Richard J. Reyes · public research · Lean 4

Mathematics, from question to proof.

The central directory for mathematical manuscripts, the main Lean 4 proof project, research questions and formal verification. Find a subject, read its results, and follow each theorem to its exact source.

27Curated exact-scope Lean theorem highlights
278.lean files in the public repositories reviewed
5Public repositories containing Lean source

Research snapshots are source-dated. The 27 results are a curated selection, not the total theorem count; the 278 count includes Lean project files. The 300 GC questions are a research agenda, not 300 certified novel open problems.

Start here · Research directory

One mathematics hub. Four ways into the work.

Explore complete analytic manuscripts, kernel-checkable proof sources, mathematical research questions, or the separate WCT object-coverage registry.

Primary formal entry point

Lean mathematics starts at MathProofs.lean.

This file imports the main mathematics proof modules. The formalization manifest tracks manuscript scope, while the WCT Lean coverage map describes a distinct 142-object registry. Neither number should be mistaken for a count of independently resolved GC problems.

Open main Lean import tree ↗

Explore by research subject

Twelve fields of mathematics.

Choose the field first. Each route opens that subject's 25 preserved mathematical questions with the full question text and research reviews visible.

Mathematical manuscripts · source-backed index

Papers, proofs, and analytic manuscripts.

All 28 PDF entries and 21 subject headings from the current math manuscript catalogue. Paper publication, analytical proof and Lean formalization are distinct claims.

Loading the mathematical manuscript catalogue…

Mathematics, explained first

What was proved, and how the results connect.

Begin with the mathematical conclusion and its assumptions. Open a Lean declaration only when you want the exact formal statement. Nothing is omitted from the source.

1. Understand the claimWhat the mathematical result actually establishes.
2. See the mathematicsA readable equation and the hypotheses it depends on.
3. Inspect the formal proofExpand the Lean declaration or open the entire source.
New to this mathematics? Open the notation and terminology guide
H² (Sobolev space)

A function space controlling the field and its weak derivatives up to second order in the square-integrable sense.

Constrained minimizer

A field attaining the lowest energy among fields with a prescribed mass or norm.

Orbital stability

Remaining close to the set of equivalent minimizing states, allowing the symmetries specified in the theorem.

Lean 4 proof

A machine-checked derivation of exactly the formal statement under its visible definitions and hypotheses.

GC question

A broader research question in the Reyes Problem Repository. A related Lean theorem need not resolve the entire question.

A connected example: fourth-order nonlinear waves

From an energy inequality to stability.

For one two-dimensional cubic–quintic model, four separate theorems establish a connected path. The parameters and function spaces matter: this is not a statement about every nonlinear wave equation.

01 · Energy bindingmath / StrictBinding.lean

Why the minimizing wave cannot gain by splitting.

Splitting a fixed mass into two positive portions always raises the sum of the minimum energies, within the proved parameter range.

\[I(m)<I(\mu)+I(m-\mu)\]

Why this matters

This strict energy gap is the binding step used when ruling out fragmentation of a minimizing sequence.

Exact assumptions

In the two-dimensional cubic–quintic setting: β, γ > 0; 0 < m ≤ mB; and every 0 < μ < m.

The key step of the proof

Apply the strict gap estimate to both portions of mass. Compare the energies per unit mass, then add them using μ + (m − μ) = m.

Scope: Two-dimensional cubic–quintic model with β, γ > 0 and 0 < m ≤ mB, not an all-mass theorem.

Read the exact Lean declaration and a key proof line

lean/MathProofs/StrictBinding.lean:531 · theorem strict_binding

theorem strict_binding (hβ : 0 < β) (hγ : 0 < γ) {m μ : ℝ} (hm : 0 < m) (hmB : m ≤ mB β γ k)
    (hμ : 0 < μ) (hμm : μ < m) :
    Imin β γ k m < Imin β γ k μ + Imin β γ k (m - μ) :=

Representative proof step from source

have h1 := gap (k := k) hβ hγ hμ hμm (by linarith)

The first code block is the theorem statement with its hypotheses. The full source contains its Lean proof body and dependencies.

02 · Existencemath / WholeSpaceMinimizers.lean

Does a lowest-energy wave actually exist?

The lowest constrained energy is achieved by at least one actual H² field, not merely approached by a sequence.

\[\exists U:\quad M(U)=m,\quad E(U)=I(m)\]

Why this matters

Attainment establishes a nonempty minimizing set on which the later orbital-stability statement is formulated.

