Documentation

TauCeti.Analysis.Calculus.Morse.HandleRetraction.Basic

Milnor's retraction of a handle onto a lower sublevel with a cell attached #

Near a nondegenerate critical point of index λ with critical value c, a Morse function has the form f = c - ‖u‖² + ‖v‖², where u ∈ Eᵤ runs over the λ negative directions and v ∈ Eₛ over the positive ones; these are the unstable and stable directions of the negative gradient flow. Passing the critical level attaches the λ-cell e = {‖u‖² ≤ ε, v = 0} to the lower sublevel {f ≤ c - ε}. In the quadratic model, the union of the lower sublevel and the cell is TauCeti.quadraticSublevelUnionCell ε, consisting of the points (u, v) with -‖u‖² + ‖v‖² ≤ -ε or with ‖u‖² ≤ ε and v = 0. Equivalently, it is the set of points with ‖v‖ ≤ √(‖u‖² - ε) (TauCeti.mem_quadraticSublevelUnionCell_iff_norm_le), using that the real square root of a negative number is 0.

This file constructs Milnor's deformation r_t of the quadratic model onto that set. The end map TauCeti.quadraticHandleRetraction ε keeps u and retracts v radially onto the closed ball of radius √(‖u‖² - ε); its fixed points are exactly the points of TauCeti.quadraticSublevelUnionCell ε. The homotopy TauCeti.quadraticHandleDeformation ε moves v along the segment from this retraction back to v, and fixes TauCeti.quadraticSublevelUnionCell ε throughout. In Milnor's three cases: points with ‖u‖² ≤ ε move along (u, t v) and end on the cell, points with ε ≤ ‖u‖² ≤ ‖v‖² + ε end on the level -‖u‖² + ‖v‖² = -ε, and points of the lower sublevel stay fixed.

At every time the deformation multiplies v by a scalar in [0, 1], namely (1 - t) min 1 (√(‖u‖² - ε) / ‖v‖) + t (TauCeti.quadraticHandleDeformation_apply_eq_smul, TauCeti.exists_quadraticHandleDeformation_apply_eq_smul). This scalar depends only on ‖u‖ and ‖v‖, which is what lets the deformation run fibrewise in a sum of normed vector bundles. Hence it preserves every set whose fibres {v | (u, v) ∈ X} are star-shaped about the origin, and X ∩ quadraticSublevelUnionCell ε is a strong deformation retract of each such X (TauCeti.exists_inter_quadraticSublevelUnionCell_deformation). Such sets include every sublevel of a function which, for fixed u, does not decrease when v is scaled up; Milnor applies the deformation to the sublevel {F ≤ c - ε} of a function F = f - μ(‖u‖² + 2‖v‖²) modified near the critical point; that sublevel is {f ≤ c - ε} together with a handle around the cell.

Main declarations #

References #

The lower sublevel with the cell attached #

def TauCeti.quadraticSublevelUnionCell {Eᵤ : Type u_1} {Eₛ : Type u_2} [NormedAddCommGroup Eᵤ] [NormedAddCommGroup Eₛ] (ε : ℝ) :
Set (Eᵤ × Eₛ)

The lower sublevel {-‖u‖² + ‖v‖² ≤ -ε} of the quadratic model -‖u‖² + ‖v‖² of a Morse function at a critical point, together with the cell {‖u‖² ≤ ε, v = 0} of radius √ε in the negative directions u.

Instances For
    @[simp]
    theorem TauCeti.mem_quadraticSublevelUnionCell {Eᵤ : Type u_1} {Eₛ : Type u_2} [NormedAddCommGroup Eᵤ] [NormedAddCommGroup Eₛ] {ε : ℝ} {z : Eᵤ × Eₛ} :
    z ∈ quadraticSublevelUnionCell ε ↔ -‖z.1‖ ^ 2 + ‖z.2‖ ^ 2 ≤ -ε ∨ ‖z.1‖ ^ 2 ≤ ε ∧ z.2 = 0
    theorem TauCeti.mem_quadraticSublevelUnionCell_iff_norm_le {Eᵤ : Type u_1} {Eₛ : Type u_2} [NormedAddCommGroup Eᵤ] [NormedAddCommGroup Eₛ] {ε : ℝ} {z : Eᵤ × Eₛ} :

    A point lies in the lower quadratic sublevel with the cell attached exactly when its positive component has norm at most √(‖u‖² - ε).

    The retraction #

    noncomputable def TauCeti.quadraticHandleRetraction {Eᵤ : Type u_1} {Eₛ : Type u_2} [NormedAddCommGroup Eᵤ] [NormedAddCommGroup Eₛ] [NormedSpace ℝ Eₛ] (ε : ℝ) :
    C(Eᵤ × Eₛ, Eᵤ × Eₛ)

    Milnor's retraction of the quadratic model onto TauCeti.quadraticSublevelUnionCell ε: it keeps u and retracts v radially onto the closed ball of radius √(‖u‖² - ε).

    Instances For
      @[simp]
      theorem TauCeti.quadraticHandleRetraction_apply {Eᵤ : Type u_1} {Eₛ : Type u_2} [NormedAddCommGroup Eᵤ] [NormedAddCommGroup Eₛ] [NormedSpace ℝ Eₛ] (ε : ℝ) (z : Eᵤ × Eₛ) :

      The retraction takes values in the lower quadratic sublevel with the cell attached.

