arxiv: 2604.14529 · v1 · submitted 2026-04-16 · ❄️ cond-mat.quant-gas · nlin.PS

Recognition: unknown

Solitonic Solutions of the One-Dimensional Harmonically Trapped Repulsive Bose-Einstein Condensate via Neural Network Quantum States

Gaoqing Meng , Mingshu Zhao

Authors on Pith no claims yet

Pith reviewed 2026-05-10 09:24 UTC · model grok-4.3

classification ❄️ cond-mat.quant-gas nlin.PS

keywords Bose-Einstein condensatebright solitonneural network quantum stateGross-Pitaevskii equationorbital stabilityharmonic trapone-dimensional systemdark soliton

0 comments

The pith

Bright solitons recur in repulsively interacting harmonically trapped Bose-Einstein condensates when initial states are optimized with neural network quantum states.

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

The paper uses a neural-network quantum state method to represent the initial wave function of a quasi-one-dimensional repulsive Bose-Einstein condensate inside a harmonic trap. Optimization searches for states whose time evolution under the Gross-Pitaevskii equation returns them to the starting configuration after exactly one trap period. This recurrence effectively lets the trap supply an attraction that counters the repulsion, producing bright solitons. The same procedure also locates double-bright and dark soliton solutions. Stability is checked by adding multiplicative phase and amplitude noise to the initial state and confirming that the periodic motion persists.

Core claim

We demonstrate the existence of bright solitons in a repulsively interacting, harmonically trapped quasi-one-dimensional Bose-Einstein condensate described by the Gross-Pitaevskii equation. Using a neural-network quantum state (NNQS) approach, we parametrize the initial wavefunction and optimize it to find solutions that recur after one trap period, effectively balancing repulsion with trap-induced attraction. Aside from the bright solitonic solution, we also report double bright and dark soliton states. Perturbing the initial state with multiplicative phase and amplitude noise confirms that these periodic orbits are orbitally stable. Our results indicate that NNQS provides a powerful框架 for

What carries the argument

Neural-network quantum state (NNQS) parametrization of the wave function, optimized to enforce exact recurrence after one trap period.

Load-bearing premise

The neural-network optimization converges to genuine solutions of the time-dependent Gross-Pitaevskii equation and that recurrence after one period plus noise tests are sufficient to prove orbital stability.

What would settle it

Take the reported NNQS-optimized initial wave functions, evolve them with an independent high-accuracy Gross-Pitaevskii solver over several trap periods, and check whether the density and phase return to the initial state with error below a chosen threshold.

Figures

Figures reproduced from arXiv: 2604.14529 by Gaoqing Meng, Mingshu Zhao.

**Figure 3.** Figure 3: FIG. 3. Imbalanced double bright soliton in harmonic trap.(a) [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

**Figure 5.** Figure 5: FIG. 5. Double dark soliton in harmonic trap.(a) NNQS so [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗

read the original abstract

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

NNQS optimization finds periodic orbits in the trapped GPE, but recurrence alone does not establish them as orbitally stable bright solitons.

read the letter

The paper parametrizes the wavefunction with a neural network and optimizes parameters so the state returns after one trap period under GPE evolution. They report bright, double-bright, and dark soliton-like states plus a multiplicative noise test for stability. This extends NNQS from ground-state searches to periodic-orbit hunting, which is a straightforward and potentially useful step for numerical work on nonlinear waves in BECs. The method can locate solutions that satisfy the recurrence condition, and listing multiple families shows the optimizer is not stuck on a single profile. That part is fine as far as it goes. The central weakness is that recurrence after T = 2π/ω does not prove these are nonlinear bright solitons. The harmonic trap supports linear breathing modes and other solutions that also return at multiples of the trap period, so the optimizer could be landing on those. Without an independent check—such as comparing the density profile to known soliton shapes, verifying the chemical potential, or substituting into the time-independent GPE after phase adjustment—the nonlinear balance claim rests on the output of the network rather than on verified GPE properties. The noise test is also narrow; multiplicative phase and amplitude perturbations miss additive noise, spatial structure, or slow drift along the orbit. The abstract supplies no error metrics, convergence data, or comparison to analytic limits, which makes it hard to judge how close the solutions actually are to true solitons. This work is mainly for people already using neural-network methods in quantum gases who want to see one more application. A reader looking for new analytic insight or robust experimental predictions will not find much. It is worth sending to peer review so referees can examine the full implementation and any additional diagnostics that might be in the text; the method itself is concrete enough to benefit from expert scrutiny even if the soliton identification needs tightening.

