ScalingStacks

7.1. 11-dimensional spaces

7.1.1. Definitions

A manifold is defined to be a topological manifold with boundary with finitely many connected components, all of which have the same dimension. A 11-dimensional manifold is a finite disjoint union of copies of S1S^{1}, 𝐑{\mathbf{R}}, 𝐑β‰₯0{\mathbf{R}}_{\geq 0} and [0,1][0,1].

Given a point xx of a topological space XX, we put C⁑(x)=CX​(x)=limUΟ€0​(Uβˆ’{x})C(x)=C_{X}(x)=\lim_{U}\pi_{0}(U-\{x\}), where UU runs over the set of open neighbourhoods of xx. If Xβ€²X^{\prime} is a subspace of XX containing an open neighbourhood of xx, then we have a canonical bijection CX′​(x)β†’βˆΌCX​(x)C_{X^{\prime}}(x)\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}C_{X}(x) and we identify those two sets.

We put T⁑(X)=∐x∈XC⁑(x)T(X)=\coprod_{x\in X}C(x) and we denote by pt:T⁑(X)β†’X\operatorname{pt}\nolimits:T(X)\to X the canonical map.

0P85

Definition 7.1.1. We define a 11-dimensional space to be a topological space that is homeomorphic to the complement of a finite set of points in a 11-dimensional finite CW-complex, and that has no connected component that is a point.

Given EE a finite subset of S1={zβˆˆπ‚|β€–zβ€–=1}S^{1}=\{z\in{\mathbf{C}}\ |\ ||z||=1\}, we put St⁑(E)=⋃e∈E𝐑β‰₯0​e\operatorname{St}\nolimits(E)=\bigcup_{e\in E}{\mathbf{R}}_{\geq 0}e and St∘⁑(E)=St⁑(E)βˆ’{0}\operatorname{St}\nolimits^{\circ}(E)=\operatorname{St}\nolimits(E)-\{0\}. These are 11-dimensional spaces. Given nβ‰₯1n\geq 1, we put St⁑(n)=St⁑({e2​i​π​r/n}0≀r<n)\operatorname{St}\nolimits(n)=\operatorname{St}\nolimits(\{e^{2i\pi r/n}\}_{0\leq r<n}).

Let XX be a 11-dimensional space. There is a finite subset EE of XX such that Xβˆ’EX-E is homeomorphic to a finite disjoint union of copies of 𝐑{\mathbf{R}}.

Let x∈Xx\in X. If UU is a small enough connected open neighbourhood of xx, then there is a homeomorphism Uβ†’βˆΌSt⁑(nx),x↦0U\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}\operatorname{St}\nolimits(n_{x}),\ x\mapsto 0 for some nx=nx,Xβ‰₯1n_{x}=n_{x,X}\geq 1. In addition, we have a canonical bijection C⁑(x)β†’βˆΌΟ€0​(Uβˆ’{x})C(x)\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}\pi_{0}(U-\{x\}) and we identify those two sets of cardinality nxn_{x}.

We define the boundary βˆ‚X={x∈X|nx=1}\partial X=\{x\in X\ |\ n_{x}=1\}. We put Xe​x​c={x∈X|nxβ‰₯3}X_{exc}=\{x\in X|n_{x}\geq 3\}.

0P86

Definition 7.1.2. We say XX is non-singular if Xe​x​c=βˆ…X_{exc}=\emptyset. Note that Xβˆ’Xe​x​cX-X_{exc} is a non-singular 11-dimensional space.

A 11-dimensional space is non-singular if and only if it is a 11-dimensional manifold.

0P87

Definition 7.1.3. We say that an open neighbourhood UU of x∈Xx\in X is small if it is homeomorphic to St⁑(nx)\operatorname{St}\nolimits(n_{x}), if |UΒ―βˆ’U|=nx|\overline{U}-U|=n_{x} and if nxβ€²=2n_{x^{\prime}}=2 for all xβ€²βˆˆUΒ―βˆ’{x}x^{\prime}\in\overline{U}-\{x\}.

Note that every point of a 11-dimensional space admits a small open neighbourhood.

