* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Chapter 5 Homotopy Theory
Surface (topology) wikipedia , lookup
Michael Atiyah wikipedia , lookup
General topology wikipedia , lookup
Grothendieck topology wikipedia , lookup
Brouwer fixed-point theorem wikipedia , lookup
Algebraic K-theory wikipedia , lookup
Homotopy type theory wikipedia , lookup
Covering space wikipedia , lookup
Chapter 5 Homotopy Theory 5.1 Homotopy Groups • Let X and Y be topological spaces and f0 : X → Y, f1 : X → Y be two continuous maps. Then the map f0 is said to be homotopic to the map f1 if there is a continuous map F : X × [0, 1] → Y such that for any x ∈ X F(x, 0) = f0 (x), F(x, 1) = f1 (x). • In other words, if one can continuously deform the map f0 to the map f1 . • Homotopy is an equivalence relation. • We fix a point x0 ∈ X. Let f : X → Y and y0 = f (x0 ) ∈ Y. The homotopy class of f at the basepoint y0 is the set of all maps g : X → Y homotopic to f such that g(x0 ) = y0 [ f, y0 ] = {g : X → Y | g ∼ f, g(x0 ) = y0 } • The homotopy classes are invariant under homeomorphisms and are topological invariants. 135 136 CHAPTER 5. HOMOTOPY THEORY • The set [X, Y; y0 ] of all homotopy classes of maps from X to Y with the basepoint y0 is a topological invariant. • Usually one fixes a standard manifold X = S n and studies the homotopy classes of continuous maps f : S n → Y. These are topological invariants of Y. • The set of homotopy classes of maps f : S n → Y is called the n-th homotopy group πn (Y) of Y, πn (Y, y0 ) = [S n , Y; y0 ] • The set of homotopy classes of maps f : S 1 → Y is called the fundamental group π1 (Y, y0 ) of Y. 5.2 Fundamental Group • Let X be a topological space. A path with an initial point x0 and an endpoint x1 is a continuous map α : [0, 1] → X such that α(0) = x0 , α(1) = x1 . • A topological space X is arcwise connected if for any two points x, y ∈ X there is a path α with the initial point x and the endpoint y. • A loop with a base point x0 is a closed path such that α(0) = α(1) = x0 • A constant path c x0 is defined by c x0 (t) = x0 , t ∈ [0, 1]. • A topological space is simply connected if any loop in it can be continuously shrunk to a point. topicsdiffgeom.tex; November 4, 2014; 11:33; p. 134 5.2. FUNDAMENTAL GROUP 137 • The product γ = α ∗ β of two paths, α and β, such that α(1) = β(0), is a path defined by � α(2t), 0 ≤ t ≤ 12 γ(t) = β(2t − 1), 12 ≤ t ≤ 1 • The inverse α−1 of a path α is the path defined by, α−1 (t) = α(1 − t), 0 ≤ t ≤ 1. • Two loops α0 , α1 with a base point x0 are said to be homotopic, α0 ∼ α1 , if there is a continuous map F : [0, 1] × [0, 1] → X (called a homotopy between α0 and α1 ) such that F(t, 0) = α0 (t), 0 ≤ t ≤ 1, F(t, 1) = α1 (t), 0 ≤ t ≤ 1, F(0, s) = F(1, s) = x0 , 0 ≤ s ≤ 1, • Proposition. The loop homotopy is an equivalence relation. • Proof. • The equivalence classes of loops are denoted by [α] and are called homotopy classes. • The set π1 (X, x0 ) of homotopy classes of loops with base point x0 is called the fundamental group (or the first homotopy group) of X at x0 . • The product of homotopy classes is defined by [α] ∗ [β] = [α ∗ β] • The inverse of the homotopy class is defined by [α]−1 = [α−1 ] • The homotopy class [c x0 ] defines the identity element. • Proposition. Let α0 , α1 , β0 , β1 be loops based at x0 ∈ X. Suppose that α0 ∼ α1 and β0 ∼ β1 . Then −1 α−1 0 ∼ α1 α0 ∗ β0 ∼ α1 ∗ β1 topicsdiffgeom.tex; November 4, 2014; 11:33; p. 135 138 CHAPTER 5. HOMOTOPY THEORY • Proof. • Proposition. The