Macroscopic diffusive fluctuations for generalized hard rods dynamics

arXiv:2305.13037 · math-ph, cond-mat.stat-mech, math.MP, math.PR · Submitted 2023-05-22 · 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: "Macroscopic diffusive fluctuations for generalized hard rods dynamics".

Mira: Here's the gist of the paper. Macroscopic diffusive fluctuations for generalized hard rods dynamics study the evolution of density fluctuations in equilibrium for a dynamics of rods with random length, proving that an initial fluctuation evolves randomly shifted by a Brownian motion <ref:2305.13037#pg2> Core Finding The result proven in the present article is that the initial fluctuations of the density of particles of velocity v, after recentering on its Euler evolution, evolve randomly shifted by a Brownian motion of variance Dpvq <ref:2305.13037#pg3>.

Kai: This diffusion coefficient Dpvq has an explicit expression depending on v and on the particular stationary measure cf.

Mira: (3.1) <ref:2305.13037#pg3>.

Kai: Stochastic Evolution Equation The evolution of the macroscopic fluctuation Ξtpy, v, rq of the density of particles of velocity v and length r at position y is described by a stochastic partial differential equation in the diffusive space-time scale <ref:2305.13037#pg3>.

Title and authors: Mira: Now let's talk about what the authors suggest as improvements to this work, because they don't just stop at proving the existence of this diffusion coefficient D p v q. They point toward things that could make the analysis more powerful.

Kai: I see them focusing on using these results to improve simulation capabilities for complex integrable many-body systems. They argue that any extended integrable system can be mapped, in a macroscopic limit, onto a collisional system, like one involving quasi-particles or solitons.

Lev: That sounds useful because if we can map it to something simpler, we don't have to simulate the whole thing at the microscopic level forever; instead, we characterize it by its scattering shift.

Mira: Exactly. The idea is that instead of simulating every single interaction, which is computationally expensive, you characterize the system macroscopically by how these quasi-particles scatter off each other. This should help AI models predict fluctuation behavior based on those scattering properties rather than needing full microscopic simulation data.

Kai: That’s a big shift from needing full microscopic details to characterizing it via macroscopic scattering shifts. It moves the focus away from tiny details towards larger scales.

Lev: From an engineering standpoint, if we can use that scattering shift as our input, the overhead for running that model on real hardware should be much lower than simulating every particle collision individually.

Mira: And they also suggest using the derived stochastic partial differential equation to enhance prediction of diffusive fluctuations in non-equilibrium regimes. They want to see how this equation evolves when things aren't in perfect equilibrium.

Kai: That means we can use the formula BtΞ t p y v q equal half D p v q B squared y Ξ t p y v q minus B y Ξ t p y v q a D p v q W nine tp v q. This equation should tell us how initial density fluctuations evolve randomly shifted by a Brownian motion of variance D p v q, even when the system is perturbed.

Lev: Modeling non-equilibrium states is always harder, and having an explicit stochastic partial differential equation gives us a framework to attack that problem systematically instead of guessing at the noise.

Mira: So what this means for AI is that we can use these derived equations to model how initial density fluctuations evolve randomly shifted by a Brownian motion of variance D p v q in non-equilibrium situations. That's an enhanced prediction capability, given the structure they found in "Macroscopic diffusive fluctuations for generalized hard rods dynamics".

Kai: And finally, they suggest developing predictive models for particle dynamics under diffusive scaling by incorporating the effective velocity derived from their analysis. They point to a term called v eff p v q which is vp one over sigma q minus pi.

Lev: Incorporating that effective velocity into dynamic models would allow AI systems to estimate the long-term trajectory of particles even when starting from macroscopic distances.

Kai: That means we can estimate how particles move in this diffusive time scale, even if we start with a coarse macroscopic view of their initial positions and velocities.

Mira: It’s about bridging the gap between the microscopic dynamics and these large-scale diffusive predictions using those effective velocity terms to predict particle motion under diffusive scaling.

Lev: That would be a useful bridge for understanding how real hardware might behave when pushed away from ideal conditions, where does that fit into our work?

The paper's summary: Kai: So, looking at the whole "Macroscopic diffusive fluctuations for generalized hard rods dynamics" paper, the main points are that we proved that initial fluctuations evolve randomly shifted by a Brownian motion with variance D p v q.

Mira: And they show how this happens through tagged quasi-particle evolution and show it leads to a functional central limit theorem, which means those particles converge to the same Brownian motion in law.

Lev: From an engineering viewpoint, that convergence is key because it validates the structure of the noise they found in equation one. It confirms that what we expect for the diffusive fluctuations in chaotic systems holds for this model.

Kai: The main implication is that this result provides a clear description of how density fluctuations behave when you're looking at it on a diffusive time scale.

Mira: And it suggests that if the initial conditions are generic, they can be smoothed into a Poisson point field with density rho p f v q over space and velocity and length.