7.1.2. Morphisms

Let Xβ€²X^{\prime} be a 11-dimensional space and let f:Xβ†’Xβ€²f:X\to X^{\prime} be a continuous map. Let Xfβ€²X^{\prime}_{f} be the set of points xβ€²βˆˆXβ€²x^{\prime}\in X^{\prime} such that there is no open neighbourhood UU of xβ€²x^{\prime} with the property that f|fβˆ’1(U):fβˆ’1(U)β†’Uf_{|f^{-1}(U)}:f^{-1}(U)\to U is a homeomorphism. Let Xf=fβˆ’1​(Xfβ€²)X_{f}=f^{-1}(X^{\prime}_{f}).

0P88

Lemma 7.1.4. The following conditions are equivalent:

  1. (1)

    there is a finite subset E1E_{1} of XX such that f⁑(Xβˆ’E1)f(X-E_{1}) is open in Xβ€²X^{\prime} and f|Xβˆ’E1:Xβˆ’E1β†’f(Xβˆ’E1)f_{|X-E_{1}}:X-E_{1}\to f(X-E_{1}) is a homeomorphism

  2. (2)

    XfX_{f} is finite

  3. (3)

    there is a finite subset E2E_{2} of XX such that f|Xβˆ’E2:Xβˆ’E2β†’f(Xβˆ’E2)f_{|X-E_{2}}:X-E_{2}\to f(X-E_{2}) is a homeomorphism

  4. (4)

    given x∈Xx\in X, there is a finite subset ExE_{x} of Xβˆ’{x}X-\{x\} such that f|Xβˆ’Exf_{|X-E_{x}} is injective

  5. (5)

    there is a finite subset E3E_{3} of XX such that f|Xβˆ’E3f_{|X-E_{3}} is injective.

0P89

Proof. The implication (1)β‡’(2)(1)\Rightarrow(2) follows from the fact that XfβŠ‚fβˆ’1​(f⁑(E1))X_{f}\subset f^{-1}(f(E_{1})). For the implication (2)β‡’(3)(2)\Rightarrow(3), take E2=XfE_{2}=X_{f}. For (3)β‡’(4)(3)\Rightarrow(4), take Ex=(Xβˆ’{x})∩(fβˆ’1​(f⁑(x))βˆͺE2)E_{x}=(X-\{x\})\cap(f^{-1}(f(x))\cup E_{2}). The implication (4)β‡’(5)(4)\Rightarrow(5) is immediate.

Let us show that (5)β‡’(1)(5)\Rightarrow(1). Note first that an injective continuous map 𝐑→𝐑{\mathbf{R}}\to{\mathbf{R}} is open and a homeomorphism onto its image. It follows that the implication holds when XX and Xβ€²X^{\prime} are homeomorphic to 𝐑{\mathbf{R}} and E3=βˆ…E_{3}=\emptyset.

Consider now the general case. There is a finite subset E1E_{1} of XX containing E3E_{3} such that Xβˆ’E1X-E_{1} and Xβ€²βˆ’f⁑(E1)X^{\prime}-f(E_{1}) are homeomorphic to a finite disjoint union of copies of 𝐑{\mathbf{R}}. By the discussion above, the restriction of ff to a connected component of Xβˆ’E1X-E_{1} is open and a homeomorphism onto its image, so the same holds for f|Xβˆ’E1f_{|X-E_{1}}.

∎

0P8A

Definition 7.1.5. We say that ff is a morphism of 11-dimensional spaces if it satisfies any of the equivalent conditions of Lemma 7.1.4.

Note that

  • β€’

    a composition of morphisms of 11-dimensional spaces is a morphism of 11-dimensional spaces

  • β€’

    a morphism of 11-dimensional spaces is invertible if and only if it is a homeomorphism.

0P8B

Definition 7.1.6. We define a 11-dimensional subspace of XX to be a subspace YY with only finitely many connected components, none of which are points.

Let us record some basic facts on subspaces.

0P8C
  1. (1)

    Lemma 7.1.7. The image of a morphism of 11-dimensional spaces is a 11-dimensional subspace.

  2. (2)

    If YY is a 11-dimensional subspace of XX, then YY is a 11-dimensional space and the inclusion map Yβ†ͺXY\hookrightarrow X is a morphism of 11-dimensional spaces.

  3. (3)

    Let f:Xβ†’Xβ€²f:X\to X^{\prime} be a morphism of 11-dimensional spaces and Yβ€²Y^{\prime} be a 11-dimensional subspace of Xβ€²X^{\prime}. Let FF be the set of connected components of fβˆ’1​(Yβ€²)f^{-1}(Y^{\prime}) that are points. Then FF is finite, Y=fβˆ’1​(Yβ€²)βˆ’FY=f^{-1}(Y^{\prime})-F is a 11-dimensional subspace of XX and f|Y:Yβ†’Yβ€²f_{|Y}:Y\to Y^{\prime} is a morphism of 11-dimensional spaces.

