Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Explicit Hamiltonian symplectic ball packing in R4\mathbb{R}^4R4 (goal)

Open
SymplecticBallPacking.explicit_symplectic_ball_packing

by LG · Nov 17, 2026 · Mathlib 0df444a (Lean v4.33.1)

analysisdifferential-geometryfrontiermathsymplectic-geometry

This is the mathematical core of Epoch AI's FrontierMath open problem Symplectic Ball Packing. Fix n=2n=2n=2. For every integer k≥10k\ge 10k≥10 and every ϵ∈(0,1)\epsilon\in(0,1)ϵ∈(0,1), let Bi⊂R4B_i\subset\mathbb R^4Bi​⊂R4 (i=0,…,k−1i=0,\dots,k-1i=0,…,k−1) be the closed Euclidean unit ball centered at (3i,0,0,0)(3i,0,0,0)(3i,0,0,0). Then there exist a smooth time-dependent Hamiltonian H:R4×[0,1]→RH:\mathbb R^4\times[0,1]\to\mathbb RH:R4×[0,1]→R, its time-111 Hamiltonian flow ϕ:R4→R4\phi:\mathbb R^4\to\mathbb R^4ϕ:R4→R4, and a radius R>0R>0R>0 such that

ϕ(⨆i=0k−1Bi)⊂B4(R),kR4>1−ϵ,\phi\Bigl(\bigsqcup_{i=0}^{k-1}B_i\Bigr)\subset B^{4}(R),\qquad \frac{k}{R^{4}}>1-\epsilon,ϕ(i=0⨆k−1​Bi​)⊂B4(R),R4k​>1−ϵ,

where B4(R)B^4(R)B4(R) is the closed Euclidean ball of radius RRR about the origin. Since the kkk unit balls have total volume k⋅vol(B4(1))k\cdot\mathrm{vol}(B^4(1))k⋅vol(B4(1)) and vol(B4(R))=R4 vol(B4(1))\mathrm{vol}(B^4(R))=R^4\,\mathrm{vol}(B^4(1))vol(B4(R))=R4vol(B4(1)), the ratio k/R4k/R^4k/R4 is exactly the fraction of the target volume occupied by the packed balls.

The existence of such packings is classically known for k≥9k\ge 9k≥9 (Biran), but the known proofs are non-explicit; the open problem asks for a single algorithm producing HHH from (k,ϵ)(k,\epsilon)(k,ϵ). This theorem states the existence content that any solution must certify, and a Lean proof would supply a fully verified construction path.

Formalization Note. Phase space is Phase 2 = Fin 2 → ℂ with zj=qj+ipjz_j=q_j+i p_jzj​=qj​+ipj​. HHH is required smooth on all of C2×R\mathbb C^2\times\mathbb RC2×R, which for this existential claim is equivalent to smoothness on the slab C2×[0,1]\mathbb C^2\times[0,1]C2×[0,1]. The flow is stated on [0,1][0,1][0,1] with one-sided time derivatives at the endpoints, and the Hamiltonian convention is ω(XH,⋅)=dH\omega(X_H,\cdot)=dHω(XH​,⋅)=dH. The benchmark's algorithmic and PyTorch/runtime requirements are meta-level constraints on the solution artifact and are not part of the proposition.

Preamble
import Mathlib
import Definitions.Def_BallPacking
import Definitions.Def_SymplecticBallPacking
open scoped ContDiff
Formal statement
namespace SymplecticBallPacking
open OAI.HigherDimensionalBallPacking
open Set

theorem explicit_symplectic_ball_packing (k : ℕ) (hk : 10 ≤ k) (ε : ℝ)
    (hε0 : 0 < ε) (hε1 : ε < 1) :
    ∃ (H : Phase 2 → ℝ → ℝ) (Φ : ℝ → Phase 2 → Phase 2) (R : ℝ),
      ContDiff ℝ ∞ (Function.uncurry H) ∧
      IsHamiltonianFlow H Φ ∧
      0 < R ∧
      MapsTo (Φ 1) (⋃ i : Fin k, euclClosedBall (inputCenter i) 1)
        (euclClosedBall 0 R) ∧
      (k : ℝ) / R ^ 4 > 1 - ε := by sorry

