Law of Large Numbers and Central Limit Theorem for random sets of solitons of the focusing nonlinear Schr"odinger equation

arXiv:2411.17036 · math-ph, math.AP, math.MP, math.PR, nlin.PS, nlin.SI · Submitted 2024-11-26 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematical Physics. The day in mathematical physics: rigorous models of physical systems.

Kai: We're looking at the paper: "Law of Large Numbers and Central Limit Theorem for random sets of solitons of the focusing nonlinear Schr"odinger equation".

Mira: Here's the gist of the paper. We study a random configuration of N soliton solutions ψN (x, t;λ) of the cubic focusing Nonlinear Schrödinger (fNLS) equation in one space dimension.

Kai: The expectation of this random measure uniquely identifies a solution ψ∞(x, t) of the fNLS equation, which can be interpreted as a soliton gas solution.

Mira: The randomness is obtained by choosing the complex eigenvalues to be i.i.d.

Kai: random variables sampled from a probability distribution with compact support in the complex plane.

Title and authors: Mira: So, moving on to what they actually found in "Law of Large Numbers and Central Limit Theorem for random sets of solitons of the focusing nonlinear Schr"odinger equation", they're establishing two major statistical results for these N soliton solutions.

Kai: They prove a Law of Large Numbers and a Central Limit Theorem for the differences between the actual N soliton solution, ψN(x, t; λ), and this deterministic limit we just talked about, ψ infinity(x, t).

Lev: That means they're saying that as N gets larger and larger—the number of solitons—the average difference between them goes to zero in a certain sense.

Kai: And the Central Limit Theorem is even more specific because it shows that when you look at the scaled differences, specifically the square root of N times those differences, they converge to some kind of random Gaussian variable.

Mira: So what this means for us is that we don't just get a rough idea of how things behave; we get a precise statistical description of the fluctuations around that average soliton gas solution.

The paper's summary: Lev: To summarize, the core contribution of this paper, "Law of Large Numbers and Central Limit Theorem for random sets of solitons of the focusing nonlinear Schr"odinger equation", is developing a rigorous statistical theory for large amplitude waves in the fNLS equation.

Kai: They are taking N soliton solutions, which we know have been interpreted as N individual particles or solitons, and they are providing a way to find their deterministic limit by averaging over random spectral data.

Mira: The key concept is that by choosing the eigenvalues randomly from a distribution with compact support in the complex plane, you can define this expectation that yields ψ infinity(x, t), which represents the soliton gas solution.

Lev: And then they rigorously prove two things: first, a Law of Large Numbers showing convergence in mean for those differences when space and time are in some compact set of R times plus R+.

Kai: And second, the Central Limit Theorem for those differences and even the difference in the squared moduli. That tells us exactly how much noise we expect around that average behavior.

Mira: It's a big step because it moves beyond just simulating these systems; it gives us a way to statistically characterize what happens when you start with random spectral data, which is often how real-world initial conditions are generated in some physical scenarios.

The paper's improvements: Kai: The paper suggests some improvements over previous work, specifically by developing techniques for error analysis. They compare the random Riemann-Hilbert problem for the N soliton solution to an averaged version.

Lev: They tackle this by looking at a specific error term, E(z), which is defined as M times N times M inverse of z, and they show that this term converges to one through a convergent Neumann series with high probability.

Mira: That convergence of the error matrix E(z) to one is crucial because it shows that the difference between the random potential psiN and the deterministic solution psi infinity is bounded under certain conditions.

Kai: So, this means they've managed to control the error term in a way that lets them prove those convergence results in mean and the Central Limit Theorem with probability bounds that tend to zero as N gets bigger.

Lev: And they also show local error estimates, like proposition three point three, which give bounds for the difference when the configuration of random points λ is in some set B delta <ref:2411.17036#pg3>.

Conclusion: Mira: So, wrapping up this discussion on "Law of Large Numbers and Central Limit Theorem for random sets of solitons of the focusing nonlinear Schr"odinger equation", the paper establishes that we can predict a macroscopic soliton gas solution from random spectral data.