We now provide a description of the local structure of morphisms of 11-dimensional spaces.

0P8D

Lemma 7.1.8. Let f:Xβ†’Xβ€²f:X\to X^{\prime} be a morphism of 11-dimensional spaces and let xβ€²βˆˆXβ€²x^{\prime}\in X^{\prime}. Let r=|fβˆ’1​(xβ€²)|r=|f^{-1}(x^{\prime})|. There exists

  • β€’

    a small open neighbourhood UU of xβ€²x^{\prime} and a homeomorphism a:St⁑(nxβ€²)β†’βˆΌUa:\operatorname{St}\nolimits(n_{x^{\prime}})\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}U with a⁑(0)=xβ€²a(0)=x^{\prime},

  • β€’

    a family of disjoint subsets I0,I1,…,IrI_{0},I_{1},\ldots,I_{r} of {e2​i​π​d/nxβ€²}0≀d<nxβ€²\{e^{2i\pi d/n_{x^{\prime}}}\}_{0\leq d<n_{x^{\prime}}} with Ilβ‰ βˆ…I_{l}\neq\emptyset for 1≀l≀r1\leq l\leq r and a homeomorphism b:St∘⁑(I0)βŠ”St⁑(I1)βŠ”β‹―βŠ”St⁑(Ir)β†’βˆΌfβˆ’1​(U)b:\operatorname{St}\nolimits^{\circ}(I_{0})\sqcup\operatorname{St}\nolimits(I_{1})\sqcup\cdots\sqcup\operatorname{St}\nolimits(I_{r})\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}f^{-1}(U)

such that f|fβˆ’1(U)=a∘g∘bβˆ’1f_{|f^{-1}(U)}=a\circ g\circ b^{-1} where g:St∘⁑(I0)βŠ”St⁑(I1)βŠ”β‹―βŠ”St⁑(Ir)β†’St⁑(nxβ€²)g:\operatorname{St}\nolimits^{\circ}(I_{0})\sqcup\operatorname{St}\nolimits(I_{1})\sqcup\cdots\sqcup\operatorname{St}\nolimits(I_{r})\to\operatorname{St}\nolimits(n_{x^{\prime}}) is the map whose restriction to St∘⁑(I0)\operatorname{St}\nolimits^{\circ}(I_{0}) and St⁑(Il)\operatorname{St}\nolimits(I_{l}) is the inclusion map.

In particular, the canonical map, still denoted by f:T⁑(X)β†’T⁑(Xβ€²)f:T(X)\to T(X^{\prime}) is injective and f⁑(Xe​x​c)βŠ‚Xe​x​cβ€²f(X_{exc})\subset X^{\prime}_{exc}.

0P8E

Proof. Let EE be a finite subset of XX such that fβˆ’1​(f​(E))=Ef^{-1}(f(E))=E, f⁑(Xβˆ’E)f(X-E) is open in Xβ€²X^{\prime} and f|Xβˆ’E:Xβˆ’Eβ†’f(Xβˆ’E)f_{|X-E}:X-E\to f(X-E) is a homeomorphism. Let UU be a small open neighbourhood of xβ€²x^{\prime} such that Uβˆ’{xβ€²}βŠ‚Xβ€²βˆ’f⁑(E)U-\{x^{\prime}\}\subset X^{\prime}-f(E). Note that f⁑(X)∩(Uβˆ’{xβ€²})f(X)\cap(U-\{x^{\prime}\}) is open in Xβ€²X^{\prime} and f|fβˆ’1(Uβˆ’{xβ€²}):fβˆ’1(Uβˆ’{xβ€²})β†’f(X)∩(Uβˆ’{xβ€²})f_{|f^{-1}(U-\{x^{\prime}\})}:f^{-1}(U-\{x^{\prime}\})\to f(X)\cap(U-\{x^{\prime}\}) is a homeomorphism.

