 1 reports that relevant expectation values can be as small as 10^-7, finite-difference truncation and cancellation errors could be comparable to the signal. This is load-bearing because any error in Re(M) or Im(V) propagates directly into the EOM and hence into all reported dynamics. Please add an epsilon-convergence test (or use exact/shift-rule derivatives) and quantify the resulting error for at least one representative trajectory.
  3. [Sec. III A, Sec. III F 3] The initial Gaussian spacing Delta x is chosen per system by experimentation ('we experimented with different values of Delta x and chose the ones that performed best'). The reported convergence with increasing Gaussian number is therefore not shown to be a property of the ansatz independent of this free parameter. A sensitivity analysis over Delta x (or an automatic selection criterion) is needed to establish that the harmonic/Morse convergence and the double-well behavior are robust rather than the result of per-system tuning. Without this, the convergence claims in Figs. 10-12 are weaker than stated.
minor comments (5)
  1. [Sec. III F 1] The shot-count estimate contains an apparent inconsistency: for a target relative error of 1% with |<sigma_z>| ~ 10^-7, the required standard error is 10^-9, hence sigma_x^2 ~ 10^-18, not 10^-9 as written. Please correct the exponent.
  2. [Sec. II A, Eq. (3)] epsilon is called 'an arbitrarily small number' but a fixed value 0.001 is used, and the same value is applied to parameters with different units (position, momentum, angle). A dimensionally consistent or parameter-specific epsilon should be described.
  3. [Sec. II B, Eq. (7)] The notation 'bin(k)_n' is used for the n-th digit of the binary representation of k, but the least/most significant convention is not stated. Please clarify.
  4. [Sec. III B] The hardware autocorrelation data are obtained with 1000 shots and no error bars or statistical uncertainty are shown. Reporting confidence intervals would help the reader judge the agreement.
  5. [Sec. III F 2] The expressivity limitation is clearly explained, but its consequence for the large-N convergence claims should be stated explicitly: the ansatz is not complete in the space of arbitrary superpositions of 2^N Gaussians, so convergence results for specific models do not imply general convergence.

Circularity Check

0 steps flagged

No significant circularity: variational EOM are derived from first principles and tested against independent split-operator dynamics.

full rationale

The claimed derivation chain is self-contained. The equations of motion (Eqs. 2 and 8) are obtained directly from the McLachlan and Kramer-Saraceno variational principles applied to the explicit ansaetze (Eqs. 1 and 7); the matrix elements M and V are defined from the ansatz, and their evaluation by finite differences (Eqs. 3-5) is an implementation choice, not a fit to the target dynamics. The reported benchmarks against split-operator exact dynamics for harmonic, Morse, and double-well potentials are external comparisons: the convergence of the superposition-of-FGs variant to the numerically exact autocorrelation is checked against a separate split-operator calculation, so the agreement is not enforced by construction. The single self-citation (Ref. [31]) is used only as background for double-well challenges and is not load-bearing. The main genuine weaknesses - the unvalidated finite-difference step epsilon=0.001, the per-system tuning of Delta x, and the hardware experiment evaluating only the autocorrelation from classically computed parameters (Sec. III B) while Sec. III F 1 concedes ~10^18 shots per matrix element on hardware - are accuracy and scope limitations, not circularity. No equation in the paper reduces a predicted quantity to an input by definition.

Axiom & Free-Parameter Ledger

2 free parameters · 4 axioms · 0 invented entities

The central claim rests on the frozen-Gaussian ansatz, the finite-difference evaluation, and the variational principle. The main hand-tuned quantity is Delta x, with epsilon as an additional chosen hyperparameter. No new physical entities are introduced.

