Bourbaki

§ 5. Van kampen theorem

1. Coequalizer of the projections of a fibre square

Let X and Y be topological spaces and let ff be a continuous mapping of X into Y. Denote by Z the fibre square X×YXX\times_YX and by p1,p2p_1,p_2 the two projections of Z into X. Denote by W the fibre product X×YX×YXX\times_YX\times_YX; for every pair (s,t)(s, t) of integers equal to 1, 2 or 3, denote by qst:W→Zq_{st}: W\rightarrow Z the mapping defined by qst(x1,x2,x3)=(xs,xt)q_{st}(x_1, x_2, x_3) = (x_s, x_t).

Denote by Coeg(ff) the groupoid Coeg(ϖ(p1),ϖ(p2)\varpi (p_1), \varpi (p_2)), coequalizer of the two morphisms ϖ(p1),ϖ(p2)\varpi (p_1),\varpi (p_2) of the Poincaré groupoid ϖ(Z)\varpi (Z) into the Poincaré groupoid ϖ(X)\varpi (X). Denote by γ:ϖ(X)→\gamma :\varpi (X)\rightarrow Coeg(ff) the canonical morphism of groupoids. Since f∘p1f\circ p_1 = f∘p2f\circ p_2, the morphisms of groupoids ϖ(f)∘ϖ(p1)\varpi (f)\circ \varpi (p_1) and ϖ(f)∘ϖ(p2)\varpi (f)\circ \varpi (p_2) of ϖ(Z)\varpi (Z) into ϖ(Y)\varpi (Y) are equal; there thus exists a unique morphism of groupoids ϖ′(f):\varpi '(f): Coeg(ff)→ϖ(Y)\rightarrow \varpi (Y) such that ϖ′(f)∘γ=ϖ(f)\varpi '(f)\circ \gamma =\varpi (f).

Definition 1 0230

One says that the mapping ff satisfies the property (VK) if it is strict, surjective and if the morphism ϖ′(f)\varpi '(f) is an isomorphism.

Example 1 0231

This property is satisfied under one of the following assumptions:

(i) The spaces X and Y are delacable, the space X×YXX\times_YX is locally arcwise connected, the mapping ff is surjective, proper, separated, with locally connected fibres;

(ii) The spaces X and Y are delacable, the mapping ff is surjective, proper, separated, with finite fibres, the diagonal ΔX\Delta_X of X×YXX\times_YX is open and its complement is locally connected;

(iii) The spaces X and Y are delacable, the mapping ff is surjective, open and possesses the path lifting property;

(iv) Every point of Y possesses a neighbourhood above which there exists a continuous section of the mapping ff.

Indeed, under each of these assumptions, the mapping ff is surjective; it is also strict by virtue of I, p. 18, example 2. Finally, the morphism ϖ′(f)\varpi '(f) is an isomorphism, both under the assumptions (i), (ii) or (iii) according to IV, p. 398, th. 1, and under assumption (iv) ( IV, p. 400, th. 2).

The prop. 5 of II, p. 208 describes the isotropy groups of the groupoid Coeg(ff). The purpose of this noo^o is to make explicit, when ff satisfies the property (VK), the description of the Poincaré groups of Y which is deduced from it by composition with the isomorphism of groupoids ϖ′(f)\varpi '(f). The following nosos^{os} will be devoted to important particular cases.

Suppose that the mapping ff satisfies the property (VK).

Put I=π0(X),J=π0(Z),K=π0(W)\mathsf{I}=\pi_0(X),\mathsf{J}=\pi_0(Z),\mathsf{K}=\pi_0(W); for j∈Jj\in \mathsf{J} and s∈{1,2}s\in \{1,2\}, put is(j)=π0(ps)(j)i_s(j) =\pi_0(p_s)(j) ; for k∈Kk\in \mathsf{K} and s,t∈{1,2,3}s,t\in \{1,2,3\}, put jst(k)=π0(qst)(k)j_{st}(k) =\pi_0(q_{st})(k). Denote by Γ\mathsf{\Gamma} the skeleton of the pair (ϖ(p1),ϖ(p2))(\varpi (p_1), \varpi (p_2)) of morphisms of groupoids from ϖ(Z)\varpi (Z) into ϖ(X)\varpi (X) (II, p. 185, def. 3). It is the quiver (I,J,π0(p1),π0(p2))(\mathsf{I},\mathsf{J}, \pi_0(p_1), \pi_0(p_2)), since the orbits of the Poincaré groupoid of a topological space are the arcwise connected components of this space.

Suppose moreover that Y is arcwise connected and nonempty. According to II, p. 200, remark 2, π0(Γ)\pi_0(\mathsf{\Gamma}) is then in bijection with the set of orbits of the groupoid Coeg(ff); since the mapping ff satisfies the property (VK), the groupoid Coeg(ff) is isomorphic to the groupoid ϖ(Y)\varpi (Y). The graph Γ\mathsf{\Gamma} is thus connected and nonempty.

Call van Kampen data of ff the data of the following elements:

(i) for every i∈Ii\in \mathsf{I}, a point a(i)\mathsf{a}(i) of the path-connected component ii of X ;

(ii) for every j∈Jj\in \mathsf{J}, a point b(j)=(b1(j),b2(j))\mathsf{b}(j) = (\mathsf{b}_1(j),\mathsf{b}_2(j)) of the path-connected component jj of Z ;

(iii) for every k∈Kk\in \mathsf{K}, a point c(k)=(c1(k),c2(k),c3(k))\mathsf{c}(k) = (\mathsf{c}_1(k),\mathsf{c}_2(k),\mathsf{c}_3(k)) of the path-connected component kk of W;

(iv) for every j∈Jj\in \mathsf{J}, the class β1(j)\beta_1(j) of a path in X joining b1(j)\mathsf{b}_1(j) to a(i1(j))\mathsf{a}(i_1(j)) and the class β2(j)\beta_2(j) of a path in X joining b2(j)\mathsf{b}_2(j) to a(i2(j))\mathsf{a}(i_2(j));

(v) for every k∈Kk\in \mathsf{K} and for every pair (s,t)(s, t) equal to (1,2)(1,2), (2,3)(2,3) or (1,3)(1,3), the class γst(k)\gamma_{st}(k) of a path in Z joining (cs(k),ct(k))(\mathsf{c}_s(k),\mathsf{c}_t(k)) to b(jst(k))\mathsf{b}(j_{st}(k));

(vi) a subquiver T\mathsf{T} of Γ\mathsf{\Gamma} whose associated graph is a maximal tree of the graph Γ~\widetilde{\mathsf{\Gamma }};

(vii) an element i0i_0 of I\mathsf{I}.

Let us choose a van Kampen datum of ff. Then, (a,b,β1,β2,T,i0)(\mathsf{a},\mathsf{b}, \beta_1, \beta_2,\mathsf{T}, i_0) is a basic equipment of the pair (ϖ(p1),ϖ(p2))(\varpi (p_1), \varpi (p_2)) of groupoid morphisms from ϖ(Z)\varpi (Z) into ϖ(X)\varpi (X) (II, p. 192, déf. 4). Moreover, the triples

z=((q12(c(k)),1),(q23(c(k)),1),(q13(c(k)),−1)) \mathsf{z}= ((q_{12}(\mathsf{c}(k)),1),(q_{23}(\mathsf{c}(k)),1),(q_{13}(\mathsf{c}(k)),-1))

and the classes of paths (γ1,2(k),γ2,3(k),γ1,3(k))(\gamma_{1,2}(k), \gamma_{2,3}(k), \gamma_{1,3}(k)), where kk describes K\mathsf{K}, define a complementary equipment of the pair (ϖ(p1),ϖ(p2))(\varpi (p_1), \varpi (p_2)) (II, p. 208, déf. 3 ; II, p. 205, exemple ; II, p. 205, remarque). We shall say that the complete equipment of the coequalizer Coeg(ff) thus defined is deduced from the van Kampen datum of ff that we have chosen.

For every j∈Jj\in \mathsf{J}, let φj:π1(Z,b(j))→π1(X,a(i1(j)))\varphi_j:\pi_1(Z,\mathsf{b}(j))\rightarrow \pi_1(X,\mathsf{a}(i_1(j))) and ψj:π1(Z,b(j))→π1(X,a(i2(j)))\psi_j:\pi_1(Z,\mathsf{b}(j))\rightarrow \pi_1(X,\mathsf{a}(i_2(j))) denote the group homomorphisms defined by

φj=\varphi_j= Int(β1(j)\beta_1(j))−1∘(p1)∗,v↦β1(j)−1((p1)∗(v))β1(j)^{-1}\circ (p_1)_*,v\mapsto \beta_1(j)^{-1}((p_1)_*(v))\beta_1(j)

(1)

ψj=\psi_j= Int(β2(j)\beta_2(j))−1∘(p2)∗,v↦β2(j)−1((p2)∗(v))β2(j)^{-1}\circ (p_2)_*,v\mapsto \beta_2(j)^{-1}((p_2)_*(v))\beta_2(j),

for v∈π1(Z,b(j))v\in \pi_1(Z,\mathsf{b}(j)). For every k∈Kk\in \mathsf{K} and every s∈{1,2,3}s\in \{1,2,3\}, let λs(k)\lambda_s(k) denote the class of a loop at the point a(is(k))\mathsf{a}(i_s(k)) in X defined by

(2)

λ1(k)=β1(j13(k))−1⋅((p1)∗(γ13(k)))−1⋅((p1)∗(γ12(k)))⋅β1(j12(k)) \lambda_1(k) =\beta_1(j_{13}(k))^{-1}\cdot ((p_1)_*(\gamma_{13}(k)))^{-1}\cdot ((p_1)_*(\gamma_{12}(k)))\cdot \beta_1(j_{12}(k))

λ2(k)=β2(j12(k))−1⋅((p2)∗(γ12(k)))−1⋅((p1)∗(γ23(k)))⋅β1(j23(k))\lambda_2(k) =\beta_2(j_{12}(k))^{-1}\cdot ((p_2)_*(\gamma_{12}(k)))^{-1}\cdot ((p_1)_*(\gamma_{23}(k)))\cdot \beta_1(j_{23}(k)),

λ3(k)=β2(j23(k))−1⋅((p2)∗(γ23(k)))−1⋅((p2)∗(γ13(k)))⋅β2(j13(k)) \lambda_3(k) =\beta_2(j_{23}(k))^{-1}\cdot ((p_2)_*(\gamma_{23}(k)))^{-1}\cdot ((p_2)_*(\gamma_{13}(k)))\cdot \beta_2(j_{13}(k))

Let τ\tau denote the unique groupoid morphism from Grp(Γ\mathsf{\Gamma}) into ϖ(Y)\varpi (Y) such that the mapping Som(τ\tau ) maps i∈Ii\in \mathsf{I} onto f(a(i))f(\mathsf{a}(i)) and Fl(τ\tau ) maps j∈Jj\in \mathsf{J} onto the class of paths f∗(β1(j))−1f∗(β2(j))f_*(\beta_1(j))^{-1}f_*(\beta_2(j)) joining f(a(i1(j)))f(\mathsf{a}(i_1(j)))

a(i1(k)) \mathsf{a}(i_1(k))
λ1(k) \lambda_1(k)
b1(j12(k))b1(j13(k)) \mathsf{b}_1(j_{12}(k))\mathsf{b}_1(j_{13}(k))
b2(j12(k))b2(j13(k)) \mathsf{b}_2(j_{12}(k))\mathsf{b}_2(j_{13}(k))
c1(k) \mathsf{c}_1(k)
c2(k)c3(k) \mathsf{c}_2(k)\mathsf{c}_3(k)
λ2(k)λ3(k) \lambda_2(k)\lambda_3(k)
a(i2(k))b1(j23(k))b2(j23(k))a(i3(k)) \mathsf{a}(i_2(k))\mathsf{b}_1(j_{23}(k))\mathsf{b}_2(j_{23}(k))\mathsf{a}(i_3(k))

to f(a(i2(j)))f(\mathsf{a}(i_2(j))) in Y. For i∈Ii\in \mathsf{I}, let di∈d_i\in Grp(Γ\mathsf{\Gamma}) be the class of the unique path without backtracking joining i0i_0 to ii in the tree T~\widetilde{\mathsf{T}} and put δi=\delta_i= τ(di)\tau (d_i).

If S is a set, recall that F(S) denotes the free group on S; the image in F(S) of an element s∈Ss\in S by the canonical mapping is denoted by [s][s], or by ss if there is no possible confusion.

Theorem 1 0232

Suppose that Y is path connected and that ff verifies the property (VK). With the preceding notations, there exists a unique group homomorphism

L:(i∗∈Iπ1(X,a(i)))∗F(J)→π1(Y,f(a(i0))) \mathsf{L}:(_i*_{\in\mathsf{I}}\pi_1(X,\mathsf{a}(i)))*F(\mathsf{J})\rightarrow \pi_1(Y, f(\mathsf{a}(i_0)))

such that

L(v)=δif∗(v)δi−1\mathsf{L}(v) =\delta_if_*(v)\delta^{-1}_i for i∈Ii\in \mathsf{I} and v∈π1(X,a(i))v\in \pi_1(X,\mathsf{a}(i)),

L(j)=δi1(j)τ(j)δi2(j)−1\mathsf{L}(j) =\delta_{i_1(j)}\tau (j)\delta^{-1}_{i_2(j)} for j∈Jj\in \mathsf{J}.

Moreover, the homomorphism L\mathsf{L} is surjective and its kernel is the smallest normal subgroup containing the following elements:

(R1R_1)r1(j)=j\mathsf{r}_1(j) =j for jj in Fl(T\mathsf{T}) ;

(R2R_2)r2(j,v)=φj(v)jψj(v)−1j−1\mathsf{r}_2(j, v) =\varphi_j(v)j\psi_j(v)^{-1}j^{-1} for j∈Jj\in \mathsf{J} and v∈π1(Z,b(j))v\in \pi_1(Z,\mathsf{b}(j));

(R3R_3)r3(k)=λ1(k)j12(k)λ2(k)j23(k)λ3(k)j13(k)−1\mathsf{r}_3(k) =\lambda_1(k)j_{12}(k)\lambda_2(k)j_{23}(k)\lambda_3(k)j_{13}(k)^{-1}, for k∈Kk\in \mathsf{K}.

The existence and uniqueness of the homomorphism L\mathsf{L} follow from the universal property of free products and free groups. Moreover, this homomorphism is the composite of the homomorphism ϖ′(f)γ(a(i0))\varpi '(f)_{\gamma(\mathsf{a}(i_0))} deduced from ϖ′(f)\varpi '(f) by passing to isotropy groups and of the group homomorphism λ\lambda defined by Prop. 5 of II, p. 208, taking into account the fact that the chosen van Kampen datum determines a complete equipment of the pair (ϖ(p1),ϖ(p2))(\varpi (p_1), \varpi (p_2)) of groupoid morphisms from ϖ(Z)\varpi (Z) into ϖ(X)\varpi (X). By that proposition, the image of the homomorphism λ\lambda is the group Coeg(ff)γ(a(i0))_{\gamma(\mathsf{a}(i_0))} and its kernel is the smallest normal subgroup containing the elements defined by relations (R1R_1),(R2), (R_2) and (R3R_3). On the other hand, the groupoid morphism ϖ′(f):\varpi '(f): Coeg(ff)→ϖ(Y)\rightarrow \varpi (Y) is an isomorphism, by definition of property (VK). The theorem is therefore proved.

2. Coverings

Let Y be a nonempty arcwise connected topological space, and let (Ai)i∈I(A_i)_{i\in I} be a covering of Y by nonempty arcwise connected subsets, indexed by a totally ordered set I. Let X=⨆i∈IAiX =\bigsqcup_{i\in I}A_i be the sum space of the family (Ai)(A_i) and let f:X→Yf: X\rightarrow Y be the mapping deduced from the family of canonical injections of each AiA_i into Y. Suppose that the mapping ff satisfies property (VK). This occurs in particular in the following two cases:

(i) the interiors of the sets AiA_i, for i∈Ii\in I, cover Y (cf. IV, p. 402, Example 2);

(ii) the space Y is contractible, as are the spaces AiA_i, for i∈Ii\in I, the family (Ai)i∈I(A_i)_{i\in I} is locally finite, the AiA_i are closed in Y and their pairwise intersections are locally arcwise connected (cf. IV, p. 399, Example 1).

