2026 FIELDS MEDAL · YU DENG

How do billiard-ball atoms become a flowing gas?

A visual route through the 125-year question linking reversible Newtonian particles, Boltzmann’s statistical equation, and the fluid laws we see.

Start from intuition Jump to the proof engine

Audience undergraduate + curious expert Reading time 75–110 min Interactive labs 14+

z_j = (x_j, v_j)

𝔼N · ε^d−1 ≈ 1

f_s → f^⊗s

Conceptual gas-particle field · not a numerical hard-sphere simulation

One physical gas.

MICROMESOMACRO

Three mathematical languages.

HOW TO READ THIS

A map of a proof—not a replacement for 192 pages

This guide makes every logical dependency visible, proves the elementary mechanics and scaling calculations in full, and gives a step-by-step reconstruction of the long-time argument. The deepest cutting-algorithm estimates are explained structurally and schematically; reproducing their full hypotheses, constants, and case analysis requires the research article.

Intuition Derivation Research-level bridge

00 · THE BIG PICTURE

Zoom out. The language changes.

Use the scale switcher. The object never changes—only what information we retain.

What are all positions and velocities?

State: (x₁,v₁), …, (xₙ,vₙ)

MODEL

Newton + elastic collisions

Follow every hard sphere. Between collisions it travels in a straight line; at collision, momentum and kinetic energy are conserved.

KEEPS

Exact particle identities and trajectories.

FORGETS

Nothing—this is why the state has about 6N numbers in 3D.

Microscopic

Hard-sphere dynamics

N particles; Newton’s laws

Kinetic limit N → ∞, ε → 0 Deng–Hani–Ma

Mesoscopic

Boltzmann equation

density f(t,x,v)

Hydrodynamic limit Kn → 0 (collision rate → ∞) Established Boltzmann → fluid limits + Deng–Hani–Ma’s periodic kinetic limit

III

Macroscopic

Fluid equations

ρ(t,x), u(t,x), T(t,x)

TWO DIFFERENT LIMITS

Stage 1: let ε→0 while α=(𝔼N)ε^d−1 is fixed; this produces Boltzmann. Stage 2: after the kinetic equation is obtained, let the Knudsen number Kn→0; in the nondimensional convention used for this bridge, Kn=α⁻¹. This is a separate hydrodynamic limit, and the companion paper specifies the permissible iterated/scaling variants.

01 · WHAT THE MEDAL RECOGNIZES

“For his work in partial differential equations, including the rigorous derivation of the Boltzmann equation from hard-sphere dynamics for rarefied gases, the derivation of wave kinetic equations from nonlinear dispersive systems, and probabilistic approaches to nonlinear Schrödinger dynamics.”

Official IMU citation ↗

Central story in this guide

Particles → Boltzmann

Long-time control of collision histories for a dilute hard-sphere gas, with Zaher Hani and Xiao Ma.

Where the method matured

Waves → kinetic theory

Rigorous statistical laws for weakly nonlinear waves at their natural kinetic timescale, with Zaher Hani.

A second probability program

Randomness → PDE control

Random tensors, probabilistic scaling, and Gibbs-measure dynamics, with Andrea Nahmod and Haitian Yue.

The hard-sphere Boltzmann theorem and its companion fluid-limit paper are joint work of Yu Deng, Zaher Hani, and Xiao Ma. The wave-kinetic results discussed here are joint work of Deng and Zaher Hani; the highlighted probabilistic NLS results are joint work of Deng, Andrea Nahmod, and Haitian Yue.

02 · START AT THE BOTTOM

One collision, derived completely

Before the probability and combinatorics, the microscopic rule is just geometry plus two conservation laws.

STATE

z_j = (x_j, v_j) ∈ ℝ^d × ℝ^d

Particle j has a center x_j and velocity v_j. In 3D that is six real numbers.

NO OVERLAP

|x_i − x_j| ≥ ε

Each sphere has diameter ε. Their centers can approach to distance ε, never less.

FREE FLIGHT

ẋ_j = v_j, ṽ_j = 0

Until a collision, velocity stays constant and position changes linearly.

INTERACTIVE LAB 1

Turn the collision normal

At impact, only the component of relative velocity parallel to the line of centers is exchanged. The perpendicular component is untouched.

Collision normal angle 28°

Total momentum before(1.20, 0.10)

Total momentum after(1.20, 0.10)

Kinetic energy error0.000000

Follow the live calculation

1 · normal componenta=(v−v*)·ω

2 · particle 1v′=v−aω

3 · particle 2v′*=v*+aω

particle 1 particle 2 collision normal ω

Exact velocity calculation; schematic replay, not a many-particle simulation.

WORKED BRIDGE · WHERE THE FORMULA COMES FROM

In the center-of-mass frame, a collision is a reflection

Define the center-of-mass velocity and relative velocity

V = (v+v_*)/2, g = v−v_*.

Momentum conservation keeps V fixed. For identical, frictionless, perfectly elastic spheres, the tangent component of g is unchanged and the normal component reverses:

g = g_tan + (g·ω)ω
g′ = g_tan − (g·ω)ω = g − 2(g·ω)ω.

Since v=V+g/2 and v_*=V−g/2, substituting g′ gives the collision rule used above.

EXACT NUMERICAL EXAMPLE

ω=(1,0), v=(2,1), v_*=(0,−1)
a=(v−v_*)·ω=2
v′=(0,1), v′_*=(2,−1)

momentum: (2,0) → (2,0)
squared-speed sum: 6 → 6
relative normal component: +2 → −2

One exact oblique example

v=(2,0), v_*=(0,1), ω=(3/5,4/5)
a=(2,−1)·(3/5,4/5)=2/5
aω=(6/25,8/25)
v′=(44/25,−8/25), v′_*=(6/25,33/25)

Momentum is (2,1) on both sides. The squared-speed sums are 5 before and 2000/625+1125/625=3.2+1.8=5 after; the normal relative velocity changes from +2/5 to −2/5.

Scope. “Exchange the parallel component” is special to equal-mass, smooth, perfectly elastic spheres. Unequal masses or rough bodies obey different formulas.