free parameters (2)
  • Delta x (initial Gaussian spacing) = tuned per system (values not listed)
    In Sec. III A, the authors state they experimented with different values of Delta x and chose the ones that performed best; this hand-tuning affects the conditioning of the overlap matrix and the achievable time steps.
  • epsilon (finite-difference shift) = 0.001
    Used in Eqs. 3-5 to approximate derivatives of the wavefunction with respect to variational parameters. The value is chosen globally without an error analysis, and the paper notes expectation values can be as small as 10^-7, making the finite-difference estimates potentially noisy.
axioms (4)
  • domain assumption The frozen Gaussian ansatz (Eq. 7) spans the relevant subspace of nuclear Hilbert space for the tested dynamics.
    The expressivity analysis (Sec. III F 2) shows the ansatz is restricted to an O(N)-dimensional subspace, so the convergence claims depend on whether the true dynamics stay in that subspace for the tested potentials.
  • ad hoc to paper Finite central differences with epsilon = 0.001 accurately approximate the derivatives in Eqs. 4 and 5.
    The paper uses this approximation without error analysis; Sec. III F 1 notes the expectation values can be as small as 10^-7, making the derivative estimates potentially noisy.
  • standard math The time-dependent variational principles (McLachlan and Kramer-Saraceno) yield dynamics that converge to exact as the ansatz becomes exact.
    This is a well-established property of variational dynamics, cited via refs 7, 16, 29.
  • domain assumption The 1D models (harmonic, Morse, double-well) are representative of the dynamics the algorithm targets.
    The paper limits itself to 1D and notes higher-dimensional extensions are future work.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of A variational hybrid continuous-variable discrete-variable quantum algorithm for adiabatic nuclear dynamics." pith.science (2026). https://pith.science/paper/GW5YKOIX

@misc{pith2026260803907,
  author       = {Pith},
  title        = {Pith review of: A variational hybrid continuous-variable discrete-variable quantum algorithm for adiabatic nuclear dynamics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GW5YKOIX}},
  note         = {Machine review of arXiv:2608.03907}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Gaussian wavepacket (GWP) methods are prevalent means of solving the nuclear time-dependent Schrodinger equation (TDSE) on classical computers. They consist of representing the nuclear wave function as a superposition of Gaussians. Here, we present a variational hybrid continuous-variable discrete-variable (CV-DV) algorithm for simulating adiabatic nuclear dynamics on qubit-oscillator quantum hardware. It encodes a superposition of frozen Gaussians (FGs) onto the qubit-oscillator register and evolves it according to the time-dependent variational principle (VP). We test two variants of the approach, one that evolves a single FG and another that evolves a superposition of FGs, on a set of prototypical 1D harmonic and anharmonic systems. We simulate the photoexcited state dynamics of SO2 using the single FG variant of the algorithm and compute its autocorrelation function on physical quantum hardware. For harmonic and Morse potentials, the variant evolving the superposition of FGs converges to the numerically exact solution as the number of Gaussians increases. In the case of the double-well potential, the algorithm simulates wavepacket bifurcation and recurrence correctly, and the ensuing irregular dynamics moderately well. Overall, this work establishes a pathway for simulating molecular dynamics on noisy intermediate-scale quantum devices.

Figures

Figures reproduced from arXiv: 2608.03907 by Anastasiia Pusenkova, Dany Lachance-Quirion, Nicholas E. Frattini, Rami Gherib, Roya Radgohar, Sara Turcotte, Scott N. Genin, Seyyed Mehdi Hosseini Jenab.

Figure 1
Figure 1. Figure 1: FIG. 1. The CV-DV circuit for approximating Re ( [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. The CV-DV circuit for approximating Im ( [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. CV-DV circuit for embedding a superposition of 2 [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. This circuit approximates Im( [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. This circuit approximates Re( [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Circuit computing [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7. (a) Photo of the hardware assembly. (b) Schematic of hardware architecture. A 3-dimensional coaxial single-mode [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: FIG. 8. (Left) Real and imaginary parts of the autocorrelation function evaluated using physical quantum hardware. (Right) [PITH_FULL_IMAGE:figures/full_fig_p010_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: FIG. 9. Circuit computing [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: FIG. 10. Panels (a), (c) and (e) compare the real and imaginary parts of [PITH_FULL_IMAGE:figures/full_fig_p011_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: FIG. 11. Panels (a), (c), (e) and (g) compare the real and imaginary parts of [PITH_FULL_IMAGE:figures/full_fig_p012_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: FIG. 12. Panels (a), (b), (c), (d), (e) and (f) compare numerically exact [PITH_FULL_IMAGE:figures/full_fig_p013_12.png] view at source ↗