Let LL be a connected component of Uβˆ’{xβ€²}U-\{x^{\prime}\}. Note that f​(fβˆ’1​(L))f(f^{-1}(L)) is an open 11-dimensional subspace of LL and LL is homeomorphic to 𝐑{\mathbf{R}}. By shrinking UU, we can assume that fβˆ’1​(L)=βˆ…f^{-1}(L)=\emptyset or f​(fβˆ’1​(L))=Lf(f^{-1}(L))=L. So, we can assume that given LL a connected component of Uβˆ’{xβ€²}U-\{x^{\prime}\} with fβˆ’1​(L)β‰ βˆ…f^{-1}(L)\neq\emptyset, the map f|fβˆ’1(L):fβˆ’1(L)β†’Lf_{|f^{-1}(L)}:f^{-1}(L)\to L is a homeomorphism.

Since UU is small, there is a homeomorphism a:St⁑(nxβ€²)β†’βˆΌU, 0↦xβ€²a:\operatorname{St}\nolimits(n_{x^{\prime}})\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}U,\ 0\mapsto x^{\prime}. Let {x1,…,xr}=fβˆ’1​(xβ€²)\{x_{1},\ldots,x_{r}\}=f^{-1}(x^{\prime}) and define

Il={e2​i​π​d/nxβ€²|0≀d<nxβ€²,xl∈fβˆ’1​(a⁑(𝐑>0​e2​i​π​d/nxβ€²))Β―}I_{l}=\{e^{2i\pi d/n_{x^{\prime}}}|0\leq d<n_{x^{\prime}},\ x_{l}\in\overline{f^{-1}(a({\mathbf{R}}_{>0}e^{2i\pi d/n_{x^{\prime}}}))}\}

for l∈{1,…,r}l\in\{1,\ldots,r\}. Define

I0={e2​i​π​d/nxβ€²|0≀d<nxβ€²,fβˆ’1(a(𝐑>0e2​i​π​d/nxβ€²))β‰ βˆ…,fβˆ’1(xβ€²)∩fβˆ’1​(a⁑(𝐑>0​e2​i​π​d/nxβ€²))Β―=βˆ…}.I_{0}=\{e^{2i\pi d/n_{x^{\prime}}}|0\leq d<n_{x^{\prime}},\ f^{-1}(a({\mathbf{R}}_{>0}e^{2i\pi d/n_{x^{\prime}}}))\neq\emptyset,\ f^{-1}(x^{\prime})\cap\overline{f^{-1}(a({\mathbf{R}}_{>0}e^{2i\pi d/n_{x^{\prime}}}))}=\emptyset\}.

Note that aa restricts to a homeomorphism St⁑(⋃0≀l≀rIr)β†’βˆΌf⁑(fβˆ’1​(U))\operatorname{St}\nolimits(\bigcup_{0\leq l\leq r}I_{r})\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}f(f^{-1}(U)).

The composition a∘ga\circ g takes values in f​(fβˆ’1​(U))f(f^{-1}(U)). Its restriction to St∘⁑(I0)\operatorname{St}\nolimits^{\circ}(I_{0}) defines a homeomorphism St∘⁑(I0)β†’βˆΌa⁑(St∘⁑(I0))\operatorname{St}\nolimits^{\circ}(I_{0})\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}a(\operatorname{St}\nolimits^{\circ}(I_{0})). Since f|fβˆ’1(a(St∘(I0))):fβˆ’1(a(St∘(I0)))β†’a(St∘(I0))f_{|f^{-1}(a(\operatorname{St}\nolimits^{\circ}(I_{0})))}:f^{-1}(a(\operatorname{St}\nolimits^{\circ}(I_{0})))\to a(\operatorname{St}\nolimits^{\circ}(I_{0})) is a homeomorphism, we have a homeomorphism b0=(f|fβˆ’1(a(St∘(I0))))βˆ’1∘(a∘g)|St∘(I0):St∘(I0)β†’βˆΌfβˆ’1(a(St∘(I0)))b_{0}=(f_{|f^{-1}(a(\operatorname{St}\nolimits^{\circ}(I_{0})))})^{-1}\circ(a\circ g)_{|\operatorname{St}\nolimits^{\circ}(I_{0})}:\operatorname{St}\nolimits^{\circ}(I_{0})\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}f^{-1}(a(\operatorname{St}\nolimits^{\circ}(I_{0}))).

