Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Definitions for Integrability of unitary representations on reproducing kernel spaces

Definition
QAlgorithms_Papers_Independent_Intakenext100_P21

by Gohan · Oct 31, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

The objects the formalized statements of Integrability of unitary representations on reproducing kernel spaces (arXiv:1406.2681) are stated in. Each was checked against the paper by a blind read-back and a moderator.

  • HasTimeDerivative: Section 2, PDF p. 6: the ordinary derivative of a locally convex valued curve. This definition does not require a Banach target.
  • CoordinateField: Sections 2--3, PDF pp. 5, 10: a tangent-section coordinate family. Only components for c in the atlas and m in c.source are used. The predicate IsSmoothVectorField imposes the tangent transition law, so this type alone does not purport to be the space of smooth vector fields.
  • IsSmoothVectorField: Section 2, PDF p. 5: smooth tangent-section coordinates. The second clause transforms each component by the derivative of the actual chart transition; no independent tangent bundle or arbitrary derivative is supplied.
  • IsIntegralCurve: Definition 2.3(b), PDF p. 6: an integral curve on an open interval containing zero. Continuity and the chartwise differential equation express differentiability into M, with no normability restriction.
  • FlowData: Definition 2.1, PDF p. 6: raw domain and a total representative of the flow. Off-domain values are never used in any identity or uniqueness claim.
  • IsSmoothFlowMap: Definition 2.1, PDF p. 6: joint smoothness in time and space in the product atlas, on the specified open flow domain.
  • IsLocalFlow: Definition 2.1, PDF p. 6: open domain containing zero times every point, interval time fibers, and the local addition law precisely wherever both sides are defined. No maximality or uniqueness is assumed.
  • HasVelocityField: Definition 2.3(a), PDF p. 6: X is the velocity at zero of this flow; the integral-curve property at other times is a theorem, not a hypothesis.
  • IsLocallyIntegrable: Definition 2.3(a), PDF p. 6: local integrability requires a smooth vector field and one actual jointly smooth local flow with this velocity.
  • CurveExtends: Definition 2.3(c), PDF p. 6: extension includes the original domain and agrees there; total representatives outside it are immaterial.
  • IsMaximalIntegralCurve: Definition 2.3(c), PDF p. 6: no integral curve on a strictly larger open interval extends the given integral curve.
  • maximalCurveDomain: Lemma 2.5(iv), PDF p. 8: union of the actual integral-curve intervals through the given initial point, not an independently chosen domain.
  • maximalCurve: Lemma 2.5(iv), PDF p. 8: glue by selecting an attained integral-curve value. Lemma 2.5 proves agreement of all choices. The default m is used only outside maximalCurveDomain, and is never part of an asserted curve.
  • maximalFlow: Theorem 2.6, PDF p. 8: the flow domain and values constructed from all integral curves. This construction assumes no smoothness of the glued map.
  • bracketCoordinate: Section 3, PDF p. 10: coordinate formula for the actual Lie bracket of two vector fields, dY(X)-dX(Y), computed in a chart.
  • pullbackCoordinate: Section 3, PDF pp. 10--11: coordinates at m of (Phi[-t])_*Y, using a chart centered at Phit. Its genuine domain is M_t; nearby times around zero suffice for the Lie derivative.
  • linearField: Proposition 2.9, PDF p. 10: X(v)=Dv in the standard one-chart atlas of the locally convex vector space itself.
  • adjointField: Definition 3.3, PDF p. 11: the linear adjoint vector field at y is [x,y], in the standard atlas of the vector space A.
  • IsAdIntegrable: Definition 3.3, PDF p. 11: every adjoint field generates a global flow. Local integrability and the entire maximal domain express the source's global-flow definition without an operator exponential on a nonnormable space.
  • adjointEvolution: Definition 3.3, PDF p. 11: e^(t ad x)y is the actual global linear flow, not a formal power series or a bounded-operator exponential.
  • IsSmoothInfinitesimalAction: Lemma 3.4, PDF p. 11: the coordinate form of a smooth Lie algebra homomorphism into vector fields. Joint evaluation is smooth into the actual tangent coordinates, and bracket preservation uses dY(X)-dX(Y).
Definition code
import Mathlib
import Definitions.Def_QAlgorithms_Papers_Independent_Claimed50_P22
import Definitions.Def_QAlgorithms_Papers_Independent_Next46_P15

/-!
Statements from Merigon--Neeb--Olafsson, arXiv:1406.2681v3, Sections 2--3.
The calculus below is jointly continuous iterated directional calculus over
Hausdorff locally convex models. The derivative tower and atlas predicate are imported from the pinned
accepted Claimed50.P22; directional limits and their selection are imported
from Next46.P15. No normability is imposed.
Coordinate representatives are compared only in actual chart domains. Flow and
curve representatives outside their domains have no mathematical meaning.
Analytic and representation-theoretic nodes deferred in this bounded stage are
listed individually in HARD.md and GAPS.json. Compilation is pending handoff.
-/

namespace QAlgorithms.Papers.Independent.Intakenext100.P21

open Set Filter Topology
open QAlgorithms.Papers.Independent.Claimed50.P22 (BastianiSmoothOn BastianiAtlas)
open QAlgorithms.Papers.Independent.Next46.P15 (HasDirectionalDerivative directionalDerivative)

attribute [local instance] Classical.propDecidable

universe u v w

section Calculus

variable {E : Type u} {F : Type v}
variable [TopologicalSpace E] [AddCommGroup E] [Module ℝ E]
variable [TopologicalSpace F] [AddCommGroup F] [Module ℝ F]

/-- The constant zero map has the zero tower, also in order zero. -/
example (U : Set E) : BastianiSmoothOn U (fun _ : E => (0 : F)) := by
  refine ⟨fun _ _ _ => 0, ?_, ?_, ?_⟩
  · intro x hx
    rfl
  · intro n
    exact continuousOn_const
  · intro n x hx a h
    simpa only [sub_self, smul_zero] using
      (tendsto_const_nhds : Tendsto (fun _ : ℝ => (0 : F))
        (𝓝[≠] (0 : ℝ)) (𝓝 (0 : F)))

/-- Section 2, PDF p. 6: the ordinary derivative of a locally convex valued
curve. This definition does not require a Banach target. -/
def HasTimeDerivative (f : ℝ → F) (t : ℝ) (y : F) : Prop :=
  Tendsto (fun s : ℝ => s⁻¹ • (f (t + s) - f t))
    (𝓝[≠] (0 : ℝ)) (𝓝 y)

/-- A constant curve has zero velocity. -/
example (v : F) (t : ℝ) : HasTimeDerivative (fun _ : ℝ => v) t 0 := by
  simpa only [HasTimeDerivative, sub_self, smul_zero] using
    (tendsto_const_nhds : Tendsto (fun _ : ℝ => (0 : F))
      (𝓝[≠] (0 : ℝ)) (𝓝 (0 : F)))

end Calculus

section Manifolds

variable {E : Type u} [TopologicalSpace E] [AddCommGroup E] [Module ℝ E]
variable {M : Type v} [TopologicalSpace M] [ChartedSpace E M]

/-- Sections 2--3, PDF pp. 5, 10: a tangent-section coordinate family.
Only components for c in the atlas and m in c.source are used. The predicate
IsSmoothVectorField imposes the tangent transition law, so this type alone
does not purport to be the space of smooth vector fields. -/
abbrev CoordinateField (E : Type u) (M : Type v)
    [TopologicalSpace E] [TopologicalSpace M] :=
  OpenPartialHomeomorph M E → M → E

