Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§III.C: induced functions preserve optimum, (63)–(67)

Proved
QAlgorithms.Papers.Independent.Advance50.P38.inducedOrbitOptima

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

prove2quantum

Let a finite-dimensional real Lie group GGG act smoothly from the right on a finite-dimensional smooth manifold MMM. Fix x∈Mx\in Mx∈M and a smooth function F:M→RF:M\to\mathbb RF:M→R. Define f(g)=F(x⋅g)f(g)=F(x\cdot g)f(g)=F(x⋅g) and f^=F∣O(x)\widehat f=F|_{\mathcal O(x)}f​=F∣O(x)​. Then

{f(g):g∈G}={f^(y):y∈O(x)}.\{f(g):g\in G\}=\{\widehat f(y):y\in\mathcal O(x)\}.{f(g):g∈G}={f​(y):y∈O(x)}.

An element ggg attains a global maximum on GGG exactly when x⋅gx\cdot gx⋅g attains one on the orbit. Also f(hg)=f(g)f(hg)=f(g)f(hg)=f(g) whenever x⋅h=xx\cdot h=xx⋅h=x, and equal orbit points give equal objective values. This expresses the stabilizer-coset descent while avoiding an unwarranted assertion that a maximum exists.

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

/-- Induced orbit objective and attained-optimum correspondence (63)--(67),
PDF p.14. Equality of value sets avoids assuming a maximum exists. -/
theorem inducedOrbitOptima
    {E F : Type*} [NormedAddCommGroup E] [NormedSpace ℝ E] [FiniteDimensional ℝ E]
    [NormedAddCommGroup F] [NormedSpace ℝ F] [FiniteDimensional ℝ F]
    {G M : Type*} [Group G] [TopologicalSpace G] [ChartedSpace E G]
    [IsManifold 𝓘(ℝ, E) ∞ G] [LieGroup 𝓘(ℝ, E) ∞ G]
    [T2Space G] [SecondCountableTopology G]
    [TopologicalSpace M] [ChartedSpace F M] [IsManifold 𝓘(ℝ, F) ∞ M]
    [T2Space M] [SecondCountableTopology M]
    (alpha : M → G → M)
    (hsmooth : ContMDiff (𝓘(ℝ, F).prod 𝓘(ℝ, E)) 𝓘(ℝ, F) ∞
      (fun p : M × G => alpha p.1 p.2))
    (hone : ∀ x, alpha x 1 = x)
    (hmul : ∀ x g h, alpha (alpha x g) h = alpha x (g * h))
    (quality : M → ℝ) (hquality : ContMDiff 𝓘(ℝ, F) 𝓘(ℝ, ℝ) ∞ quality) (x : M) :
    let f : G → ℝ := fun g => quality (alpha x g)
    let fhat : Set.range (alpha x) → ℝ := fun y => quality y.val
    Set.range f = Set.range fhat ∧
    (∀ g, (∀ h, f h ≤ f g) ↔ ∀ y : Set.range (alpha x), fhat y ≤ f g) ∧
    (∀ h g, alpha x h = x → f (h * g) = f g) ∧
    (∀ g h, alpha x g = alpha x h → f g = f h) := 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. 14, §III.C: induced functions preserve optimum, (63)–(67)
Read-back

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

Let EEE and FFF be finite-dimensional real normed vector spaces, including possibly zero-dimensional ones. Let GGG be a group equipped with a Hausdorff, second-countable smooth real manifold structure modeled on EEE, and assume that GGG is a Lie group for this structure. Let MMM be a Hausdorff, second-countable smooth real manifold modeled on FFF. Suppose α:M×G→M\alpha:M\times G\to Mα:M×G→M is smooth and satisfies α(p,1)=p\alpha(p,1)=pα(p,1)=p and α(α(p,g),h)=α(p,gh)\alpha(\alpha(p,g),h)=\alpha(p,gh)α(α(p,g),h)=α(p,gh) for every p∈Mp\in Mp∈M and g,h∈Gg,h\in Gg,h∈G. For every smooth function q:M→Rq:M\to\mathbb Rq:M→R and every x∈Mx\in Mx∈M, define the orbit Ox={α(x,g):g∈G}O_x=\{\alpha(x,g):g\in G\}Ox​={α(x,g):g∈G}, the function f:G→Rf:G\to\mathbb Rf:G→R by f(g)=q(α(x,g))f(g)=q(\alpha(x,g))f(g)=q(α(x,g)), and f^:Ox→R\widehat f:O_x\to\mathbb Rf​:Ox​→R by f^(y)=q(y)\widehat f(y)=q(y)f​(y)=q(y). Then all four assertions hold:

{f(g):g∈G}={f^(y):y∈Ox},∀g∈G,(∀h∈G, f(h)≤f(g))  ⟺  (∀y∈Ox, f^(y)≤f(g)),∀h,g∈G,α(x,h)=x  ⟹  f(hg)=f(g),∀g,h∈G,α(x,g)=α(x,h)  ⟹  f(g)=f(h).\begin{aligned} \{f(g):g\in G\}&=\{\widehat f(y):y\in O_x\},\\ \forall g\in G,\qquad \bigl(\forall h\in G,\ f(h)\le f(g)\bigr) &\iff \bigl(\forall y\in O_x,\ \widehat f(y)\le f(g)\bigr),\\ \forall h,g\in G,\qquad \alpha(x,h)=x&\implies f(hg)=f(g),\\ \forall g,h\in G,\qquad \alpha(x,g)=\alpha(x,h)&\implies f(g)=f(h). \end{aligned}{f(g):g∈G}∀g∈G,(∀h∈G, f(h)≤f(g))∀h,g∈G,α(x,h)=x∀g,h∈G,α(x,g)=α(x,h)​={f​(y):y∈Ox​},⟺(∀y∈Ox​, f​(y)≤f(g)),⟹f(hg)=f(g),⟹f(g)=f(h).​

No existence of a maximizing point 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