Consider now l∈{1,…,r}l\in\{1,\ldots,r\}. We construct as above a homeomorphism blβ€²:St∘⁑(Il)β†’βˆΌfβˆ’1​(a⁑(St∘⁑(Il)))b^{\prime}_{l}:\operatorname{St}\nolimits^{\circ}(I_{l})\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}f^{-1}(a(\operatorname{St}\nolimits^{\circ}(I_{l}))) such that (a∘g)|St∘(Il)=f∘blβ€²(a\circ g)_{|\operatorname{St}\nolimits^{\circ}(I_{l})}=f\circ b^{\prime}_{l}. The homeomorphism blβ€²b^{\prime}_{l} extends uniquely to a homeomorphism bl:St⁑(Il)β†’fβˆ’1​(a⁑(St⁑(Il)))b_{l}:\operatorname{St}\nolimits(I_{l})\to f^{-1}(a(\operatorname{St}\nolimits(I_{l}))). We define b=b0βŠ”b1βŠ”β‹―βŠ”brb=b_{0}\sqcup b_{1}\sqcup\cdots\sqcup b_{r}. We have f|fβˆ’1(U)=a∘g∘bβˆ’1f_{|f^{-1}(U)}=a\circ g\circ b^{-1}. ∎

0P8F

Example 7.1.9. Here is an example of map gg as in Lemma 7.1.8:

[Uncaptioned image]

The next two results follow immediately from Lemma 7.1.8.

0P8G

Lemma 7.1.10. Let YY be a 11-dimensional subspace of XX and let y∈Yy\in Y. Let I={e2​i​π​d/ny,X}0≀d<ny,YI=\{e^{2i\pi d/n_{y,X}}\}_{0\leq d<n_{y,Y}}. There is an open neighbourhood UU of yy in XX and a homeomorphism St⁑(ny,X)β†’βˆΌU, 0↦y\operatorname{St}\nolimits(n_{y,X})\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}U,\ 0\mapsto y whose restriction to St⁑(I)\operatorname{St}\nolimits(I) is a homeomorphism St⁑(I)β†’βˆΌU∩Y\operatorname{St}\nolimits(I)\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}U\cap Y. We have a commutative diagram

St⁑(ny,X)\textstyle{\operatorname{St}\nolimits(n_{y,X})\ignorespaces\ignorespaces\ignorespaces\ignorespaces}∼\scriptstyle{\sim}U\textstyle{U}St⁑(I)\textstyle{\operatorname{St}\nolimits(I)\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}∼\scriptstyle{\sim}U∩Y\textstyle{U\cap Y\ignorespaces\ignorespaces\ignorespaces\ignorespaces}
0P8H

Lemma 7.1.11. Let f:Xβ†’Xβ€²f:X\to X^{\prime} be a surjective morphism of 11-dimensional spaces. It induces a bijection T⁑(X)β†’βˆΌT⁑(Xβ€²)T(X)\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}T(X^{\prime}).

7.1.3. Quotients

Let X~\tilde{X} be a 11-dimensional space and ∼\sim be an equivalence relation on X~\tilde{X}.

0P8I

Definition 7.1.12. We say that ∼\sim is a finite relation if the set of points that are not alone in their equivalence class is finite.

Assume ∼\sim is a finite relation. Let q:X~β†’X=X~/∼q:\tilde{X}\to X=\tilde{X}/\!\!\sim be the quotient map. Note that XX is a 11-dimensional space with

Xe​x​c=q(X~e​x​c)βˆͺ{x∈X||qβˆ’1(x)|>2}βˆͺ{x∈X||qβˆ’1(x)|=2,qβˆ’1(x)βŠ„βˆ‚X~}X_{exc}=q(\tilde{X}_{exc})\cup\{x\in X|\ |q^{-1}(x)|>2\}\cup\{x\in X\ |\ |q^{-1}(x)|=2,\ q^{-1}(x){\not\subset}\partial\tilde{X}\}

and qq is a morphism of 11-dimensional spaces.

Given x∈Xx\in X, the quotient map induces a bijection q:∐x~∈qβˆ’1​(x)C⁑(x~)β†’βˆΌC⁑(x)q:\coprod_{\tilde{x}\in q^{-1}(x)}C(\tilde{x})\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}C(x).

Quotients have a universal property. In particular, we have the following result.