Kai: They proved convergence in mean and a Central Limit Theorem for both the difference between N soliton solutions and this limit, as well as for their squared moduli differences.

Lev: For me, what this means is that when you're actually building hardware with real noise, you need to know how these fluctuations scale to figure out the required error correction overhead.

Mira: And Kai is right about the Gaussian variables; because they converge to complex and real Gaussian random variables, we can characterize those fluctuations precisely.

Kai: And Lev’s point about scaling is important; we can't just guess the noise level when designing a system for these large amplitude waves.

MANUELA GIROTTI, TAMARA GRAVA, KENNETH D. T-R MCLAUGHLIN, JOSEPH NAJNUDEL

University of Bristol

math-ph, math.AP, math.MP, math.PR, nlin.PS, nlin.SI

Submitted: 2024-11-26

Updated: 2026-05-04

Comments: 27 pages, 2 figures

Journal ref: Duke Math. J. 175, no. 7 (2026), 1265-1304

DOI: 10.1215/00127094-2025-0050

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 82/100

The gist: We study a random configuration of N soliton solutions ψN (x, t;λ) of the cubic focusing Nonlinear Schrödinger (fNLS) equation in one space dimension.

Key concepts

Soliton Gas Solution
This is the macroscopic solution obtained by taking the average of many random N-soliton solutions. It represents an idealized, smooth wave pattern that describes how solitons behave collectively when their positions and phases are randomly distributed, moving beyond individual soliton descriptions.
Law of Large Numbers (LLN)
The LLN proves that as the number of solitons (N) increases, the average behavior of the random solution $\psi_N$ gets closer and closer to the deterministic limiting solution $\psi_{\infty}$. This means that for large N, the ensemble average accurately predicts the macroscopic soliton gas.
Central Limit Theorem (CLT)
The CLT describes how the fluctuations around the average solution scale with N. It shows that when you look at scaled versions of these fluctuations, they converge to specific Gaussian random variables (complex and real), allowing for a precise statistical description of the noise in the system.

Terminology

Summary

We study a random configuration of N soliton solutions ψN (x, t;λ) of the cubic focusing Nonlinear Schrödinger (fNLS) equation in one space dimension. p. 0

Where the paper says it

T-R MCLAUGHLIN, AND JOSEPH NAJNUDEL Abstract. We study a random configuration of N soliton solutions ψN (x, t;λ) of the cubic focusing Nonlinear Schrödinger (fNLS) equation in one space dimension.

The expectation of this random measure uniquely identifies a solution ψ∞(x, t) of the fNLS equation, which can be interpreted as a soliton gas solution. p. 0
Where the paper says it

Such expectation uniquely identifies, via the Zakharov-Shabat inverse spectral problem, a solution ψ∞(x, t) of the fNLS equation. This solution can be interpreted as a soliton gas solution.

Methodology for Solution Identification

The randomness is obtained by choosing the complex eigenvalues to be i.i.d. random variables sampled from a probability distribution with compact support in the complex plane. p. 0

Where the paper says it

The randomness is obtained by choosing the complex eigenvalues to be i. random variables sampled from a probability distribution with compact support in the complex plane.

The corresponding norming constants are interpolated by a smooth function of the eigenvalues. p. 0
Where the paper says it

random variables sampled from a probability distribution with compact support in the complex plane. The corresponding norming constants are interpolated by a smooth function of the eigenvalues.

The expectation of this random measure associated to this random spectral data uniquely identifies, via the Zakharov-Shabat inverse spectral problem, a solution ψ∞(x, t) of the fNLS equation. p. 0
Where the paper says it

Then we consider the expectation of the random measure associated to this random spectral data. Such expectation uniquely identifies, via the Zakharov-Shabat inverse spectral problem, a solution ψ∞(x, t) of the fNLS equation.

Statistical Convergence Results

The authors prove a Law of Large Numbers and a Central Limit Theorem for the differences ψN (x, t;λ)−ψ∞(x, t) and ∣ψN (x, t;λ)∣2 − ∣ψ∞(x, t)∣2 when (x, t) are in a compact set of R × R+.

