PaperlayerView exact revision on arXiv

Source-linked AI summary

Long-time continuum limits for twisted Schr\"{o}dinger lattices

Brian Choi

arXiv:2609.34081v1math.APcs.NAmath.NA

TL;DR

The paper asks whether a twisted discrete NLS lattice approximates magnetic NLS over long times. It combines Birkhoff-coordinate stability with modified energies and Fourier filtering to establish polynomial error bounds for defocusing and small-mass focusing data, while identifying a large-data focusing obstruction.

  • Problem

    The central question is whether lattice models provide a long-time continuum approximation to magnetic NLS.

    Source: §1 · p. 1

  • Method

    The proof combines Birkhoff coordinates for continuum-flow stability with modified energies for lattice Sobolev growth, and uses Fourier filtering for low regularities.

    Source: §1 · p. 2

  • Results

    Polynomial-in-time approximation bounds hold for arbitrary defocusing data and sufficiently small-mass focusing data, while modulational instability obstructs uniform polynomial bounds for general focusing data.

    Source: §1 · p. 1, §8 · p. 37

  • Takeaways & Limitations

    The approximation is rigorously justified for arbitrary defocusing and small-mass focusing data, but continuum integrability alone is insufficient for polynomial approximation of large-data focusing solutions.

    Source: §8 · p. 37

    The intermediate focusing regime between the small-mass result and the instability construction remains open.

    Source: §7.1 · p. 30

Abstract

from arXiv · show

We prove polynomial-in-time Sobolev bounds on the approximation error between the twisted discrete nonlinear Schr\"{o}dinger equation and its magnetic continuum limit. The analysis exploits complete integrability of the continuum flow through its Birkhoff representation to control approximation by a non-integrable Hamiltonian finite-difference lattice. We establish a focusing-defocusing dichotomy in uniform polynomial control: the bounds hold for arbitrary defocusing data and focusing data of sufficiently small mass, while modulational instability and aliasing yield an obstruction for large focusing data. The accompanying algebraic spatial rates are generically optimal in the Baire-category sense. The proof combines Birkhoff coordinates with modified energies for discrete Sobolev growth, while Fourier filtering recovers uniform spacetime estimates below the energy space despite degenerate lattice dispersion.

1 Introduction

The paper asks whether twisted lattice models approximate magnetic NLS over long times, and establishes polynomial-in-time error bounds with a focusing-defocusing distinction. Its analysis combines continuum integrability with discrete Sobolev control, while identifying a large-data focusing obstruction and generically optimal spatial rates.

1 Introduction

  • Uniform polynomial-in-time approximation error bounds hold for arbitrary defocusing data and focusing data with mass at most ε*, where ε*≤√π is independent of s, R, h, and a.The theorem states the same conclusion for focusing data under this small-mass condition.Source: §1 · pp. 1–2
  • Birkhoff coordinates control continuum-flow stability, while modified energies bound Sobolev growth for the non-integrable lattice.The proof uses action-dependent Birkhoff frequencies and adapts a Fourier-side modified-energy construction for fractional regularities.Source: §1 · p. 2
  • For 0 < s < 1, Fourier filtering and restriction norms recover uniform dispersive estimates by excluding degenerate lattice frequencies.The filtered scheme gives quantitative error control in terms of spatial resolution and observation time.Source: §1 · pp. 1–2
  • For sufficiently large bounded focusing data, aliasing maps continuum-stable high-frequency perturbations to low-frequency unstable lattice modes, causing order-one error at T_h ≃ |log(h)|.This obstruction shows the mass restriction is necessary for uniform polynomial error bounds.Source: §1 · p. 2
  • The spatial exponent γ_s,ρ is generically optimal in the Baire-category sense, with initialization error limiting the convergence rate to second order.The lower bound is obtained by converting the lattice-continuum phase defect into nonlinear error through high-frequency perturbations.Source: §1 · p. 3

2 Analytic framework

The analytic framework relates periodic lattice data to continuum functions and sets up magnetic continuum and lattice flows. It also records uniform linear and Duhamel estimates, alongside global well-posedness of the lattice equations.