0P8J

Lemma 7.1.13. Let f:Xβ†’Xβ€²f:X\to X^{\prime} be a morphism of 11-dimensional spaces. Define an equivalence relation on XX by x1∼x2x_{1}\sim x_{2} if f⁑(x1)=f⁑(x2)f(x_{1})=f(x_{2}). This defines a finite relation on XX and ff factors uniquely as a composition f=f¯∘qf=\bar{f}\circ q where fΒ―:X/βˆΌβ†’Xβ€²\bar{f}:X/\!\!\sim\ \to X^{\prime} is a morphism of 11-dimensional spaces and q:Xβ†’X/∼q:X\to X/\!\!\sim is the quotient map.

The next lemma shows that 11-dimensional spaces XX can be viewed (non-uniquely) as 11-dimensional manifolds with a finite relation.

0P8K

Lemma 7.1.14. Given XX a 11-dimensional space, there is a 11-dimensional manifold X^\hat{X} with a finite relation ∼\sim and an isomorphism f:X^/βˆΌβ†’βˆΌXf:\hat{X}/\!\!\sim\ \mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}X such that f⁑(X^f)=Xe​x​cf(\hat{X}_{f})=X_{exc}.

0P8L

Proof. Fix, for every x∈Xe​x​cx\in X_{exc}, a small open neighbourhood UxU_{x} of xx and a homeomorphism fx:Uxβ†’βˆΌSt⁑(Ex)f_{x}:U_{x}\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}\mathrm{St}(E_{x}), where ExE_{x} is a finite subset of S1S^{1}. We choose now an equivalence relation on ExE_{x} whose classes have cardinality at most 22. Note that fxf_{x} induces a bijection between C⁑(x)C(x) and ExE_{x}, hence the equivalence relation can be viewed on C⁑(x)C(x).

Define U^x=∐Eβ€²βˆˆEx/∼St(Eβ€²)\hat{U}_{x}=\coprod_{E^{\prime}\in E_{x}/\!\sim}\operatorname{St}\nolimits(E^{\prime}). The map fxf_{x} provides an open embedding

Uxβˆ’{x}β†’βˆΌSt∘(Ex)β†’βˆΌβˆEβ€²βˆˆEx/∼St∘(Eβ€²)β†ͺU^x.U_{x}-\{x\}\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}\mathrm{St}^{\circ}(E_{x})\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}\coprod_{E^{\prime}\in E_{x}/\!\sim}\operatorname{St}\nolimits^{\circ}(E^{\prime})\hookrightarrow\hat{U}_{x}.

We put

X^=(Xβˆ’Xe​x​c)β€‹βˆ(∐x∈Xe​x​c(Uxβˆ’{x}))(∐x∈Xe​x​cU^x).\hat{X}=(X-X_{exc})\coprod_{(\coprod_{x\in X_{exc}}(U_{x}-\{x\}))}\bigl(\coprod_{x\in X_{exc}}\hat{U}_{x}\bigr).

Note that X^\hat{X} is a 11-dimensional manifold. Let q:X^β†’Xq:\hat{X}\to X be the canonical map: it identifies XX with the quotient of X^\hat{X} by the equivalence relation given by x^1∼x^2\hat{x}_{1}\sim\hat{x}_{2} if q⁑(x^1)=q⁑(x^2)q(\hat{x}_{1})=q(\hat{x}_{2}). Up to isomorphism, X^\hat{X} depends only on the choice of an equivalence relation on C⁑(x)C(x) for x∈Xe​x​cx\in X_{exc}. ∎

7.1.4. Paths

0P8M

Lemma 7.1.15. Let EE be a finite subset of XX and Ξ³\gamma be a path in XX such that for all connected components II of [0,1]βˆ–Ξ³βˆ’1​(E)[0,1]\setminus\gamma^{-1}(E), the restriction of Ξ³\gamma to IΒ―\bar{I} is nullhomotopic. Then Ξ³\gamma is nullhomotopic.

0P8N

Proof. Given e∈Ee\in E, let UeU_{e} be a connected and simply connected open neighborhood of ee. Choose UeU_{e} small enough so that Ue∩Ueβ€²=βˆ…U_{e}\cap U_{e^{\prime}}=\emptyset for eβ‰ eβ€²e\neq e^{\prime}. Let U=⋃e∈EUeU=\bigcup_{e\in E}U_{e}. Let VV be an open subset of Xβˆ–EX\setminus E containing Xβˆ–UX\setminus U.