Convergence to the Deterministic Limit

Theorem 2.6 establishes convergence in mean, showing that lim N→∞ E[∣ψN (x, t;λ) − ψ∞(x, t)∣] = 0 and lim N→∞ E[∣ψN (x, t;λ)∣2 − ∣ψ∞(x, t)∣2] = 0 uniformly for (x, t) in a compact set of R × R+. p. 6

Where the paper says it

Then the N-soliton solution ψN (x, t;λ) and its modulus square ∣ψN (x, t;λ)∣2 converge in mean, as N → ∞, to ψ∞(x, t) and ∣ψ∞(x, t)∣2 respectively, as defined in (2. 19), namely lim N→∞ E[ ∣ψN (x, t;λ) − ψ∞(x, t)∣ ] = 0, and lim N→∞ E[∣ψN (x, t;λ)∣2 − ∣ψ∞(x, t)∣2 ] = 0, uniformly for (x, t) in a compact set of R × R + .

Fluctuation Analysis

Theorem 2.7 proves the Central Limit Theorem for the differences √N(ψN (x, t;λ) − ψ∞(x, t)) and √N(∣ψN (x, t;λ)∣2 − ∣ψ∞(x, t)∣2), showing they converge in distribution to complex and real Gaussian random variables XG1 and XG2 respectively. p. 6

Where the paper says it

7) in the domain D+, and the norming constants {ck} be interpolated by a C 1 function r according to (2. Then the random variables √ N(ψN (x, t;λ) − ψ∞(x, t)) and √ N(∣ψN (x, t;λ)∣2 − ∣ψ∞(x, t)∣2 ) converge in distribution, as N → ∞ and (x, t) in a compact set, to complex and real Gaussian random variables XG1 and XG2 respectively, with zero expectation, and covariance (i.

Correlation Function Computation

Theorem 2.8 calculates the two-point correlation function R2,N (x1, t1, x2, t2) = E[N (ψN (x1, t1;λ) − ψ∞(x1, t1)) (ψN (x2, t2;λ) − ψ∞(x2, t2))] and shows its limit is given by an integral involving the covariance function G1. p. 7

Where the paper says it

7), and the norming constants {ck} be defined as in (2. Let R2,N (x1, t1, x2, t2) ∶= E[N (ψN (x1, t1;λ) − ψ∞(x1, t1)) (ψN (x2, t2) − ψ∞(x2, t2;λ))] be the two-point correlation function.

Analyticity of the Limiting Solution

Appendix A proves that the solution ψ∞(x, t) defined in (2.19) is a classical solution to the fNLS equation and is actually analytic in both variables. p. 6

Where the paper says it

There is a unique solution M(z; x, t) to the Averaged Riemann-Hilbert problem, which determines a solution ψ∞(x, t) to the fNLS equation via ψ∞(x, t) = 2i limz→∞ z(M(z; x, t))12 , ∣ψ∞(x, t)∣2 = −2i limz→∞ z ∂x(M(z; x, t))22 . 19) Moreover, ψ∞ is a classical solution to the fNLS equation, which belongs to the class C ∞(R × R + ).

Error Bound Derivation

The error analysis compares the random RH problem 2.2 for the N-soliton solution to the Averaged RH problem 2. p. 7

Where the paper says it

Error analysis and random small norm argument In order to prove our main results we compare the random RH problem 2. 2 for the N-soliton solution to the Averaged RH problem 2.

3 by considering the error problem (3. p. 7
Where the paper says it

2 for the N-soliton solution to the Averaged RH problem 2. 3 by considering the error problem (3.

1) E(z) = MN (z)M(z)−1. The convergence of E(z) to I is shown via a convergent Neumann series with high probability. p. 9
Where the paper says it

We will now show that the integral operator in (3. 16) is invertible, thus yielding a convergent Neumann series expansion for E (except for a collection of configurations of {λj} N j=1 whose measure vanishes as N → ∞, see Proposition 4.