DERIVATION A

Why the collision formula conserves momentum and energy

1

Isolate the exchanged scalar

Let v,v_* be the velocities before impact and v′,v′_* the corresponding velocities after impact. Set a = (v − v_*) · ω, the signed normal component of relative velocity along the unit collision normal ω.

v′ = v − aω, v′_* = v_* + aω

The same vector aω is subtracted from one velocity and added to the other.
2

Add the two outgoing velocities

v′ + v′_* = (v − aω) + (v_* + aω) = v + v_*

So total momentum is exactly conserved (equal masses are normalized to 1).
3

Expand the outgoing kinetic energy

|v′|² + |v′_*|²
= |v − aω|² + |v_* + aω|²
= |v|² + |v_*|² + 2a² − 2a(v − v_*)·ω

Because |ω| = 1 and a = (v − v_*)·ω, the last two terms are 2a² − 2a² = 0.
✓

Conclude

|v′|² + |v′_*|² = |v|² + |v_*|²

The microscopic map is invertible: reverse every velocity and the spheres retrace their paths.

03 · FROM ONE TRAJECTORY TO A LAW

Randomness enters at time zero

Newton evolves each chosen configuration deterministically. Probability describes which configuration we chose—and what a typical particle does.

N-particle density

W_N(t, z₁,…,z_N)

The sequence (W_N/N!) gives the sector densities of the random-N grand-canonical law. Each N-sector lives on a 2dN-dimensional phase space.

s-particle correlation

f_s(t,z₁,…,z_s)

A rescaled factorial-moment density built from the grand-canonical sequence (W_N). It plays the role of an s-particle joint intensity, but is not generally a normalized tagged-particle marginal.

Propagation of chaos

f_s^ε(t) ≈ f(t)^⊗s𝟙_{D_s}

At finite ε, exclusion forbids overlap. For fixed s, the appropriately scaled correlation approaches the factorized density on the allowed domain as ε → 0.

f₂(z₁,z₂)≈f(z₁) f(z₂)

THE CRUCIAL WORD: “CHAOS”

It means independence, not disorder.

If knowing particle 1’s state gives asymptotically no information about particle 2’s state, their limiting correlation factorizes. Collisions do create dependence; the theorem controls the aggregate connected contribution of all relevant histories through estimates and cancellations.

Why factorization closes the equation

The exact evolution equation for f₁ depends on f₂. If f₂ ≈ f₁f₁, that unknown two-particle term becomes a quadratic expression in f₁. That quadratic expression is precisely the Boltzmann collision operator.

VISUAL CHECKPOINT · CHAOS

Equal marginals do not imply independence

Two binary variables X and Y always have 50–50 marginals here. Move only their connected correlation c, then find the one setting where knowing X tells you nothing about Y.

Correlation parameter c = 0.20

Goal: make P(Y=L | X=L) equal the marginal P(Y=L)=0.500.

Joint probability table

X=L, Y=L0.450

X=L, Y=R0.050

X=R, Y=L0.050

X=R, Y=R0.450

P(Y=L | X=L): 0.900
marginal P(Y=L): 0.500
connected cell remainder: c = +0.200
L¹ factorization error: hidden until checked

Continuous analogue: a two-particle cumulant measures f₂−f⊗f. The theorem controls this in an integrated norm, with the finite-ε exclusion indicator 𝟙_D₂.

Grand-canonical and factorial correlation—in plain language

Grand-canonical means the particle count N is itself random, concentrated around a large mean 𝔼N; this simplifies cluster bookkeeping. The function f_s counts expected ordered s-tuples near (z₁,…,z_s), so it need not integrate to 1. For a fixed-N normalized marginal, the analogous statement is the joint law of s tagged particles. Factorization means f^⊗s(z₁,…,z_s)=f(z₁)···f(z_s).

04 · THE MESOSCOPIC LAW

Read the Boltzmann equation like a balance sheet

Left: particles move through space. Right: collisions move probability into and out of velocity v.

(∂_t + v·∇_x) f = Q(f,f)

∫∫

f′ f′_* − f f_*

[(v − v_*)·ω]₊

dω dv_*

Symbol key. f=f(t,x,v), f_*=f(t,x,v_*), f′=f(t,x,v′), f′_*=f(t,x,v′_*). The partner velocity v_* runs over ℝ^d, the impact normal ω over S^d−1, and [a]₊=max(a,0). The displayed PDE normalizes the order-one particle collision rate to 1.

Transport

Along a free particle path x(t) = x₀ + tv, the operator ∂_t + v·∇_x is the total rate of change of f.

VISUAL CHECKPOINT · GAIN MINUS LOSS

Does one collision channel add or remove density at v?

For one fixed partner velocity and collision normal, predict the sign of B=f′f′_*−ff_*. Then reveal the multiplication.

f(v) 0.80 f(v_*) 0.60 f(v′) 0.40 f(v′_*) 0.50

Predict the local bracket gain, B>0 balance, B=0 loss, B<0

Pair-density products

GAIN · f′f′_*0.20

LOSS · ff_*0.48

NET · gain−losshidden

Bars show the incoming and outgoing pair products. This is one integrand channel—not the sign of the fully integrated Q(f,f)(v).

DERIVATION B

Why the collision term has “gain minus loss” form

1
Pick the state
Fix position x and velocity v. We ask how collisions change the number of particles near this state.
2
Count arrivals
A pre-collision pair (v′,v′_*) can scatter into (v,v_*). Under independence its density is f′f′_*.
3
Count departures
A present pair (v,v_*) can collide and leave velocity v. Under independence its density is ff_*.
4
Weight geometry
[(v−v_*)·ω]₊ is the positive part of the signed normal component of relative velocity. Integrate over partners and impact directions.

INTERACTIVE LAB 2

The Boltzmann–Grad balance

Make each sphere smaller. To keep roughly one collision per unit flight length, add just enough particles to compensate for the shrinking collision cross-section.

Dimension

Diameter ε = 10⁻³

Collision cross-sectionε² = 10⁻⁶

Required particle countN ≈ 10⁶

ProductN ε² ≈ 1

Mean free path (scaled)ℓ ≈ 1

Occupied volume fractionN ε³ ≈ ε = 10⁻³

The dots are a capped, subsampled illustration; the readout gives the actual scale. The unit-volume heuristic suppresses fixed geometric constants and a typical speed. Here ≍ or ≈ means “the same order up to fixed constants.”