2 Analytic framework

  • Shannon interpolation identifies the lattice Sobolev space isometrically with the band-limited subspace of continuum Sobolev space and reproduces lattice values.The cell-average discretization and Shannon interpolation define the discrete-continuum representation.Source: §2 · p. 3
  • A time-dependent gauge transformation conjugates magnetic NLS to standard NLS, while accumulated phases define the corresponding continuum and lattice linear propagators.The propagators act on Fourier coefficients by phase multiplication.Source: §2 · p. 3
  • The lattice equations are globally well-posed for every h, with absolutely continuous solutions following the ODEs almost everywhere in time.The stated basis is Carathéodory’s theorem and conservation of the discrete L2 norm.Source: §2 · p. 4
  • Restriction-norm estimates for linear evolution and Duhamel operators have constants independent of h, s, and a.Conjugating each spatial Fourier mode by its phase reduces the estimates to standard restriction-norm estimates.Source: §2 · p. 5

3 Fourier filtering and uniform local theory

Section 3 establishes uniform spacetime estimates for the regime 0 < s < 1 and uses them to obtain the nonlinear bounds needed for uniform local theory. Fourier filtering is essential: without it, uniform dispersive control fails.

3 Fourier filtering and uniform local theory

  • The section establishes a uniform spacetime estimate for the regime 0 < s < 1.This estimate is used in that regime's analysis.Source: §3 · p. 5
  • Bourgain’s embedding argument for the quadratic phase extends uniformly to the discrete phase.Source: §3 · p. 6
  • Without Fourier filtering, uniform dispersive control fails.Source: §3 · p. 8
  • The uniform L4 estimate yields the nonlinear bounds needed for uniform local theory.Source: §3 · p. 9

4 Birkhoff coordinates and tangent flow

Birkhoff coordinates represent the cubic NLS flow as coordinatewise rotations, supporting tangent-flow estimates. The Birkhoff chart is global for defocusing NLS and local around zero for focusing NLS.

4 Birkhoff coordinates and tangent flow

  • Birkhoff coordinates conjugate the cubic NLS flow to coordinatewise rotations, with a global chart for defocusing NLS and a local chart around zero for focusing NLS.Source: §4 · p. 10
  • Conjugation by the Birkhoff map reduces the tangent-flow bound's time dependence from at most exponential to linear.Source: §4 · p. 12
  • The derivative estimate also covers 0 < ρ < 1 for compact subsets of H1; in the focusing case, the relevant compact set must lie inside the local Birkhoff neighborhood.Source: §4 · pp. 12–13

5 Modified energies and polynomial Sobolev growth

This section develops polynomial-in-time control of discrete Sobolev norms through Fourier filtering and modified energies, including higher-order energies for higher regularity. It also establishes a bounded-energy regime under a threshold condition on β_h(t).

5 Modified energies and polynomial Sobolev growth

  • Discrete Sobolev norms grow at most polynomially in time, with bounds split between 0 < s < 1 and s ≥ 1.For 0 < s < 1, the stated estimates require h ≤ h0; the displayed rates distinguish 0 < s ≤ 1/2 from 1/2 < s < 1.Source: §5 · p. 13
  • The proof constructs almost-conserved coercive modified energies, adding a higher-order correction that replaces a quartic remainder with a sextic remainder and magnetic defect.For s ≥ 1, the resonance structure includes a genuinely discrete component.Source: §5 · p. 14

5.1 Fourier-filtered modified energies

  • For bandlimited f_j,F in E_h, Proposition 5.2 gives estimates uniformly in h, with implicit constants depending on s, Λ, p_j; the magnetic potential perturbs quartic cancellation by O(h|a|∞).The multiplier m_q admits a unique continuous real-valued extension to V_R.Source: §5.1 · pp. 15, 16, 18