product and the inverse of homotopy classes do not depend on the representatives. • Proof. • Theorem. The fundamental group is a group. • Proof. Prove 1) associativity, 2) unit element, 3) inverse. • A fundamental group is trivial if π1 (X, x0 ) = {e}. • Theorem. Let X be an arcwise connected topological space. Then for any two points x0 , x1 ∈ X, π1 (X, x0 ) is isomorphic to π1 (X, x1 ). • So, we can just denote it by π1 (X). • Proof. • Let α be a loop at x0 . • Let γ be a path from x0 ro x1 . • Then γ−1 ∗ α ∗ γ is a loop at x1 . • This induces a map σγ : π1 (X, x0 ) → π1 (X, x1 ). defined by σγ ([α]) = [γ−1 ∗ α ∗ γ] • It is easy to see that this map is a homomorphism. • This map has an inverse σ−1 γ : π1 (X, x1 ) → π1 (X, x0 ) defined by σγ ([β]) = [γ ∗ β ∗ γ−1 ] • Therefore, this map is bijective and is an isomorphism. topicsdiffgeom.tex; November 4, 2014; 11:33; p. 136 5.3. HOMOTOPY TYPE 5.3 139 Homotopy Type • Let X and Y be topological spaces. We say that X and Y are of the same homotopy type if there are two continuous maps f : X → Y, g:Y→X such that the maps f ◦ g : Y → Y, g◦ f : X → X are homotopic to identity maps, f ◦ g ∼ idY , g ◦ f ∼ idX . • The map f is called the homotopy equivalence and the map g is the homotopy inverse. • Proposition. Having the same homotopy type is an equivalence relation. • Proof. • Lemma. Let fi : X → Y, i = 0, 1, be maps between topological spaces X and Y and F : X × [0, 1] → Y be a homotopy between f0 and f1 . Let x0 ∈ X and γ be a path in Y with initial point f0 (x0 ) and endpoint f1 (x0 ). Let σγ : π1 (Y, f0 (x0 )) → π1 (Y, f1 (x0 )) be a group isomorphism defined by σγ ([α]) = [γ−1 ∗ α ∗ γ]. Then there are induced group homomorphisms fi,∗ : π1 (X, x0 ) → π1 (Y, fi (x0 )), defined by fi,∗ ([α], x0 ) = ([ f (α)], fi (x0 )), such that f1,∗ = σγ ◦ f0,∗ . topicsdiffgeom.tex; November 4, 2014; 11:33; p. 137 140 CHAPTER 5. HOMOTOPY THEORY • Theorem. Connected topological spaces with the same homotopy type have the same fundamental group. In other words: Let X and Y be acrwise connected topological spaces with the same homotopy type. Suppose there are maps f : X → Y, g : Y → X, f ◦ g ∼ id Y , g ◦ f ∼ id X . such that Then π1 (X, x0 ) � π1 (Y, f (x0 )), π1 (Y, y0 ) � π1 (X, g(y0 )), • Proof. We have by the Lemma, σγ ◦ (id Y )∗ = ( f ◦ g)∗ , σγ−1 ◦ (id X )∗ = (g ◦ f )∗ , • Since σγ is an isomorphism, so are ( f ◦ g)∗ and (g ◦ f )∗ . • Since ( f ◦ g)∗ = f∗ ◦ g∗ , (g ◦ f )∗ = g∗ ◦ f∗ , the maps f∗ and g∗ are isomorphisms, which proves the theorem. • Corollary. The fundamental group is a topological invariant. • Let R ⊂ X be a subspace of a topological space X. Then R is called a retract of X if there is a continuous map f : X → R (called a retraction) such that f |R = id R , that is, f (x) = x for any x ∈ R. Example. • Proposition. The retract R and the whole space X have the same fundamental group, for any x0 ∈ R π1 (R, x0 ) � π1 (X, x0 ). topicsdiffgeom.tex; November 4, 2014; 11:33; p. 138 5.3. HOMOTOPY TYPE 141 • The subspace R is called a deformation retract of X if it is a retract of X homotopic to the identity map id X . That is, there a homotopy H : X × [0, 1] → X between the retraction f and the identity map id X that leaves the points in R fixed, that is, H(·, 0) = id X , H(·, 1) = f, H(·, s)|R = id R , s ∈ [0, 1], or H(x, 0) = x, H(x, s) = x H(x, 1) = f (x) ∈ R, for any x ∈ R for any x ∈ X and any s ∈ [0, 1] • Examples. • A topological space X is contractible if a point x0 ∈ X is a deformation retract