Worked 3D numbers with geometric constants

Take ε=10⁻³ and 𝔼N=ε⁻²=10⁶ in unit volume. The center-to-center collision cross-section is σ=πε²=π·10⁻⁶. At unit relative speed, the simplified collision rate is nσ≈π and mean free path ℓ≈1/π≈0.318—still order one.

occupied fraction = (𝔼N)(π/6)ε³ = (π/6)·10⁻³ ≈ 5.24·10⁻⁴

The collision opportunity stays finite while the volume occupied by matter tends to zero. The theorem absorbs fixed geometric constants into its normalization; this is the physical order-of-magnitude calculation.

DERIVATION C

Why Nε^d−1 must stay of order one

1
Sweep a collision tube
In time Δt, one sphere with speed |v| sweeps a tube of length |v|Δt. Another center causes collision if it lies within a (d−1)-dimensional cross-section of scale ε^d−1.
swept volume ≍ |v| Δt · ε^d−1
2
Multiply by particle density
In a unit-volume normalization, the number density is of order 𝔼N; in a general volume V it is (𝔼N)/V. Multiply that by the swept volume.
expected collisions ≍ (𝔼N) ε^d−1 |v| Δt
3
Demand a nontrivial limit
If (𝔼N)ε^d−1 → 0, collisions disappear. If it tends to ∞, collisions occur infinitely fast. Keeping it near α ∈ (0,∞) preserves an order-one collision rate.
𝔼N → ∞, ε → 0, (𝔼N)ε^d−1 → α

THE EXACT BRIDGE · BBGKY

Before taking a limit, one equation never closes

The Bogoliubov–Born–Green–Kirkwood–Yvon hierarchy is the infinite ladder where “independence” must be proved—not inserted by hand. To predict particle 1, we need the chance that a partner is at contact: two-particle information f₂. The f₂ equation similarly needs f₃.

∂_tf_s^ε + ∑_i=1^s v_i·∇_{x_i} f_s^ε = C_s,s+1^ε f_s+1^ε

schematic BBGKY hierarchy; boundary collision terms are compressed into C_s,s+1^ε

WORKED BRIDGE · WHY f₁ NEEDS f₂

1
Integrate Liouville’s equation over every variable except z₁=(x₁,v₁). Interior transport terms cancel by integration by parts.
2
The exception is the contact boundary x₂=x₁+εω. Specular reflection relates incoming and outgoing traces there; for s≥2, observed-particle collisions also appear as boundary conditions on D_s^ε, while C_s,s+1^ε describes contact with an added partner.
3
Flux through a surface patch is weighted by the positive normal relative speed B=[(v₁−v₂)·ω]₊.
4
For normalized tagged marginals, contact area ε^d−1 and the order-𝔼N partner count appear as a prefactor. The paper’s rescaled factorial correlations f_s^ε absorb that Boltzmann–Grad normalization into their definition, so the displayed hierarchy has coefficient 1.
5
Formally, contact points merge and f₂^ε factorizes, producing Q(f,f). This is only the closure picture: bulk L¹ convergence alone does not control a boundary trace. The collision-history estimates justify the limiting contact flux.

(∂_t+v₁·∇_x₁)f₁^ε
= ∫ B(v₁−v₂,ω)
×[f₂^ε(z₁′,x₁+εω,v₂′)−f₂^ε(z₁,x₁+εω,v₂)] dωdv₂

Schematic formula in the site’s rescaled-correlation normalization, with z₁′=(x₁,v₁′). Prime/sign conventions vary with the orientation of ω; the reversible collision map makes the equivalent forms agree.

factorization + ε→0 ↓

(∂_t+v·∇_x)f = Q(f,f)

f₁needsf₂needsf₃needs…

propagation of chaos

f₂ ≈ f⊗f⇒∂_tf + v·∇_xf = Q(f,f)

Exact: Liouville’s transport equation for the full N-particle density preserves all microscopic information.

Form factorial correlations: integrating out particles makes f_s depend on f_s+1.

Close: only the limiting factorization f₂ ≈ f⊗f produces a one-particle kinetic equation.

05 · WHY 1975 WAS NOT ENOUGH

The short-time proof grows a forest

Lanford expanded a tagged particle’s possible collision histories backward in time. The method was revolutionary—but its absolute series converged only for a small fraction of a mean free time.

INTERACTIVE LAB 3

Grow one real Duhamel history

At each backward collision, one new possible partner is attached to one particle already in the history. Compare that actual structure with a binary branching metaphor.

Backward collisions 4

particles in one history5

possible parent sequences24

ordered-time simplext⁴ / 4!

The displayed graph is one selected history. Its rapidly growing count comes from choosing a parent at every creation and integrating labels, times, velocities, and impact directions—not from particles literally doubling.

EXACT TOY COUNT · WHY A FINITE TIME RADIUS APPEARS

Parent choices and ordered times contain opposite factorials

n creations→1·2···n = n! parent choices×Vol{0<t_n<···<t₁<t} = tⁿ/n!→(Ct)ⁿ after collision integrals

The factorials roughly cancel. Thus the obstruction is not literal binary population doubling; it is that the remaining absolute series ∑(Ct)ⁿ has only a finite radius of convergence. Time layering makes Cτ<1 locally, while cumulants carry the correlations that a naïve restart would erase.

absolute n-collision bound ≲ (Ct)ⁿ

Then ∑_n≥0(Ct)ⁿ is a geometric series and converges only when Ct<1. Time layering keeps each local width τ in that safe regime; the difficult part is carrying old correlations across the layer boundaries.

1975

Lanford’s expansion

Expand all collision histories from time t back to 0.
Bound the absolute value of every tree contribution.
Sum the resulting series.

Works only while the series stays small

2024–25

Deng–Hani–Ma’s iteration

Divide long time into locally manageable layers.
Stop expanding the factorized leading part.
Locally update the leading density; carry multi-layer history only through cumulants.

Reaches every regular Boltzmann time

06 · THE PROOF ENGINE

Keep the signal. Expand the error.

The central innovation is an inductive representation that remembers the entire relevant collision history without expanding the already-understood Boltzmann part.

Long-time Boltzmann limit, decoded

sup_0≤t≤T ‖ f_s^ε(t,z_s) − [∏_j=1^s f(t,z_j)] 𝟙_{D_s} ‖_L¹ ≤ ε^θ

uniformly for s ≤ |log ε|, provided ε is small enough and the Boltzmann solution remains regular on [0,T].

