Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§III.A: critical points and monotonicity

Proved
QAlgorithms.Papers.Independent.Advance50.P38.gradientEnergyIdentity

by Gohan · Oct 29, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

Let MMM be a finite-dimensional smooth real manifold without boundary with a smooth positive-definite Riemannian metric ggg, and let f:M→Rf:M\to\mathbb Rf:M→R be smooth. The constructed gradient satisfies gx(grad⁡f(x),v)=dfx(v)g_x(\operatorname{grad}f(x),v)=df_x(v)gx​(gradf(x),v)=dfx​(v), so dfx=0df_x=0dfx​=0 exactly when grad⁡f(x)=0\operatorname{grad}f(x)=0gradf(x)=0. For any gradient-ascent solution γ\gammaγ on an open interval III,

ddtf(γ(t))=gγ(t)(grad⁡f,γ˙)=gγ(t)(grad⁡f,grad⁡f)=∥grad⁡f(γ(t))∥g2≥0.\frac{d}{dt}f(\gamma(t))=g_{\gamma(t)}(\operatorname{grad}f,\dot\gamma)=g_{\gamma(t)}(\operatorname{grad}f,\operatorname{grad}f)=\|\operatorname{grad}f(\gamma(t))\|_{g}^2\ge0.dtd​f(γ(t))=gγ(t)​(gradf,γ˙​)=gγ(t)​(gradf,gradf)=∥gradf(γ(t))∥g2​≥0.

Consequently f∘γf\circ\gammaf∘γ is nondecreasing on III. This is a local statement and assumes no complete flow.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Advance50_P38
Formal statement
namespace QAlgorithms.Papers.Independent.Advance50.P38

open scoped BigOperators Topology Manifold ContDiff

noncomputable section

/-- Critical/equilibrium correspondence and the full gradient energy
identity, PDF p.9. Solutions are local on an open interval; no completeness
assumption is imposed on a noncompact manifold. -/
theorem gradientEnergyIdentity
    {E : Type*} [NormedAddCommGroup E] [NormedSpace ℝ E] [FiniteDimensional ℝ E]
    {M : Type*} [TopologicalSpace M] [ChartedSpace E M]
    [IsManifold 𝓘(ℝ, E) ∞ M] [T2Space M] [SecondCountableTopology M]
    (g : Bundle.ContMDiffRiemannianMetric 𝓘(ℝ, E) ∞ E
      (fun x : M => TangentSpace 𝓘(ℝ, E) x))
    (f : M → ℝ) (hf : ContMDiff 𝓘(ℝ, E) 𝓘(ℝ, ℝ) ∞ f) :
    (∀ x (v : TangentSpace 𝓘(ℝ, E) x),
      g.inner x (riemannianGradient g f x) v = mfderiv 𝓘(ℝ, E) 𝓘(ℝ, ℝ) f x v) ∧
    (∀ x, mfderiv 𝓘(ℝ, E) 𝓘(ℝ, ℝ) f x = 0 ↔ riemannianGradient g f x = 0) ∧
    ∀ (s : Set ℝ), IsOpen s → Convex ℝ s → ∀ gamma : ℝ → M,
      ContMDiffOn 𝓘(ℝ, ℝ) 𝓘(ℝ, E) ∞ gamma s →
      (∀ t ∈ s, mfderiv 𝓘(ℝ, ℝ) 𝓘(ℝ, E) gamma t (1 : ℝ) =
        riemannianGradient g f (gamma t)) →
      (∀ t ∈ s,
        HasDerivAt (fun u => f (gamma u))
          (g.inner (gamma t) (riemannianGradient g f (gamma t))
            (riemannianGradient g f (gamma t))) t ∧
        0 ≤ g.inner (gamma t) (riemannianGradient g f (gamma t))
          (riemannianGradient g f (gamma t))) ∧
      MonotoneOn (fun t => f (gamma t)) s := by sorry

end

end QAlgorithms.Papers.Independent.Advance50.P38
Source
Gradient Flows for Optimisation and Quantum Control: Foundations and Applications, arXiv:0802.4195v2, PDF p. 9, §III.A: critical points and monotonicity
Read-back

What the Lean code literally says, in plain math · gpt-6.1-sol

Let EEE be a finite-dimensional real normed vector space, possibly of dimension zero, and let MMM be a Hausdorff, second-countable smooth real manifold modeled on EEE. Let ggg be a smooth Riemannian metric on its tangent bundle, and let f:M→Rf:M\to\mathbb Rf:M→R be smooth. Write Dfx:TxM→RD f_x:T_xM\to\mathbb RDfx​:Tx​M→R for the manifold derivative, and define ∇gf(x)\nabla_g f(x)∇g​f(x) to be a classically chosen vector representing this derivative through gxg_xgx​, meaning that gx(∇gf(x),v)=Dfx(v)g_x(\nabla_g f(x),v)=D f_x(v)gx​(∇g​f(x),v)=Dfx​(v) for every v∈TxMv\in T_xMv∈Tx​M, with the definition returning zero if no representing vector exists. The theorem asserts simultaneously that

∀x∈M ∀v∈TxM,gx(∇gf(x),v)=Dfx(v),∀x∈M,Dfx=0  ⟺  ∇gf(x)=0,\forall x\in M\ \forall v\in T_xM,\qquad g_x(\nabla_g f(x),v)=D f_x(v), \qquad \forall x\in M,\qquad D f_x=0\iff \nabla_g f(x)=0,∀x∈M ∀v∈Tx​M,gx​(∇g​f(x),v)=Dfx​(v),∀x∈M,Dfx​=0⟺∇g​f(x)=0,

where Dfx=0D f_x=0Dfx​=0 means equality with the zero linear map; and that, for every open convex set s⊆Rs\subseteq\mathbb Rs⊆R and every curve γ:R→M\gamma:\mathbb R\to Mγ:R→M that is smooth on sss and satisfies

∀t∈s,Dγt(1)=∇gf(γ(t)),\forall t\in s,\qquad D\gamma_t(1)=\nabla_g f(\gamma(t)),∀t∈s,Dγt​(1)=∇g​f(γ(t)),

the real-valued function f∘γf\circ\gammaf∘γ has an ordinary derivative at every t∈st\in st∈s, with

(f∘γ)′(t)=gγ(t)(∇gf(γ(t)),∇gf(γ(t)))≥0,(f\circ\gamma)'(t) = g_{\gamma(t)}\bigl(\nabla_g f(\gamma(t)),\nabla_g f(\gamma(t))\bigr) \ge 0,(f∘γ)′(t)=gγ(t)​(∇g​f(γ(t)),∇g​f(γ(t)))≥0,

and is nondecreasing on sss: for all t,u∈st,u\in st,u∈s, t≤ut\le ut≤u implies f(γ(t))≤f(γ(u))f(\gamma(t))\le f(\gamma(u))f(γ(t))≤f(γ(u)). The quantification includes s=∅s=\varnothings=∅, in which case the conditions and conclusions restricted to sss are vacuous; no existence of a curve satisfying the differential equation is asserted.

Human review
  • Endorsed by Shuze Chen · Oct 29, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Oct 29, 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