Exact assumptions

Two-dimensional model; β, γ > 0; 0 < m ≤ mB. The result does not select a unique profile.

The key step of the proof

Construct a sequence of fields approaching the infimum. Compactness supplies a limit preserving both the mass and the minimum energy.

Scope: Whole-space minimizer set with compactness modulo translations, not profile uniqueness.

Read the exact Lean declaration and a key proof line

lean/MathProofs/WholeSpaceMinimizers.lean:1492 · theorem exists_minimizer

theorem exists_minimizer (hβ : 0 < β) (hγ : 0 < γ) {m : ℝ} (hm : 0 < m)
    (hmB : m ≤ StrictBinding.mB β γ k) :
    ∃ U : H2 2, M U = m ∧ energy β γ k U = Imin β γ k m :=

Representative proof step from source

obtain ⟨-, -, -, U, hUm, hUE, -⟩ := compactness hβ hγ hm hmB hum hE

The first code block is the theorem statement with its hypotheses. The full source contains its Lean proof body and dependencies.

03 · Evolutionmath / CubicQuinticFlow.lean

Can that wave evolve for all time?

Every allowed H² initial state has exactly one solution defined for all positive and negative times.

\[\forall u_0\in H^2(\mathbb R^2),\quad \exists!\,u:\mathbb R\to H^2(\mathbb R^2)\]

Why this matters

Global existence and uniqueness supply the evolution needed by the subsequent stability theorem.

Exact assumptions

The specific two-dimensional fourth-order cubic–quintic evolution with γ > 0 and initial data in H².

The key step of the proof

Solve the global evolution in a transformed Hilbert-space representation, map it back to H², then use uniqueness of the transformed ODE.

Scope: Model-specific flow, not a general fourth-order PDE existence theorem.

Read the exact Lean declaration and a key proof line

lean/MathProofs/CubicQuinticFlow.lean:926 · theorem global_wellposed

theorem global_wellposed (hγ : 0 < γ) (u₀ : H2 2) :
    ∃! u : ℝ → H2 2, IsSolution β γ k u ∧ u 0 = u₀ :=

Representative proof step from source

obtain ⟨W, hW0, hW⟩ := exists_global (β := β) (γ := γ) (k := k) hγ (H2.toSp u₀)

The first code block is the theorem statement with its hypotheses. The full source contains its Lean proof body and dependencies.

04 · Stabilitymath / OrbitalStability.lean

What happens under a small perturbation?

Starting sufficiently close to the set of energy minimizers keeps the solution close to that entire set for all time.

\[\operatorname{dist}_{H^2}(u(0),\mathcal G_m)<\delta\ \Longrightarrow\ \sup_t\operatorname{dist}_{H^2}(u(t),\mathcal G_m)<\varepsilon\]

Why this matters

The theorem turns variational information into a dynamical stability statement, without choosing a unique wave profile.

Exact assumptions

The specified 2D equation, β, γ > 0, 0 < m ≤ mB, with distance measured in H².

The key step of the proof

A variational stability estimate supplies the tolerance δ. Conservation of mass and energy then controls distance from the entire minimizing set for all times.

Scope: Set stability in the sufficient mass range, not uniqueness or stability of every excited state.

Read the exact Lean declaration and a key proof line

lean/MathProofs/OrbitalStability.lean:256 · theorem orbital_stability

theorem orbital_stability (hβ : 0 < β) (hγ : 0 < γ) {m : ℝ} (hm : 0 < m)
    (hmB : m ≤ StrictBinding.mB β γ k) {ε : ℝ} (hε : 0 < ε) :
    ∃ δ > 0, ∀ u : ℝ → H2 2, IsSolution β γ k u → distX (u 0) (Gset β γ k m) < δ →
      BddAbove (Set.range fun t => distX (u t) (Gset β γ k m)) ∧
      ⨆ t, distX (u t) (Gset β γ k m) < ε :=

Representative proof step from source

obtain ⟨δ, hδ, hcore⟩ := stable_core hβ hγ hm hmB (half_pos hε)

The first code block is the theorem statement with its hypotheses. The full source contains its Lean proof body and dependencies.

05 · Inverse problemsmath / V18Identifiability.lean

How much can measurements fail to identify?