Open the assumption ledger

Model: hard spheres of diameter ε in ℝ^d, d≥2; the specified grand-canonical initial ensemble, conditioned on non-overlap.

Initial profile: f₀ is a normalized nonnegative density, sufficiently regular in x, spatially localized, and bounded by a Gaussian tail in v. The paper packages these requirements in a weighted norm denoted ‖·‖_Bol,2β; β and the bound B₀ are fixed positive constants.

Effective solution: f solves the hard-sphere Boltzmann equation on [0,T] and retains a uniform Gaussian velocity bound, summarized by a fixed constant A.

Limit: the ensemble has 𝔼N·ε^d−1≈1; first choose T and its solution bounds, then take ε sufficiently small.

f_s^ε is the exact rescaled s-particle factorial correlation of the grand-canonical hard-sphere ensemble. It is a joint intensity, not generally a normalized tagged-particle probability density.

THE INVARIANT CARRIED FROM LAYER TO LAYER

Here “invariant” means an ansatz preserved by induction, not a conserved physical quantity.

f̃_K^ε(ℓτ) = ∑_H⊆K (f^A)^⊗(K∖H)(ℓτ) · E_H^ε(ℓτ) + Err_K^cumul, E_∅=1
f_K^ε(T) = f̃_K^ε(T) + f_K^trunc(T)

f^A almost-factorized leading density; stays close to Boltzmann f E_H defect/cumulant; a singleton is a one-root defect, while |H|≥2 contains genuine multi-root correlation Two errors cumulant-representation error and separately truncated rare events; both are negligible

TWO-ROOT EXPANSION

E_{1} ≈ f̃₁^ε−f₁^A
E_{1,2} ≈ f̃₁₂^ε−f₁^Af₂^A−f₁^AE_{2}−f₂^AE_{1}

RECONSTRUCTION

f̃₁₂^ε ≈ f₁^Af₂^A
+ f₁^AE_{2} + f₂^AE_{1}
+ E_{1,2} + Err

Read it carefully. The ≈ signs suppress the representation and truncation errors shown in the invariant above. E_{i} measures how one represented root differs from the chosen stable leading state f^A. Only E_{1,2} is a genuinely two-root connected term. If f^A were the exact one-particle density and errors vanished, singleton defects would vanish; the layered proof chooses f^A because it can be updated and controlled.

TWO-LAYER TOY INDUCTION

The method neither forgets correlation nor re-expands everything

WRONG 1 · DISCARD

f₂^ε ↦ f^A⊗f^A

Erasing E₁₂ assumes fresh independence at every boundary—the very statement that must be proved.

THE NARROW PATH · CARRY

f₂^ε ≈ f^A⊗f^A + E₁₂

Schematic after suppressing singleton defects and both error families. Update the product locally; expand the represented defects/cumulants backward so old correlation remains recorded.

WRONG 2 · RE-EXPAND

expand(f^A⊗f^A + E₁₂)

Re-expanding the already-understood product recreates the long divergent history forest.

δ_ℓ+1 ≤ (1+Cτ)δ_ℓ + Cε^ν

δ_L ≲ e^CT(δ₀+LCε^ν) → 0, L=T/τ fixed before ε→0

This scalar recurrence is a pedagogical stability analogy, not the paper’s estimate. The proof replaces δ_ℓ by weighted cumulants and molecule integrals with detailed support conditions.

STEP 1 OF 8 intuition

Replace one impossible interval by L small ones

Choose a short layer width τ = T/L. Local cluster expansions remain convergent inside each layer, even though a single expansion across [0,T] would diverge.

INPUT

A regular Boltzmann solution on [0,T]

OUTPUT

Short intervals [(ℓ−1)τ, ℓτ]

RISK

Correlations from earlier layers cannot be forgotten

IDENTIFICATION STEP · WHY THE MAIN TERMS ARE BOLTZMANN

No-recollision histories reproduce the kinetic Duhamel series

f(t) = U(t)f₀ + ∫₀^t U(t−s)Q(f,f)(s) ds
U(t)g(x,v)=g(x−tv,v)

Substituting this integral equation into itself attaches an earlier collision partner and builds a tree. A microscopic history with no recollision has the same four limiting ingredients: free flight, the elastic collision map, relative-speed weighting, and an independent new partner drawn from f.

microscopic
no-recollision history

term by term →

Boltzmann
Duhamel term

+ extra non-tree edgerecollision / error molecule

Recollisions make a particle meet someone already in its history. Those non-tree edges are controlled by cutting estimates; additional signed cluster terms are paired by the paper’s involution and cancel before absolute estimation.

THE COMBINATORIAL HEART

A collision history becomes a “molecule”

Atoms are collision/overlap events; bonds carry particle lines and order. In one layer, E−V+C counts independent cycles, where E is the number of edges, V vertices, and C connected components: a three-vertex path has 2−3+1=0 cycles; a triangle has 3−3+1=1. The paper’s multi-layer complexity ρ is richer, adding root/cross-layer and recollision information.

GEOMETRIC MODEL · WHERE SMALLNESS COMES FROM

A second prescribed encounter is a thin target

Fix one trajectory during a layer of length τ. Another trajectory hits its ε-neighborhood only if the transverse impact parameter lies in a (d−1)-dimensional disk of radius O(ε):

allowed transverse measure = O(ε^d−1).

An ordinary first collision uses the freedom already balanced by (𝔼N)ε^d−1≈1. A further independently prescribed encounter imposes an extra constraint. The complete proof earns a weaker but robust net ε^νρ, because grazing, nearly parallel, degenerate, and cross-layer configurations must also be covered.

INTERACTIVE LAB 4

Cut a complex molecule

Choose cuts in a legal order. Each cut is a Fubini decision: which variables will be integrated first, uniformly in the variables that remain fixed?

Normal {3}0

Good {33}0

Bad {4}0

Net ε-gainpending

Start at the top to respect time ordering.

This toy displays a nondegenerate support configuration in which the shown {33} is good. The topology label alone does not guarantee an ε-gain.

atom particle line cycle constraint chosen cut

THE NUMERICAL LOGIC OF THE PROOF

Complexity loses logarithms; geometry wins powers.