/-- Section 2, PDF p. 5: smooth tangent-section coordinates. The second
clause transforms each component by the derivative of the actual chart
transition; no independent tangent bundle or arbitrary derivative is supplied. -/
def IsSmoothVectorField (X : CoordinateField E M) : Prop :=
  (∀ c ∈ atlas E M,
    BastianiSmoothOn c.target (fun z => X c (c.symm z))) ∧
  (∀ c ∈ atlas E M, ∀ d ∈ atlas E M, ∀ m : M,
    m ∈ c.source → m ∈ d.source →
      HasDirectionalDerivative (fun z => d (c.symm z)) (c m) (X c m) (X d m))

/-- Definition 2.3(b), PDF p. 6: an integral curve on an open interval
containing zero. Continuity and the chartwise differential equation express
differentiability into M, with no normability restriction. -/
def IsIntegralCurve (X : CoordinateField E M) (I : Set ℝ) (gamma : ℝ → M) : Prop :=
  IsOpen I ∧ OrdConnected I ∧ (0 : ℝ) ∈ I ∧ ContinuousOn gamma I ∧
    ∀ t ∈ I, ∀ c ∈ atlas E M, gamma t ∈ c.source →
      HasTimeDerivative (fun s => c (gamma s)) t (X c (gamma t))

/-- A stationary point gives a constant integral curve on every open
interval containing zero. In particular this applies to the zero field. -/
example (X : CoordinateField E M) (m : M) (I : Set ℝ)
    (hOpen : IsOpen I) (hInterval : OrdConnected I) (hZero : (0 : ℝ) ∈ I)
    (hStationary : ∀ c ∈ atlas E M, m ∈ c.source → X c m = 0) :
    IsIntegralCurve X I (fun _ => m) := by
  refine ⟨hOpen, hInterval, hZero, continuousOn_const, ?_⟩
  intro t ht c hc hm
  rw [hStationary c hc hm]
  simpa only [HasTimeDerivative, sub_self, smul_zero] using
    (tendsto_const_nhds : Tendsto (fun _ : ℝ => (0 : E))
      (𝓝[≠] (0 : ℝ)) (𝓝 (0 : E)))

/-- Definition 2.1, PDF p. 6: raw domain and a total representative of the
flow. Off-domain values are never used in any identity or uniqueness claim. -/
structure FlowData (M : Type v) where
  domain : Set (ℝ × M)
  value : ℝ → M → M

/-- Definition 2.1, PDF p. 6: joint smoothness in time and space in the
product atlas, on the specified open flow domain. -/
def IsSmoothFlowMap (P : FlowData M) : Prop :=
  ContinuousOn (fun p : ℝ × M => P.value p.1 p.2) P.domain ∧
    ∀ c ∈ atlas E M, ∀ d ∈ atlas E M,
      BastianiSmoothOn
        {p : ℝ × E | p.2 ∈ c.target ∧
          (p.1, c.symm p.2) ∈ P.domain ∧
          P.value p.1 (c.symm p.2) ∈ d.source}
        (fun p : ℝ × E => d (P.value p.1 (c.symm p.2)))

/-- Definition 2.1, PDF p. 6: open domain containing zero times every
point, interval time fibers, and the local addition law precisely wherever
both sides are defined. No maximality or uniqueness is assumed. -/
def IsLocalFlow (P : FlowData M) : Prop :=
  IsOpen P.domain ∧ (∀ m : M, (0, m) ∈ P.domain) ∧
    (∀ m : M, OrdConnected {t : ℝ | (t, m) ∈ P.domain}) ∧
    IsSmoothFlowMap (E := E) P ∧
    (∀ m : M, P.value 0 m = m) ∧
    (∀ s t : ℝ, ∀ m : M,
      (s, m) ∈ P.domain → (t, P.value s m) ∈ P.domain →
      (s + t, m) ∈ P.domain →
      P.value t (P.value s m) = P.value (s + t) m)

/-- Definition 2.3(a), PDF p. 6: X is the velocity at zero of this flow;
the integral-curve property at other times is a theorem, not a hypothesis. -/
def HasVelocityField (P : FlowData M) (X : CoordinateField E M) : Prop :=
  ∀ m : M, ∀ c ∈ atlas E M, m ∈ c.source →
    HasTimeDerivative (fun t => c (P.value t m)) 0 (X c m)

/-- Definition 2.3(a), PDF p. 6: local integrability requires a smooth
vector field and one actual jointly smooth local flow with this velocity. -/
def IsLocallyIntegrable (X : CoordinateField E M) : Prop :=
  IsSmoothVectorField X ∧
    ∃ P : FlowData M, IsLocalFlow (E := E) P ∧ HasVelocityField P X

/-- Definition 2.3(c), PDF p. 6: extension includes the original domain
and agrees there; total representatives outside it are immaterial. -/
def CurveExtends (I : Set ℝ) (gamma : ℝ → M) (J : Set ℝ) (eta : ℝ → M) : Prop :=
  I ⊆ J ∧ ∀ t ∈ I, eta t = gamma t

/-- Definition 2.3(c), PDF p. 6: no integral curve on a strictly larger
open interval extends the given integral curve. -/
def IsMaximalIntegralCurve (X : CoordinateField E M) (I : Set ℝ)
    (gamma : ℝ → M) : Prop :=
  IsIntegralCurve X I gamma ∧
    ∀ J eta, IsIntegralCurve X J eta → CurveExtends I gamma J eta → J ⊆ I

/-- Lemma 2.5(iv), PDF p. 8: union of the actual integral-curve intervals
through the given initial point, not an independently chosen domain. -/
def maximalCurveDomain (X : CoordinateField E M) (m : M) : Set ℝ :=
  {t | ∃ (I : Set ℝ) (gamma : ℝ → M),
    IsIntegralCurve X I gamma ∧ gamma 0 = m ∧ t ∈ I}

/-- Lemma 2.5(iv), PDF p. 8: glue by selecting an attained integral-curve
value. Lemma 2.5 proves agreement of all choices. The default m is used only
outside maximalCurveDomain, and is never part of an asserted curve. -/
noncomputable def maximalCurve (X : CoordinateField E M) (m : M) (t : ℝ) : M :=
  if h : ∃ y : M, ∃ (I : Set ℝ) (gamma : ℝ → M),
      IsIntegralCurve X I gamma ∧ gamma 0 = m ∧ t ∈ I ∧ gamma t = y
  then Classical.choose h else m

/-- Theorem 2.6, PDF p. 8: the flow domain and values constructed from
all integral curves. This construction assumes no smoothness of the glued map. -/
noncomputable def maximalFlow (X : CoordinateField E M) : FlowData M where
  domain := {p | p.1 ∈ maximalCurveDomain X p.2}
  value := fun t m => maximalCurve X m t

/-- The constructed flow has precisely the union-of-curves time fiber. -/
example (X : CoordinateField E M) (t : ℝ) (m : M) :
    (t, m) ∈ (maximalFlow X).domain ↔ t ∈ maximalCurveDomain X m := Iff.rfl

variable [IsTopologicalAddGroup E] [ContinuousSMul ℝ E]
variable [LocallyConvexSpace ℝ E] [T2Space E] [T2Space M]

/-- Section 3, PDF p. 10: coordinate formula for the actual Lie bracket
of two vector fields, dY(X)-dX(Y), computed in a chart. -/
noncomputable def bracketCoordinate (X Y : CoordinateField E M)
    (c : OpenPartialHomeomorph M E) (m : M) : E :=
  directionalDerivative (fun z => Y c (c.symm z)) (c m) (X c m) -
    directionalDerivative (fun z => X c (c.symm z)) (c m) (Y c m)