Let J′J' be the set of triples (i,i′,V)(i, i',V), where ii and i′i' are elements of I and where V is an arcwise connected component of Ai∩Ai′A_i\cap A_{i'}. If j=(i,i′,V)∈J′j= (i, i',V)\in J', we set i1(j)=i,i2(j)=i′i_1(j) =i,i_2(j) =i' and j‾=(i′,i,V)\overline{j}= (i', i,V). Let J be the subset of J′J' formed by the triples such that i<i′i < i'. The framework of the covering is the quiver Γ whose set of vertices is I, whose set of arrows is J, and whose origin and end mappings are respectively the mappings j↦i1(j)j\mapsto i_1(j) and j↦i2(j)j\mapsto i_2(j). We identify the graph associated with Γ with the graph Γ~\widetilde{\Gamma} whose set of vertices is I, whose set of arrows is J∪J‾J\cup \overline{J}, whose origin and end mappings are the mappings j↦i1(j)j\mapsto i_1(j) and j↦i2(j)j\mapsto i_2(j), and whose involution is the mapping j↦j‾j\mapsto \overline{j}.

Let p1p_1 and p2p_2 denote the projections of the fibre square X×YXX\times_YX onto X; let Γ\mathsf{\Gamma} be the framework of the pair (ϖ(p1),ϖ(p2))(\varpi (p_1), \varpi (p_2)) of groupoid morphisms from ϖ(X×YX)\varpi (X\times_YX) into ϖ(X)\varpi (X).

Lemma 1 0233

The quiver Γ identifies with a subquiver of Γ\mathsf{\Gamma}; the quivers Γ and Γ\mathsf{\Gamma} are connected.

The arcwise connected components of X are the AiA_i, for i∈Ii\in I. The arcwise connected components of X×YXX\times_YX are the (V×{i})×Y(V\times \{i\})\times_Y (V×{i′})(V\times \{i'\}), for (i,i′,V)∈J′(i, i',V)\in J'. Consequently, the frame Γ\mathsf{\Gamma} of the pair (ϖ(p1),ϖ(p2))(\varpi (p_1), \varpi (p_2)) is isomorphic to the quiver whose set of vertices is I, whose set of arrows is J′J', the origin and term mappings being respectively the mappings j↦i1(j)j\mapsto i_1(j) and j↦i2(j)j\mapsto i_2(j). The quiver Γ\mathsf{\Gamma} is connected (IV, p. 406). Moreover, this description identifies Γ with a subquiver of Γ\mathsf{\Gamma}. Let us also observe that for every arrow jj of Γ\mathsf{\Gamma}, either j∈j\in Fl(Γ~\widetilde{\Gamma}), or there exists i∈Ii\in I such that j=(i,i,Ai)j= (i, i,A_i). It follows that the mapping π0(Γ)→π0(Γ)\pi_0(\Gamma )\rightarrow \pi_0(\mathsf{\Gamma}) deduced from the injection of Γ into Γ\mathsf{\Gamma} is bijective, so that Γ is connected.

For every element ii of I, choose a point a(i)a(i) of AiA_i.

For every element j=(i,i′,V)j= (i, i',V) of J, choose a point b(j)b(j) in V, a path B1(j)B_1(j) joining b(j)b(j) to a(i)a(i) in AiA_i and a path B2(j)B_2(j)

joiningj‾=(i′,i,b(jV)\overline{j}= (i', i,b(jV)) tobelongs toa(i′)a(i') in AJi′J^{i'}. Letand let us putj=(i,ib(′,j‾V)j= (i, ib(',\overline{j}V)) =an element ofb(j)b(j), B1(j)=BJ‾2B_1(j) = B\overline{J}_2; then(j‾)(\overline{j}) and B2(j)=B1(j‾)B_2(j) = B_1(\overline{j}). For j∈J′∪J′‾j\in J'\cup \overline{J'}, the paths B1(j)‾\overline{B_1(j)} and B2(j)B_2(j) in Y are composable. Put

B(j)=B1(j)∗B2(j)(3) B(j) = B_1(j)*B_2(j) \tag{3}

It is a path joining a(i1(j))a(i_1(j)) to a(i2(j))a(i_2(j)) in Y; one has the relation B(j‾)=B(j)B(\overline{j}) = B(j).

For every j=(i,i′,V)∈J′j= (i, i',V)\in J', denote by pj,1:V→Aip_{j,1}: V\rightarrow A_i and pj,2:V→Ai′p_{j,2}: V\rightarrow A_{i'} the canonical injections; also denote by φj:π1(V,b(j))→π1(Ai,a(i))\varphi_j:\pi_1(V, b(j))\rightarrow \pi_1(A_i, a(i)) and ψj:π1(V,b(j))→π1(Ai′,a(i′))\psi_j:\pi_1(V, b(j))\rightarrow \pi_1(A_{i'}, a(i')) the group homomorphisms defined by

φj(v)=[B1(j)]−1(pj,1)∗(v)[B1(j)] \varphi_j(v) = [B_1(j)]^{-1}(p_{j,1})_*(v)[B_1(j)]

and

ψj(v)=[B2(j)]−1(pj,2)∗(v)[B2(j)] \psi_j(v) = [B_2(j)]^{-1}(p_{j,2})_*(v)[B_2(j)]

for v∈π1(V,b(j))v\in \pi_1(V, b(j)) (cf. IV, p. 407).

Fix an element i0i_0 of I, as well as a subquiver T in the quiver Γ whose associated graph T~\widetilde{T} is a maximal tree of the graph Γ~\widetilde{\Gamma}.

For i∈Ii\in I, let (i0,j1,i1,...,jn,i)(i_0, j_1, i_1, . . . , j_n, i) be the unique path without backtracking joining i0i_0 to ii in the tree T~\widetilde{T}, and put

δ(i)=[B(j1)][B(j2)]...[B(jn)] \delta (i) = [B(j_1)][B(j_2)]. . .[B(j_n)]

this is the class of a path joining a(i0)a(i_0) to a(i)a(i) in Y. Let αi\alpha_i denote the homomorphism of π1(Ai,a(i))\pi_1(A_i, a(i)) into π1(Y,a(i))\pi_1(Y, a(i)) deduced from the canonical injection, and let μi:π1(Ai,a(i))→π1(Y,a(i0))\mu_i:\pi_1(A_i, a(i))\rightarrow \pi_1(Y, a(i_0)) be the group homomorphism defined by

μi(v)=δ(i)αi(v)δ(i)−1 \mu_i(v) =\delta (i)\alpha_i(v)\delta (i)^{-1}

Finally, let μ:F(J)→π1(Y,a(i0))\mu: F(J)\rightarrow \pi_1(Y, a(i_0)) be the unique group homomorphism such that one has

μ(j)=δ(i1(j))[B(j)]δ(i2(j))−1 \mu(j) =\delta (i_1(j))[B(j)]\delta (i_2(j))^{-1}

for every j∈Jj\in J. There exists a unique group homomorphism

M:(i∗∈Iπ1(Ai,a(i)))∗F(J)→π1(Y,a(i0)) \mathsf{M}:(_i*_{\in I}\pi_1(A_i, a(i)))*F(J)\rightarrow \pi_1(Y, a(i_0))

which coincides with μi\mu_i on π1(Ai,a(i))\pi_1(A_i, a(i)), for every i∈i\in I, and with μ\mu on F(J).

Let K′K' be the set of quadruples (i1,i2,i3,U)(i_1, i_2, i_3,U), where i1,i2,i3i_1, i_2, i_3 are elements of I and where U is a pathwise connected component of Ai1∩Ai2∩Ai3A_{i_1}\cap A_{i_2}\cap A_{i_3}. For every element k=(i1,i2,i3,U)k= (i_1, i_2, i_3,U) of K′K' and every pair (s,t)(s, t) of elements of {1,2,3}\{1,2,3\}, put jst(k)=(is,it,V)j_{st}(k) = (i_s, i_t,V), where V is the pathwise connected component of Ais∩AitA_{i_s}\cap A_{i_t} which contains U; this is an element of J′J'.

Let K be the subset of K′K' formed by the quadruples (i1,i2,i3,U)(i_1, i_2, i_3,U) such that i1<i2<i3i_1< i_2< i_3. For every element k=(i1,i2,i3,U)k= (i_1, i_2, i_3,U) of K, choose a point c(k)c(k) of U, together with paths C12(k)C_{12}(k), C23(k)C_{23}(k) and C13(k)C_{13}(k), such that Cst(k)C_{st}(k) joins c(k)c(k) to b(jst(k))b(j_{st}(k)) in Ais∩AitA_{i_s}\cap A_{i_t} for s,t∈{1,2,3}s, t\in \{1,2,3\} with s<ts < t.

Put then, for k∈Kk\in K,

L1(k)=B1(j13(k))∗C13(k)‾∗C12(k)∗B1(j12(k)) L_1(k) = B_1(j_{13}(k))*\overline{C_{13}(k)}*C_{12}(k)*B_1(j_{12}(k))

(4) LL32((kk)=B)=B22((jj2312((kk))))∗∗CC1223((kk))‾∗∗CC1323((kk))∗∗BB21((jj1323((kk))))L^2_3((kk) = B) = B^2_2((jj^{12}_{23}((kk))))**\overline{CC^{1223}((kk))}**CC^{23}_{13}((kk))**BB^1_2((jj^{23}_{13}((kk)))),.

For s∈{1,2,3}s\in \{1,2,3\}, let λs(k)\lambda_s(k) denote the class in π1(Ais,a(is))\pi_1(A_{i_s}, a(i_s)) of the loop Ls(k)L_s(k).

Proposition 1 0234

Let Y be a pathwise connected topological space and let (Ai)i∈I(A_i)_{i\in I} be a covering of Y by non-empty, pathwise connected subsets, indexed by a totally ordered set I. Suppose that the canonical mapping of the sum space of the family (Ai)i∈I(A_i)_{i\in I} into Y satisfies property (VK). Then the homomorphism M\mathsf{M} introduced above is surjective, and its kernel is the smallest normal subgroup containing the following elements:

(R1R_1)r1(j)=jr_1(j) =j for jj in Fl(T);

(R2R_2)r2(j,v)=φj(v)jψj(v)−1j−1r_2(j, v) =\varphi_j(v)j\psi_j(v)^{-1}j^{-1}

for j=(i,i′,V)∈Jj= (i, i',V)\in J and v∈π1(V,b(j))v\in \pi_1(V, b(j));

(R3R_3)r3(k)=λ1(k)j12(k)λ2(k)j23(k)λ3(k)j13(k)−1r_3(k) =\lambda_1(k)j_{12}(k)\lambda_2(k)j_{23}(k)\lambda_3(k)j_{13}(k)^{-1}

for k∈Kk\in K.

Let X be the topological sum of the family (Ai)i∈I(A_i)_{i\in I} and let f:X→Yf: X\rightarrow Y be the mapping induced by the family of canonical injections of each AiA_i into Y. The mapping ff satisfies property (VK) by hypothesis; we shall therefore apply Theorem 1 of Noo1^o1 to it. We therefore take up again the notation of this Noo^o and begin by defining a van Kampen datum for the mapping ff.

The arcwise connected components of X are the sets Xi=X_i= Ai×{i}A_i\times \{i\} for i∈Ii\in I. Thus π0(X)\pi_0(X) is identified with the set I. For each i∈Ii\in I, let a(i)\mathsf{a}(i) denote the point (a(i),i)(a(i), i) of XiX_i.

Put Z=X×YXZ = X\times_YX. The arcwise connected components of Z are the sets Zj=(V×{i})×Y(V×{i′})Z_j= (V\times \{i\})\times_Y(V\times \{i'\}), where j=(i,i′,V)j= (i, i',V) runs through J′J'. Thus π0(Z)\pi_0(Z) is identified with the set J′J'. Let J0J_0 be the set of elements of J of the form (i,i,Ai)(i, i,A_i), so that the family (J0,J,J)(J_0,J,J) is a partition of J′J'. For i∈Ii\in I and j=(i,i,Ai)∈J0j= (i, i,A_i)\in J_0, put b(j)=(a(i),a(i))\mathsf{b}(j) = (\mathsf{a}(i),\mathsf{a}(i)) and take for β1(j)\beta_1(j) and β2(j)\beta_2(j) the class of the constant path at a(i)\mathsf{a}(i). For j=(i,i′,V)∈J∪J‾j= (i, i',V)\in J\cup \overline{J}, let b(j)\mathsf{b}(j) denote the point ((b(j),i),(b(j),i′))((b(j), i),(b(j), i')) of ZjZ_j and let β1(j)\beta_1(j) and β2(j)\beta_2(j) denote the classes of the paths t↦(B1(j)(t),i)t\mapsto (B_1(j)(t), i) and t↦(B2(j)(t),i′)t\mapsto (B_2(j)(t), i') in X.

Put W=X×YX×YXW = X\times_YX\times_YX. The arcwise connected components of W are the sets Wk=(U×{i1})×Y(U×{i2})×Y(U×{i3})W_k= (U\times \{i_1\})\times_Y(U\times \{i_2\})\times_Y(U\times \{i_3\}), where k=(i1,i2,i3,U)k= (i_1, i_2, i_3,U) runs through K′K'. Thus π0(W)\pi_0(W) is identified with the set K′K'.

Let K0K_0 denote the set of elements of K′K' of the form k=(i,i,i,Ai)k= (i, i, i,A_i), for i∈Ii\in I. For such an element k∈K0k\in K_0, put c(k)=(a(i),a(i),a(i))\mathsf{c}(k) = (\mathsf{a}(i),\mathsf{a}(i),\mathsf{a}(i)) and choose for γst(k)\gamma_{st}(k) the class of the constant path at (a(i),a(i))(\mathsf{a}(i),\mathsf{a}(i)).

Let K1K_1 denote the set of elements of K′K' of the form kk = (i1,i2,i3,V)(i_1, i_2, i_3,V) for which the set {i1,i2,i3}\{i_1, i_2, i_3\} has two elements. Let kk be an element of K′K' of the form (i,i,i′,V)(i, i, i',V), so that j=(i,i′,V)j= (i, i',V) belongs to J∪J‾J\cup \overline{J}. We then put

c(k)=((b(j),i),(b(j),i),(b(j),i′)),γ12(k)=(β1(j),β1(j)) \mathsf{c}(k) = ((b(j), i),(b(j), i),(b(j), i')),\gamma_{12}(k) = (\beta_1(j), \beta_1(j))

and we take for γ13(k)\gamma_{13}(k) and γ23(k)\gamma_{23}(k) the class of the constant path at b(j)\mathsf{b}(j). One defines analogously c(k),γ12(k),γ13(k),γ23(k)c(k),\gamma_{12}(k),\gamma_{13}(k),\gamma_{23}(k) for every element kk of K′K'.

For every element k=(i1,i2,i3,U)k= (i_1, i_2, i_3,U) of K, put

c(k)=((c(k),i1),(c(k),i2),(c(k),i3)) \mathsf{c}(k) = ((c(k), i_1),(c(k), i_2),(c(k), i_3))

for every pair (s,t)(s, t) of distinct elements of {1,2,3}\{1,2,3\}, take for γst(k)\gamma_{st}(k) the image under the mapping x↦((x,s),(x,t))x\mapsto ((x, s),(x, t)) of the class of the path Cst(k)C_{st}(k) in Ais(k)∩Ait(k)A_{i_s(k)}\cap A_{i_t(k)}.

