Maps of disk pairs and relative homotopy groups #
The relative homotopy set π_{n+1}(Y, B, b) = TauCeti.RelHomotopyGroup (Fin n) _ is built from
relative cubes: maps I × Iⁿ → Y sending the face {0} × Iⁿ into B and the rest of the
boundary of the cube to b. The cells of a CW complex are instead attached along maps of pairs
(Dⁿ⁺¹, Sⁿ) → (Y, B) that need not be constant on any part of the sphere. This file connects the
two, and so states the compression lemma in its usual form (Hatcher, Lemma 4.6): if
π_n(Y, B, b) = 0 for every b ∈ B and every n such that a relative CW complex (C, D) has
n-cells, then every map (C, D) → (Y, B) is homotopic relative to D to a map into B
(TauCeti.exists_homotopicRel_range_subset_of_relHomotopyGroup).
- A map of pairs
g : (I × Iⁿ, ∂(I × Iⁿ)) → (Y, B)is homotopic, through maps of pairs, to a relative cube (TauCeti.exists_homotopicWith_relGenLoop). The open box{1} × Iⁿ ∪ I × ∂Iⁿcontracts to the centre of its top face; along the face{0} × Iⁿ, whose boundary lies in the open box, this contraction extends to a homotopy intoBbecause∂Iⁿhas the homotopy extension property inIⁿ. The resulting homotopy ofgon the boundary of the cube extends over the cube because that boundary has the homotopy extension property too (TauCeti.hasHomotopyExtensionProperty_prodBoundary). - If
π_{n+1}(Y, B, b)is trivial for everyb ∈ B, every map of pairs(Dⁿ⁺¹, Sⁿ) → (Y, B)is homotopic relative toSⁿto a map intoB(TauCeti.exists_homotopicRel_range_subset_of_subsingleton). HereDⁿ⁺¹andSⁿare the closed unit ball and the unit sphere of the sup norm onFin (n + 1) → ℝ, the domains of the characteristic maps ofTopology.RelCWComplex; they are the cubeI × Iⁿand its boundary. By the previous item and the triviality of the resulting relative cube, the map is homotopic through maps of pairs to a constant map with value inB, and sweeping that homotopy across the ball (ContinuousMap.HomotopicWith.exists_homotopicRel_range_subset) gives a homotopy relative to the sphere.
The hypothesis of the compression lemma for 0-cells is that every point of Y is joined by a
path to a point of B, which is what π_0(Y, B, b) = 0 means; relative homotopy sets are only
defined here in positive degrees.
Main declarations #
TauCeti.exists_homotopicWith_relGenLoop: a map of pairs out of the cubeI × Iⁿis homotopic through maps of pairs to a relative cube.TauCeti.exists_homotopyWith_relGenLoop_of_openBox: the same, by a homotopy that follows a prescribed contraction of the open box{1} × Iⁿ ∪ I × ∂Iⁿ, so that maps sharing a face of the cube can be deformed compatibly.TauCeti.exists_homotopicRel_range_subset_of_subsingleton: maps of disk pairs compress intoBwhen the relative homotopy sets vanish.TauCeti.exists_homotopicRel_range_subset_of_relHomotopyGroup: the compression lemma, with its hypotheses on relative homotopy sets.
References #
- A. Hatcher, Algebraic Topology, Section 4.1, Lemma 4.6 (the compression lemma) and the compression criterion preceding Theorem 4.3.
Deforming a map of pairs into a relative cube #
A map of pairs is homotopic to a relative cube along a contraction of the open box. Let
R be a deformation of the open box {1} × I^N ∪ I × ∂I^N of I × I^N inside the boundary of the
cube, starting at the identity and ending at the constant map to a point v. Every map of pairs
g : (I × I^N, ∂(I × I^N)) → (Y, B) is homotopic, through maps of pairs, to a relative cube at
g v, by a homotopy that is g ∘ R t on the open box at every time t.
A map of pairs (I × I^N, ∂(I × I^N)) → (Y, B) is homotopic, through maps of pairs, to a
relative cube. Its basepoint b ∈ B is the value of the given map at the centre of the face
{1} × I^N.
Compressing maps of disk pairs #
Compressing a map of disk pairs. Let B ⊆ Y be such that the relative homotopy set
π_{n+1}(Y, B, b) is trivial for every b ∈ B. Then every map from the closed unit ball of the
sup norm on Fin (n + 1) → ℝ to Y sending the unit sphere into B is homotopic, relative to the
unit sphere, to a map with values in B.
The compression lemma #
The compression lemma in terms of relative homotopy groups (Hatcher, Lemma 4.6). Let C
be a relative CW complex with base D, and let B ⊆ Y. Suppose that every point of Y is joined
by a path to a point of B if C has 0-cells, and that the relative homotopy set
π_{n+1}(Y, B, b) is trivial for every b ∈ B if C has (n + 1)-cells. Then every map
f : C → Y sending D into B is homotopic, relative to D, to a map into B.