5.2 Higher-order modified energies

  • For s > 1, the homological-equation argument relaxes the regularity requirement from C∞ to C2.Source: §5.2 · p. 19
  • Near vanishing factors, repeated applications of the Fundamental Theorem of Calculus bound the quotient locally, and compactness patches the estimates uniformly on K.Young’s convolution inequality supplies weighted bounds with constants uniform in h.Source: §5.2 · pp. 19–20

5.3 Proof of Proposition 5.1

  • When β_h(t) ≤ ε_R, the energy quantity Y(t) remains bounded by Y(0) ≲_R 1.The threshold ε_R depends on R.Source: §5.3 · p. 22
  • Higher Sobolev estimates close by induction on n = ⌈s − 1⌉, using a recurrence and Grönwall’s inequality.The induction is closed after applying the estimate at lower regularity.Source: §5.3 · p. 23

6 Proof of the main result

The proof combines consistency estimates for the filtered lattice approximation with local stability and bootstrap arguments to obtain polynomial-in-time Sobolev error bounds. For s>1 and 0<ρ<s, the stated bound is ∥e_h(t)∥_{H^ρ} ≲ h^{γ_{s,ρ}}⟨t⟩^{6(s+1)}.

6.1 Consistency estimates

  • Corollary 6.1 gives consistency estimates for bounded H^s initial data with 0≤ρ<s, subject to the support condition (71).Source: §6.1 · pp. 23, 25

6.2 Proof of the main theorem

  • On short intervals, a contraction argument constructs the lattice approximation, and interval-by-interval stability estimates support the polynomial-time error bound.The fixed point is identified with e_{h,N}.Source: §6.2 · pp. 26–27
  • For s>1 and 0<ρ<s, the approximation error satisfies ∥e_h(t)∥_{H^ρ} ≲ h^{γ_{s,ρ}}⟨t⟩^{6(s+1)}.The proof combines two regimes to obtain this bound.Source: §6.2 · p. 29

7 Obstructions and sharpness

The section shows that uniform polynomial approximation bounds fail for sufficiently large focusing data, while the algebraic spatial rates are generically sharp. The focusing obstruction leaves an intermediate mass regime unresolved.

7.1 Focusing instability and polynomial bounds

  • For sufficiently large focusing data, modulational instability and numerical aliasing obstruct uniform polynomial error bounds.The construction does not apply to arbitrarily small H^s-balls, and the intermediate regime between small-mass control and the instability construction remains open.Source: §7.1 · pp. 30, 32
  • The instability construction produces exponential growth of selected Fourier modes from highly oscillatory initial data projected to slowly oscillating discrete data.The mode amplitudes grow at least as e^(Λ_h t)|δ_h| over the construction's time interval.Source: §7.1 · pp. 33–34

7.2 Genericity of spatial optimality

  • For 0 ≤ ρ < s, Proposition 7.2 gives a dense Gδ set G ⊆ Hs; when s − ρ ≥ 4, each non-constant f ∈ Hs and T > 0 has a k0 ∈ Z \ {0} depending only on f.Source: §7.2 · p. 35

8 Conclusion

The paper establishes long-time approximation of magnetic NLS by a non-integrable lattice for arbitrary defocusing and small-mass focusing data. It also identifies a large-data focusing obstruction and directions for improving the estimates.

8 Conclusion

  • The analysis justifies approximation of integrable magnetic NLS by a non-integrable Hamiltonian lattice for arbitrary defocusing and small-mass focusing data.The argument relies on continuum stability and mesh-uniform discrete Sobolev estimates.Source: §8 · p. 37
  • For large focusing data, the obstruction shows that continuum integrability alone is insufficient for polynomial approximation.The paper identifies higher-order mesh-uniform normal-form reductions and polynomial approximation below the energy space without Fourier filtering as future directions.Source: §8 · p. 37
Loading 2609.34081v1…