end SymplecticBallPacking
Source
Epoch AI, FrontierMath: Open Problems — Symplectic Ball Packing, https://epoch.ai/frontiermath/open-problems/symplectic-ball-packing, Prompt section (fix n=2; k≥10; ε∈(0,1); B_i closed unit ball centered (3i−3,0,…,0); φ(⊔B_i)⊂B^{2n}(R); k/R^{2n}>1−ε).
Read-back

What the Lean code literally says, in plain math · qwen3.8-max

For every natural number kkk with 10≤k10\le k10≤k and every real ϵ\epsilonϵ with 0<ϵ<10<\epsilon<10<ϵ<1, there exist a function H:C2→R→RH:\mathbb C^2\to\mathbb R\to\mathbb RH:C2→R→R, a family Φ:R→C2→C2\Phi:\mathbb R\to\mathbb C^2\to\mathbb C^2Φ:R→C2→C2, and a real number RRR such that all of the following hold:

  1. the uncurried function (z,t)↦H(z,t)(z,t)\mapsto H(z,t)(z,t)↦H(z,t) is C∞C^\inftyC∞ on all of C2×R\mathbb C^2\times\mathbb RC2×R;
  2. Φ\PhiΦ is a Hamiltonian flow of HHH on [0,1][0,1][0,1]: Φ0(x)=x\Phi_0(x)=xΦ0​(x)=x for every xxx, and for every t∈[0,1]t\in[0,1]t∈[0,1] and every xxx the curve s↦Φs(x)s\mapsto\Phi_s(x)s↦Φs​(x) has derivative within [0,1][0,1][0,1] at ttt equal to XH(⋅,t)(Φt(x))X_{H(\cdot,t)}(\Phi_t(x))XH(⋅,t)​(Φt​(x)), where XF(z)j=∂F/∂pj−i ∂F/∂qjX_F(z)_j=\partial F/\partial p_j-i\,\partial F/\partial q_jXF​(z)j​=∂F/∂pj​−i∂F/∂qj​ (computed by the total Fréchet derivative, which is 000 where the function is not differentiable);
  3. R>0R>0R>0;
  4. Φ1\Phi_1Φ1​ maps the union ⋃i∈Fin k{z:∑j∣zj−ci,j∣2≤1}\bigcup_{i\in\mathrm{Fin}\,k}\{z:\sum_j|z_j-c_{i,j}|^2\le 1\}⋃i∈Fink​{z:∑j​∣zj​−ci,j​∣2≤1}, where cic_ici​ has first coordinate 3 i.val3\,i.\mathrm{val}3i.val andother coordinates 000, into the closed Euclidean ball {z:∑j∣zj∣2≤R2}\{z:\sum_j|z_j|^2\le R^2\}{z:∑j​∣zj​∣2≤R2};
  5. k/R4>1−ϵk/R^4>1-\epsilonk/R4>1−ϵ (real division; R>0R>0R>0 avoids division by zero).

The balls in (4) are pairwise disjoint (centers 333 apart, radius 111). The hypotheses k≥10k\ge 10k≥10 and ϵ<1\epsilon<1ϵ<1 make the ratio condition non-vacuous. The statement asserts existence only; it does not require Φ1\Phi_1Φ1​ to be a symplectomorphism (that follows from (1)–(2)) nor any algorithmic dependence on (k,ϵ)(k,\epsilon)(k,ϵ).

Human review
  • Endorsed by Shuze Chen · Nov 17, 2026

    Confirmed by the moderator at approval.

  • Endorsed by LG · Nov 17, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me