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

55 extracted references · 48 canonical work pages · 3 internal anchors

  1. [1]

    Our VQA has that same problem

    Sampling overhead A common bottleneck for VQAs [12, 49] is their potentially high sampling overhead. Our VQA has that same problem. Specifically, since matrix elements ofMandVare expectation values approximated through a finite number of measurements, in order for the algorithm to produce accurate dynamics, the statistical averages must closely approximat...

    work page

  2. [2]

    Loosely speaking, an ansatz is said to have high expressivity if, by changing its parameters, it can generate diverse quantum states throughout the Hilbert space

    Expressivity of the ansatz In the context of parameterized quantum circuits, expressivity refers to the range of quantum states an ansatz can produce as a function of its parameters [50, 51]. Loosely speaking, an ansatz is said to have high expressivity if, by changing its parameters, it can generate diverse quantum states throughout the Hilbert space. Co...

    work page

  3. [3]

    As discussed earlier in Sec

    Arbitrariness of the initial conditions The second limitation pertains to the initial conditions. As discussed earlier in Sec. III A, in our implementation, Gaussians are given initial positions determined by the parameter ∆x, which controls the initial spread between Gaussians. Because the ancilla Gaussians are not initially populated, ∆xis arbitrary. Th...

    work page

  4. [4]

    Kosloff, J

    R. Kosloff, J. Phys. Chem.92, 2087 (1988)

    2087

  5. [5]

    M. H. Beck, A. J¨ ackle, G. A. Worth, and H.-D. Meyer, Phys. Rep.324, 1 (2000)

    work page 2000

  6. [6]

    Sawada, R

    S.-I. Sawada, R. Heather, B. Jackson, and H. Metiu, J. Chem. Phys.83, 3009 (1985)

    work page 1985

  7. [7]

    Sawada and H

    S.-I. Sawada and H. Metiu, J. Chem. Phys.84, 6293 (1986)

    work page 1986

  8. [8]

    G. W. Richings, I. Polyak, K. E. Spinlove, G. A. Worth, I. Burghardt, and B. Lasorne, Int. Rev. Phys. Chem.34, 269 (2015)

    work page 2015

  9. [9]

    G. A. Worth and I. Burghardt, Chem. Phys. Lett.368, 502 (2003)

    work page 2003

  10. [10]

    Mendive-Tapia, B

    D. Mendive-Tapia, B. Lasorne, G. A. Worth, M. A. Robb, and M. J. Bearpark, J. Chem. Phys.137(2012)

    work page 2012

  11. [11]

    G. A. Worth and B. Lasorne, Quantum Chemistry and Dynamics of Excited States: Methods and Applications , 413 (2020)

    work page 2020

  12. [12]

    R. P. Feynman, inFeynman and computation(cRc Press, 2018) pp. 133–153

    work page 2018

  13. [13]

    Tacchino, A

    F. Tacchino, A. Chiesa, S. Carretta, and D. Gerace, Advanced Quantum Technologies3, 1900052 (2020)

    2020

  14. [14]

    Miessen, P

    A. Miessen, P. J. Ollitrault, F. Tacchino, and I. Tavernelli, Nat. Comput. Sci.3, 25 (2023)

    2023

  15. [15]

    Cerezo, A

    M. Cerezo, A. Arrasmith, R. Babbush, S. C. Benjamin, S. Endo, K. Fujii, J. R. McClean, K. Mitarai, X. Yuan, L. Cincio, et al., Nat. Rev. Phys.3, 625 (2021)

    work page 2021

  16. [16]

    Lee, C.-Y

    C.-K. Lee, C.-Y. Hsieh, S. Zhang, and L. Shi, J. Chem. Theory Comput.18, 2105 (2022)

    work page 2022

  17. [17]

    P. J. Ollitrault, A. Miessen, and I. Tavernelli, Acc. Chem. Res.54, 4229 (2021)

    work page 2021

  18. [18]