For every point x=(x1,x2,x3)x= (x_1, x_2, x_3) of X×YX×YXX\times_YX\times_YX and every permutation σ∈S3\sigma \in \mathfrak{S}_3, put σ(x)=(xσ−1(1),xσ−1(2),xσ−1(3))\sigma (x) = (x_{\sigma^{-1}(1)}, x_{\sigma^{-1}(2)}, x_{\sigma^{-1}(3)}). Thus an operation of the group S3\mathfrak{S}_3 on W is defined. For every k=(i1,i2,i3,U)∈K′k= (i_1, i_2, i_3,U)\in K' and every permutation σ∈S3\sigma \in \mathfrak{S}_3, put analogously σ(k)=(iσ−1(1),iσ−1(2),iσ−1(3),U)\sigma (k) = (i_{\sigma^{-1}(1)}, i_{\sigma^{-1}(2)}, i_{\sigma^{-1}(3)},U); one has

σ(Wk)=(U×{iσ−1(1)})×Y(U×{iσ−1(2)})×Y(U×{iσ−1(3)})=Wσ(k) \sigma (W_k) = (U\times \{i_{\sigma^{-1}(1)}\})\times_Y(U\times \{i_{\sigma^{-1}(2)}\})\times_Y(U\times \{i_{\sigma^{-1}(3)}\}) = W_{\sigma(k)}

Let k=(i1,i2,i3,U)k= (i_1, i_2, i_3,U) be an element of K′K' such that i1,i2,i3i_1, i_2, i_3 are pairwise distinct. There exists a unique permutation σ∈\sigma \in S3\mathfrak{S}_3 such that iσ−1(1)<iσ−1(2)<iσ−1(3)i_{\sigma^{-1}(1)}< i_{\sigma^{-1}(2)}< i_{\sigma^{-1}(3)}, so that σ(k)=\sigma (k) = (iσ−1(1),iσ−1(2),iσ−1(3),U)(i_{\sigma^{-1}(1)}, i_{\sigma^{-1}(2)}, i_{\sigma^{-1}(3)},U) belongs to K. For s∈{1,2,3}s\in \{1,2,3\}, one then puts cs(k)=cσ(s)(σ(k))c_s(k) =c_{\sigma(s)}(\sigma (k)) and c(k)=(c1(k),c2(k),c3(k))c(k) = (c_1(k), c_2(k), c_3(k)), so that c(k)=σ−1(c(σ(k)))c(k) =\sigma^{-1}(c(\sigma (k))). For (s,t)∈{(1,2),(1,3),(2,3)}(s, t)\in \{(1,2),(1,3),(2,3)\}, one defines Cst(k)=Cσ−1(s)σ−1(t)(σ(k))C_{st}(k) = C_{\sigma^{-1}(s)\sigma^{-1}(t)}(\sigma (k)); it is a path joining c(k)c(k) to b(jσ(s)σ(t)(σ(k)))=b(jst(k))b(j_{\sigma(s)\sigma(t)}(\sigma (k))) =b(j_{st}(k)).

Let gg denote the morphism of quivers from Γ into Γ\mathsf{\Gamma} which associates with a vertex i∈Ii\in I of Γ the vertex Xi=Ai×{i}X_i= A_i\times \{i\} of Γ\mathsf{\Gamma} and with an arrow j=(i,i′,V)∈J′j= (i, i',V)\in J' of Γ the arrow Zj=(V×{i})×Y(V×{i′})Z_j= (V\times \{i\})\times_Y(V\times \{i'\}) of Γ\mathsf{\Gamma}. The mapping Som(gg) is bijective; the mapping Fl(gg) is injective and the graph Γ is connected (IV, p. 410, lemma 1), and the image under gg of the subquiver T is a subquiver T\mathsf{T} of Γ\mathsf{\Gamma} whose associated graph is a maximal tree of the graph Γ~\widetilde{\mathsf{\Gamma }}.

The points a(i)\mathsf{a}(i), for i∈Ii\in I, the points b(j)\mathsf{b}(j), for j∈J′j\in J', the points c(k)\mathsf{c}(k), for k∈K′k\in K', the classes of paths β1(j)\beta_1(j) and β2(j)\beta_2(j), for j∈J′j\in J', the classes of paths γst(k)\gamma_{st}(k), for k∈K′k\in K', the subquiver g(T)g(T) of Γ\mathsf{\Gamma} and the element i0i_0 of I define a van Kampen datum for ff.

Let ρ\rho denote the unique group homomorphism

ρ:(i∗∈Iπ1(X,a(i)))∗F(J′)→(i∗∈Iπ1(Ai,a(i)))∗F(J) \rho :(_i*_{\in I}\pi_1(X,\mathsf{a}(i)))*F(J')\rightarrow (_i*_{\in I}\pi_1(A_i, a(i)))*F(J)

which induces the isomorphism of π1(X,a(i))\pi_1(X,\mathsf{a}(i)) onto π1(Ai,a(i))\pi_1(A_i, a(i)) deduced from the identification of Ai×{i}A_i\times \{i\} and AiA_i, for all i∈Ii\in I, and such that one has

ρ(j)=1\rho (j) = 1 for j∈J0j\in J_0

ρ(j)=j,ρ(j‾)=j−1\rho (j) =j,\rho (\overline{j}) =j^{-1} for j∈Jj\in J.

Let L\mathsf{L} be the group homomorphism defined in Theorem 1 of IV, p. 408. For j=(i,i,Ai)∈J0j= (i, i,A_i)\in J_0, one has L(j)=1=M∘ρ(j)\mathsf{L}(j) = 1 =\mathsf{M}\circ \rho (j). Let j=(i,i′,V)j= (i, i',V) be an element of J; one has L(j)=(M∘ρ)(j)\mathsf{L}(j) = (\mathsf{M}\circ \rho )(j) by definition. Finally, if j=j= (i,i′,V)(i, i',V) is an element of J‾,j‾∈J\overline{J},\overline{j}\in J and one verifies that

L(j)=L(j‾)−1=M(ρ(j‾))−1=M(ρ(j)) \mathsf{L}(j) =\mathsf{L}(\overline{j})^{-1}=\mathsf{M}(\rho (\overline{j}))^{-1}=\mathsf{M}(\rho (j))

Consequently, one has M∘ρ=L\mathsf{M}\circ \rho =\mathsf{L}.

The homomorphism L\mathsf{L} is surjective (loc. cit.), hence the homomorphism M\mathsf{M} is also. Since the homomorphism ρ\rho is surjective, the kernel of M\mathsf{M} is the smallest normal subgroup of (i∗∈Iπ1(Ai,a(i)))∗F(J1)(_i*_{\in I}\pi_1(A_i, a(i)))*F(J_1) which contains the images under ρ\rho of the elements defined by relations (R1R_1), (R2R_2),(R3), (R_3) of theorem 1 (IV, p. 408). The proof will be complete once we have verified that these images are, besides the elements defined by relations (R1R_1),(R2),(R3), (R_2), (R_3) of Prop. 1, elements conjugate to them, or conjugate to their inverses, together with the identity element.

Elements R1R_1. — An arrow of the oriented tree T\mathsf{T} is of the form ZjZ_j, with j=(i,i′,V)∈Jj= (i, i',V)\in J; its image is the element jj of F(J).

Elements R2R_2. — Let j=(i,i′,V)∈J′j= (i, i',V)\in J'. If i=i′i=i', we have ρ(r2(j,v))=1\rho (\mathsf{r}_2(j, v)) = 1 for every v∈π1(Ai,a(i))v\in \pi_1(A_i,\mathsf{a}(i)). If j∈Jj\in J, the image of r2(j,v)\mathsf{r}_2(j, v) is the element r2(j,v)=φj(v)jψj(v)−1j−1r_2(j, v) =\varphi_j(v)j\psi_j(v)^{-1}j^{-1}, for every v∈π1(Z,b(j))v\in \pi_1(Z,\mathsf{b}(j)). In the remaining case, we have j‾∈J\overline{j}\in J and the equality

ρ(r2(j,v))=ρ(φj(v)jψj(v)−1j−1) \rho (\mathsf{r}_2(j, v)) =\rho (\varphi_j(v)j\psi_j(v)^{-1}j^{-1})
=[B1(j)]−1v[B1(j)]ρ(j)[B2(j)]−1v−1[B2(j)]ρ(j)−1 = [B_1(j)]^{-1}v[B_1(j)]\rho (j)[B_2(j)]^{-1}v^{-1}[B_2(j)]\rho (j)^{-1}
=[B2(j‾)]−1v[B2(j‾)]j‾−1[B1(j‾)]−1v−1[B1(j‾)]j‾ = [B_2(\overline{j})]^{-1}v[B_2(\overline{j})]\overline{j}^{-1}[B_1(\overline{j})]^{-1}v^{-1}[B_1(\overline{j})]\overline{j}

implies that ρ(r2(j,v))\rho (\mathsf{r}_2(j, v)) is conjugate to ρ(r2(j,v−1))\rho (\mathsf{r}_2(j, v^{-1})).

Elements R3R_3. — Let k=(i1,i2,i3,U)k= (i_1, i_2, i_3,U) be an element of K′K'.

If k∈K0,i1k\in K_0,i_1 = i2i_2 = i3,λs(k)i_3,\lambda_s(k) is the class of the trivial path for every s∈{1,2,3},jst(k)∈J0s\in \{1,2,3\},j_{st}(k)\in J_0 for every pair (s,t)∈(s, t)\in {(1,2),(1,3),(2,3)}\{(1,2),(1,3),(2,3)\}. Then ρ(r3(k))\rho (\mathsf{r}_3(k)) is the identity element.

Suppose k∈K1k\in K_1. If i1=i2i_1=i_2, then j=(i1,i3,U)∈J∪J‾j= (i_1, i_3,U)\in J\cup \overline{J} and we have

r3(k)=β1(j)−1j12β1(j)j23β2(j)−1β2(j)j13−1 \mathsf{r}_3(k) =\beta_1(j)^{-1}j_{12}\beta_1(j)j_{23}\beta_2(j)^{-1}\beta_2(j)j_{13}^{-1}

whose image under ρ\rho is the identity element. The other cases are dealt with analogously.

Suppose that i1,i2,i3i_1, i_2, i_3 are pairwise distinct. If i1<i2<i3i_1< i_2< i_3, k∈Kk\in K and the image of ρ(r3(k))\rho (\mathsf{r}_3(k)) is the element r3(k)r_3(k).

Let σ∈S3\sigma \in \mathfrak{S}_3 be the permutation carrying 1 to 2 and 2 to 3. We have

λ1(σ(k))=β1(j13(σ(k)))−1⋅p1,∗(γ13(σ(k)))−1⋅ \lambda_1(\sigma (k)) =\beta_1(j_{13}(\sigma (k)))^{-1}\cdot p_{1,*}(\gamma_{13}(\sigma (k)))^{-1}\cdot
⋅p1,∗(γ12(σ(k)))⋅β1(j12(σ(k))) \cdot p_{1,*}(\gamma_{12}(\sigma (k)))\cdot \beta_1(j_{12}(\sigma (k)))
=β1(j32(k))−1⋅p1,∗(γ32(k))−1⋅p1,∗(γ31(k))⋅β1(j31(k)) =\beta_1(j_{32}(k))^{-1}\cdot p_{1,*}(\gamma_{32}(k))^{-1}\cdot p_{1,*}(\gamma_{31}(k))\cdot \beta_1(j_{31}(k))
=β2(j23(k))−1⋅p2,∗(γ2,3(k))−1⋅p2,∗(γ13(k))⋅β2(j13(k)) =\beta_2(j_{23}(k))^{-1}\cdot p_{2,*}(\gamma_{2,3}(k))^{-1}\cdot p_{2,*}(\gamma_{13}(k))\cdot \beta_2(j_{13}(k))
=λ3(k) =\lambda_3(k)

One verifies analogously that λ2(σ(k))=λ1(k)\lambda_2(\sigma (k)) =\lambda_1(k) and λ3(σ(k))=λ2(k)\lambda_3(\sigma (k)) =\lambda_2(k). Hence,

ρ(r3(σ(k)))=λ1(σ(k))ρ(j12(σ(k)))λ2(σ(k))⋅ \rho (\mathsf{r}_3(\sigma (k))) =\lambda_1(\sigma (k))\rho (j_{12}(\sigma (k)))\lambda_2(\sigma (k))\cdot
⋅ρ(j23(σ(k)))λ3(σ(k))ρ(j13(σ(k)))−1 \cdot \rho (j_{23}(\sigma (k)))\lambda_3(\sigma (k))\rho (j_{13}(\sigma (k)))^{-1}
=λ3(k)ρ(j31(k))λ1(k)ρ(j12(k))λ2(k)ρ(j32(k))−1 =\lambda_3(k)\rho (j_{31}(k))\lambda_1(k)\rho (j_{12}(k))\lambda_2(k)\rho (j_{32}(k))^{-1}
=λ3(k)j13(k)−1λ1(k)j12(k)λ2(k)j23(k) =\lambda_3(k)j_{13}(k)^{-1}\lambda_1(k)j_{12}(k)\lambda_2(k)j_{23}(k)

which proves that ρ(r3(σ(k)))\rho (\mathsf{r}_3(\sigma (k))) is conjugate to

λ1(k)j12(k)λ2(k)j23(k)λ3(k)j13(k)−1=ρ(r3(k)) \lambda_1(k)j_{12}(k)\lambda_2(k)j_{23}(k)\lambda_3(k)j_{13}(k)^{-1}=\rho (\mathsf{r}_3(k))

Let τ∈S3\tau \in \mathfrak{S}_3 be the transposition with support {1,2}\{1,2\}. We have

λ1(τ(k))=λ2(k)−1,λ2(τ(k))=λ1(k)−1\lambda_1(\tau (k)) =\lambda_2(k)^{-1},\lambda_2(\tau (k)) =\lambda_1(k)^{-1} and λ3(τ(k))=λ3(k)−1\lambda_3(\tau (k)) =\lambda_3(k)^{-1}. The equalities

ρ(r3(τ(k)))=λ2(k)−1ρ(j21(k))λ1(k)−1ρ(j13(k))λ3(k)−1ρ(j23(k))−1 \rho (\mathsf{r}_3(\tau (k))) =\lambda_2(k)^{-1}\rho (j_{21}(k))\lambda_1(k)^{-1}\rho (j_{13}(k))\lambda_3(k)^{-1}\rho (j_{23}(k))^{-1}
=(ρ(j23(k))λ3(k)ρ(j13(k))−1λ1(k)ρ(j12(k))λ2(k))−1 =(\rho (j_{23}(k))\lambda_3(k)\rho (j_{13}(k))^{-1}\lambda_1(k)\rho (j_{12}(k))\lambda_2(k))^{-1}

show that ρ(r3(τ(k)))\rho (\mathsf{r}_3(\tau (k))) is conjugate to the inverse of

λ1(k)ρ(j12(k))−1λ2(k)ρ(j23(k))λ3(k)ρ(j13(k))−1=ρ(r3(k)) \lambda_1(k)\rho (j_{12}(k))^{-1}\lambda_2(k)\rho (j_{23}(k))\lambda_3(k)\rho (j_{13}(k))^{-1}=\rho (\mathsf{r}_3(k))

Since the group S3\mathfrak{S}_3 is generated by the permutations τ\tau and σ\sigma, it follows that, for every k∈Kk\in K and every σ∈S3,ρ(r3(σ(k)))\sigma \in \mathfrak{S}_3,\rho (\mathsf{r}_3(\sigma (k))) is conjugate to ρ(r3(k))\rho (\mathsf{r}_3(k)) or to its inverse.

Thus Proposition 1 is proved.

Corollary 1 0235

Under the assumptions of proposition 1, suppose moreover that, for every i∈i\in I, the image of the homomorphism from π1(Ai,a(i))\pi_1(A_i, a(i)) into π1(Y,a(i))\pi_1(Y, a(i)), induced by the canonical injection of AiA_i into Y, is trivial. The homomorphism M′:F(J)→π1(Y,a(i0))\mathsf{M}': F(J)\rightarrow \pi_1(Y, a(i_0)) induced by M\mathsf{M} by restriction is then surjective, and its kernel is the smallest distinguished subgroup containing the elements j∈j\in Fl(T) and the elements j12(k)j23(k)j13(k)−1j_{12}(k)j_{23}(k)j_{13}(k)^{-1} for k∈Kk\in K.

Let π:(i∗∈Iπ1(Ai,a(i)))∗F(J)→F(J)\pi : (_i*_{\in I}\pi_1(A_i, a(i)))*F(J)\rightarrow F(J) be the unique homomorphism which induces the trivial homomorphism on each π1(Ai,a(i))\pi_1(A_i, a(i)) and the identity on F(J); it is surjective. Let i∈Ii\in I. The definition of M\mathsf{M} and the assumption that the image of the homomorphism from π1(Ai,a(i))\pi_1(A_i, a(i)) into π1(Y,a(i))\pi_1(Y, a(i)) induced by the canonical injection is trivial imply that, for every v∈π1(Ai,a(i)),M(v)v\in \pi_1(A_i, a(i)),\mathsf{M}(v) is the identity element of π1(Y,a(i0))\pi_1(Y, a(i_0)). Hence M=M′∘π\mathsf{M}=\mathsf{M}'\circ \pi. It follows that the homomorphism M′\mathsf{M}' is surjective, and its kernel is the smallest distinguished subgroup containing the images under π\pi of the elements r1(j),r2(j,v)r_1(j),r_2(j, v) and r3(k)r_3(k) defined in Prop. 1. For j∈j\in Fl(T), we have π(r1(j))=j\pi (r_1(j)) =j. For every j∈j\in J and every v∈π1(V,b(j))v\in \pi_1(V, b(j)), we have π(r2(j,v))=e\pi (r_2(j, v)) =e. Finally, for every k=(i1,i2,i3,U)∈Kk= (i_1, i_2, i_3,U)\in K and every s∈s\in {1,2,3},λs(k)\{1,2,3\},\lambda_s(k) is the class of a loop in AisA_{i_s}; hence π(λs(k))=e\pi (\lambda_s(k)) =e, so that π(r3(k))=j12(k)j23(k)j13(k)−1\pi (r_3(k)) =j_{12}(k)j_{23}(k)j_{13}(k)^{-1}. The corollary follows.