Let CC be the set of connected components II of [0,1]βˆ–Ξ³βˆ’1​(E)[0,1]\setminus\gamma^{-1}(E) such that IΒ―\bar{I} is not contained in Ξ³βˆ’1​(U)\gamma^{-1}(U) nor in Ξ³βˆ’1​(V)\gamma^{-1}(V). By Lebesgue’s number Lemma, that set is finite. Since the restriction of Ξ³\gamma to IΒ―\bar{I} is nullhomotopic for I∈CI\in C, it follows that Ξ³\gamma is homotopic to a path Ξ³β€²\gamma^{\prime} that is constant on IΒ―\bar{I} for I∈CI\in C and that coincides with Ξ³\gamma on [0,1]βˆ’β‹ƒI∈CI[0,1]-\bigcup_{I\in C}I. Let Iβ€²I^{\prime} be a connected component of [0,1]βˆ–Ξ³βˆ’1​(E)[0,1]\setminus\gamma^{-1}(E) with Iβ€²βˆ‰CI^{\prime}{\not\in}C. We have IΒ―βˆ©Ξ³βˆ’1​(E)β‰ βˆ…\bar{I}\cap\gamma^{-1}(E)\neq\emptyset, hence IΒ―βŠ‚Ξ³βˆ’1​(U)\bar{I}\subset\gamma^{-1}(U). We deduce that γ′​([0,1])βŠ‚U\gamma^{\prime}([0,1])\subset U, hence Ξ³β€²\gamma^{\prime} is nullhomotopic. ∎

0P8P

Lemma 7.1.16. Let EE be a finite subset of XX and Ξ³\gamma a path in XX. Let BB be the set of connected components II of [0,1]βˆ–Ξ³βˆ’1​(E)[0,1]\setminus\gamma^{-1}(E) such that Ξ³|IΒ―\gamma_{|\bar{I}} is not nullhomotopic. Then BB is finite and there are paths Ξ³β€²\gamma^{\prime} and Ξ³β€²β€²\gamma^{\prime\prime} homotopic to Ξ³\gamma such that

  • β€’

    Ξ³\gamma and Ξ³β€²\gamma^{\prime} coincide on ⋃I∈BIΒ―\bigcup_{I\in B}\bar{I} and γ′​([0,1]βˆ–β‹ƒI∈BIΒ―)βŠ‚E\gamma^{\prime}([0,1]\setminus\bigcup_{I\in B}\bar{I})\subset E

  • β€’

    Ξ³β€²β€²βˆ’1(E)\gamma^{\prime\prime-1}(E) is finite.

0P8Q

Proof. Let 𝒰{\mathcal{U}} be an open covering of XX by connected and simply connected subsets, each of which contain at most one element of EE. By Lebesgue’s number Lemma, there are only finitely many IβˆˆΟ€0​([0,1]βˆ–Ξ³βˆ’1​(E))I\in\pi_{0}([0,1]\setminus\gamma^{-1}(E)) such that IΒ―\bar{I} is not contained in an element of Ξ³βˆ’1​(𝒰)\gamma^{-1}({\mathcal{U}}). So, BB is finite.

We can write Ξ³\gamma as a finite composition of its restrictions to IΒ―\bar{I} for I∈BI\in B interlaced with finitely many paths that satisfy the assumptions of Lemma 7.1.15. Thanks to that lemma, we obtain a path Ξ³β€²\gamma^{\prime} satisfying the requirements of the lemma. By shrinking the intervals on which Ξ³β€²\gamma^{\prime} is constant to points, we obtain a path Ξ³β€²β€²\gamma^{\prime\prime} as desired. ∎

0P8R

Definition 7.1.17. We say that a path Ξ³\gamma in a 11-dimensional space XX is minimal if there is a finite covering of [0,1][0,1] by open subsets such that the restriction of Ξ³\gamma to any of those open subsets is injective.

Given a continuous map f:Xβ†’Xβ€²f:X\to X^{\prime} and a path Ξ³:[0,1]β†’X\gamma:[0,1]\to X, we will usually denote by f⁑(Ξ³)f(\gamma) the path f∘γf\circ\gamma.

We denote by [γ][\gamma] the homotopy class of a path γ\gamma. Note that we always consider homotopies relative to the endpoints. We denote by Π⁑(X)\Pi(X) the fundamental groupoid of XX.

Given x0,x1∈Xx_{0},x_{1}\in X such that there is a unique homotopy class of paths from x0x_{0} to x1x_{1} in XX, we denote by [x0β†’x1][x_{0}\to x_{1}] that homotopy class.