Local Error Estimates

Proposition 3.3 provides bounds for the difference ∣ψN (x, t;λ) − ψ∞(x, t) < when the configuration of random points λ = λ1,. p. 0

Where the paper says it

T-R MCLAUGHLIN, AND JOSEPH NAJNUDEL Abstract. We study a random configuration of N soliton solutions ψN (x, t;λ) of the cubic focusing Nonlinear Schrödinger (fNLS) equation in one space dimension.

.., λN is in the set Bδ. p. 9
Where the paper says it

20) Bδ ∶= B α=1 δ = ⋂ z∈γ+ ⎧⎪⎪ ⎨ ⎪⎪⎩ {λ1, . , λN } ∶ R R R R R R R R R R R X f N (z) N R R R R R R R R R R R < δ ⎫⎪⎪ ⎬ ⎪⎪⎭ .

Convergence to Gaussian Variables

The proof of Theorem 2.7 relies on Lemma 5. p. 6

Where the paper says it

19) Moreover, ψ∞ is a classical solution to the fNLS equation, which belongs to the class C ∞(R × R + ). Existence of the solution ψ∞ can be proved via an application of the vanishing lemma approach of Zhou [53, Theorem 9.

1, which relates the linear statistics XG N to the integral of G1(z; x, t) over D+. p. 1
Where the paper says it

This has similarities to the work [18], where a finite Toda lattice with random spectral data was used to study the statistics of deflation times. In a sense, we are introducing randomness in the linear setting of the scattering data, and studying random large-amplitude nonlinear dynamics.

The convergence in distribution is guaranteed by the Central Limit Theorem [28]. p. 0
Where the paper says it

For linear partial differential equations, random initial data is usually constructed from a superposition of uncorrelated linear waves (Fourier modes) with random phases and amplitudes satisfying the Central Limit Theorem. Thanks to the linearity of the differential equation, as time evolves this superposition of linear waves remains uncorrelated and unchanged in distribution.

Correlation Function Limit

The limit of the correlation function R2,N (x1, t1, x2, t2) as N→∞ is given by the integral of G1(s; x1, t1)G1(s; x2, t2) over D+.

Uniform Bound on the Solution

Lemma 4.2 establishes an upper bound for the N-soliton solution ψN with spectrum λk satisfying ψN (x, t;λ) ≤ 4 N ∑ k=1 Im (λk) ∀(x, t) ∈ R × R+. p. 2

Where the paper says it

1) consists of the real z axis where one defines the reflection coefficient ρ(z), and a finite collection of L 2 -eigenvalues {λ1, . , λN } which are (generically) in the upper half-plane C+, and for each eigenvalue λk there is an associated normalization constant ck ∈ C ∖ {0}.

Convergence of the Error Term

The convergence of the error term in Theorem 2.7 is shown by bounding the Neumann series terms using estimates derived from the small norm RH theory. p. 9

Where the paper says it

16) is invertible, thus yielding a convergent Neumann series expansion for E (except for a collection of configurations of {λj} N j=1 whose measure vanishes as N → ∞, see Proposition 4. We will resort to a small norm argument [33].

Probability Bounds for Convergence

The probability bounds P((Bδ ∩ Fc)) tend to zero as N → ∞, which is crucial for proving convergence in mean and the Central Limit Theorem.

Final Result on Fluctuations

The final result shows that √N(ψN (x, t;λ)−ψ∞(x, t)) converges to a complex Gaussian random variable. The Central Limit Theorem for the difference of the moduli is also established. p. 6

Where the paper says it

Next, we consider the fluctuations of the difference between the random and the deterministic solutions. 7 (Central Limit Theorem).

p. 6
Where the paper says it

Next, we consider the fluctuations of the difference between the random and the deterministic solutions. 7 (Central Limit Theorem).

Correlation Function Convergence

The correlation function calculation concludes that lim N→∞ R2,N (x1, t1, x2, t2) = ∬D+ G1(s; x1, t1)G1(s; x2, t2)dµ(s) − ∬D+ G1(s; x1, t1)dµ(s)∬D+ G1(s′; x2, t2)dµ(s′), which is the covariance of the linear Gaussian random variables XG(ξ1) and XG(ξ2).