    P. J. Ollitrault, S. Jandura, A. Miessen, I. Burghardt, R. Martinazzo, F. Tacchino, and I. Tavernelli, Quantum7, 1139 (2023)

    work page 2023

  19. [19]

    McLachlan, Mol

    A. McLachlan, Mol. Phys.8, 39 (1964)

    work page 1964

  20. [20]

    Li and S

    Y. Li and S. C. Benjamin, Phys. Rev. X7, 021050 (2017)

    2017

  21. [21]

    Simulations of Many-Body Quantum Systems by a Quantum Computer

    S. Wiesner, arXiv preprint quant-ph/9603028 (1996)

    work page internal anchor Pith review Pith/arXiv arXiv 1996

  22. [22]

    Zalka, Proc

    C. Zalka, Proc. R. Soc. A: Math. Phys. Eng. Sci.454, 313 (1998)

    work page 1998

  23. [23]

    Somma, G

    R. Somma, G. Ortiz, E. Knill, and J. Gubernatis, Int. J. Quantum Inf.1, 189 (2003)

    work page 2003

  24. [24]

    Kassal, S

    I. Kassal, S. P. Jordan, P. J. Love, M. Mohseni, and A. Aspuru-Guzik, Proc. Natl. Acad. Sci. U.S.A.105, 18681 (2008)

    work page 2008

  25. [25]

    P. J. Ollitrault, G. Mazzola, and I. Tavernelli, Phys. Rev. Lett.125, 260511 (2020)

    2020

  26. [26]

    Y. Liu, S. Singh, K. C. Smith, E. Crane, J. M. Martyn, A. Eickbusch, A. Schuckert, R. D. Li, J. Sinanan-Singh, M. B. Soley,et al., PRX Quantum7, 010201 (2026)

    work page 2026

  27. [27]

    R. J. MacDonell, C. E. Dickerson, C. J. Birch, A. Kumar, C. L. Edmunds, M. J. Biercuk, C. Hempel, and I. Kassal, Chem. Sci.12, 9794 (2021)

    work page 2021

  28. [28]

    Malpathak, S

    S. Malpathak, S. D. Kallullathil, and A. F. Izmaylov, J. Phys. Chem. Lett.16, 1855 (2025)

    work page 2025

  29. [29]

    R. J. MacDonell, T. Navickas, T. F. Wohlers-Reichel, C. H. Valahu, A. D. Rao, M. J. Millican, M. A. Currington, M. J. Biercuk, T. R. Tan, C. Hempel, and I. Kassal, Chem. Sci.14, 9439 (2023)

    work page 2023

  30. [30]

    C. H. Valahu, V. C. Olaya-Agudelo, R. J. MacDonell, T. Navickas, A. D. Rao, M. J. Millican, J. B. P´ erez-S´ anchez, J. Yuen-Zhou, M. J. Biercuk, C. Hempel,et al., Nat. Chem.15, 1503 (2023)

    work page 2023

  31. [31]

    Navickas, R

    T. Navickas, R. J. MacDonell, C. H. Valahu, V. C. Olaya-Agudelo, F. Scuccimarra, M. J. Millican, V. G. Matsos, H. L. Nourse, A. D. Rao, M. J. Biercuk,et al., J. Am. Chem. Soc.147, 23566 (2025)

    work page 2025

  32. [32]

    Kramer and M

    P. Kramer and M. Saraceno,Geometry of the time-dependent variational principle in quantum mechanics(Springer, 1981)

    work page 1981

  33. [33]

    Begusic, E

    T. Begusic, E. Tapavicza, and J. Vanicek, J. Chem. Theory Comput.18, 3065 (2022). 18

    work page 2022

  34. [34]

    I. G. Ryabinkin, R. Gherib, and S. N. Genin, arXiv preprint arXiv:2405.01729 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  35. [35]

    Variational Adaptive Gaussian Decomposition: Scalable Quadrature-Free Time-Sliced Thawed Gaussian Dynamics

    R. Sharma and A. Bose, arXiv preprint arXiv:2603.10931 (2026)

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  36. [36]

    Lasser and C

    C. Lasser and C. Su, J. Math. Phys.63(2022)

    work page 2022

  37. [37]