Corollary 2 0236

Under the assumptions of proposition 1, suppose moreover that for every i∈Ii\in I, the group π1(Ai,a(i))\pi_1(A_i, a(i)) is reduced to the identity element and that, for every triple (i1,i2,i3)(i_1, i_2, i_3) of pairwise distinct elements of I, the set Ai1∩Ai2∩Ai3A_{i_1}\cap A_{i_2}\cap A_{i_3} is empty. The homomorphism M′′:\mathsf{M}'': F(J-Fl(T)) →π1(Y,a(i0))\rightarrow \pi_1(Y, a(i_0)) induced by M\mathsf{M} by restriction is then an isomorphism.

Let π′:F(J)→\pi ': F(J)\rightarrow F(J - Fl(T)) be the homomorphism which maps [j][j] to [j][j] if j∈j\in J-Fl(T) and which maps [j][j] to the identity element if j∈j\in Fl(T); it is surjective and one has M′′∘π′=M′\mathsf{M}''\circ \pi '=\mathsf{M}', where M′\mathsf{M}' is the surjective homomorphism defined in corollary 1. It follows that M′′\mathsf{M}'' is surjective and that its kernel is the smallest distinguished subgroup of F(J-Fl(T)) which contains the images under π′\pi ' of the elements described in loc. cit. But, by construction, π′(j)=e\pi '(j) =e if j∈j\in Fl(T) and the set K is empty, by hypothesis. The homomorphism M′′\mathsf{M}'' is therefore an isomorphism.

Example 1 0237

For the case of a covering formed of two sets, see No. 3.

Example 2 0238

Let G be a graph (II, p. 155, definition 1); let S denote the set of vertices of G, A the set of its oriented edges, oo and tt the origin and term mappings of A into S; for every oriented edge a∈Aa\in A, let a‾\overline{a} denote the opposite oriented edge. Endow the sets S and A with the discrete topology; let X be the sum space of the space S and of the space I×A\mathbf{I}\times A, and let ∼\sim be the finest equivalence relation in X for which (u,a)∼(1−u,a)(u, a)\sim (1-u, a), (0,a)∼o(a)(0, a)\sim o(a) and (1,a)∼t(a)(1, a)\sim t(a) for every u∈Iu\in \mathbf{I} and every oriented edge a∈Aa\in A. The quotient space ∣G∣=X/∼|G|= X/\sim is called the geometric realization of the graph G. Let pp denote the canonical projection of X onto ∣G∣|G|.

Let us prove that ∣G∣|G| is locally contractible. Let s∈Ss\in S. Let XsX_s denote the union of {s}\{s\} and of the sets [0,1[×{a}[0,1[\times \{a\} for a∈o−1(s)a\in \overset{-1}{o}(s) and of the

sets ]0,1]×{a}]0,1]\times \{a\} for a∈t−1(s)a\in \overset{-1}{t}(s). Let UsU_s be the image of XsX_s in ∣G∣|G|; it is an open neighbourhood of p(s)p(s) in ∣G∣|G| because XsX_s is an open saturated neighbourhood of ss in X. Let ff be the mapping of Xs×IX_s\times \mathbf{I} into XsX_s defined, for u,v∈Iu, v\in \mathbf{I} and a∈Aa\in A, by the relations

f(s,v)=s f(s, v) =s

((1−v)u,a)((1-v)u, a) if 0⩽u<10\leqslant u <1 and o(a)=so(a) =s,

f((u,a),v)= f((u, a), v) =

(1−(1−v)(1−u),a)(1-(1-v)(1-u), a) if 0<u⩽10< u\leqslant 1 and t(a)=st(a) =s.

It is continuous and compatible with the equivalence relation ∼\sim. It therefore defines by passing to the quotient a mapping φs:Us×I→Us\varphi_s: U_s\times \mathbf{I}\rightarrow U_s which is continuous, since I\mathbf{I} is locally compact (I, p. 19, Prop. 10). It is a strong contraction of UsU_s onto p(s)p(s) (III, p. 237, Def. 6).

Let moreover x=(τ,a)∈]0,1[×Ax= (\tau , a)\in ]0,1[\times A. Let us denote by Xx=]0,1[×{a,a}X_x= ]0,1[\times \{a, a\} and let UxU_x be its image in ∣G∣|G| by pp; it is a neighbourhood of p(x)p(x) in ∣G∣|G|, homeomorphic to ]0,1[]0,1[. The mapping of Xx×IX_x\times \mathbf{I} into XxX_x given by ((u,a),v)↦((1−v)u+vτ,a)((u, a), v)\mapsto ((1-v)u+v\tau , a) and ((u,a),t)↦((1−v)u+v(1−τ),a)((u, a), t)\mapsto ((1-v)u+v(1-\tau ), a) is continuous. It defines by passing to the quotient a mapping φx:Ux×\varphi_x: U_x\times I→Ux\mathbf{I}\rightarrow U_x which is continuous (loc. cit.) and is a strong contraction at xx.

Every point of ∣G∣|G| is the image of a point s∈Ss\in S or of a point of I×A\mathbf{I}\times A of the form (τ,a)(\tau , a) where 0<τ<10< \tau <1. It follows that every point of ∣G∣|G| has a neighbourhood contractible at that point. In other words, ∣G∣|G| is locally contractible and, in particular, unravelable (IV, p. 346, Prop. 5).

Let us choose a total order on S and consider the open covering (Us)s∈S(U_s)_{s\in S} of ∣G∣|G|. Let ss and s′s' be distinct elements of S. The intersection Us∩Us′U_s\cap U_{s'} is the union of the p(]0,1[,a)p(]0,1[, a) where aa runs through the set of oriented edges of G whose extremities are ss and s′s'. Consequently, the framework of the covering (Us)s∈S(U_s)_{s\in S} is identified with the quiver G. Moreover, for every s∈Ss\in S, UsU_s is contractible at ss, hence π1(Us,p(s))\pi_1(U_s, p(s)) is reduced to the identity element. It then follows from Cor. 2 of IV, p. 416 that for every s∈S,π1(∣G∣,p(s))s\in S,\pi_1(|G|, p(s)) is a free group (Nielsen-Schreier theorem).

Corollary 3 0239

Under the hypotheses of Prop. 1, suppose in addition that the framework of the covering (Ai)i∈I(A_i)_{i\in I} is an oriented tree. The homomorphism N:i∗∈Iπ1(Ai,a(i))→π1(Y,a(i0))\mathsf{N}:_i*_{\in I}\pi_1(A_i, a(i))\rightarrow \pi_1(Y, a(i_0)) deduced from M\mathsf{M} by restriction is then surjective; its kernel is the smallest normal subgroup of i∗∈Iπ1(Ai,a(i))_i*_{\in I}\pi_1(A_i, a(i)) which contains the elements φj(v)ψj(v)−1\varphi_j(v)\psi_j(v)^{-1}, for

every j=(i1,i2,V)∈Jj= (i_1, i_2,V)\in J and every v∈π1(V,b(j))v\in \pi_1(V, b(j)).

If the arcwise connected components of the intersections Ai∩Ai′A_i\cap A_{i'}, for i≠i′i=\not i' are moreover simply connected, the homomorphism N\mathsf{N} is then an isomorphism.

Under the hypotheses of the corollary, the graph associated with the quiver Γ is a tree, whence T = Γ. It follows that the image of the group F(J) by the homomorphism M\mathsf{M} is reduced to the identity element.

Let

ρ:(i∗∈Iπ1(Ai,a(i)))∗F(J)→i∗∈Iπ1(Ai,a(i)) \rho :(_i*_{\in I}\pi_1(A_i, a(i)))*F(J)\rightarrow_i*_{\in I}\pi_1(A_i, a(i))

be the unique group homomorphism which induces the identity homomorphism on π1(Ai,a(i))\pi_1(A_i, a(i)) and whose kernel contains F(J). We have M=N∘ρ\mathsf{M}=\mathsf{N}\circ \rho. Consequently, the homomorphism N\mathsf{N} is surjective and its kernel is the smallest normal subgroup of i∗∈Iπ1(Ai,a(i))_i*_{\in I}\pi_1(A_i, a(i)) which

contains the images under ρ\rho of the elements defined by relations (R1R_1), (R2R_2) and (R3R_3) of Prop. 1. We have ρ(j)=1\rho (j) = 1 for every j∈j\in Fl(Γ). Since the framework of the covering (Ai)i∈I(A_i)_{i\in I} is a tree, one has Ai1∩Ai2∩Ai3=∅A_{i_1}\cap A_{i_2}\cap A_{i_3}=\emptyset for every triple (i1,i2,i3)(i_1, i_2, i_3) of distinct elements of I. It follows that the kernel of N\mathsf{N} is the smallest normal subgroup containing the elements φj(v)ψj(v)−1\varphi_j(v)\psi_j(v)^{-1} for every j=(i1,i2,V)∈Jj= (i_1, i_2,V)\in J and every v∈π1(V,b(j))v\in \pi_1(V, b(j)).

Example 3 (Plane deprived of nn points) 023A

The fundamental group of R2\mathbf{R}^2 {0}\{0\} is isomorphic to Z\mathbf{Z}, and the class of the path t↦e2πitt\mapsto e^{2\pi it} is a generator of it (IV, p. 347, corollary). More generally, let nn be a natural number and let A={z1,...,zn}A =\{z_1, . . . , z_n\} be a set of nn points of R2\mathbf{R}^2. Let Y be the space R2−\mathbf{R}^2- A; we shall prove that the fundamental group of Y is isomorphic to the free group FnF_n on nn generators. For every ii, put zi=(ui,vi)z_i= (u_i, v_i). Replacing ziz_i if necessary by f(zi)f(z_i), where f:R2→R2f:\mathbf{R}^2\rightarrow \mathbf{R}^2 is a homeomorphism of the form (u,v)↦(u+αv,v)(u, v)\mapsto (u+\alpha v, v), one may suppose that the abscissae of the ziz_i are pairwise distinct. There is likewise no loss in supposing that one has u1<⋯<unu_1<\cdots < u_n.

Put V1=]−∞,u2[×R,Vi=]ui−1,ui+1[×RV_1= ]-\infty , u_2[\times \mathbf{R},V_i= ]u_{i-1}, u_{i+1}[\times \mathbf{R} for 2⩽i⩽n−12\leqslant i\leqslant n-1, Vn=]un−1,+∞[×RV_n= ]u_{n-1},+\infty [\times \mathbf{R}. For 1⩽i⩽n1\leqslant i\leqslant n, the set Ui=Vi−{zi}U_i= V_i-\{z_i\} is open in the plane and homeomorphic to R2−{0}\mathbf{R}^2-\{0\}. The family (Ui)1⩽i⩽n(U_i)_{1\leqslant i\leqslant n} is an open covering of the space Y=R2−Y =\mathbf{R}^2- A. The intersection Ui∩UjU_i\cap U_j is empty for ∣i−j∣⩾2|i-j|\geqslant 2, homeomorphic to R2\mathbf{R}^2 for ∣i−j∣=1|i-j|= 1. By Corollary 3, the fundamental group of Y is isomorphic to the free group FnF_n.

Let aa be a point of Y. For every integer i∈{1,...,n}i\in \{1, . . . , n\}, let rir_i be a real number strictly positive and strictly less than the distances from ziz_i to the points zjz_j, for j=i̸j=\not i. Let us denote by viv_i the class of the loop t↦t\mapsto zi+rie2πitz_i+r_ie^{2\pi it} at the point zi+riz_i+r_i of Y. Let θi\theta_i be the class of a path γi\gamma_i joining the point aa to the point zi+riz_i+r_i in Y. If the paths γi\gamma_i are injective and if their images meet only at aa, one can prove that the unique homomorphism of the free group F(t1,...,tn)F(t_1, . . . , t_n) into π1(Y,a)\pi_1(Y, a) such that φ(ti)=θiviθi−1\varphi (t_i) =\theta_iv_i\theta^{-1}_i for every i∈{1,...,n}i\in \{1, . . . , n\} is a group isomorphism.

An analogous reasoning makes it possible to prove that for every discrete closed subset A of the plane, the fundamental group of R2−\mathbf{R}^2- A is isomorphic to F(A) (IV, p. 463, Exercise 1).

Corollary 4 023B

Under the hypotheses of Proposition 1, suppose that there exists a subset A of Y, connected by arcs and nonempty, such that the intersection Ai∩Ai′A_i\cap A_{i'} is equal to A for every pair (i,i′)(i, i') of distinct elements of I. Let aa be a point of A. There exists a unique homomorphism φ\varphi of the sum of the family of groups (π1(Ai,a))i∈I(\pi_1(A_i, a))_{i\in I} amalgamated by π1(A,a)\pi_1(A, a) into π1(Y,a)\pi_1(Y, a) which coincides with the homomorphism induced by the canonical injection of AiA_i into Y, for every i∈Ii\in I. The homomorphism φ\varphi is an isomorphism.

In particular, if the group π1(A,a)\pi_1(A, a) is reduced to the identity element, the canonical homomorphism of the free product of the family of groups (π1(Ai,a))i∈I(\pi_1(A_i, a))_{i\in I} into π1(Y,a)\pi_1(Y, a) is an isomorphism.

For i∈i\in I, let us denote by gi:π1(A,a)→π1(Ai,a)g_i:\pi_1(A, a)\rightarrow \pi_1(A_i, a) and fi:π1(Ai,a)→f_i:\pi_1(A_i, a)\rightarrow π1(Y,a)\pi_1(Y, a) the canonical homomorphisms induced by the inclusions of A into AiA_i and of AiA_i into Y. Let us denote by ∗π1(Ai,a)*\pi_1(A_i, a) the sum of the groups π1(Ai,a)\pi_1(A_i, a) amalgamated by π1(A,a)\pi_1(A, a). Let alsoAp^Ap be the unique homomorphism of i∗∈Iπ1(Ai,a)_i*_{\in I}\pi_1(A_i, a) into ∗Aπ1(Ai,a)*_A\pi_1(A_i, a) which induces the identity on π1(Ai,a)\pi_1(A_i, a) (A, I, p. 80, prop. 4). The homomorphism pp is surjective and it follows from the definition of the monoid ∗π1(Ai,a)*\pi_1(A_i, a) that its kernel is the smallest distinguished subgroup ofiA∗∈Iπ1(Ai,a)^A_i*_{\in I}\pi_1(A_i, a) which contains the

elements gi(v)gi′(v)−1g_i(v)g_{i'}(v)^{-1}, for i,i′∈Ii, i'\in I and v∈π1(A,a)v\in \pi_1(A, a).

The homomorphisms fi∘gi:π1(A,a)→π1(Y,a)f_i\circ g_i:\pi_1(A, a)\rightarrow \pi_1(Y, a), for i∈Ii\in I, are equal. It therefore follows from the universal property of amalgamated sums of monoids (A, I, p. 80, Prop. 4) that there exists a unique group homomorphism φ\varphi (resp. ff) from ∗Aπ1(Ai,a)*_A\pi_1(A_i, a) (resp. from i∗∈Iπ1(Ai,a)_i*_{\in I}\pi_1(A_i, a)) into π1(Y,a)\pi_1(Y, a) which induces the homomorphism fif_i on π1(Ai,a)\pi_1(A_i, a). We have f=f= φ∘p\varphi \circ p.