/-- Section 3, PDF pp. 10--11: coordinates at m of (Phi[-t])_*Y,
using a chart centered at Phi[t](m). Its genuine domain is M_t; nearby
times around zero suffice for the Lie derivative. -/
noncomputable def pullbackCoordinate (P : FlowData M) (Y : CoordinateField E M)
    (t : ℝ) (c : OpenPartialHomeomorph M E) (m : M) : E :=
  let d := chartAt E (P.value t m)
  directionalDerivative (fun z => c (P.value (-t) (d.symm z)))
    (d (P.value t m)) (Y d (P.value t m))

end Manifolds

section LinearFields

variable {V : Type u} [TopologicalSpace V] [AddCommGroup V] [Module ℝ V]

/-- Proposition 2.9, PDF p. 10: X(v)=Dv in the standard one-chart atlas
of the locally convex vector space itself. -/
def linearField (D : V →L[ℝ] V) : CoordinateField V V := fun _ v => D v

/-- The zero continuous linear map gives the zero field. -/
example (c : OpenPartialHomeomorph V V) (v : V) :
    linearField (0 : V →L[ℝ] V) c v = 0 := rfl

end LinearFields

section AdjointFlows

variable {A : Type u} [TopologicalSpace A] [LieRing A] [Module ℝ A]
variable [LieAlgebra ℝ A]

/-- Definition 3.3, PDF p. 11: the linear adjoint vector field
at y is [x,y], in the standard atlas of the vector space A. -/
def adjointField (x : A) : CoordinateField A A := fun _ y => ⁅x, y⁆

/-- Definition 3.3, PDF p. 11: every adjoint field generates a global
flow. Local integrability and the entire maximal domain express the source's
global-flow definition without an operator exponential on a nonnormable space. -/
def IsAdIntegrable : Prop :=
  ∀ x : A, IsLocallyIntegrable (adjointField x) ∧
    (maximalFlow (adjointField x)).domain = Set.univ

/-- Definition 3.3, PDF p. 11: e^(t ad x)y is the actual global linear
flow, not a formal power series or a bounded-operator exponential. -/
noncomputable def adjointEvolution (x y : A) (t : ℝ) : A :=
  (maximalFlow (adjointField x)).value t y

/-- The adjoint field at zero vanishes, including for the zero Lie algebra. -/
example (c : OpenPartialHomeomorph A A) (y : A) : adjointField (0 : A) c y = 0 := by
  exact zero_lie _

variable {E : Type v} [TopologicalSpace E] [AddCommGroup E] [Module ℝ E]
variable {M : Type w} [TopologicalSpace M] [ChartedSpace E M]

/-- Lemma 3.4, PDF p. 11: the coordinate form of a smooth Lie algebra
homomorphism into vector fields. Joint evaluation is smooth into the actual
tangent coordinates, and bracket preservation uses dY(X)-dX(Y). -/
def IsSmoothInfinitesimalAction
    (beta : A →ₗ[ℝ] CoordinateField E M) : Prop :=
  (∀ x : A, IsSmoothVectorField (beta x)) ∧
    (∀ x y : A, ∀ c ∈ atlas E M, ∀ m ∈ c.source,
      beta ⁅x, y⁆ c m = bracketCoordinate (beta x) (beta y) c m) ∧
    (∀ c ∈ atlas E M,
      BastianiSmoothOn (Set.univ ×ˢ c.target)
        (fun p : A × E => beta p.1 c (c.symm p.2)))

end AdjointFlows

end QAlgorithms.Papers.Independent.Intakenext100.P21
Source
Integrability of unitary representations on reproducing kernel spaces, arXiv:1406.2681v3, https://arxiv.org/abs/1406.2681v3
Read-back

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

QAlgorithms.Papers.Independent.Intakenext100.P21.CoordinateField

For topological spaces EEE and MMM, a coordinate field is an arbitrary function assigning an element X(c,m)∈EX(c,m)\in EX(c,m)∈E to every open partial homeomorphism c:M⇢Ec:M\dashrightarrow Ec:M⇢E and every point m∈Mm\in Mm∈M. An open partial homeomorphism supplies mutually inverse homeomorphisms between specified open source and target sets, represented by total functions on the ambient spaces. The assignment is defined for every such ccc, including those outside any chosen atlas, and for every mmm, including points outside the source of ccc. This abbreviation imposes no continuity, smoothness, or compatibility conditions on the assignment.

QAlgorithms.Papers.Independent.Intakenext100.P21.CurveExtends

For any type MMM, sets I,J⊆RI,J\subseteq\mathbb RI,J⊆R, and total functions γ,η:R→M\gamma,\eta:\mathbb R\to Mγ,η:R→M, this predicate means that I⊆JI\subseteq JI⊆J and η(t)=γ(t)\eta(t)=\gamma(t)η(t)=γ(t) for every t∈It\in It∈I. It imposes no continuity, interval, differential, or initial-value conditions. Equality of the domains is allowed, values outside III need not agree, and if I=∅I=\varnothingI=∅, the predicate holds for every J,γ,ηJ,\gamma,\etaJ,γ,η of these types.

QAlgorithms.Papers.Independent.Intakenext100.P21.FlowData

For any type MMM, an object of this structure consists of a set D⊆R×MD\subseteq\mathbb R\times MD⊆R×M, called its domain, and a total function φ:R→M→M\varphi:\mathbb R\to M\to Mφ:R→M→M, called its value. The structure imposes no relation between DDD and φ\varphiφ, and no openness, continuity, identity, or composition condition. In particular, φ(t,m)\varphi(t,m)φ(t,m) is supplied even when (t,m)∉D(t,m)\notin D(t,m)∈/D; neither DDD nor MMM is required to be nonempty.

QAlgorithms.Papers.Independent.Intakenext100.P21.HasTimeDerivative

For a real vector space FFF equipped with a topology, a function f:R→Ff:\mathbb R\to Ff:R→F, a time t∈Rt\in\mathbb Rt∈R, and a vector y∈Fy\in Fy∈F, this predicate means that

lim⁡s→0s≠01s(f(t+s)−f(t))=y.\lim_{\substack{s\to0\\s\ne0}}\frac1s\bigl(f(t+s)-f(t)\bigr)=y.s→0s=0​lim​s1​(f(t+s)−f(t))=y.

The limit uses the topology of FFF and approaches zero through all nonzero real increments, from either side. The definition itself imposes no additional continuity, completeness, or separation assumption on FFF.

QAlgorithms.Papers.Independent.Intakenext100.P21.HasVelocityField

Let EEE be a real vector space equipped with a topology, let MMM be a topological space with a specified atlas modeled on EEE, let PPP consist of D⊆R×MD\subseteq\mathbb R\times MD⊆R×M and a total function φ:R→M→M\varphi:\mathbb R\to M\to Mφ:R→M→M, and let X(c,m)∈EX(c,m)\in EX(c,m)∈E be an assignment for every open partial homeomorphism c:M⇢Ec:M\dashrightarrow Ec:M⇢E and every m∈Mm\in Mm∈M. This predicate means that for every m∈Mm\in Mm∈M and every atlas chart ccc whose source contains mmm,

lim⁡t→0t≠0c(φ(t,m))−c(φ(0,m))t=X(c,m).\lim_{\substack{t\to0\\t\ne0}} \frac{c(\varphi(t,m))-c(\varphi(0,m))}t=X(c,m).t→0t=0​lim​tc(φ(t,m))−c(φ(0,m))​=X(c,m).

The limit uses all nonzero real increments. The set DDD does not occur in this condition, and the predicate itself does not assume (0,m)∈D(0,m)\in D(0,m)∈D, φ(0,m)=m\varphi(0,m)=mφ(0,m)=m, or any local-flow property.