COUNT

# molecules of complexity ρ ≤ C^|M||log ε|^Cρ

Each unit of multi-layer complexity incurs a logarithmic counting loss.

COST

‖I_M‖ ≤ τ^|M| ε^νρ

The full cutting/root budget combines good thin-set restrictions, compensates bad pieces, and earns a net ε^νρ. It is not literally one gain per graph cycle.

→

NET

|log ε|^Cρ ε^νρ → 0

Any fixed positive power of ε beats every logarithmic power.

First sum molecule sizes. Choose τ so Cτ<1; then ∑_m≥0(Cτ)^m converges for m=|M|. What remains per unit of complexity is |log ε|^Cε^ν→0. Here ν>0 and c>0 below are fixed small constants depending on dimension; ν is not a velocity.

Research-level precision: the cutting budget

THE CUTTING THEOREM’S BUDGET · SCHEMATIC

½υ · #_good + (d−1)(|H| − #_bad) ≥ cρ

Good elementary restrictions and the root normalization jointly pay for bad pieces and for a positive fraction of the multi-layer complexity. The paper fixes υ=3^−d−1; this auxiliary bookkeeping constant is distinct from the final gain exponent ν. Constants, topology subtypes, and logarithmic factors are suppressed here.

VISUAL CHECKPOINT · ASYMPTOTIC BUDGET

When does the ε-power actually win?

Explore the explicitly pedagogical bound R_ρ(ε)=(|log ε|^Aε^ν)^ρ, with A=1 and ν=0.15. These are chosen for visibility, not claimed constants from the paper.

ε = 10⁻³ complexity ρ = 3

At this ε, increasing ρ will… improve the toy bound worsen the toy bound

Base-10 exponent ledger

smaller bound ←larger bound →

log-count loss+2.52

ε-power gain−1.35

log₁₀ R_ρ+1.17

3[log₁₀(3 log 10) − 0.15·3] = +1.17

Positive log₁₀R means this illustrative numerical bound exceeds one; negative means it is small. The theorem is asymptotic and does not claim practical accuracy at moderate ε.

NORMAL

{2}

Two fixed ends determine the remaining collision variables. No power gained or lost.

NORMAL

{3}

One free incoming trajectory supplies exactly the ordinary collision freedom.

GOOD

{33}

A second prescribed encounter constrains a free trajectory to a thin tube: a positive ε-power gain.

BAD

{4}

No end is fixed, so an integration degree is wasted. The algorithms ensure good pieces compensate these losses.

Research-level precision: topology alone does not determine “good”

The four cards are the intuition used in the exposé. In the paper, “good” depends on support restrictions created during splitting, not topology alone: a {3} or {33A} piece may be normal or good; {33B} supplies important good cases; a {4} is counted as bad but can also carry a good restriction; and good {44} pieces handle strongly degenerate configurations.

Boltzmann stability keeps f^A(ℓτ) close to f(ℓτ).

Cutting/root gains make every cumulant E_H small after the full budget is summed.

Truncation estimates make Err negligible.

∴

Induction closes for ℓ = 1,…,L.

07 · COMPLETE THE CHAIN

From velocity clouds to fields

At very high collision rate, velocity distributions rapidly approach local Maxwellians. Their few parameters become the fluid variables.

Temperature T=1.3 Flow speed u=0.7

LOCAL EQUILIBRIUM

One bell curve, three fields

M(t,x,v) = ρ(t,x)[2πT(t,x)]^d/2 exp(−|v−u(t,x)|² / 2T(t,x))

ρ: mass density—the area under the velocity bell
u: bulk velocity—the bell’s center
T: temperature—the bell’s width

∫M dv=ρ, ρ⁻¹∫vM dv=u, (dρ)⁻¹∫|v−u|²M dv=T

So the three fields really are moments of one velocity bell. The canvas shows a one-dimensional velocity slice; particle mass and Boltzmann’s constant are normalized to 1.

Two macroscopic zooms. Compressible Euler observes order-one density changes on an advective (hyperbolic) time scale. Incompressible NSF zooms into small fluctuations over a slower diffusive scale, so viscosity and heat conduction survive.

Fast collisions + small perturbations

Incompressible Navier–Stokes–Fourier

∂_tu + u·∇u − μ₁Δu = −∇p
∂_tϑ + u·∇ϑ − μ₂Δϑ = 0
∇·u = 0

ϑ is the coupled temperature/density fluctuation. The companion paper writes an equivalent scalar ρ under its Boussinesq relation ρ + ϑ = 0.

Local Maxwellian + hyperbolic scaling

Compressible Euler

∂_tρ + ∇·(ρu) = 0
∂_t(ρu) + ∇·(ρu⊗u) + ∇p = 0
∂_tE + ∇·[(E+p)u] = 0
E = ½ρ|u|² + ½dρT, p=ρT

What “resolves Hilbert’s sixth problem” means here

PROVED

Deng, Hani, and Ma derive the Boltzmann equation from dilute identical hard spheres on ℝ^d, d≥2, through every fixed interval on which the required regular Boltzmann solution exists. Their companion paper extends the kinetic limit to 𝕋^d, d=2,3, and combines it with established Boltzmann-to-fluid limit theorems to derive incompressible Navier–Stokes–Fourier and compressible Euler under the paper’s stated prepared-data, regularity, and iterated/scaling limits.

NOT CLAIMED

A complete axiomatization of physics; dense gases or liquids; arbitrary interactions, shapes, domains, or boundary conditions; every hydrodynamic scaling; one unrestricted simultaneous limit; or global regularity for the Boltzmann or three-dimensional Navier–Stokes equations.

THE ARROW OF TIME

Reversible rules, irreversible statistics

Visual analogy: colored dots mix under an invertible playback. This is not a correlation simulation.

1
The hard-sphere flow preserves phase-space volume and can be reversed.
2
The initial ensemble selects “chaotic” states: particles are almost independent.
3
The time-reversed state after mixing has extraordinarily organized correlations; it is not a fresh chaotic initial state.
4
Thus an irreversible effective law can emerge for the prescribed forward-chaotic ensemble in the kinetic limit without contradicting reversible microscopic mechanics.

The theorem justifies the Boltzmann description across its regular lifespan. The entropy interpretation additionally uses Boltzmann’s H-theorem and the statistical choice of initial ensemble.