Lev: So for us, that means we have a starting point to build models for how these fluctuations behave in non-equilibrium situations using this Poisson structure as our baseline.

Kai: The paper provides the full framework for predicting those fluctuations, from the initial state to the final evolution under diffusive scaling.

Mira: And it’s a solid piece of work that connects many different areas of statistical mechanics together in a very organized way.

Lev: It gives us something concrete to test against when we look at experimental data, which is what we need for real hardware development.

Kai: So that's our take on "Macroscopic diffusive fluctuations for generalized hard rods dynamics". We've got the core result, the evolution by Brownian motion, and a Poisson baseline for understanding fluctuations.

The paper's improvements: Kai: So, we just talked about how initial fluctuations evolve randomly shifted by Brownian motion in this hard rod system. Now let's look at what the authors suggest they could do next to make this analysis more useful for real things.

Mira: They're talking about using these results to improve how we simulate complex systems. The idea is that any extended integrable system can be mapped onto a collisional system, like one with quasi-particles or solitons.

Kai: That means instead of simulating every single tiny collision at the microscopic level, we can characterize the whole thing by how those quasi-particles scatter off each other. That sounds like a big deal for computation.

Lev: It does. If we can use that scattering shift as our input, it lowers the computational overhead significantly compared to running full microscopic simulations on real hardware. It moves the focus from tiny details to larger scales, which is what you need when you're dealing with massive systems.

Mira: And they also suggest using that stochastic partial differential equation we derived earlier to predict fluctuations even when things aren't perfectly in equilibrium. They give us a formula for how the initial density fluctuations evolve randomly shifted by that Brownian motion of variance D p v q.

Kai: So, we can use this equation to model how systems change randomly under noise, even when they are out of balance. That’s better than just assuming everything stays the same after a little bit.

Lev: It's useful for non-equilibrium setups because it gives us a structured way to handle that noise term instead of having to guess what the noise looks like in those complex states. It helps define the limits of what we can run on real hardware reliably, too.

Mira: They also introduce something called an effective velocity, v eff p v q, which is derived from their analysis. This term helps us predict particle dynamics under diffusive scaling even when we start with a coarse macroscopic view of the initial positions and velocities.

Kai: So we can use this effective velocity to estimate how particles move in a diffusive time scale, even if our starting points are just broad macroscopic averages. That bridges the gap between the microscopic world and these large-scale predictions.

Lev: That bridge is important because it tells us how particle trajectories behave when you're operating under diffusive scaling constraints on real hardware. We need those kinds of estimates for error correction codes, for instance.

Mira: So they're suggesting ways to use these mathematical tools not just to describe equilibrium, but to predict evolution in non-equilibrium and under scaling conditions using effective parameters.

Conclusion: Kai: So we've covered how initial density fluctuations in generalized hard rod dynamics evolve randomly shifted by Brownian motion, and how they map onto a Poisson point field in the long run.

Mira: Right, it’s about taking those microscopic dynamics and seeing them through a very large-scale lens where the system behaves like a simple point process over time.

Lev: From an engineering standpoint, knowing that we can model this noise structure helps us define the stability limits for any real hardware we build using error correction. It gives us a baseline for what's expected in those noisy environments.

Kai: Exactly, and they even provided a stochastic equation that lets us track these fluctuations when the system isn't perfectly still. one/two D p v q B squared y plus minus B y is that noise term driving everything.

Mira: That’s the main assumption, though we need to be careful about which specific stationary measure they use, because that choice really defines what D p v q actually means for the system's behavior.

Lev: The caveat there is that the noise correlation Gamma p v q has a limit; it only decorrelates when two velocities are at most present, which means in a real system with many different speeds, this simple diffusion picture breaks down quickly.

Kai: So for someone just listening to the show, it means we can use this framework to predict how density bumps move around randomly over time, even if the initial conditions were messy.

Mira: It suggests that for these hard rod systems, you don't need a super detailed simulation if you only care about the overall fluctuation behavior. You can rely on these macroscopic scaling results instead.

Lev: I think that’s the practical value; it lets us use analytic tools to predict long-term behavior without having to run simulations that take longer than we have compute time for real hardware testing.

Kai: It's a solid piece of work, this paper on macroscopic diffusive fluctuations for generalized hard rods dynamics. It really grounds the theory in how systems behave over large scales.

Mira: Yeah, it sets a strong foundation for how we think about non-equilibrium dynamics in these types of models moving forward.

Lev: Next time, we’ll look at how we take those scaling ideas and try to apply them to something more complex, maybe the Egorov theorem on analytic observables.