QAlgorithms.Papers.Independent.Intakenext100.P21.IsAdIntegrable

For the given real vector space AAA with topology, bracket operation, and self-charted structure, this proposition requires the following for every x∈Ax\in Ax∈A. Define Zcx(y)=[x,y]Z^x_c(y)=[x,y]Zcx​(y)=[x,y] for every open partial homeomorphism c:A⇄Ac:A\rightleftarrows Ac:A⇄A and y∈Ay\in Ay∈A. Here smoothness of f:W→Tf:W\to Tf:W→T on UUU means existence of functions dn:W×Wn→Td_n:W\times W^n\to Tdn​:W×Wn→T for every integer n≥0n\ge0n≥0, jointly continuous on U×WnU\times W^nU×Wn, with d0(y,())=f(y)d_0(y,())=f(y)d0​(y,())=f(y) on UUU and

lim⁡s→0s≠0s−1(dn(y+sh,a)−dn(y,a))=dn+1(y,(a,h))\lim_{\substack{s\to0\\s\ne0}}s^{-1}\bigl(d_n(y+sh,a)-d_n(y,a)\bigr) =d_{n+1}(y,(a,h))s→0s=0​lim​s−1(dn​(y+sh,a)−dn​(y,a))=dn+1​(y,(a,h))

for every n,y∈U,a∈Wn,h∈Wn,y\in U,a\in W^n,h\in Wn,y∈U,a∈Wn,h∈W; multilinearity of the dnd_ndn​ is not required. First, z↦[x,c−1(z)]z\mapsto[x,c^{-1}(z)]z↦[x,c−1(z)] must be smooth on the target of each atlas chart ccc, and, for every pair of atlas charts c,dc,dc,d whose sources contain yyy, the directional difference quotient of d∘c−1d\circ c^{-1}d∘c−1 at c(y)c(y)c(y) in direction [x,y][x,y][x,y] must tend to [x,y][x,y][x,y]. Second, there must exist a domain Ω⊆R×A\Omega\subseteq\mathbb R\times AΩ⊆R×A and a total function P:R×A→AP:\mathbb R\times A\to AP:R×A→A such that Ω\OmegaΩ is open, contains all (0,y)(0,y)(0,y), has order-connected time sections, PPP is continuous on Ω\OmegaΩ, every map (t,z)↦d(P(t,c−1(z)))(t,z)\mapsto d(P(t,c^{-1}(z)))(t,z)↦d(P(t,c−1(z))) is smooth on {(t,z):z∈c.target, (t,c−1(z))∈Ω, P(t,c−1(z))∈d.source}\{(t,z):z\in c.\mathrm{target},\ (t,c^{-1}(z))\in\Omega,\ P(t,c^{-1}(z))\in d.\mathrm{source}\}{(t,z):z∈c.target, (t,c−1(z))∈Ω, P(t,c−1(z))∈d.source}, P(0,y)=yP(0,y)=yP(0,y)=y, and P(t,P(s,y))=P(s+t,y)P(t,P(s,y))=P(s+t,y)P(t,P(s,y))=P(s+t,y) whenever (s,y),(t,P(s,y)),(s+t,y)∈Ω(s,y),(t,P(s,y)),(s+t,y)\in\Omega(s,y),(t,P(s,y)),(s+t,y)∈Ω; for every atlas chart ccc and y∈c.sourcey\in c.\mathrm{source}y∈c.source, the time difference quotient of c(P(t,y))c(P(t,y))c(P(t,y)) at t=0t=0t=0 must tend to [x,y][x,y][x,y]. Third, for every t∈Rt\in\mathbb Rt∈R and y∈Ay\in Ay∈A, there must exist an open, order-connected I⊆RI\subseteq\mathbb RI⊆R containing 0,t0,t0,t and a total function γ:R→A\gamma:\mathbb R\to Aγ:R→A, continuous on III, such that γ(0)=y\gamma(0)=yγ(0)=y and

lim⁡s→0s≠0s−1(c(γ(r+s))−c(γ(r)))=[x,γ(r)]\lim_{\substack{s\to0\\s\ne0}} s^{-1}\bigl(c(\gamma(r+s))-c(\gamma(r))\bigr) =[x,\gamma(r)]s→0s=0​lim​s−1(c(γ(r+s))−c(γ(r)))=[x,γ(r)]

for every r∈Ir\in Ir∈I and every atlas chart ccc containing γ(r)\gamma(r)γ(r) in its source. This last requirement is precisely that the domain of the defined maximal flow of ZxZ^xZx equals R×A\mathbb R\times AR×A; the witnesses I,γI,\gammaI,γ may depend on x,y,tx,y,tx,y,t. Order-connected means containing every real number between any two members.

QAlgorithms.Papers.Independent.Intakenext100.P21.IsIntegralCurve

Let EEE be a real vector space equipped with a topology, let MMM be a topological space with a specified atlas modeled on EEE, let X(c,m)∈EX(c,m)\in EX(c,m)∈E be an arbitrary assignment for every open partial homeomorphism c:M⇢Ec:M\dashrightarrow Ec:M⇢E and m∈Mm\in Mm∈M, let I⊆RI\subseteq\mathbb RI⊆R, and let γ:R→M\gamma:\mathbb R\to Mγ:R→M be a total function. This predicate means that III is open, contains 000, and is order-connected—whenever a,b∈Ia,b\in Ia,b∈I and a≤r≤ba\le r\le ba≤r≤b, one has r∈Ir\in Ir∈I—that γ\gammaγ is continuous on III, and that for every t∈It\in It∈I and every atlas chart ccc whose source contains γ(t)\gamma(t)γ(t),

lim⁡s→0s≠0c(γ(t+s))−c(γ(t))s=X(c,γ(t)).\lim_{\substack{s\to0\\s\ne0}} \frac{c(\gamma(t+s))-c(\gamma(t))}s =X(c,\gamma(t)).s→0s=0​lim​sc(γ(t+s))−c(γ(t))​=X(c,γ(t)).

Thus III is a nonempty open interval, possibly unbounded or all of R\mathbb RR. No initial value for γ(0)\gamma(0)γ(0) is prescribed, no smoothness assumption on XXX is included, and no condition on γ\gammaγ is imposed outside III.

QAlgorithms.Papers.Independent.Intakenext100.P21.IsLocalFlow

Let EEE be a real vector space equipped with a topology, let MMM be a topological space with a specified atlas modeled on EEE, and let PPP consist of D⊆R×MD\subseteq\mathbb R\times MD⊆R×M and a total function φ:R→M→M\varphi:\mathbb R\to M\to Mφ:R→M→M. This predicate means that DDD is open, (0,m)∈D(0,m)\in D(0,m)∈D for every m∈Mm\in Mm∈M, each time section Dm={t:(t,m)∈D}D_m=\{t:(t,m)\in D\}Dm​={t:(t,m)∈D} contains every real number between any two of its elements, (t,m)↦φ(t,m)(t,m)\mapsto\varphi(t,m)(t,m)↦φ(t,m) is continuous on DDD, and φ(0,m)=m\varphi(0,m)=mφ(0,m)=m for every mmm. It additionally requires coordinate smoothness as follows: for every pair of atlas charts c,dc,dc,d, put W=R×EW=\mathbb R\times EW=R×E,

Uc,d={(t,z):z∈target⁡(c), (t,c−1(z))∈D, φ(t,c−1(z))∈source⁡(d)},gc,d(t,z)=d(φ(t,c−1(z))).U_{c,d}=\{(t,z):z\in\operatorname{target}(c),\ (t,c^{-1}(z))\in D,\ \varphi(t,c^{-1}(z))\in\operatorname{source}(d)\}, \qquad g_{c,d}(t,z)=d(\varphi(t,c^{-1}(z))).Uc,d​={(t,z):z∈target(c), (t,c−1(z))∈D, φ(t,c−1(z))∈source(d)},gc,d​(t,z)=d(φ(t,c−1(z))).