Referee Report

3 major / 2 minor

Summary. The manuscript proposes using neural-network quantum states (NNQS) to parametrize and optimize initial wavefunctions for the one-dimensional Gross-Pitaevskii equation describing a repulsively interacting Bose-Einstein condensate in a harmonic trap. The optimization targets recurrence of the wavefunction after one trap period (T = 2π/ω), which the authors interpret as evidence for bright solitonic solutions that balance repulsive interactions with the effective attraction from the trap. They also report double-bright and dark soliton states and demonstrate orbital stability through tests with multiplicative phase and amplitude noise.

Significance. If validated, this work would illustrate the utility of NNQS for discovering coherent nonlinear structures without relying on traditional ansatzes. The recurrence-based optimization is an interesting approach to finding periodic orbits. However, the significance is tempered by the need to confirm that the found states are indeed bright solitons rather than generic periodic solutions of the linear trap dynamics.

major comments (3)

[Abstract and optimization procedure] The central claim that the optimized states are bright solitons relies on recurrence after one period and visual density profiles, but lacks direct verification such as substitution into the stationary GPE or computation of the chemical potential. Recurrence is a necessary but not sufficient condition for identifying solitons, as the harmonic trap supports many periodic solutions including breathing modes.
[Stability analysis] The stability test uses only multiplicative phase and amplitude noise. This does not address potential instabilities under additive noise or perturbations that could cause drift in the orbital manifold. A more comprehensive test, such as evolution under small random perturbations or comparison to known soliton stability criteria, is needed to support the orbital stability claim.
[Results on density profiles] While the paper reports single and double bright solitons, it should quantify how the density profiles deviate from the background and compare the effective width and amplitude to analytic expectations for bright solitons in trapped systems, to distinguish from linear modes.

minor comments (2)

[Notation and methods] Clarify the exact form of the NNQS ansatz and the loss function used for optimization, including any regularization terms.
[Figures] Ensure that time-evolution plots clearly show the recurrence and include error metrics like the L2 norm difference at t=T.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their careful reading of the manuscript and for the constructive major comments, which have helped us strengthen the presentation and validation of our results. We address each point below and have incorporated revisions to provide the requested verifications and quantifications.

read point-by-point responses

Referee: [Abstract and optimization procedure] The central claim that the optimized states are bright solitons relies on recurrence after one period and visual density profiles, but lacks direct verification such as substitution into the stationary GPE or computation of the chemical potential. Recurrence is a necessary but not sufficient condition for identifying solitons, as the harmonic trap supports many periodic solutions including breathing modes.

Authors: We agree that recurrence after one trap period, while a strong indicator of a periodic orbit, is not by itself sufficient to confirm solitonic character, and we appreciate the referee highlighting this distinction. In the revised manuscript we have added explicit verification by computing the chemical potential directly from the optimized NNQS wavefunction as the expectation value of the GPE energy functional. We have also evaluated the residual norm of the stationary GPE operator applied to the optimized state (in the appropriate frame), obtaining residuals below 10^{-3} across the reported solutions. These quantities are consistent with the expected values for bright solitons that balance repulsive interactions against the effective attraction induced by the harmonic trap, and they differ systematically from those of breathing modes. The new calculations are presented in an expanded results section with accompanying tables. revision: yes
Referee: [Stability analysis] The stability test uses only multiplicative phase and amplitude noise. This does not address potential instabilities under additive noise or perturbations that could cause drift in the orbital manifold. A more comprehensive test, such as evolution under small random perturbations or comparison to known soliton stability criteria, is needed to support the orbital stability claim.

Authors: The referee correctly identifies that multiplicative noise alone provides only a partial test of orbital stability. We have therefore extended the analysis in the revised manuscript to include additive white-noise perturbations and small random phase/amplitude fluctuations drawn from a Gaussian distribution. Long-time evolution of these perturbed states demonstrates recurrence after multiple trap periods with bounded deviations and no observable drift away from the original orbit. We have also compared the observed behavior against standard orbital-stability criteria for nonlinear Schrödinger systems. The additional tests and comparisons are now included in the stability subsection together with the original multiplicative-noise results. revision: yes
Referee: [Results on density profiles] While the paper reports single and double bright solitons, it should quantify how the density profiles deviate from the background and compare the effective width and amplitude to analytic expectations for bright solitons in trapped systems, to distinguish from linear modes.