Departamento de Matemática, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires · CEREMADE, Université Paris Dauphine - PSL Research University and Institut Universitaire de France and GSSI, L’Aquila

math-ph, cond-mat.stat-mech, math.MP, math.PR

Submitted: 2023-05-22

Updated: 2026-11-12

Comments: In the published version the linearized Euler equation was missing a term. This has been corrected in section 2. The main result, contained in Section 3, is not affected. AN erratum will appear in AAP

Journal ref: Annals Applied Probability (2025)

DOI: 10.1214/24-AAP2137

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

Importance score: 79/100

The gist: Macroscopic diffusive fluctuations for generalized hard rods dynamics study the evolution of density fluctuations in equilibrium for a dynamics of rods with random length…

Key concepts

Macroscopic Diffusive Fluctuations
This refers to studying how large-scale density variations in a system of hard rods change over time. The research proves that these initial fluctuations do not evolve deterministically but are randomly shifted by Brownian motion, providing a statistical description of the system's behavior at a macroscopic level.
Stochastic Evolution Equation
This is a mathematical equation describing the random movement of density fluctuations ($\Xi_{tpy, v, rq}$). It shows that these fluctuations evolve according to a specific stochastic process involving diffusion and noise terms, contrasting with simpler models where noise is purely additive.
Quasi-particle Interpretation
The derivation uses the concept of tagged quasi-particles—pairs of rods with the same velocity. By tracking how these pairs interact and shift due to collisions, researchers can map the complex many-body dynamics onto a simpler evolution process, leading to a functional central limit theorem.
Diffusive Scaling
This technique simplifies the analysis by looking at the system over very long time scales where fluctuations become dominant. It involves defining flow fields and fluctuation fields in a scaled space-time regime, allowing for the identification of simpler limiting behaviors, such as convergence to a Poisson point process.

Terminology

Summary

Macroscopic diffusive fluctuations for generalized hard rods dynamics study the evolution of density fluctuations in equilibrium for a dynamics of rods with random length, proving that an initial fluctuation evolves randomly shifted by a Brownian motion <ref:2305. p. 0

Where the paper says it

This includes the classical hard rod elastic collisions, when rod lengths are constant and equal to a positive value. We prove that in the diffusive space-time scaling, an initial fluctuation of density of particles of velocity v, after recentering on its Euler evolution, evolve randomly shifted by a Brownian motion of variance D p v q .

13037#pg2>

Core Finding

The result proven in the present article is that the initial fluctuations of the density of particles of velocity v, after recentering on its Euler evolution, evolve randomly shifted by a Brownian motion of variance Dpvq <ref:2305.13037#pg3>. This diffusion coefficient Dpvq has an explicit expression depending on v and on the particular stationary measure cf. p. 1

Where the paper says it

The result we prove in the present article is that the initial fluctuations of the density of particles of velocity v, after recentering on its Euler evolution, evolve randomly shifted by a Brownian motion of variance Dpvq. This diffusion coefficient Dpvq has an explicit expression depending on v and on the particular stationary measure (cf.

(3.1) <ref:2305. p. 0
Where the paper says it

We will discuss initial inhomogeneous non-stationary state in Section 4 . 1 arXiv:2305.

p. 0
Where the paper says it

We will discuss initial inhomogeneous non-stationary state in Section 4 . 1 arXiv:2305.

p. 0
Where the paper says it

We will discuss initial inhomogeneous non-stationary state in Section 4 . 1 arXiv:2305.

p. 0
Where the paper says it

We will discuss initial inhomogeneous non-stationary state in Section 4 . 1 arXiv:2305.

p. 0
Where the paper says it

We will discuss initial inhomogeneous non-stationary state in Section 4 . 1 arXiv:2305.

13037#pg3>.

Stochastic Evolution Equation

The evolution of the macroscopic fluctuation Ξtpy, v, rq of the density of particles of velocity v and length r at position y is described by a stochastic partial differential equation in the diffusive space-time scale <ref:2305.13037#pg3>. This equation is given by BtΞtpy, v, rq “ 1/2 DpvqB 2yΞtpy, v, rq - ByΞtpy, v, rq a DpvqW9tpvq (1. p. 1

Where the paper says it

This result corresponds to the following stochastic partial differential equation for the evolution of the macroscopic fluctuation Ξtpy, v, rq of the density of particles of velocity v and length r at position y: BtΞtpy, v, rq “ 1 2 DpvqB2 yΞtpy, v, rq ` ByΞtpy, v, rq a DpvqW9 tpvq, (1. 1) where pW9 tpvq : t, v P Rq is a centered Gaussian field with covariance E ` W9 tpvqW9 spwq ˘ “ δpt ´ sq Γpv, wq a DpvqDpwq with Γpv, wq given in (3.

p. 1
Where the paper says it

This result corresponds to the following stochastic partial differential equation for the evolution of the macroscopic fluctuation Ξtpy, v, rq of the density of particles of velocity v and length r at position y: BtΞtpy, v, rq “ 1 2 DpvqB2 yΞtpy, v, rq ` ByΞtpy, v, rq a DpvqW9 tpvq, (1. 1) where pW9 tpvq : t, v P Rq is a centered Gaussian field with covariance E ` W9 tpvqW9 spwq ˘ “ δpt ´ sq Γpv, wq a DpvqDpwq with Γpv, wq given in (3.