DERIVATION D · THE H-THEOREM

Where the irreversible sign appears

Formal calculation for positive smooth f, with periodic x or sufficient spatial/velocity decay. Rigorous entropy identities or inequalities require the corresponding integrability and trace hypotheses.

1 · DEFINE

H[f] = ∫ f log f dx dv

Up to physical constants and a sign, −H is entropy.

2 · DIFFERENTIATE

dH/dt = ∫ Q(f,f) log f dx dv

The transport term integrates to zero on a periodic domain or with sufficient decay. The “+1” from differentiating f log f vanishes because ∫Q(f,f)dv=0: collisions conserve mass.

3 · SYMMETRIZE

dH/dt = −¼∫ [(v−v_*)·ω]₊ (A−C) log(A/C) dx dv dv_* dω

Here A=f′f′_* and C=ff_*. The factor ¼ comes from averaging the equivalent expressions obtained by swapping particles and swapping pre/post-collision variables; with the positive-part kernel, the latter change is accompanied by ω↦−ω.

4 · SIGN

(a−c) log(a/c) ≥ 0 ⇒ dH/dt ≤ 0

Because log is increasing. Equality characterizes local collision equilibrium under the usual hypotheses.

NUMERICAL SIGN CHECK

B=1, A=4, C=1 ⇒ −¼B(A−C)log(A/C)=−¾log4≈−1.040

One symmetrized channel is negative. If A and C are interchanged, both A−C and log(A/C) change sign, so their product stays nonnegative and the leading minus sign still makes entropy production nonpositive.

08 · THE BIGGER PICTURE

One idea migrates across fields

Deng’s programs share a theme: randomness is not noise to discard—it is structure that can survive nonlinear evolution in a controlled form.

WAVES

Nonlinear Schrödinger equation

(i∂_t − Δ_β)u + λ²|u|²u = 0

Fourier modes interact weakly. Δ_β is an anisotropic Laplacian with dispersion ω(k)=|k|_β². Picard iterates produce paired trees and Feynman diagrams.

SHARED TEMPLATE layer long time preserve statistical history organize nonlinear diagrams make smallness beat their count

PARTICLES

Hard-sphere Newtonian flow

ẋ_j=v_j, ṽ_j=0 between impacts; v_j jumps by the elastic rule at impact

Collision histories create clusters and recollisions. The wave strategy inspired the architecture; particle geometry required new algorithms.

WAVE-SPECIFIC MACHINERY

Paired interaction trees, layered “gardens,” oscillatory resonance estimates, twist/vine cancellations, and rigidity.

PARTICLE-SPECIFIC MACHINERY

Cumulants, collision/overlap molecules, hard-sphere recollision geometry, truncation, and cutting into elementary integrals.

THE SAME META-PROBLEM IN THREE DIALECTS

Propagation of randomness

Nonlinearity creates correlations. Deng’s work finds enough structure inside those correlations to retain a statistical law—or construct a rough random flow—over meaningful times. Both wave pillars start with random Fourier data, but wave kinetics averages over realizations to derive a closed law for a spectrum, whereas probabilistic PDE fixes almost every realization and constructs its canonical rough trajectory. This table compares proof problems; it does not claim the theorems are instances of one formal framework.

Comparison categoryParticlesWeak wavesRough random PDE

Microscopic system

Newtonian hard spheres

random-mode cubic NLS

Wick-ordered NLS flow

Object retained

marginals f_s; density f

spectrum n(k)=𝔼|û_k|²

Gibbs-typical solution law

Naïve obstacle

recollisions destroy exact independence

resonant modes build correlations

rough products are not classically defined

Memory object

cumulants & molecules

paired trees & layered gardens

random averaging operators & tensors

Source of gain

thin recollision geometry ε^ν

oscillation, pairings, cancellations

Gaussian contraction & operator norms

Output

quadratic Boltzmann closure

cubic wave-kinetic closure

strong flow + invariant Gibbs measure

VISUAL CHECKPOINT · THE BIGGER PICTURE

Match each source of smallness to its program

The three pillars share a count-versus-structure problem, but the small parameter and output object are different. Choose a program for each mechanism.

Thin prescribed recollision tube Oscillatory phase and resonant pairing Gaussian tensor/operator-norm cancellation

INTERACTIVE LAB 5 · WAVE PILLAR

When four Fourier modes resonate

On a periodic box, u(x)=∑_kû_ke^ik·x. A wavevector k labels a Fourier wave’s direction and wavelength; points here are wavevector tips, not physical positions. Momentum matching determines k=k₁−k₂+k₃. On the square torus, lasting exact resonance also matches ω(k)=|k|².

Angle between k₁−k₂ and k₃−k₂ 90°

Goal: find the perpendicular rectangle where Δω=−2(k₁−k₂)·(k₃−k₂)=0.

Wavevector mismatch0 · enforced

Normalized mismatch Δω/|k₁−k₂|²0.000

Statusexact resonance

k₁ − k₂ + k₃ − k = 0 Δω = |k₁|² − |k₂|² + |k₃|² − |k|²

Why phase mismatch matters: ∫₀^te^iΔωsds=(e^iΔωt−1)/(iΔω), whose size is at most 2/|Δω| when Δω≠0; at Δω=0 it equals t and accumulates.

Scope. This slider uses continuum Euclidean geometry. Finite-torus Fourier modes lie on a lattice, so an arbitrary angle need not be an allowed discrete quartet. The α=L⁻¹ theorem in the 2023 paper uses generic anisotropic β; the lab explains resonance algebra, not its number-theoretic estimates.

EXACT INTEGER RESONANCE

k₂=(0,0), k₁=(1,0), k₃=(0,2), k=(1,2)
k₁−k₂+k₃−k=(0,0)
Δω=1−0+4−5=0

OFF RESONANCE

move k₃ to (1,1), so k=(2,1)
Δω=1+2−5=−2
∫₀^πe^−2isds=0, while resonance gives π

Momentum matching builds a parallelogram. For Euclidean quadratic dispersion, writing u=k₁−k₂ and w=k₃−k₂ gives Δω=−2u·w, so a rectangle is exactly resonant.