For i∈i\in I, let uiu_i be the canonical injection of A into AiA_i and viv_i the canonical injection of AiA_i into Y. Let also ww be the canonical injection of A into Y. For every i∈i\in I, one has vi∘uiv_i\circ u_i = ww, hence π1(vi,a)∘π1(ui,a)=π1(w,a)\pi_1(v_i, a)\circ \pi_1(u_i, a) =\pi_1(w, a). It then follows from the universal property of amalgamated sums of monoids (A, I, p. 80, Prop. 4) that there exists a unique group homomorphism φ\varphi from ∗iπ1(Ai,a)*_i\pi_1(A_i, a) into π1(Y,a)\pi_1(Y, a) which induces the homomorphism π1(vi,a)\pi_1(v_i, a) on π1(Ai,a)\pi_1(A_i, a). It remains to prove that φ\varphi is an isomorphism.

The set J is identified with the set of pairs (i,i′)(i, i') of elements of I such that i<i′i < i'. The set K is identified with the set of triples (i1,i2,i3)(i_1, i_2, i_3) of elements of I such that i1<i2<i3i_1< i_2< i_3. Let us choose all the base points a(i),b(j)a(i),b(j) and c(k)c(k) equal to aa and all the paths B(j)B(j), Cst(k)C_{st}(k) equal to the constant path with image aa. Let us also fix a point i0∈Ii_0\in I.

For every pair (i,i′)(i, i') of distinct points of I, the framework Γ of the covering (Ai)(A_i) has exactly one arrow with endpoints ii and i′i'. The arrows of Γ one of whose endpoints is equal to i0i_0 are the arrows of an oriented maximal tree T.

For every element k∈Kk\in K, one has r3(k)=j12(k)j23(k)j13(k)−1r_3(k) =j_{12}(k)j_{23}(k)j_{13}(k)^{-1}; let R be the normal subgroup of F(J) generated by the elements j∈j\in Fl(T) and the elements r3(k),k∈r_3(k),k\in K. Let us prove that R = F(J). It is enough to show that every element j=(i1,i2)j= (i_1, i_2) of J belongs to R. This is true, by hypothesis, if i1=i0i_1=i_0 or i2=i0i_2=i_0. Suppose i0<i1i_0< i_1 and put k=(i0,i1,i2)k= (i_0, i_1, i_2). This is an element of K such that j=j23(k)=jj=j_{23}(k) =j. Moreover, j12(k)j_{12}(k) and j13(k)j_{13}(k) belong to Fl(T). It follows that jj belongs to R. The cases where i1<i0<i2i_1< i_0< i_2 or i2<i0i_2< i_0 are treated in an analogous manner.

It then follows from Prop. 1 (IV, p. 412) that the homomorphism ff is surjective and that its kernel is the smallest distinguished subgroup containing the relators r2(j,v)=gi(v)gi′(v)−1r_2(j, v) =g_i(v)g_{i'}(v)^{-1}, for j=(i,i′,A)∈Jj= (i, i',A)\in J and v∈π1(A,a)v\in \pi_1(A, a). In other words, Ker(ff) == Ker(pp). It follows that φ\varphi is an isomorphism, as was to be proved.

If π1(A,a)\pi_1(A, a) is reduced to the identity element, pp is an isomorphism, whence the second assertion.

Example 4 023C

Let ((Xi,xi))i∈I((X_i, x_i))_{i\in I} be a family of pointed topological spaces. The bouquet of the family ((Xi,xi))i∈I((X_i, x_i))_{i\in I} is by definition, and is denoted by ⋁i∈I(Xi,xi)\bigvee_{i\in I}(X_i, x_i), the quotient topological space of the sum space of the family (Xi)i∈I(X_i)_{i\in I} by the equivalence relation which identifies all the points (xi,i)(x_i, i) with one another, for i∈Ii\in I. Let X denote this topological space and xx the common image of the xix_i. Suppose that, for every i∈i\in I, the point xix_i is closed in XiX_i and that the spaces XiX_i are simply connected. If I is finite, Corollary 4 implies that the canonical homomorphism i∗∈Iπ1(Xi,xi)→π1(X,x)_i*_{\in I}\pi_1(X_i, x_i)\rightarrow \pi_1(X, x) is an isomorphism.

Remark 1 of IV, p. 429 and Exercise 3, IV, p. 463 give less restrictive conditions under which this homomorphism is an isomorphism. See nevertheless Exercise 4, IV, p. 464.

3. Particular case of a covering formed of two parts

Let X be an arcwise connected topological space, and let B and C be non-empty arcwise connected subsets of X. Assume further one of the two following hypotheses:

(i) The interiors of the sets B and C cover X;

(ii) The sets B and C are closed in X, their union is equal to X, the spaces X, B, C are simply connected, and the space B∩CB\cap C is locally arcwise connected.

Under these hypotheses, the canonical mapping of the sum space of the family (B,C)(B,C) onto the space X satisfies property (VK) (cf. IV, p. 409).

Put A=B∩CA = B\cap C. Since the space X is connected, the set A is not empty. Let aa be a point of A; let j0j_0 denote the arcwise connected component of aa in A. For every arcwise connected component jj of A distinct from j0j_0, choose a point aja_j of jj, the class βj\beta_j of a path in B and the class γj\gamma_j of a path in C, both joining aja_j to aa; let φj:π1(A,aj)→π1(B,a)\varphi_j:\pi_1(A, a_j)\rightarrow \pi_1(B, a) and ψj:π1(A,aj)→π1(C,a)\psi_j:\pi_1(A, a_j)\rightarrow \pi_1(C, a) denote the group homomorphisms defined by

φj(v)=βj−1vβj\varphi_j(v) =\beta_j^{-1}v\beta_j and ψj(v)=γj−1vγj\psi_j(v) =\gamma_j^{-1}v\gamma_j

for v∈π1(A,aj)v\in \pi_1(A, a_j). Let φ0\varphi_0 and ψ0\psi_0 also denote the homomorphisms of π1(A,a)\pi_1(A, a) into π1(B,a)\pi_1(B, a) and π1(C,a)\pi_1(C, a) respectively induced by the canonical injections. Let ιB\iota_B and ιC\iota_C denote the homomorphisms of π1(B,a)\pi_1(B, a) and π1(C,a)\pi_1(C, a) respectively into π1(X,a)\pi_1(X, a) induced by the canonical injections. Finally, let μ\mu be the unique homomorphism of the free group F(π0(A)−{j0})F(\pi_0(A)-\{j_0\}) into π1(X,a)\pi_1(X, a) such that μ(j)=βj−1γj\mu(j) =\beta_j^{-1}\gamma_j for every j∈π0(A)−{j0}j\in \pi_0(A)-\{j_0\}.

Proposition 2 023D

There exists a unique group homomorphism

M:π1(B,a)∗π1(C,a)∗F(π0(A)−{j0})→π1(X,a) \mathsf{M}:\pi_1(B, a)*\pi_1(C, a)*F(\pi_0(A)-\{j_0\})\rightarrow \pi_1(X, a)

which coincides with ιB\iota_B in π1(B,a)\pi_1(B, a), ιC\iota_C in π1(C,a)\pi_1(C, a) and μ\mu in F(π0(A)F(\pi_0(A) {j0})\{j_0\}). This homomorphism is surjective and its kernel is the smallest normal subgroup containing the elements

φj(v)jψj(v)−1j−1 \varphi_j(v)j\psi_j(v)^{-1}j^{-1}

for j∈π0(A)−{j0}j\in \pi_0(A)-\{j_0\} and v∈π1(A,aj)v\in \pi_1(A, a_j), and the elements

φ0(v)ψ0(v)−1 \varphi_0(v)\psi_0(v)^{-1}

for v∈π1(A,a)v\in \pi_1(A, a).

The framework Γ of the covering of X defined by the family (B,C)(B,C) has two vertices bb and cc corresponding to the two sets B and C. The set of its arrows is equal to π0(A)\pi_0(A); they join the point bb to the point cc. The graph associated with the subquiver of Γ whose only arrow is j0j_0 is a maximal tree of Γ~\widetilde{\Gamma}. The proposition then results from IV, p. 412, prop. 1.

Example 023E

For n⩾1n\geqslant 1, the sphere Sn\mathbf{S}_n is the union of two closed hemispheres, homeomorphic to the closed ball Bn−1\mathbf{B}_{n-1} (TG, VI, p. 12), and whose intersection is identified with the sphere Sn−1\mathbf{S}_{n-1}. For n⩾2n\geqslant 2, the sphere Sn−1\mathbf{S}_{n-1} is arcwise connected; it follows that the Poincaré group of Sn\mathbf{S}_n is trivial (cf. I, p. 127, example 3).

The sphere S0\mathbf{S}_0 has two arcwise connected components; thus one recovers that the Poincaré group of the circle S1\mathbf{S}_1 is isomorphic to a free group on one generator. More precisely, let B and C be the intersections of S1\mathbf{S}_1 with the half-planes of equations y⩾0y\geqslant 0 and y⩽0y\leqslant 0 in the plane R2\mathbf{R}^2. Put a=(1,0),a′=(−1,0)a= (1,0),a'= (-1,0); one has B∩C={a,a′}B\cap C =\{a, a'\}; its connected components are j0={a}j_0=\{a\} and j={a′}j=\{a'\}. Let β\beta be the class of the path t↦eπitt\mapsto e^{\pi it} in C\mathbf{C}; if one identifies C\mathbf{C} with R2\mathbf{R}^2, it joins aa to a′a' in B. Analogously, let γ\gamma be the class of the path t↦e−πitt\mapsto e^{-\pi it} joining aa to a′a' in C. The path βγ−1\beta \gamma^{-1} is a loop at aa, given by t↦e2πitt\mapsto e^{2\pi it}. By proposition 2, its class generates the group π1(S1,a)\pi_1(\mathbf{S}_1, a).

Corollary 1 023F

The homomorphism μ\mu is injective. More precisely, there exists a retraction associated with μ\mu which is a group homomorphism.

Let ρ\rho be the unique homomorphism of π1(B,a)∗π1(C,a)∗F(π0(A)\pi_1(B, a)*\pi_1(C, a)*F(\pi_0(A) {j0})\{j_0\}) into F(π0(A)F(\pi_0(A) - {j0})\{j_0\}) which induces the trivial homomorphism on π1(B,a)\pi_1(B, a) and π1(C,a)\pi_1(C, a) and the identity on F(π0(A)−{j0})F(\pi_0(A)-\{j_0\}). Let N be the kernel of the homomorphism M\mathsf{M}. By proposition 2,ρ(N)2,\rho (N) is reduced to the identity element. There therefore exists a unique homomorphism r\mathsf{r} of π1(X,a)\pi_1(X, a) into F(π0(A)−{j0})F(\pi_0(A)-\{j_0\}) such that ρ=r∘M\rho =\mathsf{r}\circ \mathsf{M}. For every v∈F(π0(A)−{j0})v\in F(\pi_0(A)-\{j_0\}), one has M(v)=μ(v)\mathsf{M}(v) =\mu(v), hence r∘μ\mathsf{r}\circ \mu is the identity homomorphism. The corollary follows.

Corollary 2 023G

If the group π1(X,a)\pi_1(X, a) is trivial, the set A = B∩CB\cap C is arcwise connected; if it is commutative, the set A has at most two arcwise connected components.

In fact, if S is a set, the free group F(S) is trivial only if S is empty and is commutative only if Card S⩽1S\leqslant 1.

Corollary 3 023H

If the groups π1(X,a)\pi_1(X, a) and π1(A,a)\pi_1(A, a) are trivial, the same is true of the groups π1(B,a)\pi_1(B, a) and π1(C,a)\pi_1(C, a).

By corollary 2, the set A is arcwise connected. The group π1(X,a)\pi_1(X, a) is therefore isomorphic to the free product of the groups π1(B,a)\pi_1(B, a) and π1(C,a)\pi_1(C, a). In particular it contains subgroups isomorphic to the groups π1(B,a)\pi_1(B, a) and π1(C,a)\pi_1(C, a) (A, I, p. 83). These two groups are therefore trivial if π1(X,a)\pi_1(X, a) is.

4. Quotient Spaces

Let X be an arcwise connected topological space endowed with a proper (right) operation of a discrete group G. Put Y=X/GY = X/G and denote by f:X→Yf: X\rightarrow Y the canonical mapping. If g∈Gg\in G and c:I→Xc:\mathbf{I}\rightarrow X is a path in X, let g∗cg^*c denote the path t↦c(t)⋅gt\mapsto c(t)\cdot g and g∗[c]g^*[c] its strict homotopy class.

Let oo be a point of X. For each g∈Gg\in G, let βg\beta_g be the class of a path joining o⋅go\cdot g to oo in X. For each g∈Gg\in G, let XgX^g be the set of points x∈Xx\in X such that x⋅g=xx\cdot g=x; for each arcwise connected component jj of XgX^g, let aja_j be a point of jj and let γj\gamma_j be the class of a path in X joining aja_j to oo. Let ν:F(G)→π1(Y,f(o))\nu : F(G)\rightarrow \pi_1(Y, f(o)) be the unique group homomorphism such that ν(g)=f∗(βg)\nu (g) =f_*(\beta_g) for g∈Gg\in G. Let N:π1(X,o)∗F(G)→π1(Y,f(o))\mathsf{N}:\pi_1(X, o)*F(G)\rightarrow \pi_1(Y, f(o)) be the unique group homomorphism which coincides with π1(f,o)\pi_1(f, o) on π1(X,o)\pi_1(X, o) and with ν\nu on F(G).

Proposition 3 023I

Suppose that X is unlaceable. The homomorphism N\mathsf{N} is then surjective, and its kernel is the smallest distinguished subgroup of π1(X,o)∗F(G)\pi_1(X, o)*F(G) containing the elements

(R2R_2)r2(k,v)=[k]−1v[k](βk−1k∗(v)−1βk)r_2(k, v) = [k]^{-1}v[k](\beta_k^{-1}k^*(v)^{-1}\beta_k)

for k∈Gk\in G and v∈π1(X,o)v\in \pi_1(X, o);

(R3′R'_3)r3′(k,j)=[k](βk−1k∗(γj)−1γj)r'_3(k, j) = [k](\beta_k^{-1}k^*(\gamma_j)^{-1}\gamma_j)

for k∈Gk\in G and j∈π0(Xk)j\in \pi_0(X^k);

(R3′′R''_3)r3′′(k,h)=[kh]−1[k][h](βh−1h∗(βk−1)βkh)r''_3(k, h) = [kh]^{-1}[k][h](\beta_h^{-1}h^*(\beta_k^{-1})\beta_{kh})

for kk and h∈Gh\in G.

The groupoid morphism ϖ(f)\varpi (f) induces a morphism

ϖ′′(f):ϖ(X)/G→ϖ(Y) \varpi ''(f):\varpi (X)/G\rightarrow \varpi (Y)

which is an isomorphism by Theorem 3 (IV, p. 403), since the space X is supposed unlaceable. The proposition then results from II, p. 211, prop. 6.

The three corollaries below follow from the immediately corresponding corollaries of prop. 6 of II, p. 211.

Corollary 1 023J

Suppose that X is unlaceable and that the group G is generated by the stabilizers of the points of X. The canonical morphism π1(f,o):π1(X,o)→π1(Y,f(o))\pi_1(f, o):\pi_1(X, o)\rightarrow \pi_1(Y, f(o)) is then surjective. In particular, if X is simply connected by arcs, the same is true of Y.

Remark 023K

If X is unlaceable, so is Y (IV, p. 349, prop. 8). Since a connected unlaceable space is simply connected by arcs if and only if it is simply connected (IV, p. 344, corollary 1 of Theorem 1), one thus recovers prop. 11 of I, p. 137.

Example 1 023L

Let X be an arcwise connected, unlaceable, separated topological space, and let aa be a point of X. Let nn be an integer ⩾2\geqslant 2 and let Y be the quotient of the space XnX^n by the action of the group Sn\mathfrak{S}_n operating by permutation of the factors; let f:Xn→Yf: X^n\rightarrow Y denote the canonical mapping; let g:X→Yg: X\rightarrow Y denote the mapping x↦f(x,a,...,a)x\mapsto f(x, a, . . . , a). It follows from the proposition that, for every ii, the homomorphism π1(g,a)\pi_1(g, a) of π1(X,a)\pi_1(X, a) into π1(Y,g(a))\pi_1(Y, g(a)) is surjective and that its kernel is the derived group of π1(X,a)\pi_1(X, a). In particular, the group π1(Y,g(a))\pi_1(Y, g(a)) is abelian.