1) <ref:2305.13037#pg3>. This equation contrasts with the typical diffusive evolution of fluctuations in chaotic systems which is expected an additive space-time white noise driving the fluctuations and the equation would be of the type (1. p. 1
Where the paper says it

Since Ξt is a distribution in py, v, rq, (1) should be understood in the weak sense. Notice that in (1) This is in contrast with the typical diffusive evolution of fluctuations in chaotic systems, where it is expected an additive space-time white noise driving the fluctuations and the equation would be of the type [21] BtΞstpy, v, rq “ 1 2 DpvqB2 yΞtpy, v, rq ` a DpvqByW9 y,tpvq, (1.

2) <ref:2305. p. 0
Where the paper says it

We will discuss initial inhomogeneous non-stationary state in Section 4 . 1 arXiv:2305.

p. 0
Where the paper says it

We will discuss initial inhomogeneous non-stationary state in Section 4 . 1 arXiv:2305.

13037#pg3>.

Quasi-particle Interpretation

The derivation follows the behavior of two tagged quasi-particles with the same velocity, which are defined by the exchange of positions at collision <ref:2305.13037#pg5>. The standard technique involves mapping quasi-particles to points and evolving them without interaction, and then shifting by collisions with other quasi-particles of different velocity <ref:2305. p. 1

Where the paper says it

The standard technique to study such dynamics is to go to the reduced description where quasi-particles are mapped to points and evolve without interaction. Then the evolution of the tagged quasi-particle is obtained by the trivial evolution of the corresponding point, shifted by the collisions with quasi-particles of different velocity.

13037#pg5>. Recentering the position of the tagged quasi-particle on the Euler deterministic behavior leads to a functional central limit theorem so that its position converges in law to a Brownian motion of variance Dpvq <ref:2305. p. 2
Where the paper says it

tagged quasi-particle on the Euler deterministic behavior, we have then a functional central limit theorem so that the position converges in law to a Brownian motion of variance Dpvq. Now consider two tagged quasi-particles with the same velocity and initially located at macroscopic distance: this means that there are typically ε ´1 particles in between, where ε is the scaling parameter going to 0.

13037#pg5>.

Diffusive Scaling Results

In the diffusive scaling, for any point px, vq P R squared and t ϵ 0 define the flow j epsilon px, v q: “ epsilon ypx˜v˜r˜qPXε r 1rv a s1rx a x pv´v˜qts - 1rv a s1rx´pv´v˜qt a x xs (2.17) <ref:2305. p. 0

Where the paper says it

We will discuss initial inhomogeneous non-stationary state in Section 4 . 1 arXiv:2305.

p. 0
Where the paper says it

We will discuss initial inhomogeneous non-stationary state in Section 4 . 1 arXiv:2305.

13037#pg7>. The limit in (3. p. 0
Where the paper says it

The density of particles of each given velocity is conserved and in the Euler scaling the macroscopic evolution of such densities have been studied by the pioneristic work of Percus [18] and Boldrighini, Dobrushin and Suhov [5]. Fluctuations around this Euler limit have been studied by Boldrighini and Wick in [ 7].

p. 0
Where the paper says it

The density of particles of each given velocity is conserved and in the Euler scaling the macroscopic evolution of such densities have been studied by the pioneristic work of Percus [18] and Boldrighini, Dobrushin and Suhov [5]. Fluctuations around this Euler limit have been studied by Boldrighini and Wick in [ 7].

1) is equivalent to (3. p. 2
Where the paper says it

Since this macroscopic behaviour is due to the scattering shift that the quasi particles have when crossing particles with a different velocity, we believe that this equations are universal for any extended integrable system. In fact any extended integrable system is equivalent (in a macroscopic limit) to a collisional system (of quasi-particle or impulsions or solitons) and it is characterized macroscopically by its scattering shift [22].

p. 2
Where the paper says it

Since this macroscopic behaviour is due to the scattering shift that the quasi particles have when crossing particles with a different velocity, we believe that this equations are universal for any extended integrable system. In fact any extended integrable system is equivalent (in a macroscopic limit) to a collisional system (of quasi-particle or impulsions or solitons) and it is characterized macroscopically by its scattering shift [22].

1) <ref:2305.13037#pg7>. The fluctuation field at diffusive scaling is defined as ΞY,εt pφq: “ epsilon 1/2 ” epsilon ypx,v,rqPXε rφ “ y epsilon vepsilon t pxq ´ veffpvqepsilon t, v, r‰´ 1 1 σxxφyyı (3. p. 5
Where the paper says it

ÝÑεÑ0 ρ ¡ rφ` x ` xσ, v, r˘ dx dµpv, rq “ ρ 1 ` σ ¡ rφpy, v, rqdy dµpv, rq “ 1 1 ` σ xxφyy. Static CLT for the dilated configuration We define the fluctuation field ξ Y,εpφq “ ε ´1{2 ` Kεφ ´ EKεφ ˘ .