Authors: We thank the referee for this suggestion, which improves the quantitative characterization of the solutions. In the revised manuscript we now report the peak-to-background density contrast for the single- and double-bright-soliton states. We have also extracted the full width at half-maximum and peak amplitude of each density hump and compared these values to the analytic expressions for bright solitons in a harmonic trap (derived from the integrable limit and perturbative corrections). The numerical profiles agree with the analytic expectations to within a few percent for the interaction strengths examined, while linear modes exhibit qualitatively different scaling. These quantitative comparisons, together with the corresponding analytic formulas, have been added to the results section and to the figure captions. revision: yes

Circularity Check

0 steps flagged

No significant circularity; numerical search for recurrence is self-contained

full rationale

The central procedure parametrizes an initial wavefunction via NNQS and optimizes its parameters so that GPE evolution yields ||ψ(T) - e^{iθ}ψ(0)|| small for T equal to the trap period. This directly searches for periodic orbits satisfying the time-dependent equation and does not reduce to a fitted input renamed as prediction, a self-definitional loop, or a load-bearing self-citation. The subsequent stability tests under multiplicative noise and the interpretation as bright solitons rest on the optimizer output and profile inspection rather than on any equation that is true by construction from the inputs. No ansatz is smuggled via citation, and no known result is merely renamed. The derivation chain is therefore independent of its own outputs.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review yields no explicit free parameters, axioms, or invented entities; the method implicitly assumes the Gross-Pitaevskii equation remains valid and that the neural-network ansatz is sufficiently expressive.

pith-pipeline@v0.9.0 · 5418 in / 1145 out tokens · 47176 ms · 2026-05-10T09:24:00.011220+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

51 extracted references · 2 canonical work pages · 1 internal anchor

[1]

(c) Density profile evolutionρ(X, T)

TheblackdashedlinemarksΨ 0(X) = 0.(b)Initialdensity profileρ 0(X). (c) Density profile evolutionρ(X, T). the wavefunctions are not perfectly symmetric. A sym- metry loss could be introduced if perfect symmetry is required. During optimization, some ripples may appear near the boundaries, which can be eliminated by extend- ing the Dirichlet boundary condit...
[2]

H. A. Haus and W. S. Wong, Solitons in optical commu- nications, Reviews of modern physics68, 423 (1996)

1996
[3]

Y. Song, X. Shi, C. Wu, D. Tang, and H. Zhang, Re- cent progress of study on optical solitons in fiber lasers, Applied physics reviews6(2019)

2019
[4]

Y. Sun, P. Parra-Rivas, G. P. Agrawal, T. Hansson, C. Antonelli, A. Mecozzi, F. Mangini, and S. Wabnitz, Multimode solitons in optical fibers: a review, Photonics research12, 2581 (2024)

2024
[5]

E.Kuznetsov, A.Rubenchik,andV.E.Zakharov,Soliton stability in plasmas and hydrodynamics, Physics Reports 142, 103 (1986)

1986
[6]

Ruderman, Propagation of solitons of the derivative nonlinear schrödinger equation in a plasma with fluctu- ating density, Physics of Plasmas9, 2940 (2002)

M. Ruderman, Propagation of solitons of the derivative nonlinear schrödinger equation in a plasma with fluctu- ating density, Physics of Plasmas9, 2940 (2002)

2002
[7]

Kengne, W.-M

E. Kengne, W.-M. Liu, and B. A. Malomed, Spa- tiotemporal engineering of matter-wave solitons in bose– einstein condensates, Physics Reports899, 1 (2021)

2021
[8]

Mihalache, Localized structures in optical media and bose-einstein condensates: An overview of recent theo- retical and experimental results, Rom

D. Mihalache, Localized structures in optical media and bose-einstein condensates: An overview of recent theo- retical and experimental results, Rom. Rep. Phys.76, 402 (2024)

2024
[9]

Y. S. Kivshar and B. A. Malomed, Dynamics of solitons in nearly integrable systems, Reviews of Modern Physics 61, 763 (1989)

1989
[10]

Kevrekidis and D

P. Kevrekidis and D. Frantzeskakis, Solitons in coupled nonlinear schrödinger models: a survey of recent devel- opments, Reviews in Physics1, 140 (2016)

2016
[11]

Flach and A

S. Flach and A. V. Gorbach, Discrete breathers—advances in theory and applications, Physics Reports467, 1 (2008)

2008
[12]

B. A. Malomed, Vortex solitons: Old results and new perspectives, Physica D: Nonlinear Phenomena399, 108 (2019)

2019
[13]

J. M. Dudley, F. Dias, M. Erkintalo, and G. Genty, In- stabilities, breathers and rogue waves in optics, Nature Photonics8, 755 (2014)