    R. D. Coalson and M. Karplus, J. Chem. Phys.93, 3919 (1990)

    work page 1990

  38. [38]

    Petzold, SIAM J

    L. Petzold, SIAM J. Sci. Stat. Comput.4, 136 (1983)

    work page 1983

  39. [39]

    Virtanen, R

    P. Virtanen, R. Gommers, T. E. Oliphant, M. Haberland, T. Reddy, D. Cournapeau, E. Burovski, P. Peterson, W. Weckesser, J. Bright, S. J. van der Walt, M. Brett, J. Wilson, K. J. Millman, N. Mayorov, A. R. J. Nelson, E. Jones, R. Kern, E. Larson, C. J. Carey, ˙I. Polat, Y. Feng, E. W. Moore, J. VanderPlas, D. Laxalde, J. Perktold, R. Cimrman, I. Henriksen,...

    work page 2020

  40. [40]

    A. N. Tikhonovet al., inDokl. akad. nauk sssr, Vol. 39 (1943) pp. 195–198

    work page 1943

  41. [41]

    A. A. Granovsky, Firefly version 8, http://classic.chem.msu.su/gran/firefly/index.html, accessed: 2024-12-12

    work page 2024

  42. [42]

    A. D. Becke, J. Chem. Phys.98, 1372 (1993)

    work page 1993

  43. [43]

    C. Lee, W. Yang, and R. G. Parr, Phys. Rev. B37, 785 (1988)

    1988

  44. [44]

    Krishnan, J

    R. Krishnan, J. S. Binkley, R. Seeger, and J. A. Pople, J. Chem. Phys.72, 650 (1980)

    work page 1980

  45. [45]

    Blais, A

    A. Blais, A. L. Grimsmo, S. M. Girvin, and A. Wallraff, Reviews of Modern Physics93, 025005 (2021)

    2021

  46. [46]

    Lachance-Quirion, M.-A

    D. Lachance-Quirion, M.-A. Lemonde, J. O. Simoneau, L. St-Jean, P. Lemieux, S. Turcotte, W. Wright, A. Lacroix, J. Fr´ echette-Viens, R. Shillito,et al., Phys. Rev. Let.132, 150607 (2024)

    work page 2024

  47. [47]

    Eickbusch, V

    A. Eickbusch, V. Sivak, A. Z. Ding, S. S. Elder, S. R. Jha, J. Venkatraman, B. Royer, S. M. Girvin, R. J. Schoelkopf, and M. H. Devoret, Nat. Phys.18, 1464 (2022)

    work page 2022

  48. [48]

    Holland, M

    D. Holland, M. MacDonald, M. Hayes, P. Baltzer, L. Karlsson, M. Lundqvist, B. Wannberg, and W. Von Niessen, Chem. Phys.188, 317 (1994)

    work page 1994

  49. [49]

    I. G. Ryabinkin, L. Joubert-Doriol, and A. F. Izmaylov, J. Chem. Phys.140(2014)

    work page 2014

  50. [50]

    Kol/os and L

    W. Kol/os and L. Wolniewicz, J. Chem. Phys.45, 509 (1966)

    work page 1966

  51. [51]

    Swalen and J

    J. Swalen and J. A. Ibers, J. Chem. Phys.36, 1914 (1962)

    work page 1914

  52. [52]

    Scriva, N

    G. Scriva, N. Astrakhantsev, S. Pilati, and G. Mazzola, Phys. Rev. A109, 032408 (2024)

    work page 2024

  53. [53]

    S. Sim, P. D. Johnson, and A. Aspuru-Guzik, Adv. Quantum Technol.2, 1900070 (2019)

    work page 2019

  54. [54]

    Ragone, B

    M. Ragone, B. N. Bakalov, F. Sauvage, A. F. Kemper, C. Ortiz Marrero, M. Larocca, and M. Cerezo, Nat. Commun.15, 7172 (2024)

    work page 2024

  55. [55]

    Lubich, Math

    C. Lubich, Math. Comput.74, 765 (2005)

    work page 2005

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.