The solution ψ∞(x, t) to the fNLS equation is analytic in x and t. The authors show that R depends analytically on x and t. p. 1

Where the paper says it

Although originally formulated for a dilute gas of solitons for the KdV equation [49], the kinetic theory has been extended to the more general case of a dense gas [24] and to soliton gasses for other equations [25, 26, 27], including the fNLS equation. The two fundamental ingredients in this kinetic theory are (1) a collection of solitons that are so abundant that they can be described in terms of an evolving "space-time density function" f(z; x, t), and (2) a separate, easily identifiable "tracer soliton" whose velocity, s(z; x, t) depends on the spectral parameter z, and is assumed to evolve in t due to the interaction with the gas of solitons.

Convergence of the Squared Modulus

The convergence of the difference in squared moduli is established by showing that its scaled version converges to a real Gaussian random variable. The variance is explicitly computed. p. 6

Where the paper says it

21) E[X Gi (x, t) 2 ] = ∬D+ Gi(w; x, t) 2 dµ(w) − (∬D+ Gi(w; x, t)dµ(w))2 , where the functions G1 and G2 are defined in (5. The expectation of the variance (i.

Final Conclusion on Fluctuations

The final result shows that √N(ψN (x, t;λ)−ψ∞(x, t)) converges to a complex Gaussian random variable. The Central Limit Theorem for the difference of the moduli is also established.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved solution to random initial data problems for nonlinear Schrödinger equations by establishing a Law of Large Numbers (Theorem 2.6) and a Central Limit Theorem (Theorem 2.7). This allows the system to predict macroscopic soliton gas solutions from random spectral data, moving beyond qualitative kinetic theories.

  2. Rigorous analysis of the difference between random N-soliton solutions and their deterministic counterparts by proving that this difference converges in mean (Theorem 2.6) and in distribution to a complex Gaussian random variable (Theorem 2.7). This capability enables precise statistical characterization of fluctuations around large amplitude waves.

  3. Computation of correlation functions for the differences between random and deterministic solutions, which are explicitly given by integrals involving the linear statistics X G1 and X G2 (Theorem 2.8). This allows researchers to quantify spatial and temporal correlations in the soliton gas behavior across different points in space-time.

  4. Development of a small norm Riemann-Hilbert problem theory for random N-soliton solutions, showing that the error matrix E has a convergent Neumann series with high probability (Lemma 3.2). This enables the comparison between the random potential psiN and the deterministic solution psi infinity by showing that psiN(x, t; λ) - ψ∞(x, t) is bounded under appropriate conditions.

  5. Analytical proof of the differentiability of solutions to Riemann-Hilbert problems in space and time using singular integral equations (Lemma 3.4). This allows for the derivation of analytic properties for N-soliton solutions, confirming that psi infinity is an analytic solution to the fNLS equation.

Abstract

We study a random configuration of N soliton solutions ψ N(x,t; λ) of the cubic focusing Nonlinear Schrödinger (fNLS) equation in one space dimension. The N soliton solutions are parametrized by 2N complex numbers (λ,) where λ in C+ N are the eigenvalues of the Zakharov-Shabat linear operator, and in C N 0 are the norming constants of the corresponding eigenfunctions. The randomness is obtained by choosing the complex eigenvalues to be i.i.d. random variables sampled from a probability distribution with compact support in the complex plane. The corresponding norming constants are interpolated by a smooth function of the eigenvalues. Then we consider the expectation of the random measure associated to this random spectral data. Such expectation uniquely identifies, via the Zakharov-Shabat inverse spectral problem, a solution ψ infinity(x,t) of the fNLS equation. This solution can be interpreted as a soliton gas solution. We prove a Law of Large Numbers and a Central Limit Theorem for the differences ψ N(x,t; λ)-ψ infinity(x,t) and ψ N(x,t; λ) 2-ψ infinity(x,t) squared when (x,t) are in a compact set of R times R+; we additionally compute the correlation functions.

Sources

Related papers