There must exist functions An:W×Wn→EA_n:W\times W^n\to EAn​:W×Wn→E, for all natural numbers n≥0n\ge0n≥0, with A0(p,())=gc,d(p)A_0(p,())=g_{c,d}(p)A0​(p,())=gc,d​(p) on Uc,dU_{c,d}Uc,d​, jointly continuous on Uc,d×WnU_{c,d}\times W^nUc,d​×Wn, and satisfying

lim⁡r→0r≠0An(p+rh,a)−An(p,a)r=An+1(p,(a,h))\lim_{\substack{r\to0\\r\ne0}}\frac{A_n(p+rh,a)-A_n(p,a)}r =A_{n+1}(p,(a,h))r→0r=0​lim​rAn​(p+rh,a)−An​(p,a)​=An+1​(p,(a,h))

for every p∈Uc,dp\in U_{c,d}p∈Uc,d​, a∈Wna\in W^na∈Wn, and h∈Wh\in Wh∈W, where (a,h)(a,h)(a,h) appends the last direction and the zero-length tuple is empty. Finally, for all real s,ts,ts,t and all m∈Mm\in Mm∈M,

(s,m)∈D,(t,φ(s,m))∈D,(s+t,m)∈D⟹φ(t,φ(s,m))=φ(s+t,m).(s,m)\in D,\quad (t,\varphi(s,m))\in D,\quad (s+t,m)\in D \quad\Longrightarrow\quad \varphi(t,\varphi(s,m))=\varphi(s+t,m).(s,m)∈D,(t,φ(s,m))∈D,(s+t,m)∈D⟹φ(t,φ(s,m))=φ(s+t,m).

All three domain memberships are hypotheses of this implication; membership of (s+t,m)(s+t,m)(s+t,m) is not asserted to follow from the first two. Each DmD_mDm​ is a nonempty open interval containing zero, while values of φ\varphiφ outside DDD remain part of the supplied total function.

QAlgorithms.Papers.Independent.Intakenext100.P21.IsLocallyIntegrable

Let EEE be a real vector space equipped with a topology, let MMM be a topological space with a specified atlas modeled on EEE, and let X(c,m)∈EX(c,m)\in EX(c,m)∈E be an assignment for every open partial homeomorphism c:M⇢Ec:M\dashrightarrow Ec:M⇢E and every m∈Mm\in Mm∈M. To state the predicate explicitly, for a topological real vector space WWW, a set U⊆WU\subseteq WU⊆W, and a total function g:W→Eg:W\to Eg:W→E, write B(U,g)\mathcal B(U,g)B(U,g) for the following condition: there exist functions An:W×Wn→EA_n:W\times W^n\to EAn​:W×Wn→E, for every natural number n≥0n\ge0n≥0, with A0(p,())=g(p)A_0(p,())=g(p)A0​(p,())=g(p) on UUU, each AnA_nAn​ jointly continuous on U×WnU\times W^nU×Wn, and

lim⁡r→0r≠0An(p+rh,a)−An(p,a)r=An+1(p,(a,h))\lim_{\substack{r\to0\\r\ne0}}\frac{A_n(p+rh,a)-A_n(p,a)}r =A_{n+1}(p,(a,h))r→0r=0​lim​rAn​(p+rh,a)−An​(p,a)​=An+1​(p,(a,h))

for every n≥0n\ge0n≥0, p∈Up\in Up∈U, a∈Wna\in W^na∈Wn, and h∈Wh\in Wh∈W, where (a,h)(a,h)(a,h) appends the last direction and W0W^0W0 contains only the empty tuple. The predicate requires B(target⁡(c),z↦X(c,c−1(z)))\mathcal B(\operatorname{target}(c),z\mapsto X(c,c^{-1}(z)))B(target(c),z↦X(c,c−1(z))) for every atlas chart ccc, and, for every pair of atlas charts c,dc,dc,d and mmm in both sources,

lim⁡r→0r≠0d(c−1(c(m)+rX(c,m)))−d(m)r=X(d,m).\lim_{\substack{r\to0\\r\ne0}} \frac{d(c^{-1}(c(m)+rX(c,m)))-d(m)}r=X(d,m).r→0r=0​lim​rd(c−1(c(m)+rX(c,m)))−d(m)​=X(d,m).

It further requires the existence of a set D⊆R×MD\subseteq\mathbb R\times MD⊆R×M and a total function φ:R→M→M\varphi:\mathbb R\to M\to Mφ:R→M→M such that DDD is open, contains every (0,m)(0,m)(0,m), each section {t:(t,m)∈D}\{t:(t,m)\in D\}{t:(t,m)∈D} contains every real number between any two of its elements, φ\varphiφ is jointly continuous on DDD, and φ(0,m)=m\varphi(0,m)=mφ(0,m)=m. For every pair of atlas charts c,dc,dc,d, it requires B(Uc,d,gc,d)\mathcal B(U_{c,d},g_{c,d})B(Uc,d​,gc,d​), where

Uc,d={(t,z):z∈target⁡(c), (t,c−1(z))∈D, φ(t,c−1(z))∈source⁡(d)},gc,d(t,z)=d(φ(t,c−1(z))).U_{c,d}=\{(t,z):z\in\operatorname{target}(c),\ (t,c^{-1}(z))\in D,\ \varphi(t,c^{-1}(z))\in\operatorname{source}(d)\}, \qquad g_{c,d}(t,z)=d(\varphi(t,c^{-1}(z))).Uc,d​={(t,z):z∈target(c), (t,c−1(z))∈D, φ(t,c−1(z))∈source(d)},gc,d​(t,z)=d(φ(t,c−1(z))).

For all real s,ts,ts,t and m∈Mm\in Mm∈M, the three hypotheses (s,m)∈D(s,m)\in D(s,m)∈D, (t,φ(s,m))∈D(t,\varphi(s,m))\in D(t,φ(s,m))∈D, and (s+t,m)∈D(s+t,m)\in D(s+t,m)∈D must imply φ(t,φ(s,m))=φ(s+t,m)\varphi(t,\varphi(s,m))=\varphi(s+t,m)φ(t,φ(s,m))=φ(s+t,m). Finally, for every atlas chart ccc and m∈source⁡(c)m\in\operatorname{source}(c)m∈source(c),

lim⁡t→0t≠0c(φ(t,m))−c(m)t=X(c,m).\lim_{\substack{t\to0\\t\ne0}}\frac{c(\varphi(t,m))-c(m)}t=X(c,m).t→0t=0​lim​tc(φ(t,m))−c(m)​=X(c,m).

All displayed limits use unrestricted nonzero real increments. Only existence of such D,φD,\varphiD,φ is required; uniqueness is not included.

QAlgorithms.Papers.Independent.Intakenext100.P21.IsMaximalIntegralCurve

Let EEE be a real vector space equipped with a topology, let MMM be a topological space with a specified atlas modeled on EEE, let X(c,m)∈EX(c,m)\in EX(c,m)∈E be an assignment for every open partial homeomorphism c:M⇢Ec:M\dashrightarrow Ec:M⇢E and every m∈Mm\in Mm∈M, let I⊆RI\subseteq\mathbb RI⊆R, and let γ:R→M\gamma:\mathbb R\to Mγ:R→M be a total function. This predicate first requires that III be open, contain zero, and contain every real number between any two of its elements, that γ\gammaγ be continuous on III, and that for every t∈It\in It∈I and every atlas chart ccc with γ(t)∈source⁡(c)\gamma(t)\in\operatorname{source}(c)γ(t)∈source(c),

lim⁡s→0s≠0c(γ(t+s))−c(γ(t))s=X(c,γ(t)).\lim_{\substack{s\to0\\s\ne0}} \frac{c(\gamma(t+s))-c(\gamma(t))}s=X(c,\gamma(t)).s→0s=0​lim​sc(γ(t+s))−c(γ(t))​=X(c,γ(t)).