2) <ref:2305.13037#pg8>.

Main Theorem and Conclusion

Theorem 3.2 states that ΞY,εt pφq law ÝÑεÑ0 ΞY,t pφq “ ΞY,0’ φBt q where Bt is a family of Wiener processes with covariance E Btpvq Btpwq = tΓpv, wq with Γ defined in (3. p. 7

Where the paper says it

15) Putting together (2. 1) we have shown that ξ Y,εpφq law ÝÑεÑ0 ξ Y pφq “ ξ X ´ φq ´ Pφq ¯ “ ξ X ` Cφq ˘ , (2.

p. 7
Where the paper says it

15) Putting together (2. 1) we have shown that ξ Y,εpφq law ÝÑεÑ0 ξ Y pφq “ ξ X ´ φq ´ Pφq ¯ “ ξ X ` Cφq ˘ , (2.

1) <ref:2305.13037#pg9>. This implies that the weak sense ξY,t satisfies the equation BtξY,t veffBxξY,t “ 0, which is the expected equation <ref:2305. p. 1
Where the paper says it

Since Ξt is a distribution in py, v, rq, (1) should be understood in the weak sense. Notice that in (1) This is in contrast with the typical diffusive evolution of fluctuations in chaotic systems, where it is expected an additive space-time white noise driving the fluctuations and the equation would be of the type [21] BtΞstpy, v, rq “ 1 2 DpvqB2 yΞtpy, v, rq ` a DpvqByW9 y,tpvq, (1.

13037#pg9>.

Inhomogeneous Initial Conditions

For generic initial conditions where the initial density f is absolutely continuous in the x and v coordinates, it satisfies that the limit as t Ñ 8 of ftpx, v, rq is constant in x, i.e., ftpx, v, rq ÝÑ ρfpv, rq for some ρ P R and fpv, rq <ref:2305. p. 0

Where the paper says it

We will discuss initial inhomogeneous non-stationary state in Section 4 . 1 arXiv:2305.

13037#pg10>. This suggests that in a diffusive time scale the system essentially behaves like if it is a stationary state determined by a Poisson point field ρfpv, rqdx dv dr, and the analysis for the macroscopic fluctuations of Section 3 applies <ref:2305. p. 3
Where the paper says it

Essentially, the only tools we use are the law of large numbers and the central limit theorem for a Poisson field. In Section 2 we prove the macroscopic evolution of the fluctuations in the Euler scaling (recovering the result of [7]).

13037#pg10>.

Limit Theorem for Poisson Process

Lemma 5.1 states that let φpx, v, rq a smooth function on R cubed with compact support in x, and tBepsilon pxq, x P Ru a process converging in law to tBpxq, x P Ru. p. 3

Where the paper says it

Equilibrium Fluctuations in the Euler scaling In the following µ is a probability on R 2 with finite second moments, and ρ ą 0 is a density parameter. For ε ą 0, let Xε be the Poisson process on R 3 with intensity ε ´1ρdx dµpv, rq.

p. 3
Where the paper says it

Equilibrium Fluctuations in the Euler scaling In the following µ is a probability on R 2 with finite second moments, and ρ ą 0 is a density parameter. For ε ą 0, let Xε be the Poisson process on R 3 with intensity ε ´1ρdx dµpv, rq.

Then lim epsilon 0 epsilon ypx,v,rqPXepsilon rφpx vrq Bepsilon pxq law “ ρ 1/r φpx vrq Bpxq dx dµpv, rq <ref:2305. p. 9
Where the paper says it

ÝÑεÑ0 ρ ¡ rφpx ` vt, v, rq dx dµpv, rq “ xxφyy. The X-fluctuation field at time t satisfies ξ X,ε t pφq :“ ε ´1{2 ` N ε t φ ´ xxφyy˘ law ÝÑεÑ0 ξ Xpφtq, (2.

13037#pg10>. This shows that the empirical measure Mepsilon on R 0, 1s defined by Mepsilon pψq “ epsilon ypx,v,rqPXepsilon 1r0 a s ψpxq, ψ P Cp0, 1q converges a. p. 4
Where the paper says it

Considering φpx, v, rq “ 1r0,1spxq, (2) implies that ε times the number of points in Xε with the x coordinate in r0, 1s converges to the density σ, Nε pr0, 1sˆ R 2 q a. ÝÑεÑ0 σ.

p. 4
Where the paper says it

Considering φpx, v, rq “ 1r0,1spxq, (2) implies that ε times the number of points in Xε with the x coordinate in r0, 1s converges to the density σ, Nε pr0, 1sˆ R 2 q a. ÝÑεÑ0 σ.

s. to the Lebesgue measure dx on r0, 1s <ref:2305. p. 0
Where the paper says it

We will discuss initial inhomogeneous non-stationary state in Section 4 . 1 arXiv:2305.

13037#pg10>.