of X. • If X is contractible to a point x0 then there is a homotopy H : X × [0, 1] → X (called a contraction) such that H(x, 0) = x0 , H(x, 1) = id X (x) = x, H(x0 , t) = x0 for any x ∈ X for any t ∈ [0, 1] • Example. • An arcwise connected topological space is called simply connected if it has a trivial fundamental group. • Theorem. A contractible space is simply connected, that is, it has a trivial fundamental group. • Examples. • Theorem. The fundamental group of the circle S 1 = R/2πZ is isomorphic to Z, π1 (S 1 ) � Z. topicsdiffgeom.tex; November 4, 2014; 11:33; p. 139 142 CHAPTER 5. HOMOTOPY THEORY • One can identify the maps f : S 1 → S 1 such that f (1) = 1 with the maps f˜ : R → R such that f˜(0) = 0 and f˜(x + 2π) = f˜(x) + 2πn The integer n = deg f is equal to the degree of f . • Two maps f, g : S 1 → S 1 such that f (1) = g(1) are homotopic if and only if they have the same degree. • For any n ∈ Z there is a map f : S 1 → S 1 with degree n. • Therefore, there is a bijection, in fact, isomorphism, between π1 (S 1 , 1) and Z. • Theorem. Let X and Y be arcwise connected topological spaces and x0 ∈ X, y0 ∈ Y. Then π1 (X × Y, (x0 , y0 )) � π1 (X, x0 ) ⊕ π1 (Y, y0 ) • Examples. n π1 (T ) = n � Z i=1 π1 (S 1 × R) = Z ⊕ {e} � Z 5.4 Fundamental Groups of Polyhedra • If the space X is triangulable then one can compute π1 (X) by a routine procedure. • Let X be an arcwise connected topological space. • Let f : |K| → X be a triangulation of X, where K is a simplicial complex and |K| is a polyhedron |K|. • Then π1 (X) = π1 (|K|) • There is a systematic way to compute π1 (|K|). topicsdiffgeom.tex; November 4, 2014; 11:33; p. 140 5.4. FUNDAMENTAL GROUPS OF POLYHEDRA 143 • Any loop in |K| is made of 1-simplexes. • An edge path in a simplicial complex K is a sequence of vertices v0 v1 · · · vk , such that each consecutive pair vi vi+1 is either a 0-simplex or a 1-simplex. • An edge path is an edge loop at v0 if v0 = vk . • Two edge loops are equivalent if one is obtained from another by the following operations: 1. The edge path uvw is equivalent to the path uw if u, v, w span a 2simplex. 2. The edge path uvu is equivalent to the 0-simplex u. • the equivalence class of edge loops at a vertex v is denoted by {vv1 · · · vk−1 v} • We define the product of equivalence classes of edge loops at a vertex v by {vv1 · · · vk−1 v} ∗ {vu1 · · · u j−1 v} = {vv1 · · · vk−1 vu1 · · · u j−1 v} • The unit element is the equivalence class {v}. • The inverse equivalence class is defined by {vv1 · · · vk−1 v}−1 = {vvk−1 · · · v1 v} • Theorem. The set of equivalence classes of edge loops at a vertex v forms a group E(K, v), called the edge group of the simplicial complex K. • Theorem. The edge group of a simplicial complex K is isomorphic to the fundamental group of the polyhedron |K| E(K, v) � π1 (|K|, v) • Let L be a simplicial subcomplex of K such that L contains all vertices of K and the polyhedron |L| is simply connected. • For any path connected simplicial complex K there is such a subcomplex. topicsdiffgeom.tex; November 4, 2014; 11:33; p. 141 144 CHAPTER 5. HOMOTOPY THEORY • A one-dimensional simply connected simplicial complex is called a tree. • A tree is called a maximal tree if it is not a proper subset of other trees. • Proposition. A maximal tree of a simplicial complex contains all its vertices. • Let v0 , v1 , . . . vn be vertices of K. • For each ordered pair of vertices (vi , v j ) such that �vi v j � is a 1-simplex in K we assign an element gi j of