🔬 condensed matter

Derivation of the Boltzmann equation with no "molecular chaos"-type approximation

This paper derives a closed evolution equation for the $s$ -particle distribution function from the Liouville equation without using the "molecular chaos" approximation by employing a projection operator to account for initial correlations, ultimately showing that these correlations vanish at large timescales to recover the linear and nonlinear Boltzmann equations.

Original authors: Victor F. Los

Published 2026-07-23

📖 7 min read🧠 Deep dive

CC BY 4.0

Original authors: Victor F. Los

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are watching a massive, chaotic dance floor packed with thousands of people. Every single person is moving, bumping into neighbors, changing direction, and shouting. If you wanted to predict exactly where every single dancer would be in ten minutes, you would need a supercomputer and a rulebook for every possible collision. This is the world of statistical mechanics, the branch of physics that tries to understand how huge groups of tiny particles behave.

The big puzzle scientists have been wrestling with for over a century is this: How do we get from the chaotic, reversible rules that govern every single particle (like a billiard ball bouncing off another) to the smooth, one-way flow of time we see in the real world, like heat spreading out or gas filling a room? Usually, to solve this, physicists have to make a big, messy guess called "molecular chaos." It's like assuming that every time two dancers bump into each other, they instantly forget everything about their previous partners and dance completely randomly. This guess makes the math work, but it feels a bit like cheating because, in reality, particles do remember their collisions for a tiny split second.

Now, a team of researchers from Ukraine has come up with a clever new way to look at this dance floor. Instead of forcing the dancers to forget their past, they found a mathematical trick that lets them keep the memory of every bump and still get a clean, simple equation. They managed to derive the famous Boltzmann equation—the master rule for how gases evolve—without ever having to tell the particles to "forget" their initial connections. It's as if they found a way to predict the future of the whole dance floor by tracking the history of just a few dancers, proving that you don't need to pretend chaos exists to understand how order emerges.

The Problem: The "Forgetfulness" Rule

For a long time, the standard way to describe a gas was to use a chain of equations known as the BBGKY hierarchy. Think of this as a game of "telephone." To know how one particle moves, you need to know about two particles. To know about two, you need three, and so on, all the way up to the entire crowd. This chain is impossible to solve in real life.

To break the chain, the physicist Ludwig Boltzmann made a bold move in the 1870s. He assumed that right before any two particles collide, they are completely uncorrelated. He called this "molecular chaos." In our dance floor analogy, it's like saying that before two people bump, they have no idea who the other person is or where they came from. This assumption allowed him to write a single, closed equation (the Boltzmann equation) that describes how the gas changes over time.

However, this assumption has a flaw. If two particles collide, they do become correlated; they know about each other now. If they collide again later, they aren't random strangers. The "molecular chaos" rule assumes that this correlation disappears instantly, which isn't strictly true. For decades, mathematicians have tried to prove that this rule works, but usually only for very short periods of time or under very specific, idealized conditions. If you want to know what happens over a long time, or if the system starts with a specific pattern of connections, the old method gets messy.

The New Trick: The "Memory Keeper"

The authors of this paper, Victor F. Los and V.G. Baryakhtar, decided to stop trying to prove that particles forget. Instead, they asked: "What if we keep the memory, but hide it in the math?"

They used a tool called a "projection operator." Imagine you have a giant, high-resolution photo of the entire dance floor at the start of the party ( $t=0$ ). This photo captures every single person's position and mood. Usually, when scientists try to simplify the problem to just look at one person, they throw away the rest of the photo, assuming the rest doesn't matter.

The authors introduced a special mathematical "lens" (their projection operator) that looks at the whole system but is designed in a very specific way. This lens has a superpower: it takes the full, complex photo of the entire crowd at the start and projects it onto the single person you are interested in, without changing the starting photo at all.

By using this lens, they transformed the messy, open-ended equations (which usually require a "source" term to account for the initial mess) into a perfectly closed, self-contained equation. The "memory" of the initial crowd isn't thrown away; it's baked into the "kernel" of the equation. Think of the kernel as the engine of a car. Usually, the engine needs a constant stream of fuel (the initial conditions) to run. The authors' new engine has the fuel tank built right into the design, so it runs smoothly on its own, carrying the history of the start with it.