Covariance Operator

The covariance operator C is identified as C = ρ 1/r 2C squared = ρ 1/r 2´I ´ ´ σ squared p1 1/r 2 σ 1/¯P <ref:2305.13037#pg9>. This corresponds to the Fourier transforms φppk, v, rq “ ż e i 2πkyφpy, v, rqdy EpξYpφqξYpψqq = ρ 1/r 2Cφppk, v, rq˚Cψppk, vrqdkdµpv <ref:2305. p. 7

Where the paper says it

, this identifies ξ Y as the centered Gaussian field with covariance Epξ Y pφqξ Y pψqq “ Epξ XpCφqqξ XpCψqqq “ ρ ¡ r 2Cφpxp1 ` σq, v, rqCψpxp1 ` σq, v, rqdxdµpv, rq, “ ρ 1 ` σ ¡ r 2Cφpy, v, rqCψpy, v, rqdydµpv, rq “ ρ 1 ` σ xxCφCψyy2. That means for the Fourier transforms φppk, v, rq “ ż e i2πkyφpy, v, rqdy Epξ Y pφqξ Y pψqq “ ρ 1 ` σ ¡ r 2Cφppk, v, rq ˚Cψppk, v, rqdkdµpv, rq.

13037#pg9>.

Correlation in Velocity

The correlation at different velocities v ă v is given by Γpv, vq = 1/2´Dpvq Dpvq ´ pv ´ vqρ 1/r 2dµpv, rq <ref:2305.13037#pg8>. This implies that Γpv, vq ě 0 <ref:2305. p. 4

Where the paper says it

Considering φpx, v, rq “ 1r0,1spxq, (2) implies that ε times the number of points in Xε with the x coordinate in r0, 1s converges to the density σ, Nε pr0, 1sˆ R 2 q a. ÝÑεÑ0 σ.

13037#pg9>. Typically this decorrelation happens only when two velocities at most are present

Summary of References

[1] Aizenman, M., Goldstein, S. and Lebowitz, J. L. (1975). Ergodic properties of an infinite one dimensional hard rod system <ref:2305. p. 0

Where the paper says it

Introduction The mechanical system of one-dimensional hard rods is the simplest non trivial completely integrable dynamics where the macroscopic behavior can be described by generalized hydrodynamics. A rod configuration Y is a set of points py, v, rq, viewed as a segment/rod ry, y ` rq travelling at bare speed v, satisfying ry, y ` rq X ry, ty ˜ ` r˜q is empty for all py, v, rq,py, ˜ v, ˜ r˜q in Y .

13037#pg3>

[2] Aizenman, M., Lebowitz, J. and Marro, J. (1978). Time-displaced correlation functions in an infinite one-dimensional mixture of hard rods with different diameters <ref:2305. p. 3

Where the paper says it

the diffusive scale of multipoint correlation functions, mentioned in [10] as an open problem. As we believe that our approach is more elementary than the one used in the previous literature, we have written this article to be completely independent of results on hard rods prior to the paper [12], which is our starting point.

13037#pg3>

[4] Boldrighini, C., Dobrushin, R. p. 0

Where the paper says it

Rods travel ballistically with its bare velocity until collision with other rod, when they interchange speed and length. The density of particles of each given velocity is conserved and in the Euler scaling the macroscopic evolution of such densities have been studied by the pioneristic work of Percus [18] and Boldrighini, Dobrushin and Suhov [5].

p. 0
Where the paper says it

Rods travel ballistically with its bare velocity until collision with other rod, when they interchange speed and length. The density of particles of each given velocity is conserved and in the Euler scaling the macroscopic evolution of such densities have been studied by the pioneristic work of Percus [18] and Boldrighini, Dobrushin and Suhov [5].

and Suhov, Y. p. 0
Where the paper says it

Rods travel ballistically with its bare velocity until collision with other rod, when they interchange speed and length. The density of particles of each given velocity is conserved and in the Euler scaling the macroscopic evolution of such densities have been studied by the pioneristic work of Percus [18] and Boldrighini, Dobrushin and Suhov [5].

p. 0
Where the paper says it

Rods travel ballistically with its bare velocity until collision with other rod, when they interchange speed and length. The density of particles of each given velocity is conserved and in the Euler scaling the macroscopic evolution of such densities have been studied by the pioneristic work of Percus [18] and Boldrighini, Dobrushin and Suhov [5].

p. 0
Where the paper says it

Rods travel ballistically with its bare velocity until collision with other rod, when they interchange speed and length. The density of particles of each given velocity is conserved and in the Euler scaling the macroscopic evolution of such densities have been studied by the pioneristic work of Percus [18] and Boldrighini, Dobrushin and Suhov [5].

M. (1990). OneDimensional Hard-Rod Caricature of Hydrodynamics: Navier-Stokes Correction Technical Report <ref:2305. p. 2
Where the paper says it

We recently discovered the article by Presutti and Wick [19] that concerns the diffusive behavior of travelling wave initial conditions in hard rods systems, where in the remark after Theorem 1 they comment about possible diffusive behaviour of fluctuations and they write: “spatially separated fluctuations in the density of rods with the same velocity move with the same Brownian component”. Strangely [19] is never quoted in the following literature about Navier-Stokes corrections for the hard rods hydrodynamics (cf.