a group. • Let G(K, L) be the group generated by gi j with the following relations 1. gi j = 1 2. if the �vi v j � ∈ L , gi j g jk gki = 1 if �vi v j vk � is a 2-simplex of K and there are no nontrivial loops around vi v j vk . • Proposition. gii = 1 (gi j )−1 = g ji • Theorem. The group G(K, L) is isomorphic to the edge group E(K, v) and, therefore, to the fundamental group π1 (|K|, v), G(K, L) � E(K, v) � π1 (|K|, v) . • Summary. To compute the fundamental group of X we need: 1. Find a triangulation f : |K| → X. 2. Find a simply connected subcomplex L of K that contains all vertices of K. 3. Assign a generator gi j to each 1-simplex �vi v j � ∈ K − L with i < j. topicsdiffgeom.tex; November 4, 2014; 11:33; p. 142 5.4. FUNDAMENTAL GROUPS OF POLYHEDRA 145 4. Impose a relation gi j g jk = gik if there is a 2-simplex �vi v j vk � such that i < j < k. If two of the vertices vi , v j , vk form a 1-simplex of L the generator should be set equal to 1. 5. The fundamental group π1 (X) is isomorphic to the group G(K, L) generated by {gi j } with the above relations. • The set K (p) of all i-simplexes for i = 0, 1, . . . , p of a simplicial complex K is called a p-skeleton of K. • It should be clear that the fundamental group of a polyhedron is determined by its 2-skeleton π1 (|K|) � π1 (|K (2) |) • For any n ≥ 2, a (n + 1)-simplex σn+1 and its boundary ∂σn+1 have the same 2-skeleton. • Examples. • Since σn+1 is contractible and ∂σn+1 is a polyhedron of S n , we obtain π1 (S n ) = {e} n≥2 The n-sphere for n ≥ 2 is simply connected • A triangulation of the circle is K = {v1 , v2 , v3 , �v1 v2 �, �v1 v3 �, �v2 v3 �} For the circle there is only one generator with no relations π1 (S 1 ) � (x; ∅) � Z • A n-bouquet M is the one-point union of n circles. π1 (M) = (x1 , . . . , xn ; ∅) • The 2-disk is simply connected π1 (D2 ) � {e} topicsdiffgeom.tex; November 4, 2014; 11:33; p. 143 146 CHAPTER 5. HOMOTOPY THEORY • For the 2-torus there are two generators with one relation π1 (T 2 ) � (x, y; xyx−1 y−1 ) � Z ⊕ Z • For the surface of genus g there are 2g generators with one relation g � π1 (Σg ) � x1 , . . . xg , y1 , . . . yg ; xi yi xi−1 y−1 i i=1 • For the real projective plane there is one generator with one relation π1 (RP2 ) � (x; x2 ) � Z2 The same is true for any n > 1 π1 (RPn ) � (x; x2 ) � Z2 • Exercises. • Show that for the Möbius strip M there is one generator with no relations π1 (M 2 ) � (x; ∅) � Z • Show that for the Klein bottle there are two generators with one relation π1 (K 2 ) � (x, y; xyxy−1 ) • Theorem. Let K be a connected simplicial complex. Then the first homology group H1 (K) is isomorphic to the quotient of the fundamental group π1 (|K|) by its commutator subgroup F H1 (K) � π1 (|K|)/F. • Corollary. The first homology group of a connected topological space is isomorphic to its fundamental group H1 (K) � π1 (|K|) if and only if the fundamental group is commutative. • Corollary. The first homology groups of two topological spaces X and Y of the same homotopy type are the same, H1 (X) = H1 (Y) topicsdiffgeom.tex; November 4, 2014; 11:33; p. 144 5.5. HIGHER HOMOTOPY GROUPS 5.5 147 Higher Homotopy Groups • Let I n be the unit cube [0, 1] × · · · × [0, 1] = {t1 , . . . , tn | ti ∈ [0, 1]} I n = �������������������������������������� n • The boundary of I n is ∂I n = {t1 , . . . , tn | ti ∈ [0, 1], t j = 0 or 1, for some j} • A map α : I n → X is called an n-loop at a point x0 ∈ X if it maps the boundary ∂I n to x0 . • A constant n-loop