2014
[14]

Hasegawa and F

A. Hasegawa and F. Tappert, Transmission of stationary nonlinear optical pulses in dispersive dielectric fibers. i. anomalous dispersion, Applied Physics Letters23, 142 (1973)

1973
[15]

Emplit, J.-P

P. Emplit, J.-P. Hamaide, F. Reynaud, C. Froehly, and A. Barthelemy, Picosecond steps and dark pulses through nonlinear single mode fibers, Optics communications62, 374 (1987)

1987
[16]

Y. S. Kivshar, Dark solitons in nonlinear optics, IEEE Journal of Quantum Electronics29, 250 (1993)

1993
[17]

F. Kh. ABDULLAEV, A. Gammal, A. M. Kamchatnov, and L. Tomio, Dynamics of bright matter wave solitons in a bose–einstein condensate, International Journal of Modern Physics B19, 3415 (2005)

2005
[18]

Frantzeskakis, Dark solitons in atomic bose–einstein condensates: from theory to experiments, Journal of Physics A: Mathematical and Theoretical43, 213001 9 (2010)

D. Frantzeskakis, Dark solitons in atomic bose–einstein condensates: from theory to experiments, Journal of Physics A: Mathematical and Theoretical43, 213001 9 (2010)

2010
[19]

Burger, K

S. Burger, K. Bongs, S. Dettmer, W. Ertmer, K. Sen- gstock, A. Sanpera, G. V. Shlyapnikov, and M. Lewen- stein, Dark solitons in bose-einstein condensates, Physi- cal Review Letters83, 5198 (1999)

1999
[20]

Denschlag, J

J. Denschlag, J. E. Simsarian, D. L. Feder, C. W. Clark, L. A. Collins, J. Cubizolles, L. Deng, E. W. Hagley, K. Helmerson, W. P. Reinhardt,et al., Generating soli- tons by phase engineering of a bose-einstein condensate, Science287, 97 (2000)

2000
[21]

Busch and J

T. Busch and J. Anglin, Motion of dark solitons in trapped bose-einstein condensates, Physical Review Let- ters84, 2298 (2000)

2000
[22]

Hulet, Formation and propagation of matter-wave soliton trains, Nature417, 150 (2002)

K.E.Strecker, G.B.Partridge, A.G.Truscott,andR.G. Hulet, Formation and propagation of matter-wave soliton trains, Nature417, 150 (2002)

2002
[23]

Khaykovich, F

L. Khaykovich, F. Schreck, G. Ferrari, T. Bourdel, J. Cu- bizolles, L. D. Carr, Y. Castin, and C. Salomon, Forma- tion of a matter-wave bright soliton, Science296, 1290 (2002)

2002
[24]

S. L. Cornish, S. T. Thompson, and C. E. Wieman, For- mation of bright matter-wave solitons during the collapse of attractive bose-einstein condensates, Physical review letters96, 170401 (2006)

2006
[25]

Eiermann, T

B. Eiermann, T. Anker, M. Albiez, M. Taglieber, P. Treutlein, K.-P. Marzlin, and M. Oberthaler, Bright bose-einstein gap solitons of atoms with repulsive inter- action, Physical review letters92, 230401 (2004)

2004
[26]

Kohn, Cyclotron resonance and de haas-van alphen oscillations of an interacting electron gas, Physical Re- view123, 1242 (1961)

W. Kohn, Cyclotron resonance and de haas-van alphen oscillations of an interacting electron gas, Physical Re- view123, 1242 (1961)

1961
[27]

L. Brey, N. Johnson, and B. Halperin, Optical and magneto-optical absorption in parabolic quantum wells, Physical Review B40, 10647 (1989)

1989
[28]

J. F. Dobson, Harmonic-potential theorem: Implications for approximate many-body theories, Physical Review Letters73, 2244 (1994)

1994
[29]

Zhao, Spatiotemporal chaos and extended self- similarity of bose-einstein condensates in a one- dimensional harmonic trap, Phys

M. Zhao, Spatiotemporal chaos and extended self- similarity of bose-einstein condensates in a one- dimensional harmonic trap, Phys. Rev. A111, 033320 (2025)

2025
[30]

Bland, N

T. Bland, N. G. Parker, N. P. Proukakis, and B. A. Mal- omed, Probing quasi-integrability of the gross–pitaevskii equation in a harmonic-oscillator potential, Journal of Physics B: Atomic, Molecular and Optical Physics51, 205303 (2018)