13037#pg3>

[5] Boldrighini, C., Dobrushin, R. L. and Suhov, Y. M. (1983). Onedimensional hard rod caricature of hydrodynamics

[6] Boldrighini, C. and Suhov, Y. M. (1997). One-dimensional hard-rod caricature of hydrodynamics: “Navier-Stokes correction” for local equilibrium initial states

[7] Boldrighini, C. and Wick, W. p. 0

Where the paper says it

The density of particles of each given velocity is conserved and in the Euler scaling the macroscopic evolution of such densities have been studied by the pioneristic work of Percus [18] and Boldrighini, Dobrushin and Suhov [5]. Fluctuations around this Euler limit have been studied by Boldrighini and Wick in [ 7].

D. (1988). Fluctuations in a onedimensional mechanical system. p. 0
Where the paper says it

Keywords and phrases: Hard Rods dynamics, completely integrable systems, generalized hydrodynamic limits, diffusive fluctuations. Introduction The mechanical system of one-dimensional hard rods is the simplest non trivial completely integrable dynamics where the macroscopic behavior can be described by generalized hydrodynamics.

I. The Euler limit

[9] De Nardis, J., Bernard, D. and Doyon, B. (2019). Diffusion in generalized hydrodynamics and quasiparticle scattering

[18] Percus, J. K. (1969). Exact Solution of Kinetics of a Model Classical Fluid

[20] Spohn, H. (1982). Hydrodynamical theory for equilibrium time correlation functions of hard rods

[21] Spohn, H. (1991). Large Scale Dynamics of Interacting Particles

[22] Spohn, H. (2023). Hydrodynamic Scales of Integrable Many-Particle Systems

[24] van der Vaart, A. W. (1998). Asymptotic statistics

--- Page 1 ---

Macroscopic diffusive fluctuations for generalized hard rods dynamics study the evolution of density fluctuations in equilibrium for a dynamics of rods with random length, proving that an initial fluctuation evolves randomly shifted by a Brownian motion <ref:2305.13037#pg3>.

The result proven in the present article is that the initial fluctuations of the density of particles of velocity v, after recentering on its Euler evolution, evolve randomly shifted by a Brownian motion of variance Dpvq <ref:2305.13037#pg3>.

The evolution of the macroscopic fluctuation Ξtpy, v, rq of the density of particles of velocity v and length r at position y is described by a stochastic partial differential equation in the diffusive space-time scale <ref:2305.13037#pg3>. This equation is given by BtΞtpy, v, rq “ 1/2 DpvqB 2yΞtpy, v, rq - ByΞtpy, v, rq a DpvqW9tpvq (1.1) <ref:2305.13037#pg3>.

The derivation follows the behavior of two tagged quasi-particles with the same velocity, which are defined by the exchange of positions at collision. The standard technique involves mapping quasi-particles to points and evolving them without interaction, and then shifting by collisions with other quasi-particles of different velocity. p. 1

Where the paper says it

The standard technique to study such dynamics is to go to the reduced description where quasi-particles are mapped to points and evolve without interaction. Then the evolution of the tagged quasi-particle is obtained by the trivial evolution of the corresponding point, shifted by the collisions with quasi-particles of different velocity.

In the diffusive scaling, for any point px, vq P R squared and t ϵ 0 define the flow j epsilon px, v q: “ epsilon ypx˜v˜r˜qPXε r 1rv a s1rx a x pv´v˜qts - 1rv a s1rx´pv´v˜qt a x xs (2.17) <ref:2305.13037#pg7>. The limit in (3.1) is equivalent to (3.1).

Theorem 3.2 states that ΞY,εt pφq law ÝÑεÑ0 ΞY,t pφq “ ΞY,0’ φBt q where Bt is a family of Wiener processes with covariance E Btpvq Btpwq = tΓpv, wq with Γ defined in (3.1). This implies that the weak sense ξY,t satisfies the equation BtξY,t veffBxξY,t “ 0, which is the expected equation. p. 1

Where the paper says it

Since Ξt is a distribution in py, v, rq, (1) should be understood in the weak sense. Notice that in (1) This is in contrast with the typical diffusive evolution of fluctuations in chaotic systems, where it is expected an additive space-time white noise driving the fluctuations and the equation would be of the type [21] BtΞstpy, v, rq “ 1 2 DpvqB2 yΞtpy, v, rq ` a DpvqByW9 y,tpvq, (1.