      The fixed points of the retraction are exactly the points of the lower quadratic sublevel with the cell attached.

      The deformation #

      Milnor's deformation of the quadratic model onto TauCeti.quadraticSublevelUnionCell ε: at time t it keeps u and replaces v by (1 - t) r(v) + t v, where r(v) is the radial retraction of v onto the closed ball of radius √(‖u‖² - ε). It is a homotopy from TauCeti.quadraticHandleRetraction ε to the identity which fixes TauCeti.quadraticSublevelUnionCell ε.

      Instances For
        @[simp]
        theorem TauCeti.quadraticHandleDeformation_apply {Eᵤ : Type u_1} {Eₛ : Type u_2} [NormedAddCommGroup Eᵤ] [NormedAddCommGroup Eₛ] [NormedSpace ℝ Eₛ] (ε : ℝ) (t : ↑unitInterval) (z : Eᵤ × Eₛ) :
        (quadraticHandleDeformation ε) (t, z) = (z.1, (1 - ↑t) • radialRetraction (√(‖z.1‖ ^ 2 - ε)) z.2 + ↑t • z.2)
        theorem TauCeti.quadraticHandleDeformation_apply_eq_smul {Eᵤ : Type u_1} {Eₛ : Type u_2} [NormedAddCommGroup Eᵤ] [NormedAddCommGroup Eₛ] [NormedSpace ℝ Eₛ] (ε : ℝ) (t : ↑unitInterval) (z : Eᵤ × Eₛ) :
        (quadraticHandleDeformation ε) (t, z) = (z.1, ((1 - ↑t) * min 1 (√(‖z.1‖ ^ 2 - ε) / ‖z.2‖) + ↑t) • z.2)

        At time t the deformation multiplies v by (1 - t) min 1 (√(‖u‖² - ε) / ‖v‖) + t.

        theorem TauCeti.mem_Icc_quadraticHandleDeformation_scalar {Eᵤ : Type u_1} {Eₛ : Type u_2} [NormedAddCommGroup Eᵤ] [NormedAddCommGroup Eₛ] (ε : ℝ) (t : ↑unitInterval) (z : Eᵤ × Eₛ) :
        (1 - ↑t) * min 1 (√(‖z.1‖ ^ 2 - ε) / ‖z.2‖) + ↑t ∈ Set.Icc 0 1

        The scalar (1 - t) min 1 (√(‖u‖² - ε) / ‖v‖) + t by which the deformation multiplies v lies in [0, 1].

        theorem TauCeti.exists_quadraticHandleDeformation_apply_eq_smul {Eᵤ : Type u_1} {Eₛ : Type u_2} [NormedAddCommGroup Eᵤ] [NormedAddCommGroup Eₛ] [NormedSpace ℝ Eₛ] (ε : ℝ) (t : ↑unitInterval) (z : Eᵤ × Eₛ) :
        ∃ s ∈ Set.Icc 0 1, (quadraticHandleDeformation ε) (t, z) = (z.1, s • z.2)

        At every time, the deformation keeps u and multiplies v by a scalar in [0, 1].

        theorem TauCeti.norm_snd_quadraticHandleDeformation_apply_le {Eᵤ : Type u_1} {Eₛ : Type u_2} [NormedAddCommGroup Eᵤ] [NormedAddCommGroup Eₛ] [NormedSpace ℝ Eₛ] (ε : ℝ) (t : ↑unitInterval) (z : Eᵤ × Eₛ) :

        The deformation never increases the norm of the positive component v.

        theorem TauCeti.quadraticHandleDeformation_mem_of_starConvex {Eᵤ : Type u_1} {Eₛ : Type u_2} [NormedAddCommGroup Eᵤ] [NormedAddCommGroup Eₛ] {z : Eᵤ × Eₛ} [NormedSpace ℝ Eₛ] {X : Set (Eᵤ × Eₛ)} (hX : ∀ (u : Eᵤ), StarConvex ℝ 0 {v : Eₛ | (u, v) ∈ X}) (ε : ℝ) (t : ↑unitInterval) (hz : z ∈ X) :

        The deformation preserves every set whose fibres {v | (u, v) ∈ X} are star-shaped about the origin.

        theorem TauCeti.exists_inter_quadraticSublevelUnionCell_deformation {Eᵤ : Type u_1} {Eₛ : Type u_2} [NormedAddCommGroup Eᵤ] [NormedAddCommGroup Eₛ] [NormedSpace ℝ Eₛ] {X : Set (Eᵤ × Eₛ)} (hX : ∀ (u : Eᵤ), StarConvex ℝ 0 {v : Eₛ | (u, v) ∈ X}) (ε : ℝ) :

        Milnor's handle retraction. If every fibre {v | (u, v) ∈ X} of X is star-shaped about the origin, then X ∩ quadraticSublevelUnionCell ε is a strong deformation retract of X: there is a retraction of X onto it, and a homotopy in X from this retraction to the identity which fixes X ∩ quadraticSublevelUnionCell ε. Both are restrictions of TauCeti.quadraticHandleRetraction ε and TauCeti.quadraticHandleDeformation ε.