It further requires that for every set J⊆RJ\subseteq\mathbb RJ⊆R and total function η:R→M\eta:\mathbb R\to Mη:R→M, if JJJ is open, contains zero, contains every real number between any two of its elements, η\etaη is continuous on JJJ, and

lim⁡s→0s≠0c(η(t+s))−c(η(t))s=X(c,η(t))\lim_{\substack{s\to0\\s\ne0}} \frac{c(\eta(t+s))-c(\eta(t))}s=X(c,\eta(t))s→0s=0​lim​sc(η(t+s))−c(η(t))​=X(c,η(t))

for every t∈Jt\in Jt∈J and every atlas chart ccc whose source contains η(t)\eta(t)η(t), then the additional hypotheses I⊆JI\subseteq JI⊆J and η(t)=γ(t)\eta(t)=\gamma(t)η(t)=γ(t) for all t∈It\in It∈I imply J⊆IJ\subseteq IJ⊆I. Consequently, every such extending curve has the same domain III. The predicate does not prescribe γ(0)\gamma(0)γ(0), require XXX to be smooth, or require extending total functions to agree outside III.

QAlgorithms.Papers.Independent.Intakenext100.P21.IsSmoothFlowMap

Let EEE be a real vector space equipped with a topology, let MMM be a topological space with a specified atlas modeled on EEE, and let PPP consist of a set D⊆R×MD\subseteq\mathbb R\times MD⊆R×M and a total function φ:R→M→M\varphi:\mathbb R\to M\to Mφ:R→M→M. This predicate means that (t,m)↦φ(t,m)(t,m)\mapsto\varphi(t,m)(t,m)↦φ(t,m) is continuous on DDD, and that the following holds for every pair of atlas charts c,dc,dc,d. Define

Uc,d={(t,z)∈R×E:z∈target⁡(c), (t,c−1(z))∈D, φ(t,c−1(z))∈source⁡(d)},gc,d(t,z)=d(φ(t,c−1(z))).U_{c,d}=\{(t,z)\in\mathbb R\times E: z\in\operatorname{target}(c),\ (t,c^{-1}(z))\in D,\ \varphi(t,c^{-1}(z))\in\operatorname{source}(d)\}, \qquad g_{c,d}(t,z)=d(\varphi(t,c^{-1}(z))).Uc,d​={(t,z)∈R×E:z∈target(c), (t,c−1(z))∈D, φ(t,c−1(z))∈source(d)},gc,d​(t,z)=d(φ(t,c−1(z))).

Writing W=R×EW=\mathbb R\times EW=R×E, there must exist functions An:W×Wn→EA_n:W\times W^n\to EAn​:W×Wn→E, for every natural number n≥0n\ge0n≥0, such that A0(p,())=gc,d(p)A_0(p,())=g_{c,d}(p)A0​(p,())=gc,d​(p) on Uc,dU_{c,d}Uc,d​, each AnA_nAn​ is jointly continuous on Uc,d×WnU_{c,d}\times W^nUc,d​×Wn, and for every p∈Uc,dp\in U_{c,d}p∈Uc,d​, a∈Wna\in W^na∈Wn, and h∈Wh\in Wh∈W,

lim⁡s→0s≠0An(p+sh,a)−An(p,a)s=An+1(p,(a,h)).\lim_{\substack{s\to0\\s\ne0}}\frac{A_n(p+sh,a)-A_n(p,a)}s =A_{n+1}(p,(a,h)).s→0s=0​lim​sAn​(p+sh,a)−An​(p,a)​=An+1​(p,(a,h)).

Here (a,h)(a,h)(a,h) appends the final direction, and W0W^0W0 contains only the empty tuple. These limits use unrestricted nonzero real increments and total ambient functions. The predicate does not separately require DDD to be open or nonempty; an empty DDD satisfies it.

QAlgorithms.Papers.Independent.Intakenext100.P21.IsSmoothInfinitesimalAction

Let AAA be a real Lie algebra equipped with a topology, let EEE be a real vector space equipped with a topology, and let MMM be a topological space with charts modelled on EEE. A coordinate field assigns a vector Zc(m)∈EZ_c(m)\in EZc​(m)∈E to every open partial homeomorphism c:M→Ec:M\to Ec:M→E and every m∈Mm\in Mm∈M; this assignment is defined even for charts outside the atlas and points outside a chart’s source. Let β\betaβ be a real-linear map from AAA to these coordinate fields. To describe the predicate, write Dg(z;h)Dg(z;h)Dg(z;h) for the classically chosen vector satisfying lim⁡r→0, r≠0r−1(g(z+rh)−g(z))=Dg(z;h)\lim_{r\to0,\ r\ne0}r^{-1}(g(z+rh)-g(z))=Dg(z;h)limr→0, r=0​r−1(g(z+rh)−g(z))=Dg(z;h), if any such vector exists; if none exists, this notation still denotes an arbitrarily chosen vector. Also, say that a map g:B→Cg:B\to Cg:B→C has the required smoothness on U⊆BU\subseteq BU⊆B when there exist maps dn:B×Bn→Cd_n:B\times B^n\to Cdn​:B×Bn→C for every integer n≥0n\ge0n≥0, with d0(z;())=g(z)d_0(z;())=g(z)d0​(z;())=g(z) for z∈Uz\in Uz∈U, each dnd_ndn​ jointly continuous on U×BnU\times B^nU×Bn, and, for every n≥0n\ge0n≥0, z∈Uz\in Uz∈U, a∈Bna\in B^na∈Bn, and h∈Bh\in Bh∈B, lim⁡r→0, r≠0r−1(dn(z+rh;a)−dn(z;a))=dn+1(z;a,h)\lim_{r\to0,\ r\ne0}r^{-1}(d_n(z+rh;a)-d_n(z;a))=d_{n+1}(z;a,h)limr→0, r=0​r−1(dn​(z+rh;a)−dn​(z;a))=dn+1​(z;a,h). These limits use the globally defined maps, without restricting z+rhz+rhz+rh to UUU; order zero uses the unique empty tuple. The predicate asserts the conjunction of three conditions: first, for every a∈Aa\in Aa∈A and every atlas chart ccc, the map z↦β(a)c(c−1z)z\mapsto\beta(a)_c(c^{-1}z)z↦β(a)c​(c−1z) has this smoothness on c.targetc.\mathrm{target}c.target, and for every pair of atlas charts c,dc,dc,d and every m∈c.source∩d.sourcem\in c.\mathrm{source}\cap d.\mathrm{source}m∈c.source∩d.source, the limit lim⁡r→0, r≠0r−1(d(c−1(c(m)+rβ(a)c(m)))−d(m))\lim_{r\to0,\ r\ne0}r^{-1}\bigl(d(c^{-1}(c(m)+r\beta(a)_c(m)))-d(m)\bigr)limr→0, r=0​r−1(d(c−1(c(m)+rβ(a)c​(m)))−d(m)) exists and equals β(a)d(m)\beta(a)_d(m)β(a)d​(m); second, for all a,b∈Aa,b\in Aa,b∈A, every atlas chart ccc, and every m∈c.sourcem\in c.\mathrm{source}m∈c.source, β([a,b])c(m)=D(z↦β(b)c(c−1z))(c(m);β(a)c(m))−D(z↦β(a)c(c−1z))(c(m);β(b)c(m))\beta([a,b])_c(m)=D(z\mapsto\beta(b)_c(c^{-1}z))(c(m);\beta(a)_c(m))-D(z\mapsto\beta(a)_c(c^{-1}z))(c(m);\beta(b)_c(m))β([a,b])c​(m)=D(z↦β(b)c​(c−1z))(c(m);β(a)c​(m))−D(z↦β(a)c​(c−1z))(c(m);β(b)c​(m)); and third, for every atlas chart ccc, the map (a,z)↦β(a)c(c−1z)(a,z)\mapsto\beta(a)_c(c^{-1}z)(a,z)↦β(a)c​(c−1z), defined on all of A×EA\times EA×E, has the stated smoothness on A×c.targetA\times c.\mathrm{target}A×c.target. Conditions restricted to empty chart domains are vacuous.