Corollary 2 023M

Suppose that X is unlaceable and that the group G operates freely in X. There exists a unique group homomorphism p:π1(Y,f(o))→Gp:\pi_1(Y, f(o))\rightarrow G whose kernel contains the image of π1(X,o)\pi_1(X, o) and such that p(N(g))=gp(\mathsf{N}(g)) =g for every g∈Gg\in G. Moreover, π1(X,o)→π1(Y,f(o))−→pG\pi_1(X, o)\rightarrow \pi_1(Y, f(o))-\overset{p}{\rightarrow}G is an extension of G by π1(X,o)\pi_1(X, o).

Corollary 3 023N

Suppose that X is simply connected by arcs. The mapping of G into π1(Y,f(o))\pi_1(Y, f(o)) which, to g∈g\in G, associates the path class f∗(βg)f_*(\beta_g) is a surjective group homomorphism; its kernel is the subgroup of G generated by the stabilizers of the points of X.

5. Cones; contraction of a subspace

Let X and Y be non-empty topological spaces, and let f:X→Yf: X\rightarrow Y be a continuous mapping. Let Cone(f)(f) be the cone of the mapping ff, and let ss be its vertex. Let αf′:X×I→\alpha '_f: X\times \mathbf{I}\rightarrow Cone(f)(f) and βf′:Y→\beta '_f: Y\rightarrow Cone(f)(f) denote the canonical mappings. The restriction of αf′\alpha '_f to the subspace X×{0}X\times \{0\} of X×IX\times \mathbf{I} is the constant mapping with image {s}\{s\}. The mapping βf′\beta '_f induces a homeomorphism of Y onto the base of the cone Cone(f)(f), by which we shall identify these two spaces. Also denote by

σf′:(\sigma '_f: (Cone(f)−{s})×I→(f)-\{s\})\times \mathbf{I}\rightarrow Cone(f)−{s}(f)-\{s\}

the canonical contraction, and by ρf′:\rho '_f: Cone(f)−{s}→Y(f)-\{s\} \rightarrow Y the canonical retraction of the cone deprived of its vertex onto its base.

Let J=π0(X)J =\pi_0(X); for each element jj of J, let XjX_j denote the component jj of X, let bjb_j be a point of XjX_j, and let γj\gamma_j denote the class of the path t↦αf′(bj,t)t\mapsto \alpha '_f(b_j, t) in Côn(f)(f) joining ss to f(bj)f(b_j).

Let I be the image of the mapping π0(f)\pi_0(f); this is the set of arcwise connected components of Y which meet f(X)f(X); let φ:J→I\varphi : J\rightarrow I denote the mapping deduced from ff by passing to arcwise connected components. For each element i∈Ii\in I, let YiY_i denote the component ii of Y and choose a point aia_i in YiY_i.

For each element jj of J, choose a path BjB_j joining the point f(bj)f(b_j) to the point aφ(j)a_{\varphi(j)} in Yφ(j)Y_{\varphi(j)}, and let βj\beta_j denote its class. Let ψj\psi_j denote the homomorphism from π1(X,bj)\pi_1(X, b_j) into π1(Y,aφ(j))\pi_1(Y, a_{\varphi(j)}) defined by

ψj(v)=βj−1f∗(v)βj \psi_j(v) =\beta_j^{-1}f_*(v)\beta_j

for v∈π1(X,aj)v\in \pi_1(X, a_j).

Let σ:I→J\sigma : I\rightarrow J be a section of the mapping φ\varphi. Put T=σ(I)T =\sigma (I) and τ=σ∘φ\tau =\sigma \circ \varphi; observe that φ∘τ=φ\varphi \circ \tau =\varphi.

Proposition 4 023O

Suppose that the arcwise connected components of Y are open. For each i∈i\in I, let GiG_i be the quotient of the group π1(Yi,bi)\pi_1(Y_i, b_i) by the smallest normal subgroup containing the image of the homomorphisms ψj\psi_j, for j∈φ−1(i)j\in \overset{-1}{\varphi}(i) ; let pip_i denote the canonical surjection of π1(Yi,bi)\pi_1(Y_i, b_i) onto GiG_i.

There exists a unique group homomorphism

P:i∗∈IGi∗F(J−T)→π1(\mathsf{P}:_i*_{\in I}G_i*F(J-T)\rightarrow \pi_1(Côn(f),s)(f), s)

such that

P(pi(v))=\mathsf{P}(p_i(v)) = Int(γσ(i)βσ(i)\gamma_{\sigma(i)}\beta_{\sigma(i)})(v)(v) for i∈Ii\in I and v∈π1(Yi,bi)v\in \pi_1(Y_i, b_i),

P(j)=γjβjβτ(j)−1γτ(j)−1\mathsf{P}(j) =\gamma_j\beta_j\beta_{\tau(j)}^{-1}\gamma_{\tau(j)}^{-1} for j∈J−Tj\in J-T.

The homomorphism P\mathsf{P} is an isomorphism.

Let Y′Y' denote the union of the arcwise connected components of Y which meet f(X)f(X), and let f′:X→Y′f': X\rightarrow Y' be the mapping given by x↦f(x)x\mapsto f(x). The set Y′Y' is an open subset of Y; its arcwise connected components are open. The cone Cone(f′)(f') is identified with the arcwise connected component of ss in Cone(f)(f). This makes it possible to suppose that Y=Y′Y = Y', in other words that the mapping π0(f)\pi_0(f) is surjective and I=π0(Y)I =\pi_0(Y).

For each j∈Jj\in J, let us put Vj=αf′(Xj×]0,1[)V_j=\alpha '_f(X_j\times ]0,1[). On passing to subspaces, the mapping αf′\alpha '_f induces a homeomorphism of X×]0,1[X\times ]0,1[ onto the complement of Y∪{s}Y\cup \{s\} in Cone(f)(f). Hence the sets VjV_j are the arcwise connected components of Cone(f)−(Y∪{s})(f)-(Y\cup \{s\}).

For each i∈Ii\in I, let us put Ui=(ρf′)−1(Yi)U_i= (\rho '_f)^{-1}(Y_i); this is an open subset of Cone(f)(f), since YiY_i is open in Y by hypothesis. For each j∈φ−1(i)j\in \overset{-1}{\varphi}(i), one has f(Xj)⊂Yif(X_j)\subset Y_i and

Vj∪Yi=αf′(Xj×]0,1])∪Yi V_j\cup Y_i=\alpha '_f(X_j\times ]0,1])\cup Y_i

so that Vj∪YiV_j\cup Y_i is an arcwise connected subset of Cone(f)(f) containing YiY_i. Since UiU_i is the union of YiY_i and of the sets VjV_j, for j∈φ−1(i)j\in \overset{-1}{\varphi}(i), it follows that UiU_i is arcwise connected.

Finally, the set C′(X)=C'(X) = Cone(f)(f) - Y is an open subset of Cone(f)(f); it is contractible to ss, hence arcwise connected.

The set C′(X)C'(X) and the sets UiU_i, for i∈Ii\in I, constitute an open arcwise connected covering of Cone(f)(f), to which we shall apply prop. 1 of IV, p. 412. Let I′I' be the set obtained by adjoining ss to I; let it be provided with a total ordering for which ss is its smallest element.

For distinct elements i,i′i, i' of I, one has Ui∩Ui′=∅U_i\cap U_{i'}=\emptyset. For i∈Ii\in I, C′(X)∩UiC'(X)\cap U_i is the union of the sets VjV_j, for j∈φ−1(i)j\in \overset{-1}{\varphi}(i); they are connected and pairwise disjoint. The intersection of any three distinct sets of this covering is empty.

The framework Γ of the covering considered has the set I′I' for its vertices. Its edges are the triplets (s,i,Vj)(s, i,V_j), for j∈Jj\in J and i=φ(j)i=\varphi (j); thus the set of edges of Γ will be identified with the set J.

For i∈Ii\in I, one chooses as base point a(i)=ai∈Ui\mathsf{a}(i) =a_i\in U_i; one also puts a(s)=s∈C′(X)\mathsf{a}(s) =s\in C'(X).

For j∈j\in J, put b(j)=αf′(bj,12)\mathsf{b}(j) =\alpha '_f(b_j,\frac{1}{2}). Let B1(j)B_1(j) be the path in C′(X)C'(X) with origin b(j)\mathsf{b}(j) and endpoint a(s)\mathsf{a}(s) given by t↦αf′(bj,(1−t)/2)t\mapsto \alpha '_f(b_j,(1-t)/2). Let B2(j)B_2(j) be the path in Uφ(j)U_{\varphi(j)} with origin b(j)\mathsf{b}(j) and endpoint a(φ(j))=aφ(j)\mathsf{a}(\varphi (j)) =a_{\varphi(j)}, obtained by juxtaposing the path t↦αf′(bj,(1+t)/2)t\mapsto \alpha '_f(b_j,(1 +t)/2) and the path BjB_j. Then the class of the path B(j)=B‾1(j)∗B2(j)B(j) =\overline{B}_1(j)*B_2(j) is equal to γjβj\gamma_j\beta_j.

We choose i0=si_0=s.

We take for maximal oriented tree T the unique oriented tree of Γ whose set of arrows is σ(I)\sigma (I). We have δ(s)=e\delta (s) =e, whereas for i∈I,δ(i)=[B(σ(i))]=γσ(i)βσ(i)i\in I,\delta (i) = [B(\sigma (i))] =\gamma_{\sigma(i)}\beta_{\sigma(i)}.

Let j∈Jj\in J and let i=φ(j)i=\varphi (j).

The homomorphism φj\varphi_j of π1(Vj,b(j))\pi_1(V_j,\mathsf{b}(j)) into π1(C′(X),s)\pi_1(C'(X), s) is the trivial homomorphism, since C′(X)C'(X) is contractible to ss.

The mapping αf′\alpha '_f induces a homeomorphism of Xj×]0,1[X_j\times ]0,1[ onto VjV_j; this homeomorphism induces an isomorphism of the group π1(Vj,b(j))\pi_1(V_j,\mathsf{b}(j)) onto the group π1(Xj,bj)=π1(X,bj)\pi_1(X_j, b_j) =\pi_1(X, b_j). By passing to subspaces, the mapping σf′\sigma '_f induces a strong contraction of UiU_i onto YiY_i, which induces an isomorphism of the group π1(Ui,a(i))\pi_1(U_i,\mathsf{a}(i)) onto the group π1(Y,ai)\pi_1(Y, a_i). By means of these isomorphisms, the homomorphism

ψj:π1(Vj,b(j))→π1(Uφ(j),a(φ(j))) \psi_j:\pi_1(V_j,\mathsf{b}(j))\rightarrow \pi_1(U_{\varphi(j)},\mathsf{a}(\varphi (j)))

is identified with the homomorphism Int(βφ(j)−1\beta_{\varphi(j)}^{-1})∘f∗\circ f_* of π1(X,bj)\pi_1(X, b_j) into π1(Y,ai)\pi_1(Y, a_i).

Since C′(X)C'(X) is contractible to ss, the homomorphism μs\mu_s is the trivial homomorphism.

Let i∈Ii\in I. The homomorphism μi\mu_i of π1(Ui,a(i))\pi_1(U_i,\mathsf{a}(i)) into π1(\pi_1(Côn(f),s)(f), s) is identified with the homomorphism of π1(Y,ai)\pi_1(Y, a_i) into π1(\pi_1(Côn(f),s)(f), s) obtained by composing the homomorphism Int(δ(i)\delta (i)) and the homomorphism of π1(Ui,s)\pi_1(U_i, s) into π1(\pi_1(Côn(f),s)(f), s) deduced from the canonical injection of UiU_i into Côn(f)(f).

Finally, the homomorphism μ:F(J)→π1(\mu: F(J)\rightarrow \pi_1(Côn(f),s)(f), s) is given by

μ(j)=γjβjβτ(j)−1γτ(j)−1 \mu(j) =\gamma_j\beta_j\beta_{\tau(j)}^{-1}\gamma_{\tau(j)}^{-1}

Let P′\mathsf{P}' be the unique group homomorphism

P′:i∗∈Iπ1(Yi,ai)∗F(J)→π1(\mathsf{P}':_i*_{\in I}\pi_1(Y_i, a_i)*F(J)\rightarrow \pi_1(Cone(f),s)(f), s)

which coincides with the homomorphism Int(δ(i)−1\delta (i)^{-1}) on π1(Yi,ai)\pi_1(Y_i, a_i) and with the homomorphism μ\mu on F(J). By Prop. 1 of IV, p. 412, the homomorphism P′\mathsf{P}' is surjective and its kernel is the smallest normal subgroup of ∗iπ1(Yi,ai)∗F(J)*_i\pi_1(Y_i, a_i)*F(J) which contains the elements r1(j)\mathsf{r}_1(j), for j∈Tj\in T, and the elements r2(j,v)\mathsf{r}_2(j, v), for j∈Jj\in J and v∈π1(Vj,b(j))v\in \pi_1(V_j,\mathsf{b}(j)). (There are no elements r3(k)\mathsf{r}_3(k), since the set K is empty.)

For j∈Tj\in T, one has r1(j)=jr_1(j) =j. Let j∈Jj\in J and let v∈π1(Vj,b(j))v\in \pi_1(V_j,\mathsf{b}(j)). In view of the identification of π1(Vj,b(j))\pi_1(V_j,\mathsf{b}(j)) with π1(X,bj)\pi_1(X, b_j), one has r2(j,v)=jψj(v)j−1\mathsf{r}_2(j, v) =j\psi_j(v)j^{-1}. Let pp denote the canonical surjective homomorphism of ∗π1(Yi,ai)∗F(J)*\pi_1(Y_i, a_i)*F(J) onto ∗Gi∗F(J*G_i*F(J - T). For j∈Tj\in T, one has p(r1(j))=iep(\mathsf{r}_1(j)) =^ie; for j∈Jj\in J and v∈π1(Xi,bj)v\in \pi_1(X^i, b_j), one has p(ψj(v))=ep(\psi_j(v)) =e. Hence there exists a unique group homomorphism P\mathsf{P} of ∗Gi∗F(J−T)*G_i*F(J-T) into π1(\pi_1(Cone(f),s)(f), s) such that P′∘p=P\mathsf{P}'\circ p=\mathsf{P}; it is an isomorphism.i^i

Corollary 1 023P

Suppose moreover that the spaces X and Y are arcwise connected and let aa be a point of X. The canonical mapping of π1(Y,f(a))\pi_1(Y, f(a)) into π1(\pi_1(Cone(f),f(a))(f), f(a)) is surjective, and its kernel is the smallest normal subgroup which contains the image of the homomorphism π1(f,a)\pi_1(f, a).

Corollary 2 023Q

Suppose moreover that the arcwise connected components of Y are simply arcwise connected. The homomorphism μ:F(J−T)→π1(\mu: F(J-T)\rightarrow \pi_1(Cone(f),s)(f), s) is then an isomorphism.

Remark 1 023R

Let X be a topological space whose arcwise connected components are open. Let A be a closed subspace of X; let ι:A→\iota : A\rightarrow X denote the canonical injection, let X/AX/A denote the space deduced from X by contracting A to a point oo, and let p:X→X/Ap: X\rightarrow X/A denote the canonical mapping. Suppose further that the pair (X,A)(X,A) has the homotopy extension property. The canonical mapping ρ‾:\overline{\rho}: Cone(ι)→X/A(\iota )\rightarrow X/A is then a homotopy equivalence (III, p. 255, Remark 1), and one deduces from Prop. 4 the calculation of the Poincaré group of X/AX/A at its base point oo. In particular, if the arcwise connected components of X are simply arcwise connected, the group π1(X/A,o)\pi_1(X/A, o) is a free group.

6. Blowing up and gluing back together

Let C be a topological space and let (Bℓ)ℓ∈L(B_{\ell})_{\ell\in L} be a finite family of closed subsets of C, pairwise disjoint. Let B be a topological space and, for each ℓ∈L\ell \in L, let hℓh_{\ell} be a homeomorphism of B onto BℓB_{\ell}. We denote by BLB_L the union of the family (Bℓ)ℓ∈L(B_{\ell})_{\ell\in L}. We shall suppose that B and L are non-empty. Let R be the equivalence relation on C defined as follows. The class of an element xx of C−BLC-B_L is the set {x}\{x\}; if xx is an element of BℓB_{\ell}, where ℓ∈L\ell \in L, the class of xx is

the set of elements hk(hℓ−1(x))h_k(\overset{-1}{h_{\ell}}(x)), where kk runs through L. Let A be the quotient topological space C/RC/R and f:C→Af: C\rightarrow A the canonical surjection. One says that the space A is obtained from the space C by identification of the sets BℓB_{\ell} by means of the homeomorphisms hℓh_{\ell}.