The Results: From Memory to the Boltzmann Equation

The paper shows that this new, closed equation works for any starting condition, whether the particles were correlated or not. They then simplified this equation for a gas that isn't too crowded (low density), which is a common situation for gases.

Here is where the story gets exciting. They looked at how this equation behaves over time:

The Short Time ( $t \sim t_{cor}$ ): In the very beginning, right after the particles start interacting, the "memory" of the initial correlations is very strong. The equation shows that the particles' past connections are actively influencing how they move. The system is not yet behaving like a simple gas.
The Long Time ( $t \gg t_{cor}$ ): As time goes on, the particles collide and move around. The authors show that the influence of those initial "memories" fades away. The complex terms in their equation that represent the initial correlations cancel each other out.
The Result: Once the memory fades, the equation simplifies perfectly into the linear Boltzmann equation. This is the standard equation used to describe gases, but here it was derived without assuming the particles were uncorrelated at the start. The math proved that the "forgetfulness" isn't an assumption you have to make; it's a natural result that happens after enough time has passed.

The "Non-Linear" Surprise

The paper goes one step further. It asks: "What if we look at a time that is long enough for the memory to fade, but short enough that the gas hasn't changed its overall shape yet?"

In this specific window of time (between the correlation time and the relaxation time), the authors show that the linear equation can be rewritten as the non-linear Boltzmann equation. This is the famous version where the probability of a collision depends on the current state of both particles colliding.

Usually, getting this non-linear version requires the "molecular chaos" assumption. But in this paper, it appears naturally because the initial correlations have died out, and the particles are effectively moving freely between collisions. The authors prove that if the gas is very thin (meaning the average distance a particle travels before hitting another is very large), this non-linear equation holds true for all finite times after the initial moment.

Why This Matters

This work is a significant step in the foundations of physics. It resolves a long-standing debate about whether the Boltzmann equation is a fundamental truth or just a lucky guess based on an approximation. The authors show that the equation is robust. You don't need to pretend particles have amnesia to get the right answer. If you start with a messy, correlated system, the math itself will wash out those correlations over time, leaving you with the clean, predictable laws of thermodynamics.

The paper doesn't just suggest this; it derives the equations exactly. It shows that the "molecular chaos" isn't a rule you force on nature, but a state nature arrives at on its own. For anyone curious about how the chaotic dance of atoms turns into the smooth flow of heat and wind, this paper offers a new, clearer map of the journey.

Technical Summary: Derivation of the Boltzmann Equation without "Molecular Chaos" Approximation

Problem Statement
The paper addresses a fundamental challenge in nonequilibrium statistical mechanics: deriving a completely closed, time-local evolution equation for an $s$ -particle distribution function $F_s(t)$ (where $s \ll N$ ) starting from the reversible Liouville equation for an $N$ -particle system ( $N \gg 1$ ). Traditional approaches, such as the BBGKY hierarchy, require decoupling the chain of equations, typically necessitating approximations like "molecular chaos" (factorization of the two-particle distribution function) or the "principle of weakening of initial correlations." These approximations often assume uncorrelated initial states or the rapid vanishing of correlations, which limits the rigor of the derivation for arbitrary initial conditions and prevents a fully closed description that accounts for initial correlations on equal footing with collision dynamics.

Methodology
The authors employ a projection operator formalism to resolve the issue of the inhomogeneous source term (representing initial correlations) found in the standard Nakajima-Zwanzig Generalized Master Equation (GME).