is a map c x0 (t1 , . . . , tn ) = x0 , for any ti ∈ [0, 1]. • The cube I n whose boundary ∂I n is shrunk to a point is homeomorphic to the sphere I n /∂I n � S n • Two n-loops at x0 , α0 and α1 , are homotopic if there is a continuous map, called a homotopy between α0 and α1 , F : In × I → X such that F(t, 0) = α0 (t), F(t, s) = x0 where t = (t1 , . . . , tn ). F(t, 1) = α1 (t) for t ∈ ∂I n , s ∈ [0, 1] • Homotopy is an equivalence relation. • The corresponding equivalence classes are called the homotopy classes. • The product of two n-loops at x0 is defined by � α(2t1 , t2 . . . , tn ), (α ∗ β)(t1 , t2 , . . . , tn ) = β(2t1 − 1, t2 . . . , tn ), 0 ≤ t1 ≤ 12 1 ≤ t1 ≤ 1 2 topicsdiffgeom.tex; November 4, 2014; 11:33; p. 145 148 CHAPTER 5. HOMOTOPY THEORY • The inverse of an n-loop is defined by α−1 (t1 , t2 , . . . , tn ) = α(1 − t1 , t2 , . . . , tn ) • Then α−1 ∗ α ∼ α ∗ α−1 ∼ c x0 • For any n ≥ 1 the set πn (X, x0 ) of all homotopy classes of n-loops at x0 is called the n-th homotopy group of X at x0 . • The product of homotopy classes is defined by [α] ∗ [β] = [α ∗ β] • This product does not depend on the representatives of the homotopy classes. • Theorem. The n-th homotopy groups are groups. • The set π0 (X) is not a group; it is rather the number of independent component of X. • Proposition. Higher homotopy groups πn (X, x0 ), for any n ≥ 2, are Abelian. • Proposition. For a connected space X the homotopy groups πn (X, x0 do not depend on the choice of the point x0 . • Proposition. If the (connected) spaces X and Y have the same homotopy type then the their homotopy groups are isomorphic πn (X) � πn (Y) • Corollary. The homotopy groups of a contractible space X are trivial πn (X) � {e} • Proposition. For any connected topological spaces X and Y πn (X × Y) � πn (X) ⊕ πn (Y) topicsdiffgeom.tex; November 4, 2014; 11:33; p. 146 5.6. UNIVERSAL COVERING SPACES 5.6 149 Universal Covering Spaces • A connected topological space X̃ is called the covering space of a topological space X if there is a continuous map f : X̃ → X such that 1. f is surjective, 2. for each x ∈ X there is a connected open set U ⊂ X containing x such that the inverse image f −1 (U) is a disjoint union of open sets Vi in X̃, � Vi , f −1 (U) = i each of which is homeomorphic to U, that is, f : Vi → U are homeomorphisms. • If the covering space X̃ is simply connected, then it is called the universal covering space of X. • If the spaces X and X̃ are topological groups and the map f is a group homomorphism, then X̃ is called the covering group. • The universal covering of the circle S 1 (or the group U(1)) is the real line R. • Theorem. Let X̃ be a universal covering space of a connected space X and f : X̃ → X be the covering map. Let x̃0 ∈ X̃ and x0 = f ( x̃0 ) ∈ X. Then for any n ≥ 2, the induced homomorphism f∗ : πn (X̃, x̃0 ) → πn (X, x0 ) is an isomorphism. • As an example, for any n ≥ 2. πn (S 1 ) = {e} topicsdiffgeom.tex; November 4, 2014; 11:33; p. 147 150 CHAPTER 5. HOMOTOPY THEORY • The sphere S n is a covering space of the real projective space RPn for n ≥ 2. • For any n ≥ 2 πn (RPn ) = {e} • Examples. RP3 = S O(3) • S 3 = S U(2) • Proposition. For any n ≥ 2 πn (S O(3)) = πn (RP3 ) = πn (S 3 ) = πn (S U(2)) • The universal covering group of the special orthogonal group S O(n) is called the spin group. • Examples. Spin (3) = S U(2) Spin (4) = S U(2) × S U(2) Spin (6) = S U(4) • Examples. πn (S n ) = πn (RPn ) = Z, n≥2 topicsdiffgeom.tex; November 4, 2014; 11:33; p. 148