Nearby, genuinely different parameter choices can give the same eleven observations.

\[\dim(\text{local observation fiber})=20\]

Why this matters

The local twenty-dimensional fiber shows structural ambiguity, beyond numerical ill-conditioning.

Exact assumptions

The frozen 31-parameter / 11-observation nonlinear map, with α ≥ 0.

The key step of the proof

The frozen Jacobian has an eleven-dimensional image and twenty-dimensional kernel. The local level-set theorem turns that kernel into a nonlinear family of indistinguishable parameters.

Scope: Local to the frozen model, not a global identification classification.

Read the exact Lean declaration and a key proof line

lean/MathProofs/V18Identifiability.lean:32 · theorem observed_local_level_set

theorem observed_local_level_set (alpha : ℝ) (ha : 0 ≤ alpha) :
    HasTwentyDimensionalLocalFiber alpha :=

Representative proof step from source

exists_c1_level_set_chart observed (frozen alpha 5) (jacobian (frozen alpha 5))

The first code block is the theorem statement with its hypotheses. The full source contains its Lean proof body and dependencies.

06 · Geometric obstructionmath / ManifoldScalarMetric.lean

Can one complex scalar define a four-dimensional metric?

A complex scalar's real-gradient Gram tensor has rank at most two on a four-dimensional tangent space.

\[\operatorname{rank}(g_\psi)\leq2,\qquad\dim\ker(g_\psi)\geq2\]

Why this matters

This particular construction cannot define a nondegenerate Lorentzian metric, without ruling out other metric constructions.

Exact assumptions

A C¹ complex scalar on the specified four-dimensional smooth manifold.

The key step of the proof

The theorem combines Gram symmetry and nonnegativity with a rank-at-most-two bound. In dimension four that leaves a nontrivial kernel, so this tensor is degenerate.

Scope: Applies to this specific tensor construction, not all scalar-derived effective metrics.

Read the exact Lean declaration and a key proof line

lean/MathProofs/ManifoldScalarMetric.lean:247 · theorem main

theorem main (ψ : M → ℂ) (_hψ : ContMDiff I 𝓘(ℝ, ℂ) 1 ψ)
    (h4 : Module.finrank ℝ E = 4) (p : M) :
    (tensor I ψ p).IsSymm ∧
    (∀ v, 0 ≤ tensor I ψ p v v) ∧
    LinearMap.ker (tensor I ψ p) =
      LinearMap.ker (mfderiv I 𝓘(ℝ, ℂ) ψ p).toLinearMap ∧
    Module.finrank ℝ (LinearMap.range (tensor I ψ p)) ≤ 2 ∧
    2 ≤ Module.finrank ℝ (LinearMap.ker (tensor I ψ p)) ∧
    ¬ (tensor I ψ p).Nondegenerate :=

Representative proof step from source

exact ⟨gram_symmetric _, gram_nonnegative _, gram_ker _, gram_rank_le_two _,
    gram_nullity _ h4, gram_degenerate _ h4⟩

The first code block is the theorem statement with its hypotheses. The full source contains its Lean proof body and dependencies.

Cross-repository proof explorer

Twenty-seven results, explained.

Every result has a readable explanation, its mathematical form, assumptions and limitations. The exact Lean theorem remains one click away, with a direct link to the full proof source.

Loading proof declarations… · ·

Loading the public proof inventory. Read the source index on GitHub ↗

The displayed Lean signatures state the encoded theorem, not the proof body. A formal proof proves only the declaration under its hypotheses. The public 20-module selected index also references three currently private-repository files; these are deliberately excluded from this public-source showcase. Individual modules may overlap in mathematical subject.

Public source inventory · checked November 2, 2026

Follow the .lean files.

All 30 public repositories on the owner account were checked. These five contain Lean source. Counts include entry-point and project files, not only theorem declarations.

Loading public repository inventory…

Complete public code index

Every public .lean file, searchable.

Source paths from the 30-repository public audit. A file appearing here does not by itself certify a theorem.

Loading 278 source files…

The repository links are primary sources. The highlighted 27 declarations are a guided reading path through these files, not an exhaustive verification audit. The source-code and manuscript evidence standards remain separate.

The research continues

There are 300 questions to work on.

The Reyes Problem Repository is the separate call for rigorous mathematics: choose a model, evaluate prior art, prove a statement or find a counterexample, and contribute the exact scope.

Open the problem roadmap