QAlgorithms.Papers.Independent.Intakenext100.P21.IsSmoothVectorField

Let EEE be a real vector space equipped with a topology, let MMM be a topological space with a specified charted-space structure modeled on EEE, and let X(c,m)∈EX(c,m)\in EX(c,m)∈E be an assignment for every open partial homeomorphism c:M⇢Ec:M\dashrightarrow Ec:M⇢E and every m∈Mm\in Mm∈M. This predicate requires two conditions. First, for each chart ccc in the specified atlas, there exist functions Dn:E×En→ED_n:E\times E^n\to EDn​:E×En→E, for every natural number n≥0n\ge0n≥0, such that D0(z,())=X(c,c−1(z))D_0(z,())=X(c,c^{-1}(z))D0​(z,())=X(c,c−1(z)) for zzz in the target of ccc, each DnD_nDn​ is jointly continuous on target⁡(c)×En\operatorname{target}(c)\times E^ntarget(c)×En, and, for every n≥0n\ge0n≥0, z∈target⁡(c)z\in\operatorname{target}(c)z∈target(c), tuple a∈Ena\in E^na∈En, and h∈Eh\in Eh∈E,

lim⁡s→0s≠0Dn(z+sh,a)−Dn(z,a)s=Dn+1(z,(a,h)),\lim_{\substack{s\to0\\s\ne0}}\frac{D_n(z+sh,a)-D_n(z,a)}s =D_{n+1}(z,(a,h)),s→0s=0​lim​sDn​(z+sh,a)−Dn​(z,a)​=Dn+1​(z,(a,h)),

where (a,h)(a,h)(a,h) appends hhh as the last entry and E0E^0E0 consists of the empty tuple. Second, for all atlas charts c,dc,dc,d and every m∈source⁡(c)∩source⁡(d)m\in\operatorname{source}(c)\cap\operatorname{source}(d)m∈source(c)∩source(d),

lim⁡s→0s≠0d ⁣(c−1(c(m)+sX(c,m)))−d(m)s=X(d,m).\lim_{\substack{s\to0\\s\ne0}} \frac{d\!\left(c^{-1}(c(m)+sX(c,m))\right)-d(m)}s =X(d,m).s→0s=0​lim​sd(c−1(c(m)+sX(c,m)))−d(m)​=X(d,m).

All limits use unrestricted nonzero real increments; the functions involved are defined on their whole ambient spaces. Values of XXX at charts outside the atlas or at points outside the relevant chart sources receive no additional conditions.

QAlgorithms.Papers.Independent.Intakenext100.P21.adjointEvolution

For the given space AAA with topology, bracket operation, and self-charted structure, and for every x,y∈Ax,y\in Ax,y∈A and t∈Rt\in\mathbb Rt∈R, the value adjointEvolution⁡(x,y,t)\operatorname{adjointEvolution}(x,y,t)adjointEvolution(x,y,t) is a classically chosen z∈Az\in Az∈A for which there exist an open, order-connected set I⊆RI\subseteq\mathbb RI⊆R containing 0,t0,t0,t and a total function γ:R→A\gamma:\mathbb R\to Aγ:R→A, continuous on III, such that γ(0)=y\gamma(0)=yγ(0)=y, γ(t)=z\gamma(t)=zγ(t)=z, and, for every r∈Ir\in Ir∈I and every atlas chart ccc whose source contains γ(r)\gamma(r)γ(r),

lim⁡s→0s≠0s−1(c(γ(r+s))−c(γ(r)))=[x,γ(r)].\lim_{\substack{s\to0\\s\ne0}} s^{-1}\bigl(c(\gamma(r+s))-c(\gamma(r))\bigr) =[x,\gamma(r)].s→0s=0​lim​s−1(c(γ(r+s))−c(γ(r)))=[x,γ(r)].

If no such zzz exists, the defined value is yyy. Order-connected means that III contains every real number between any two of its members. This definition does not assume the proposition that all adjoint fields are locally integrable with full maximal-flow domain, and it does not assume uniqueness of the eligible zzz; values of γ\gammaγ outside III are unrestricted.

QAlgorithms.Papers.Independent.Intakenext100.P21.adjointField

For the given space AAA with bracket operation [ ⋅,⋅ ][\,\cdot,\cdot\,][⋅,⋅], and for every x∈Ax\in Ax∈A, this definition assigns to every open partial homeomorphism c:A⇄Ac:A\rightleftarrows Ac:A⇄A and every y∈Ay\in Ay∈A the vector

(adjointField⁡x)c(y)=[x,y].(\operatorname{adjointField}x)_c(y)=[x,y].(adjointFieldx)c​(y)=[x,y].

The assignment is independent of ccc and is defined for all yyy, including points outside c.sourcec.\mathrm{source}c.source.

QAlgorithms.Papers.Independent.Intakenext100.P21.bracketCoordinate

Let EEE be a real vector space with topology, MMM a topological space, and X,YX,YX,Y coordinate assignments giving vectors Xc(m),Yc(m)∈EX_c(m),Y_c(m)\in EXc​(m),Yc​(m)∈E for every open partial homeomorphism c:M⇄Ec:M\rightleftarrows Ec:M⇄E and m∈Mm\in Mm∈M. For a function f:E→Ef:E\to Ef:E→E, write δhf(x)\delta_hf(x)δh​f(x) for the classically selected vector vvv satisfying

lim⁡s→0s≠0s−1(f(x+sh)−f(x))=v\lim_{\substack{s\to0\\s\ne0}}s^{-1}\bigl(f(x+sh)-f(x)\bigr)=vs→0s=0​lim​s−1(f(x+sh)−f(x))=v

when at least one such vector exists, and for an arbitrary selected vector otherwise. For every such ccc and every m∈Mm\in Mm∈M, the definition is

bracketCoordinate⁡(X,Y,c,m)=δXc(m) ⁣(z↦Yc(c−1(z)))(c(m))−δYc(m) ⁣(z↦Xc(c−1(z)))(c(m)).\operatorname{bracketCoordinate}(X,Y,c,m) =\delta_{X_c(m)}\!\left(z\mapsto Y_c(c^{-1}(z))\right)(c(m)) -\delta_{Y_c(m)}\!\left(z\mapsto X_c(c^{-1}(z))\right)(c(m)).bracketCoordinate(X,Y,c,m)=δXc​(m)​(z↦Yc​(c−1(z)))(c(m))−δYc​(m)​(z↦Xc​(c−1(z)))(c(m)).

There is no requirement here that X,YX,YX,Y be smooth, that the displayed limits exist, that ccc belong to the atlas, or that m∈c.sourcem\in c.\mathrm{source}m∈c.source; the total functions underlying the partial homeomorphism and its inverse are used even outside their specified source and target.

QAlgorithms.Papers.Independent.Intakenext100.P21.linearField

For a real vector space VVV with its given topology and a continuous real-linear map D:V→VD:V\to VD:V→V, this definition assigns the vector DvDvDv to every pair consisting of an open partial homeomorphism c:V⇄Vc:V\rightleftarrows Vc:V⇄V and a point v∈Vv\in Vv∈V:

(linearField⁡D)c(v)=Dv.(\operatorname{linearField}D)_c(v)=Dv.(linearFieldD)c​(v)=Dv.