THE WAVE KINETIC COLLISION OPERATOR

Near-resonances concentrate onto a resonant manifold

Write n=n(τ,k) and n_i=n(τ,k_i), where τ=t/T_kin. At finite L, near-resonant quartets accumulate; in the kinetic limit their contribution concentrates onto the continuum manifold represented by the two Dirac deltas. For d≥3 and the admissible scalings, n satisfies a closed cubic law. The β-subscript covers the anisotropic dispersion; on the square torus β is Euclidean.

∂_τn(k) = ∫ [n₁n₂n₃ − nn₂n₃ + nn₁n₃ − nn₁n₂]
× δ(k₁−k₂+k₃−k)
× δ(|k₁|_β²−|k₂|_β²+|k₃|_β²−|k|_β²) dk₁dk₂dk₃

The Dirac deltas impose wavevector and frequency matching. The bracket is the wave analogue of gain minus loss—cubic rather than Boltzmann’s quadratic closure.

WHY RANDOM NLS NEEDS NEW OBJECTS

A function too rough to multiply by itself

u₀(x) = ∑_k∈ℤ² g_k / ⟨k⟩ · e^ik·x

The coefficients g_k are independent complex Gaussians and ⟨k⟩=(1+|k|²)^1/2. Almost surely u₀ lies in H^s for every s<0 (written H⁰⁻), but not generally in L². Naïve cutoff powers fail to converge as classical products; Wick ordering subtracts divergent self-contractions before the cutoff is removed.

u = u_lin + 𝒬(u_lin) + Y (schematic)

Wick ordering subtracts divergent self-contractions before the cutoff limit. The random averaging correction 𝒬 isolates rough linear randomness; Y is smoother. Random tensors record how Gaussian inputs pair through nonlinear iterations.

SCALING THRESHOLDS FOR p-POWER NLS

s_prob = −1/(p−1), s_det = d/2 − 2/(p−1)

For odd nonlinear degree p≥3 (excluding the exceptional general-theorem case (d,p)=(1,3)), randomness can lower the relevant threshold from s_det toward s_prob; the full probabilistically subcritical range s>s_prob is covered. A tensor “flattening” groups input/output indices into a matrix; operator norms give stability while Hilbert–Schmidt bounds capture square-root cancellation.

INTERACTIVE LAB 6 · RANDOM-PDE PILLAR

Why H⁰⁻ converges while L² diverges

Truncate u at |k|≤K in two dimensions. Its expected squared Sobolev norm is ∑⟨k⟩^2s−2. A dyadic shell |k|≈2^j has ≍2^2j modes, so—after suppressing fixed lattice constants—its normalized shell contribution is ≍2^2sj.

dyadic cutoff K = 2⁸

Which shell pattern gives a convergent norm? 2^−j/2 for s=−1/4 1 for s=0

normalized L² shell model: 8.00 · grows like log K
normalized H^−1/4 shell model: 2.26 · approaches 2.41

This plots an asymptotically equivalent, normalized dyadic-shell model—not exact lattice sums and not a random sample. Standard independence/Gaussian arguments upgrade the threshold to the almost-sure statement.

INTELLECTUAL LINEAGE

From a physical guess to a long-time theorem

1872

A statistical equation is born

Ludwig Boltzmann formulates a kinetic equation using “molecular chaos”: particles about to collide are treated as statistically independent.

What it suppliedA correct and extraordinarily successful target equation—but not a theorem deriving it from Newtonian particles.

WAVE-TURBULENCE LINEAGE

Peierls, Hasselmann, and Zakharov formulated kinetic turbulence; Spohn, Lukkarinen–Spohn, Collot–Germain, and Buckmaster–Germain–Hani–Shatah established rigorous partial regimes before the full kinetic-time program.

RANDOM-PDE LINEAGE

Lebowitz–Rose–Speer and Bourgain built Gibbs–NLS foundations; Burq–Tzvetkov developed random-data well-posedness. Parabolic work such as Da Prato–Debussche and Hairer is adjacent, but dispersive NLS lacks parabolic smoothing.

WHAT CAME BEFORE THE LONG-TIME BREAKTHROUGH?

Earlier theorems were profound—but lived in different regimes

No single row “supersedes” all the others. Each controls a different combination of time, perturbation size, and closeness to equilibrium.

1975 · LANFORD

Non-perturbative regular dilute data, very short time

Collision-tree convergence for a small fraction of a mean-free time. This established the microscopic Boltzmann derivation.

1989 · ILLNER–PULVIRENTI

Long time, near vacuum

Global-in-time results in a dispersive low-density regime where particles separate strongly.

2010s–23 · EQUILIBRIUM PROGRAM

Long kinetic times at exact equilibrium

Bodineau, Gallagher, Saint-Raymond, Simonella and collaborators obtained fluctuation and linearized regimes where equilibrium cancellations are available—not nonlinear propagation of chaos for arbitrary near-equilibrium data.

2024–25 · DENG–HANI–MA

Long time, nonlinear off-equilibrium

Every fixed time in the lifespan of the required regular nonlinear Boltzmann solution, without a near-vacuum or equilibrium restriction.

KNOWLEDGE MAP

Prerequisites for the particle theorem

WHERE THE MACHINERY GOES NEXT

Extensions, bridges, and open terrain

These statuses matter: a related use of the machinery is not automatically a consequence of the hard-sphere theorem.

EXTENSION

Forced and inhomogeneous waves

Damped-driven wave turbulence and inhomogeneous kinetic limits bring forcing, dissipation, and space dependence into the picture.

Grande–Hani ↗

ACTIVE BRIDGE

Gravity water waves

Deng–Ionescu–Pusateri establish deterministic estimates and propagation of randomness. This is progress toward—not yet a full quasilinear wave-kinetic derivation.

Part II ↗

Hyperbolic Φ⁴₃ and Hartree

Random-tensor and averaging ideas contribute to Gibbs dynamics and singular dispersive/wave equations beyond the original 2D NLS model.

Φ⁴₃ paper ↗

OPEN · PARTICLES

Beyond ideal hard spheres

Long-time limits for smooth or long-range potentials, boundaries, mixtures, non-spherical particles, off-equilibrium fluctuations, large deviations, and weak Boltzmann solutions.

OPEN · WAVES

Harder resonance worlds

Quasilinear water waves, lower-dimensional degeneracy, condensation or blow-up, and general quantum Boltzmann limits.