The mapping f∘hℓf\circ h_{\ell} of B into A is independent of the element ℓ\ell of L; it is closed and injective; it therefore induces a homeomorphism of B onto a closed subset of A. Thus we identify B with f∘hℓ(B)f\circ h_{\ell}(B) by means of the homeomorphism f∘hℓf\circ h_{\ell}; the mapping ff induces a homeomorphism of C−BLC-B_L onto A-B.

Suppose moreover that there exists a family (Nℓ)ℓ∈L(N_{\ell})_{\ell\in L} of open subsets of C, pairwise disjoint, such that NℓN_{\ell} contains BℓB_{\ell} for each ℓ∈L\ell \in L. The union U of the family (f(Nℓ))ℓ∈L(f(N_{\ell}))_{\ell\in L} is open in A and contains B. The set U-B is the union of the pairwise disjoint open sets f(Nℓ−Bℓ)f(N_{\ell}-B_{\ell}), for ℓ∈L\ell \in L.

Lemma 2 023S

The mapping f:C→Af: C\rightarrow A is proper and separated; its fibres are finite.

Let us prove that ff is closed. Let X be a closed subset of C. Let us prove that its image is closed in A. For each ℓ∈L\ell \in L, X∩BℓX\cap B_{\ell} is closed in BℓB_{\ell}, hence the space Y=⋃ℓ∈Lhℓ−1(X∩Bℓ)Y =\bigcup_{\ell\in L}h^{-1}_{\ell}(X\cap B_{\ell}) is closed in B since L is finite. The saturation X∗X^* of X for the equivalence relation R is then equal to X∪⋃ℓ∈Lhk(Y)X\cup \bigcup_{\ell\in L}h_k(Y), hence is closed in C. Consequently, f(X)=f(X∗)f(X) =f(X^*) is closed in A, which was to be proved.

The fibres of ff are the equivalence classes of the relation R; they are finite. Consequently, the mapping ff is proper (TG, I, p. 75, Theorem 1).

Lastly, let us show that ff is separated. Let xx and yy be distinct points of C which have the same image under ff. There therefore exists a point b∈Bb\in B and distinct elements ℓ\ell and m∈Lm\in L such that x=hℓ(b)x=h_{\ell}(b) and y=hm(b)y=h_m(b). Consequently, NℓN_{\ell} and NmN_m are disjoint neighbourhoods of xx and yy in C, whence the required assertion by virtue of Prop. 1 of I, p. 25.

Assume in addition the following hypotheses:

– the space A is simply connected and connected;

– the space C is simply connected;

– the space B is connected and locally arcwise connected.

Let us put I=π0(C)I =\pi_0(C); if ii is an element of I, let CiC_i denote the connected component ii of C. Let η:L→I\eta : L\rightarrow I denote the mapping which associates to an element ℓ∈L\ell \in L the connected component of C containing BℓB_{\ell}. The mapping η\eta is surjective. For, arguing by contradiction, consider a connected component X of C which meets no set BℓB_{\ell}. It is an open and closed subset of C, saturated with respect to the equivalence relation R. Hence its image f(X)f(X) is an open and closed subset of A, disjoint from B. Since A is assumed connected and B nonempty, f(X)f(X) is empty, whence a contradiction.

Let σ:I→L\sigma : I\rightarrow L be a section of the mapping η\eta; put τ=σ∘η\tau =\sigma \circ \eta and T=σ(I)T =\sigma (I).

Choose a point bb of B. For every ℓ∈L\ell \in L, put bℓ=hℓ(b)b_{\ell}=h_{\ell}(b) and choose the class βℓ\beta_{\ell} of a path joining bℓb_{\ell} to bτ(ℓ)b_{\tau(\ell)} in C; if ℓ=τ(ℓ)\ell =\tau (\ell ), choose for βℓ\beta_{\ell} the class of the constant path whose image is bℓb_{\ell}. For every ℓ∈L\ell \in L, let ϑℓ\vartheta_{\ell} denote the homomorphism from π1(B,b)\pi_1(B, b) into π1(C,bτ(ℓ))\pi_1(C, b_{\tau(\ell)}) defined by

ϑℓ(v)=\vartheta_{\ell}(v) = Int(βℓ\beta_{\ell})−1((hℓ)∗(v))=βℓ−1(hℓ)∗(v)βℓ^{-1}((h_{\ell})_*(v)) =\beta_{\ell}^{-1}(h_{\ell})_*(v)\beta_{\ell},

for every v∈π1(B,b)v\in \pi_1(B, b). Finally, fix an element ℓ0\ell_0 of L such that ℓ0=\ell_0= τ(ℓ0)\tau (\ell_0).

Let Q\mathsf{Q} be the unique group homomorphism

Q:(i∗∈Iπ1(Ci,bσ(i)))∗F(L−T)→π1(A,b) \mathsf{Q}:(_i*_{\in I}\pi_1(C_i, b_{\sigma(i)}))*F(L-T)\rightarrow \pi_1(A, b)

such that Q(ℓ)=f∗(βℓ)\mathsf{Q}(\ell ) =f_*(\beta_{\ell}) for every ℓ∈\ell \in L - T and which coincides with π1(f,bσ(i))\pi_1(f, b_{\sigma(i)}) on π1(Ci,bσ(i))\pi_1(C_i, b_{\sigma(i)}) for every i∈Ii\in I.

Proposition 5 (Van Kampen) 023T

The homomorphism Q\mathsf{Q} is surjective; its kernel is the smallest normal subgroup containing the elements

ϑℓ0(v)ϑℓ(v)−1\vartheta_{\ell_0}(v)\vartheta_{\ell}(v)^{-1} for v∈π1(B,b)v\in \pi_1(B, b) and ℓ∈T\ell \in T,

ϑℓ0(v)ℓϑℓ(v)−1ℓ−1\vartheta_{\ell_0}(v)\ell \vartheta_{\ell}(v)^{-1}\ell^{-1} for v∈π1(B,b)v\in \pi_1(B, b) and ℓ∈L−T\ell \in L-T.

The mapping f:C→f: C\rightarrow A is proper, separated, with finite fibres (IV, p. 430, lemma 2); the spaces A and C are unlaceable. Moreover, C×ACC\times_AC is the union of the diagonal ΔC\Delta_C and of the disjoint subsets Bℓ×ABk=B_{\ell}\times_AB_k= (hℓ,hk)(B)(h_{\ell}, h_k)(B), for (ℓ,k)∈L2(\ell , k)\in L^2 with ℓ=k̸\ell =\not k. For such a pair (ℓ,k)(\ell , k), we have Bℓ×ABk=Nℓ×ANkB_{\ell}\times_AB_k= N_{\ell}\times_AN_k; hence this subset is open and closed in C×ACC\times_AC. Thus the diagonal ΔC\Delta_C is open, and its complement in C×ACC\times_AC, a finite union of disjoint subsets homeomorphic to B, is locally connected. This proves that the mapping ff satisfies property (VK) (case (ii) of IV, p. 405). With a view to applying theorem 1 of IV, p. 408, we are going to define van Kampen data for ff.

For each i∈Ii\in I, let us choose as base point in CiC_i the point a(i)=\mathsf{a}(i) = bσ(i)b_{\sigma(i)}.

Let J be the set of path-connected components of C×ACC\times_AC. The sets ΔCi\Delta_{C_i}, for i∈Ii\in I, are the connected components of the diagonal ΔC\Delta_C, which is open and closed in C×ACC\times_AC, therefore these sets belong to J. Analogously, the sets [ℓ1,ℓ2]=Bℓ1×A[\ell_1, \ell_2] = B_{\ell_1}\times_A Bℓ2B_{\ell_2} for ℓ1\ell_1 and ℓ2\ell_2 in L with ℓ1≠ℓ2\ell_1=\not\ell_2, belong to J. Since these sets form a partition of C×ACC\times_AC, they describe the set J.

Let jj be an element of J of the form ΔCi\Delta_{C_i}, for i∈Ii\in I. Let us choose as base point b(j)\mathsf{b}(j) the point (a(i),a(i))=(bσ(i),bσ(i))(\mathsf{a}(i),\mathsf{a}(i)) = (b_{\sigma(i)}, b_{\sigma(i)}) and let us take for path classes β1(j)\beta_1(j) and β2(j)\beta_2(j) the class of the constant path at bσ(i)b_{\sigma(i)}. Let jj be an element of J of the form [ℓ1,ℓ2][\ell_1, \ell_2], where ℓ1\ell_1 and ℓ2\ell_2 are distinct elements of L. Let us then put b([ℓ1,ℓ2])=(bℓ1,bℓ2),β1([ℓ1,ℓ2])=βℓ1\mathsf{b}([\ell_1, \ell_2]) = (b_{\ell_1}, b_{\ell_2}),\beta_1([\ell_1, \ell_2]) =\beta_{\ell_1} and β2([ℓ1,ℓ2])=βℓ2\beta_2([\ell_1, \ell_2]) =\beta_{\ell_2}.

Let K=π0(C×AC×AC)K =\pi_0(C\times_AC\times_AC).

Let ΔC′\Delta '_C denote the diagonal of the space C×AC×ACC\times_AC\times_AC; this is a closed subset of C×AC×ACC\times_AC\times_AC, because ff is separated, and it is homeomorphic to C. For every i∈i\in I, let ΔCi′\Delta '_{C_i} also denote the image of CiC_i under the diagonal mapping of C into C×AC×ACC\times_AC\times_AC; these are open and closed subsets of ΔC′\Delta '_C. For every triplet (ℓ1,ℓ2,ℓ3)(\ell_1, \ell_2, \ell_3) of elements of L, put also [ℓ1,ℓ2,ℓ3]=Bℓ1×ABℓ2×ABℓ3[\ell_1, \ell_2, \ell_3] = B_{\ell_1}\times_AB_{\ell_2}\times_AB_{\ell_3}; these are closed subsets of C×AC\times_A C×ACC\times_AC, homeomorphic to B. If the set {ℓ1,ℓ2,ℓ3}\{\ell_1, \ell_2, \ell_3\} has at least two elements, one has [ℓ1,ℓ2,ℓ3]=Nℓ1×ANℓ2×ANℓ3[\ell_1, \ell_2, \ell_3] = N_{\ell_1}\times_AN_{\ell_2}\times_AN_{\ell_3}, which implies that [ℓ1,ℓ2,ℓ3][\ell_1, \ell_2, \ell_3] is also open in C×AC×ACC\times_AC\times_AC. Moreover, C×AC×ACC\times_AC\times_AC is the union of the pairwise disjoint subsets ΔC′\Delta '_C and [ℓ1,ℓ2,ℓ3][\ell_1, \ell_2, \ell_3], where (ℓ1,ℓ2,ℓ3)(\ell_1, \ell_2, \ell_3) runs through the set of triplets of elements of L which are not all equal to one and the same element. Thus, ΔC′\Delta_{C'}, then the ΔCi′\Delta '_{C_i}, for i∈Ii\in I, are open and closed in C×AC×ACC\times_AC\times_AC. This implies that the set K of connected components of this space is the union of the following disjoint sets K0K_0 and K1K_1.

The set K0K_0 is the set of components of the form ΔCi′\Delta '_{C_i}. Let i∈Ii\in I and let k=ΔCi′k= \Delta '_{C_i}. Put c(k)=(bσ(i),bσ(i),bσ(i))\mathsf{c}(k) = (b_{\sigma(i)}, b_{\sigma(i)}, b_{\sigma(i)}). For (s,t)∈{(1,2),(2,3),(1,3)}(s, t)\in \{(1,2),(2,3),(1,3)\}, choose for γst(k)\gamma_{st}(k) the class of the constant path at (bσ(i),bσ(i))(b_{\sigma(i)}, b_{\sigma(i)}).

The set K1K_1 consists of the components of the form kk = [ℓ1,ℓ2,ℓ3][\ell_1, \ell_2, \ell_3], where ℓ1,ℓ2,ℓ3\ell_1, \ell_2, \ell_3 are three elements of B such that the set {ℓ1,ℓ2,ℓ3}\{\ell_1, \ell_2, \ell_3\} is of cardinal ⩾2\geqslant 2. Put c(k)=(bℓ1,bℓ2,bℓ3)\mathsf{c}(k) = (b_{\ell_1}, b_{\ell_2}, b_{\ell_3}). Let (s,t)∈{(1,2),(1,3),(2,3)}(s, t)\in \{(1,2),(1,3),(2,3)\}. If ℓs=ℓt\ell_s=\ell_t, take for γst(k)\gamma_{st}(k) the class (βℓs,βℓt)(\beta_{\ell_s}, \beta_{\ell_t}); if ℓs≠ℓt\ell_s=\not\ell_t, take for γst(k)\gamma_{st}(k) the class of the constant path at (bℓs,bℓt)(b_{\ell_s}, b_{\ell_t}).

It is then verified that, for every k∈Kk\in K and for every s∈{1,2,3}s\in \{1,2,3\}, the class of loops λs(k)\lambda_s(k) defined by relation (2) of IV, p. 407, is trivial.

The framework Γ\mathsf{\Gamma} of the pair of groupoid morphisms (ϖ(p1),ϖ(p2))(\varpi (p_1), \varpi (p_2)) from ϖ(C×AC)\varpi (C\times_AC) into ϖ(C)\varpi (C) has as vertices the set I and as oriented edges the set J. If i∈Ii\in I, the arrow j=ΔCij= \Delta_{C_i} has origin and end ii; if (ℓ1,ℓ2)(\ell_1, \ell_2) is a pair of distinct elements of L, the arrow j=[ℓ1,ℓ2]j= [\ell_1, \ell_2] has origin η(ℓ1)\eta (\ell_1) and end η(ℓ2)\eta (\ell_2).

Let T\mathsf{T} be the subquiver of Γ\mathsf{\Gamma} whose set of vertices is I and whose arrows are those of the form [ℓ0,ℓ][\ell_0, \ell ], for ℓ∈T−{ℓ0}\ell \in T-\{\ell_0\}. The associated graph of T\mathsf{T} is a maximal tree of Γ~\widetilde{\mathsf{\Gamma }}.

Put i0=η(ℓ0)i_0=\eta (\ell_0).

If i∈I−{i0}i\in I-\{i_0\}, the unique path in T\mathsf{T} joining i0i_0 to ii is (i0,[ℓ0,σ(i)],i)(i_0,[\ell_0, \sigma (i)], i). The groupoid morphism from Grp(Γ\mathsf{\Gamma}) into ϖ(Y)\varpi (Y) defined on p. 407 (and denoted by τ\tau loc. cit.) maps jj to the identity element if j=ΔCij= \Delta_{C_i}, for i∈Ii\in I, and maps j=[ℓ1,ℓ2]j= [\ell_1, \ell_2] to f∗(βℓ1)−1f∗(βℓ2)f_*(\beta_{\ell_1})^{-1}f_*(\beta_{\ell_2}), if ℓ1\ell_1 and ℓ2\ell_2 are distinct elements of L. For every i∈Ii\in I, the path class δi\delta_i defined loc. cit. and joining b=f(bσ(i0))b=f(b_{\sigma(i_0)}) to b=f(bσ(i))b=f(b_{\sigma(i)}) is given by

δi=f∗(βℓ0)−1f∗(βσ(i))=e \delta_i=f_*(\beta_{\ell_0})^{-1}f_*(\beta_{\sigma(i)}) =e

because βℓ=e\beta_{\ell}=e if ℓ∈T\ell \in T.

Thus a van Kampen datum of the mapping ff has been defined. Consider then the unique group homomorphism

Q′:(i∗∈Iπ1(Ci,bσ(i)))∗F(J)→π1(A,b) \mathsf{Q}':(_i*_{\in I}\pi_1(C_i, b_{\sigma(i)}))*F(J)\rightarrow \pi_1(A, b)

which coincides with π1(f,bσ(i))\pi_1(f, b_{\sigma(i)}) on π1(C,bσ(i))\pi_1(C, b_{\sigma(i)}) and such that

Q′(j)=f∗(β1(j))−1f∗(β2(j)) \mathsf{Q}'(j) =f_*(\beta_1(j))^{-1}f_*(\beta_2(j))

for j∈Jj\in J. By IV, p. 408, Theorem 1, this homomorphism is surjective and its kernel is the smallest normal subgroup containing the relators r1(j)\mathsf{r}_1(j) (for j∈j\in Fl(T\mathsf{T})), r2(j,v)\mathsf{r}_2(j, v) (for j∈j\in J and v∈v\in π1(C×AC,b(j))\pi_1(C\times_AC,\mathsf{b}(j))) and r3(k)\mathsf{r}_3(k) (for k∈Kk\in K) defined, loc. cit., by equations (R1R_1),(R2), (R_2) and (R3R_3).