For generic initial conditions where the initial density f is absolutely continuous in the x and v coordinates, it satisfies that the limit as t Ñ 8 of ftpx, v, rq is constant in x, i.e., ftpx, v, rq ÝÑ ρfpv, rq for some ρ P R and fpv, rq. This suggests that in a diffusive time scale the system essentially behaves like if it is a stationary state determined by a Poisson point field ρfpv, rqdx dv dr, and the analysis for the macroscopic fluctuations of Section 3 applies. p. 3

Where the paper says it

Essentially, the only tools we use are the law of large numbers and the central limit theorem for a Poisson field. In Section 2 we prove the macroscopic evolution of the fluctuations in the Euler scaling (recovering the result of [7]).

Lemma 5.1 states that let φpx, v, rq a smooth function on R cubed with compact support in x, and tBepsilon pxq, x P Ru a process converging in law to tBpxq, x P Ru. Then lim epsilon 0 epsilon ypx,v,rqPXepsilon rφpx vrq Bepsilon pxq dx dµpv, rq. p. 9

Where the paper says it

We have that, by (2), the free gas empirical length measure at time t satisfies N ε t φ :“ ε ÿ px,v,rqPXε rφpx ` vt, v, rq a. ÝÑεÑ0 ρ ¡ rφpx ` vt, v, rq dx dµpv, rq “ xxφyy.

This shows that the empirical measure Mepsilon on R 0, 1s defined by Mepsilon pψq “ epsilon ypx,v,rqPXepsilon 1r0 a s ψpxq, ψ P Cp0, 1q converges a.s. to the Lebesgue measure dx on r0, 1s. p. 1
Where the paper says it

The result we prove in the present article is that the initial fluctuations of the density of particles of velocity v, after recentering on its Euler evolution, evolve randomly shifted by a Brownian motion of variance Dpvq. This diffusion coefficient Dpvq has an explicit expression depending on v and on the particular stationary measure (cf.

The covariance operator C is identified as C = ρ 1/r 2C squared = ρ 1/r 2´I ´ ´ σ squared p1 1/r 2 σ 1/¯P. This corresponds to the Fourier transforms φppk, v, rq “ ż e i 2πkyφpy, v, rqdy EpξYpφqξYpψqq = ρ 1/r 2Cφppk, vrqdkdµpv. p. 7

Where the paper says it

, this identifies ξ Y as the centered Gaussian field with covariance Epξ Y pφqξ Y pψqq “ Epξ XpCφqqξ XpCψqqq “ ρ ¡ r 2Cφpxp1 ` σq, v, rqCψpxp1 ` σq, v, rqdxdµpv, rq, “ ρ 1 ` σ ¡ r 2Cφpy, v, rqCψpy, v, rqdydµpv, rq “ ρ 1 ` σ xxCφCψyy2. That means for the Fourier transforms φppk, v, rq “ ż e i2πkyφpy, v, rqdy Epξ Y pφqξ Y pψqq “ ρ 1 ` σ ¡ r 2Cφppk, v, rq ˚Cψppk, v, rqdkdµpv, rq.

The correlation at different velocities v ă v is given by Γpv, vq = 1/2´Dpvq Dpvq ´ pv ´ vqρ 1/r 2dµpv, rq. This implies that Γpv, vq ě 0. p. 4

Where the paper says it

Considering φpx, v, rq “ 1r0,1spxq, (2) implies that ε times the number of points in Xε with the x coordinate in r0, 1s converges to the density σ, Nε pr0, 1sˆ R 2 q a. ÝÑεÑ0 σ.

Typically this decorrelation happens only when two velocities at most are present. p. 1
Where the paper says it

About the uncorrelation in r, this persists in the macroscopic fluctuations as consequence of the decorrelation at the microscopic scale. In order to understand why (1) arise, we follow the behaviour of two tagged quasi-particles with the same velocity.

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 simulation capability for complex integrable many-body systems by utilizing results showing that any extended integrable system is equivalent (in a macroscopic limit) to a collisional system (of quasi-particle or impulsions or solitons) and it is characterized macroscopically by its scattering shift. This allows AI models to predict macroscopic fluctuation behavior based on scattering properties rather than full microscopic simulation.

  2. Enhanced prediction of diffusive fluctuations in non-equilibrium regimes by using the derived stochastic partial differential equation: BtΞtpy, v, rq “ 1/2 DpvqB squared yΞtpy, v, rq - ByΞtpy, v, rq a DpvqW9 tpvq. This enables the AI to model how initial density fluctuations evolve randomly shifted by a Brownian motion of variance Dpvq.

  3. Development of predictive models for particle dynamics under diffusive scaling by incorporating the effective velocity derived from the analysis: v eff pv q: vp1 / σq - π. This allows AI systems to estimate the long-term trajectory of particles in a diffusive time scale, even when starting from macroscopic distances.

Abstract

We study the fluctuations in equilibrium for a dynamics of rods with random length. This includes the classical hard rod elastic collisions, when rod lengths are constant and equal to a positive value. We prove that in the diffusive space-time scaling, an initial fluctuation of density of particles of velocity v, after recentering on its Euler evolution, evolve randomly shifted by a Brownian motion of variance D(v).

Sources

Related papers