OPEN · PROOF THEORY

How universal is “memory + cutting”?

Which many-body systems admit a structured history object whose geometric or oscillatory gain beats combinatorial growth?

NOTATION WITHOUT FEAR

Search the glossary

sphere diameter

The small parameter sent to zero in the dilute-gas limit.

𝔼N

expected particle count

In the grand-canonical ensemble, grows while ε shrinks, balanced by (𝔼N)ε^d−1 ≈ α.

f(t,x,v)

one-particle density

Expected density near position x and velocity v at time t.

f_s

s-particle correlation

Rescaled factorial-moment density of the grand-canonical ensemble; a joint intensity, not generally a normalized marginal.

chaos

asymptotic independence

f_s converges to the product f^⊗s for fixed s.

E_H

cumulant

The genuinely connected, non-factorized correlation among roots H.

molecule

A layered graph encoding collisions, overlaps, particle lines, and time order.

multi-layer complexity

Built from per-layer roots/crossings and recollision ranks; ordinary circuit rank E−V+C is the one-layer prototype.

layer width

T/L, chosen short enough for local cluster expansion.

particle collision rate

The limiting value of (𝔼N)ε^d−1; unrelated to the weak-coupling notation used in some wave papers.

collision normal

Unit vector along the centers of two spheres at impact.

𝟙_{D_s}

exclusion indicator

Zero when any of the s spheres overlap, one otherwise.

Knudsen number

Mean free path divided by macroscopic length. Kn→0 is the fluid limit after the kinetic equation has been obtained.

geometric gain exponent

A fixed positive number depending on dimension; ε^ν is the smallness earned by a good collision restriction.

f^A

leading one-particle state

The locally updated, almost-factorized density carried across time layers. A is a label, not an exponent.

n(τ,k), δ

wave spectrum and constraint

n is expected Fourier-mode energy; a Dirac delta restricts the kinetic integral to a matching manifold.

H^s

Sobolev regularity

A scale measuring derivatives in an averaged sense. H⁰⁻ means membership for every s<0, generally rougher than L².

Wick ordering

renormalized product

Subtract divergent self-contractions before removing a frequency cutoff, producing a meaningful random nonlinear term.

Gibbs measure

energy-weighted random field law

A nonlinear energy weight applied to a Gaussian reference field; invariance means the PDE flow preserves this probability law.

≍

same order

Two quantities bound each other up to fixed positive multiplicative constants.

CHECK YOUR MODEL

Ten questions that expose the common confusions

Score: 0 / 10 · answered 0 / 10

1 Where is probability introduced?

2 Why not expand the full solution on every time layer?

3 What pays for the many possible recollision diagrams?

4 Does the theorem prove all of Hilbert’s sixth problem in every interpretation?

5 Why does exact resonance matter for the wave limit?

6 What is the output of the highlighted random-PDE program?

7 In d=3, ε shrinks by a factor of 10. How must N change in Boltzmann–Grad scaling?

8 Which limiting substitution closes the first BBGKY equation?

9 One backward Duhamel history has three collision creations. How many particles does it contain?

10 If A=f′f′_*=9 and C=ff_*=4, what is the H-theorem sign?

09 · READ THE ORIGINALS

Source library

Primary papers first, expert exposition second, accessible context third. Dates and scopes are shown so claims remain auditable.

Editorial standard

The displayed main theorem is a pedagogical compression, checked against arXiv:2408.07818 v3; consult the paper for the complete hypotheses and normalization. The proof architecture was also compared with the 2026 Bourbaki exposé. “Resolution of Hilbert’s sixth problem” is always qualified by model, limit, domain, regularity, and scope. Interactive simulations are conceptual illustrations, not numerical evidence for the theorem.

Last research verification: 24 July 2026.

How do billiard-ball atoms become a flowing gas?

A map of a proof—not a replacement for 192 pages

Zoom out. The language changes.

Newton + elastic collisions

Hard-sphere dynamics

Boltzmann equation

Fluid equations

Particles → Boltzmann

Waves → kinetic theory

Randomness → PDE control

One collision, derived completely

Turn the collision normal

In the center-of-mass frame, a collision is a reflection

Why the collision formula conserves momentum and energy

Isolate the exchanged scalar

Add the two outgoing velocities

Expand the outgoing kinetic energy

Conclude

Randomness enters at time zero

N-particle density

s-particle correlation

Propagation of chaos

It means independence, not disorder.

Equal marginals do not imply independence

Joint probability table

Read the Boltzmann equation like a balance sheet

Does one collision channel add or remove density at v?

Pair-density products

Why the collision term has “gain minus loss” form

Pick the state

Count arrivals

Count departures

Weight geometry

The Boltzmann–Grad balance

Why Nεd−1 must stay of order one

Sweep a collision tube

Multiply by particle density

Demand a nontrivial limit

Before taking a limit, one equation never closes

The short-time proof grows a forest

Grow one real Duhamel history

Parent choices and ordered times contain opposite factorials

Lanford’s expansion

Deng–Hani–Ma’s iteration

Keep the signal. Expand the error.

Long-time Boltzmann limit, decoded

The method neither forgets correlation nor re-expands everything

Replace one impossible interval by L small ones

No-recollision histories reproduce the kinetic Duhamel series

A collision history becomes a “molecule”

A second prescribed encounter is a thin target

Cut a complex molecule

Complexity loses logarithms; geometry wins powers.

When does the ε-power actually win?

Base-10 exponent ledger

{2}

{3}

{33}

{4}

From velocity clouds to fields

One bell curve, three fields

Incompressible Navier–Stokes–Fourier

Compressible Euler

What “resolves Hilbert’s sixth problem” means here

Reversible rules, irreversible statistics

Where the irreversible sign appears

One idea migrates across fields

Nonlinear Schrödinger equation

Hard-sphere Newtonian flow

Propagation of randomness

Match each source of smallness to its program

From nonlinear waves to a wave kinetic equation

Random data can make rough equations meaningful

When four Fourier modes resonate

Near-resonances concentrate onto a resonant manifold

A function too rough to multiply by itself

Why H0− converges while L² diverges

From a physical guess to a long-time theorem

A statistical equation is born

Earlier theorems were profound—but lived in different regimes

Why Nε^d−1 must stay of order one

Why H⁰⁻ converges while L² diverges