Let q′q' be the unique homomorphism of F(L) into F(L - T) such that q′(ℓ)=ℓq'(\ell ) =\ell if ℓ∈L−T\ell \in L-T and q′(ℓ)=eq'(\ell ) =e otherwise. Let

q:(i∗∈Iπ1(Ci,bσ(i)))∗F(J)→(i∗∈Iπ1(Ci,bσ(i)))∗F(L−T) q:(_i*_{\in I}\pi_1(C_i, b_{\sigma(i)}))*F(J)\rightarrow (_i*_{\in I}\pi_1(C_i, b_{\sigma(i)}))*F(L-T)

be the unique group homomorphism which coincides with the identity on π1(Ci,bσ(i))\pi_1(C_i, b_{\sigma(i)}), for i∈i\in I and such that one has q(j)=eq(j) =e if j=ΔCij= \Delta_{C_i}, q([ℓ,ℓ′])=q′(ℓ)−1q′(ℓ′)q([\ell , \ell ']) =q'(\ell )^{-1}q'(\ell ') if ℓ\ell and ℓ′\ell ' are distinct elements of L. The homomorphism qq is surjective.

If i∈Ii\in I and j=ΔCij= \Delta_{C_i}, one has Q′(j)=eb=Q′(q(j))\mathsf{Q}'(j) =e_b=\mathsf{Q}'(q(j)). If j=[ℓ,ℓ′]j= [\ell , \ell '], for ℓ\ell and ℓ′\ell ' in L, distinct, one has

Q′(j)=f∗(βℓ)−1f∗(βℓ′)=Q(q′(ℓ))−1Q(q′(ℓ′)) \mathsf{Q}'(j) =f_*(\beta_{\ell})^{-1}f_*(\beta_{\ell'}) =\mathsf{Q}(q'(\ell ))^{-1}\mathsf{Q}(q'(\ell '))
=Q(q′(ℓ)−1q′(ℓ′))=Q∘q([ℓ,ℓ′]) =\mathsf{Q}(q'(\ell )^{-1}q'(\ell ')) =\mathsf{Q}\circ q([\ell , \ell '])

Therefore, Q′\mathsf{Q}' = Q∘q\mathsf{Q}\circ q. It follows that the homomorphism Q\mathsf{Q} is surjective and that its kernel is the smallest normal subgroup of (i∗∈Iπ1(Ci,bσ(i)))∗F(L−T)(_i*_{\in I}\pi_1(C_i, b_{\sigma(i)}))*F(L-T) containing the images under qq of the relators r1(j)\mathsf{r}_1(j) (for j∈j\in Fl(T\mathsf{T})),r2(j,v),\mathsf{r}_2(j, v) (for j∈Jj\in J and v∈π1(C×AC,b(j))v\in \pi_1(C\times_AC,\mathsf{b}(j))) and r3(k)\mathsf{r}_3(k) (for k∈Kk\in K).

If ℓ∈T−{ℓ0}\ell \in T-\{\ell_0\} and j=[ℓ0,ℓ]j= [\ell_0, \ell ], one has r1(j)=j\mathsf{r}_1(j) =j and q(r1(j))=eq(\mathsf{r}_1(j)) =e.

Let k∈Kk\in K. One has r3(k)=j12(k)j23(k)j13(k)−1\mathsf{r}_3(k) =j_{12}(k)j_{23}(k)j_{13}(k)^{-1}. If k=ΔCi′k= \Delta '_{C_i}, for i∈Ii\in I, let us put j=ΔCij= \Delta_{C_i}; then one has

q(r3(k))=q(jjj−1)=q(j)=e q(\mathsf{r}_3(k)) =q(jjj^{-1}) =q(j) =e

Let ℓ\ell and ℓ′\ell ' be distinct elements of L. If k=[ℓ,ℓ,ℓ′]k= [\ell , \ell , \ell '], one has therefore

q(r3(k))=q(ΔCη(ℓ)[ℓ,ℓ′][ℓ,ℓ′]−1)=q(ΔCη(ℓ))=e q(\mathsf{r}_3(k)) =q(\Delta_{C_{\eta(\ell)}}[\ell , \ell '][\ell , \ell ']^{-1}) =q(\Delta_{C_{\eta(\ell)}}) =e

If k=[ℓ,ℓ′,ℓ]k= [\ell , \ell ', \ell ], one gets

q(r3(k))=q([ℓ,ℓ′][ℓ′,ℓ]ΔCη−1(ℓ)) q(\mathsf{r}_3(k)) =q([\ell , \ell '][\ell ', \ell ]\Delta^{-1_{(\ell)}}_{C_{\eta}})
=q′(ℓ)−1q′(ℓ′)q′(ℓ′)−1q′(ℓ)q(ΔCη(ℓ))−1=e =q'(\ell )^{-1}q'(\ell ')q'(\ell ')^{-1}q'(\ell )q(\Delta_{C_{\eta}(\ell)})^{-1}=e

Moreover, if k=[ℓ′,ℓ,ℓ]k= [\ell ', \ell , \ell ], one has

q(r3(k))=q([ℓ′,ℓ]ΔCη(ℓ)[ℓ′,ℓ]−1) q(\mathsf{r}_3(k)) =q([\ell ', \ell ]\Delta_{C_{\eta(\ell)}}[\ell ', \ell ]^{-1})
=q(ℓ′)−1q(ℓ)q(ΔCη(ℓ))q(ℓ)−1q(ℓ′)=e =q(\ell ')^{-1}q(\ell )q(\Delta_{C_{\eta(\ell)}})q(\ell )^{-1}q(\ell ') =e

Finally, if k=[ℓ1,ℓ2,ℓ3]k= [\ell_1, \ell_2, \ell_3], where ℓ1,ℓ2,ℓ3\ell_1, \ell_2, \ell_3 are elements of L, pairwise distinct, one has

q(r3(k))=q([ℓ1,ℓ2])q([ℓ2,ℓ3])q([ℓ1,ℓ3])−1=e q(\mathsf{r}_3(k)) =q([\ell_1, \ell_2])q([\ell_2, \ell_3])q([\ell_1, \ell_3])^{-1}=e

Let i∈Ii\in I and put j=ΔCij= \Delta_{C_i}. The group homomorphisms φj\varphi_j and ψj\psi_j, from π1(C×AC,b(j))=π1(Ci,a(i))\pi_1(C\times_AC,\mathsf{b}(j)) =\pi_1(C_i,\mathsf{a}(i)) into π1(C,a(i))\pi_1(C,\mathsf{a}(i)) defined by relations (1) of IV, p. 407, are equal respectively to (p1)∗(p_1)_* and to (p2)∗(p_2)_*. We have then, for v∈π1(Ci,a(i)),r2(j,v)=vjv−1j−1v\in \pi_1(C_i,\mathsf{a}(i)),\mathsf{r}_2(j, v) =vjv^{-1}j^{-1}, whence q(r2(j,v))=eq(\mathsf{r}_2(j, v)) =e.

Finally, let j=[ℓ,ℓ′]j= [\ell , \ell '], where ℓ\ell and ℓ′\ell ' are distinct elements of L. The group homomorphism φj:π1(Bℓ×ABℓ′,b(j))→π1(Cη(ℓ),bτ(ℓ))\varphi_j:\pi_1(B_{\ell}\times_AB_{\ell'},\mathsf{b}(j))\rightarrow \pi_1(C_{\eta(\ell)}, b_{\tau(\ell)}) defined loc. cit. is the homomorphism ϑℓ\vartheta_{\ell}, and the homomorphism ψj\psi_j is the homomorphism ϑℓ′\vartheta_{\ell'}. We have then, for v∈π1(Bℓ×ABℓ′,b(j))v\in \pi_1(B_{\ell}\times_AB_{\ell'},\mathsf{b}(j))

q(r2(j,v))=ϑℓ(v)q′(ℓ)−1q′(ℓ′)ϑℓ′(v)−1q′(ℓ′)−1q′(ℓ) q(\mathsf{r}_2(j, v)) =\vartheta_{\ell}(v)q'(\ell )^{-1}q'(\ell ')\vartheta_{\ell'}(v)^{-1}q'(\ell ')^{-1}q'(\ell )

Let us distinguish four cases. If ℓ\ell and ℓ′\ell ' both belong to T, we have

q(r2(j,v))=(ϑℓ0(v)ϑℓ(v)−1)−1(ϑℓ0(v)ϑℓ′(v)−1) q(\mathsf{r}_2(j, v)) =(\vartheta_{\ell_0}(v)\vartheta_{\ell}(v)^{-1})^{-1}(\vartheta_{\ell_0}(v)\vartheta_{\ell'}(v)^{-1})

When ℓ′=ℓ0\ell '=\ell_0, we obtain the inverse of the element ϑℓ0(v)ϑℓ(v)−1\vartheta_{\ell_0}(v)\vartheta_{\ell}(v)^{-1}. If ℓ∈T\ell \in T but ℓ′∉T\ell '\notin T, we have

q(r2(j,v))=ϑℓ(v)ℓ′ϑℓ′(v)−1(ℓ′)−1 q(\mathsf{r}_2(j, v)) =\vartheta_{\ell}(v)\ell '\vartheta_{\ell'}(v)^{-1}(\ell ')^{-1}
=(ϑℓ0(v)ϑℓ(v)−1)−1ϑℓ0(v)ℓ′ϑℓ′(v)−1(ℓ′)−1 =(\vartheta_{\ell_0}(v)\vartheta_{\ell}(v)^{-1})^{-1}\vartheta_{\ell_0}(v)\ell '\vartheta_{\ell'}(v)^{-1}(\ell ')^{-1}

Analogously, if ℓ∉T\ell \notin T and ℓ′∈T\ell '\in T, we have

q(r2(j,v))=ϑℓ(v)ℓ−1ϑℓ′(v)−1ℓ q(\mathsf{r}_2(j, v)) =\vartheta_{\ell}(v)\ell^{-1}\vartheta_{\ell'}(v)^{-1}\ell
=ℓ−1(ϑℓ0(v)ℓϑℓ(v)−1ℓ−1)−1(ϑℓ0(v)ϑℓ′(v)−1)ℓ =\ell^{-1}(\vartheta_{\ell_0}(v)\ell \vartheta_{\ell}(v)^{-1}\ell^{-1})^{-1}(\vartheta_{\ell_0}(v)\vartheta_{\ell'}(v)^{-1})\ell

Taking ℓ′=ℓ0\ell '=\ell_0, we obtain an element conjugate to the inverse of ϑℓ0(v)ℓϑℓ(v)−1ℓ−1\vartheta_{\ell_0}(v)\ell \vartheta_{\ell}(v)^{-1}\ell^{-1}. Finally, if neither ℓ\ell nor ℓ′\ell ' belongs to T, we have

q(r2(j,v))=(ϑℓ(v)ℓ−1ϑℓ0(v)−1ℓ)ℓ−1(ϑℓ0(v)ℓ′ϑℓ′(v)−1(ℓ′)−1)ℓ q(\mathsf{r}_2(j, v)) =(\vartheta_{\ell}(v)\ell^{-1}\vartheta_{\ell_0}(v)^{-1}\ell )\ell^{-1}(\vartheta_{\ell_0}(v)\ell '\vartheta_{\ell'}(v)^{-1}(\ell ')^{-1})\ell

These relations prove, on the one hand, that the elements announced in the proposition belong to the kernel of the homomorphism Q\mathsf{Q}, and, on the other hand, that these elements q(r2(j,v))q(\mathsf{r}_2(j, v)) all belong to the smallest normal subgroup containing the elements announced in the statement. The proposition follows.

Remark 023U

Let A be a topological space, let B be a closed subset of A, and let U be an open neighbourhood of B in A. Suppose that the set U-B is the union of a finite family (Mℓ)ℓ∈L(M_{\ell})_{\ell\in L} of pairwise disjoint open sets. For each ℓ∈L\ell \in L, put Nℓ′=Mℓ∪BN'_{\ell}= M_{\ell}\cup B. Let C′C' denote the topological sum of the spaces A-B and Nℓ′N'_{\ell}, for ℓ∈L\ell \in L; let also φ:A−B→C′\varphi : A-B\rightarrow C' and φℓ:Nℓ′→C′\varphi_{\ell}: N'_{\ell}\rightarrow C', for ℓ∈L\ell \in L, denote the canonical injections. Let C be the quotient topological space of C′C' by the finest equivalence relation S for which the points φ(x)\varphi (x) and φℓ(x)\varphi_{\ell}(x) are equivalent, for each ℓ∈L\ell \in L and each x∈Mℓx\in M_{\ell}; let ρ:C′→C\rho : C'\rightarrow C be the canonical mapping.

Let us prove that the space A is recovered from the space C by identification of the sets BℓB_{\ell} by means of the homeomorphisms ρ∘\rho \circ (φℓ∣B):B→Bℓ(\varphi_{\ell}|B): B\rightarrow B_{\ell}.

For every ℓ∈L\ell \in L, the set MℓM_{\ell} is open in A-B and in Nℓ′N'_{\ell}. The mappings ρ∘φ\rho \circ \varphi and ρ∘φℓ\rho \circ \varphi_{\ell} are homeomorphisms of the spaces A-B and Nℓ′N'_{\ell}, for ℓ∈L\ell \in L, onto open subsets of C (TG, I, p. 17, Prop. 9). For every ℓ∈L\ell \in L, the set φℓ(B)\varphi_{\ell}(B) is closed in C′C' and saturated for the equivalence relation S. Hence the set Bℓ=ρ(φℓ(B))B_{\ell}=\rho (\varphi_{\ell}(B)) is closed in C and the mapping ρ∘φℓ\rho \circ \varphi_{\ell} induces a homeomorphism of B onto Bℓ(B_{\ell}(loc. cit.). Analogously, for every ℓ∈L\ell \in L, the set Nℓ=ρ(φℓ(Nℓ′))N_{\ell}=\rho (\varphi_{\ell}(N'_{\ell})) is an open neighbourhood of BℓB_{\ell}; the sets NℓN_{\ell} are mutually disjoint.

By passing to the quotient, the mapping f′:C′→Af': C'\rightarrow A deduced from the canonical injections into A of the spaces A-B and NℓN_{\ell}, for ℓ∈L\ell \in L, induces a

continuous mapping f:C→Af: C\rightarrow A. If xx is a point of B, the fibre f−1(x)\overset{-1}{f}(x) is the set of points ρ(φℓ(x))\rho (\varphi_{\ell}(x)), for ℓ∈L\ell \in L. The equivalence relation on C associated with the mapping ff is the relation R defined at the beginning of the number. Let BLB_L denote the union of the family (Bℓ)ℓ∈L(B_{\ell})_{\ell\in L}. By construction, the mapping ff induces a homeomorphism of C−BLC-B_L onto A-B and, for ℓ∈L\ell \in L, a homeomorphism of Nℓ′N'_{\ell} onto NℓN_{\ell}. We shall prove that the mapping ff is closed; the topology of A will therefore be the quotient topology of C by the equivalence relation R (I, p. 18, Example 2). For this, let us prove that if F is a subset of A such that F∩(A−B)F\cap (A-B) is closed in A-B and such that F∩Nℓ′F\cap N'_{\ell} is closed in Nℓ′N'_{\ell}, for ℓ∈L\ell \in L, the set F is closed in A. The set U, union of the family (Nℓ′)ℓ∈L(N'_{\ell})_{\ell\in L}, is open in A and the sets Nℓ′N'_{\ell}, for ℓ∈L\ell \in L, constitute a finite closed covering of it. Hence the set F∩UF\cap U is closed in U (TG, I, p. 18, Prop. 3). The sets U and A-B form an open covering of A, therefore the set F is closed in A (loc. cit.).

Exercises

See the exercises for § 5.

Machine translation of the English, by gpt-5-6-mini, gpt-5-6, gpt-5.4, against glossary version 34. Not checked by a person. That model is a cut down version of the one the rest of this language was written by.