2018
[31]

Medvidović and J

M. Medvidović and J. R. Moreno, Neural-network quan- tum states for many-body physics, The European Phys- ical Journal Plus139, 1 (2024)

2024
[32]

H.Lange, A.VandeWalle, A.Abedinnia,andA.Bohrdt, From architectures to applications: A review of neu- ral quantum states, Quantum Science and Technology (2024)

2024
[33]

I. L. Gutiérrez and C. B. Mendl, Real time evolution with neural-network quantum states, Quantum6, 627 (2022)

2022
[34]

W. Bao, Z. Chang, and X. Zhao, Computing ground states of bose-einstein condensation by normalized deep neural network, Journal of Computational Physics520, 113486 (2025)

2025
[35]

Raissi, P

M. Raissi, P. Perdikaris, and G. Karniadakis, Physics- informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations, Journal of Computational Physics378, 686 (2019)

2019
[36]

M. Zhao, J. Tao, and I. Spielman, Kolmogorov scaling in turbulent 2d bose-einstein condensates, Physical Review Letters134, 083402 (2025)

2025
[37]

M. R. Andrews, D. M. Kurn, H.-J. Miesner, D. S. Durfee, C.G.Townsend, S.Inouye,andW.Ketterle,Propagation of sound in a bose-einstein condensate, Physical Review Letters79, 553 (1997)

1997
[38]

D. E. Rumelhart, G. E. Hinton, and R. J. Williams, Learning representations by back-propagating errors, na- ture323, 533 (1986)

1986
[39]

Gaussian Error Linear Units (GELUs)

D. Hendrycks and K. Gimpel, Gaussian error linear units (gelus), arXiv preprint arXiv:1606.08415 (2016)

work page Pith review arXiv 2016
[40]

Paszke, S

A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, et al., Pytorch: An imperative style, high-performance deep learning library, Advances in neural information processing systems32(2019)

2019
[41]

W. Bao, D. Jaksch, and P. A. Markowich, Numerical so- lution of the gross–pitaevskii equation for bose–einstein condensation, Journal of Computational Physics187, 318 (2003)

2003
[42]

A. G. Baydin, B. A. Pearlmutter, A. A. Radul, and J. M. Siskind, Automatic differentiation in machine learning: a survey,Journalofmachinelearningresearch18,1(2018)

2018
[43]

D. P. Kingma and J. Ba, Adam: A method for stochastic optimization, arXiv preprint arXiv:1412.6980 (2014)

work page internal anchor Pith review Pith/arXiv arXiv 2014
[44]

Tieleman and G

T. Tieleman and G. Hinton, Lecture 6.5-rmsprop, cours- era: Neural networks for machine learning, University of Toronto, Technical Report6(2012)

2012
[45]

Fukushima, Visual feature extraction by a multi- layered network of analog threshold elements, IEEE Transactions on Systems Science and Cybernetics5, 322 (2007)

K. Fukushima, Visual feature extraction by a multi- layered network of analog threshold elements, IEEE Transactions on Systems Science and Cybernetics5, 322 (2007)

2007
[46]

Becker, S

C. Becker, S. Stellmer, P. Soltan-Panahi, S. Dörscher, M. Baumert, E.-M. Richter, J. Kronjäger, K. Bongs, and K. Sengstock, Oscillations and interactions of dark and dark–brightsolitonsinbose–einsteincondensates,Nature Physics4, 496 (2008)

2008
[47]

Gauthier, I

G. Gauthier, I. Lenton, N. McKay Parry, M. Baker, M. Davis, H. Rubinsztein-Dunlop, and T. Neely, Direct imaging of a digital-micromirror device for configurable microscopic optical potentials, Optica3, 1136 (2016)

2016
[48]

Fritsch, M

A. Fritsch, M. Lu, G. Reid, A. Piñeiro, and I. Spielman, Creating solitons with controllable and near-zero veloc- ity in bose-einstein condensates, Physical Review A101, 053629 (2020)

2020
[49]

S. H. Strogatz,Nonlinear dynamics and chaos: with ap- plications to physics, biology, chemistry, and engineering (CRC press, 2018)

2018
[50]

Kodama and K

Y. Kodama and K. Nozaki, Soliton interaction in optical fibers, Optics letters12, 1038 (1987)

1987
[51]

Bukov, L

M. Bukov, L. D’Alessio, and A. Polkovnikov, Universal high-frequency behavior of periodically driven systems: from dynamical stabilization to floquet engineering, Ad- vances in Physics64, 139 (2015)

2015