The assignment is independent of ccc and is defined for every vvv, including points outside c.sourcec.\mathrm{source}c.source.

QAlgorithms.Papers.Independent.Intakenext100.P21.maximalCurve

Let EEE be the real vector space with topology serving as the chart model for the topological space MMM. A coordinate field XXX assigns a vector Xc(p)∈EX_c(p)\in EXc​(p)∈E to every open partial homeomorphism c:M⇄Ec:M\rightleftarrows Ec:M⇄E and every p∈Mp\in Mp∈M. For m∈Mm\in Mm∈M and t∈Rt\in\mathbb Rt∈R, this definition chooses a point y∈My\in My∈M for which there exist a set I⊆RI\subseteq\mathbb RI⊆R and a total function γ:R→M\gamma:\mathbb R\to Mγ:R→M such that III is open and order-connected, 0,t∈I0,t\in I0,t∈I, γ\gammaγ is continuous on III, γ(0)=m\gamma(0)=mγ(0)=m, γ(t)=y\gamma(t)=yγ(t)=y, and, for every r∈Ir\in Ir∈I and every atlas chart ccc with γ(r)∈c.source\gamma(r)\in c.\mathrm{source}γ(r)∈c.source,

lim⁡s→0s≠0s−1(c(γ(r+s))−c(γ(r)))=Xc(γ(r)).\lim_{\substack{s\to0\\s\ne0}} s^{-1}\bigl(c(\gamma(r+s))-c(\gamma(r))\bigr) =X_c(\gamma(r)).s→0s=0​lim​s−1(c(γ(r+s))−c(γ(r)))=Xc​(γ(r)).

Here order-connected means that every real number between two members of III also belongs to III. If no such yyy exists, the defined value is mmm. The choice is made separately for each m,tm,tm,t; the definition assumes neither existence of such curves nor uniqueness of their values. Values of γ\gammaγ outside III are unrestricted.

QAlgorithms.Papers.Independent.Intakenext100.P21.maximalCurveDomain

Let EEE be a real vector space equipped with a topology, let MMM be a topological space with a specified atlas modeled on EEE, let X(c,m)∈EX(c,m)\in EX(c,m)∈E be an assignment for every open partial homeomorphism c:M⇢Ec:M\dashrightarrow Ec:M⇢E and every m∈Mm\in Mm∈M, and fix m∈Mm\in Mm∈M. This definition gives the set of all t∈Rt\in\mathbb Rt∈R for which there exist a set I⊆RI\subseteq\mathbb RI⊆R and a total function γ:R→M\gamma:\mathbb R\to Mγ:R→M such that III is open, contains zero and ttt, contains every real number between any two of its elements, γ\gammaγ is continuous on III, γ(0)=m\gamma(0)=mγ(0)=m, and for every r∈Ir\in Ir∈I and every atlas chart ccc whose source contains γ(r)\gamma(r)γ(r),

lim⁡s→0s≠0c(γ(r+s))−c(γ(r))s=X(c,γ(r)).\lim_{\substack{s\to0\\s\ne0}} \frac{c(\gamma(r+s))-c(\gamma(r))}s=X(c,\gamma(r)).s→0s=0​lim​sc(γ(r+s))−c(γ(r))​=X(c,γ(r)).

The witnesses I,γI,\gammaI,γ may depend on ttt; neither maximality nor uniqueness of a witnessing curve is required. The definition asserts no existence of any such curve, so the set can be empty; if any witnessing curve exists, the set contains zero.

QAlgorithms.Papers.Independent.Intakenext100.P21.maximalFlow

Let EEE be the real vector space with topology modeling the topological charted space MMM, and let Xc(p)∈EX_c(p)\in EXc​(p)∈E be a coordinate assignment for every open partial homeomorphism c:M⇄Ec:M\rightleftarrows Ec:M⇄E and every p∈Mp\in Mp∈M. This definition produces a pair consisting of a domain ΩX⊆R×M\Omega_X\subseteq\mathbb R\times MΩX​⊆R×M and a total value function ΦX:R×M→M\Phi_X:\mathbb R\times M\to MΦX​:R×M→M. A pair (t,m)(t,m)(t,m) belongs to ΩX\Omega_XΩX​ exactly when there exist an open, order-connected set I⊆RI\subseteq\mathbb RI⊆R containing 0,t0,t0,t and a total function γ:R→M\gamma:\mathbb R\to Mγ:R→M, continuous on III, such that γ(0)=m\gamma(0)=mγ(0)=m and, for every r∈Ir\in Ir∈I and every atlas chart ccc containing γ(r)\gamma(r)γ(r) in its source,

lim⁡s→0s≠0s−1(c(γ(r+s))−c(γ(r)))=Xc(γ(r)).\lim_{\substack{s\to0\\s\ne0}} s^{-1}\bigl(c(\gamma(r+s))-c(\gamma(r))\bigr)=X_c(\gamma(r)).s→0s=0​lim​s−1(c(γ(r+s))−c(γ(r)))=Xc​(γ(r)).

Order-connected means containing all points between any two of its points. For (t,m)∈ΩX(t,m)\in\Omega_X(t,m)∈ΩX​, the value ΦX(t,m)\Phi_X(t,m)ΦX​(t,m) is a classically chosen yyy that equals γ(t)\gamma(t)γ(t) for some such I,γI,\gammaI,γ; for (t,m)∉ΩX(t,m)\notin\Omega_X(t,m)∈/ΩX​, it is mmm. The definition supplies this domain and total function without imposing flow axioms or assuming uniqueness of the possible yyy.

QAlgorithms.Papers.Independent.Intakenext100.P21.pullbackCoordinate

Let EEE be the real vector space with topology modeling the topological charted space MMM. Let PPP consist of an arbitrary domain Ω⊆R×M\Omega\subseteq\mathbb R\times MΩ⊆R×M and a total value function P:R×M→MP:\mathbb R\times M\to MP:R×M→M, and let Yc(m)∈EY_c(m)\in EYc​(m)∈E be a coordinate assignment indexed by every open partial homeomorphism c:M⇄Ec:M\rightleftarrows Ec:M⇄E and every m∈Mm\in Mm∈M. For any t∈Rt\in\mathbb Rt∈R, any such ccc, and any m∈Mm\in Mm∈M, put p=P(t,m)p=P(t,m)p=P(t,m) and let ddd be the distinguished chart at ppp. The defined value is

δYd(p) ⁣(z↦c(P(−t,d−1(z))))(d(p)),\delta_{Y_d(p)}\!\left(z\mapsto c(P(-t,d^{-1}(z)))\right)(d(p)),δYd​(p)​(z↦c(P(−t,d−1(z))))(d(p)),

where δhf(x)\delta_hf(x)δh​f(x) is a classically selected vector vvv for which lim⁡s→0, s≠0s−1(f(x+sh)−f(x))=v\lim_{s\to0,\ s\ne0}s^{-1}(f(x+sh)-f(x))=vlims→0, s=0​s−1(f(x+sh)−f(x))=v, if such a vector exists, and is an arbitrary selected vector otherwise. The domain Ω\OmegaΩ is not consulted in this formula: no local-flow property, membership of (t,m)(t,m)(t,m) or (−t,p)(-t,p)(−t,p) in Ω\OmegaΩ, membership of mmm in c.sourcec.\mathrm{source}c.source, smoothness of YYY, or existence of the directional limit is required.

Human review
  • Endorsed by Shuze Chen · Oct 31, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Oct 31, 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