Special Projection Operator: Instead of discarding the irrelevant initial condition term $\rho_i(0)$ , the authors introduce a specific, time-independent projection operator $P_{s\Sigma}$ defined as:
$P_{s\Sigma} = F_N(0) \frac{1}{F_s(0) V^s} \int \dots \int dx_\Sigma$
where $F_N(0)$ is the arbitrary initial $N$ -particle distribution function and $F_s(0)$ is the corresponding $s$ -particle marginal. Crucially, this operator satisfies $P_{s\Sigma} F_N(0) = F_N(0)$ , meaning it leaves the arbitrary initial state unchanged.
Transformation to Homogeneous GME: By applying this operator to the Liouville equation, the authors derive an exact, completely closed (homogeneous) linear Generalized Master Equation for the relevant part of the distribution function. The initial correlations, which usually appear as an inhomogeneous source term, are instead "hidden" within the memory kernel governing the time evolution of $F_s(t)$ .
Linear Density Approximation: The derived exact equation is specialized for a system with two-body interactions. The authors expand the kernel in the linear approximation with respect to the particle density $n = N/V$ . This yields explicit evolution equations for the one-particle ( $F_1$ ) and two-particle ( $F_2$ ) distribution functions that are exact to first order in density but valid for arbitrary initial correlations and time scales.

Key Results
The paper derives the following specific outcomes:

Exact Closed Equations: Equations (35) and (36) provide the exact evolution for $F_1(t)$ and $F_2(t)$ in the linear density approximation. These equations explicitly include terms dependent on the initial correlation function $g_2(x_1, x_2; 0)$ (or the dimensionless parameter $C_2$ ), demonstrating how initial correlations influence the system at arbitrary timescales.
Large Timescale Behavior ( $t \gg t_{cor}$ ): In the spatially homogeneous case, the authors analyze the equation for $F_1(t)$ on timescales much larger than the correlation time $t_{cor}$ but smaller than the relaxation time $t_{rel}$ . They demonstrate that the terms describing the influence of initial correlations in the collision integral cancel out exactly.
Derivation of the Linear Boltzmann Equation: As a result of the cancellation of initial correlation terms at large timescales, the evolution of $F_1(t)$ is governed by a linear Boltzmann equation (Eq. 74). Notably, this equation is derived without assuming "molecular chaos" at $t=0$ ; rather, the initial correlations naturally decay or compensate within the kernel, leaving a linear equation where the second particle's distribution is fixed at its initial value $F_2(v_2; 0)$ .
Recovery of the Nonlinear Boltzmann Equation: The authors show that in the time interval $t_{cor} \ll t \ll t_{rel}$ , the difference between $F_1(v; t)$ and $F_1(v; 0)$ is negligible. Consequently, the linear equation can be approximated by the conventional nonlinear Boltzmann equation (Eq. 77), where the distribution functions are evaluated at time $t$ .
The Limit of Infinite Mean Free Path: The paper concludes that in the limit where the relaxation time $t_{rel} \to \infty$ (equivalent to the mean free path $l \to \infty$ ), the nonlinear Boltzmann equation holds exactly for all finite times $t \gg t_{cor}$ , without requiring the "molecular chaos" assumption.

Significance and Claims
The paper claims to solve the problem of deriving a closed kinetic equation for reduced distribution functions for an arbitrary initial state of interacting classical particles. Its primary significance lies in:

Rigorous Treatment of Initial Correlations: It demonstrates that initial correlations do not need to be discarded or assumed to vanish initially. Instead, they are incorporated into the evolution kernel and naturally cease to influence the macroscopic evolution of the one-particle distribution function at large timescales.
Elimination of "Molecular Chaos" Assumption: The derivation of the Boltzmann equation is achieved without invoking the "molecular chaos" approximation (factorization of $F_2$ at all times) or the "propagation of chaos" hypothesis. The factorization emerges as a consequence of the dynamics in the limit of large mean free paths, rather than being an imposed initial condition.
Subdynamics Formalism: The work realizes the concept of "subdynamics" (originally proposed by Prigogine) by defining a projection operator that isolates the dynamics of an $s$ -particle subsystem while exactly accounting for the initial state of the full $N$ -particle system, resulting in a homogeneous evolution equation.

The authors assert that this approach provides a more rigorous foundation for kinetic theory, particularly for understanding the emergence of irreversibility from reversible dynamics without relying on unrealistic assumptions about the initial uncorrelated state of the system.