The following lemma is classical for 11-dimensional finite CW-complexes.

0P8S

Lemma 7.1.18. Let XX be a 11-dimensional space. A homotopy class of paths in XX contains a minimal path if and only if it is not an identity.

Given Ξ³,Ξ³β€²\gamma,\gamma^{\prime} two homotopic minimal paths in XX, there is a homeomorphism Ο•:[0,1]β†’βˆΌ[0,1]\phi:[0,1]\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}[0,1] with ϕ⁑(0)=0\phi(0)=0 and ϕ⁑(1)=1\phi(1)=1 such that Ξ³β€²=Ξ³βˆ˜Ο•\gamma^{\prime}=\gamma\circ\phi.

0P8T

Proof. Let Ξ³1\gamma_{1}, Ξ³2\gamma_{2} be two minimal paths in XX with Ξ³1​(1)=Ξ³2​(0)\gamma_{1}(1)=\gamma_{2}(0). The path Ξ³2∘γ1\gamma_{2}\circ\gamma_{1} is minimal if and only if there are t1,t2∈(0,1)t_{1},t_{2}\in(0,1) such that Ξ³1​((t1,1))∩γ2​((0,t2))=βˆ…\gamma_{1}((t_{1},1))\cap\gamma_{2}((0,t_{2}))=\emptyset. If Ξ³2∘γ1\gamma_{2}\circ\gamma_{1} is not minimal, then there are unique elements t1∈[0,1)t_{1}\in[0,1) and t2∈(0,1]t_{2}\in(0,1] such that (Ξ³2)|[0,t2]∘(Ξ³1)|[t1,1](\gamma_{2})_{|[0,t_{2}]}\circ(\gamma_{1})_{|[t_{1},1]} is homotopic to a constant path and (Ξ³2)|[t2,1]∘(Ξ³1)|[0,t1](\gamma_{2})_{|[t_{2},1]}\circ(\gamma_{1})_{|[0,t_{1}]} is minimal (if t2β‰ 1t_{2}\neq 1 or t1β‰ 0t_{1}\neq 0).

We deduce by induction that a composition of minimal paths is homotopic to a minimal path or to a constant path.

Let Ξ³\gamma be a path in XX. If XX is homeomorphic to an interval of 𝐑{\mathbf{R}}, then Ξ³\gamma is homotopic to a minimal path or a constant path. In general there is a finite subset EE of XX such that given UU a connected component of Xβˆ–EX\setminus E, the space UΒ―\bar{U} is homeomorphic to an interval of 𝐑{\mathbf{R}}. By Lemma 7.1.16 there is a path Ξ³β€²\gamma^{\prime} homotopic to Ξ³\gamma and such that Ξ³β€²βˆ’1​(E)\gamma^{\prime-1}(E) is finite. So, Ξ³β€²\gamma^{\prime} is a composition of paths contained in subspaces of XX that are homeomorphic to intervals of 𝐑{\mathbf{R}}. Consequently, Ξ³β€²\gamma^{\prime} is a composition of minimal paths. It follows that Ξ³β€²\gamma^{\prime}, hence Ξ³\gamma, is homotopic to a minimal or constant path.

Let Ξ³\gamma be a path homotopic to a constant path. The image Ξ³Β―\bar{\gamma} of Ξ³\gamma in XΒ―=X/(Xe​x​cβˆͺ{γ⁑(0),γ⁑(1)})\bar{X}=X/(X_{exc}\cup\{\gamma(0),\gamma(1)\}) is homotopic to a constant path. Since XΒ―\bar{X} is homotopy equivalent to a wedge of circles, its fundamental group is free and Ξ³Β―\bar{\gamma} cannot be a minimal path. It follows that Ξ³\gamma is not minimal.

Let Ξ³\gamma be a minimal path. Let {0=t0<t1<…<tn=1}={0,1}βˆͺΞ³βˆ’1(Xe​x​c)\{0=t_{0}<t_{1}<\ldots<t_{n}=1\}=\{0,1\}\cup\gamma^{-1}(X_{exc}). Note that γ⁑((ti,ti+1))\gamma((t_{i},t_{i+1})) is contained in a connected component UiU_{i} of Xβˆ–Xe​x​cX\setminus X_{exc} and it is a connected component if γ⁑(ti),γ⁑(ti+1)∈Xe​x​c\gamma(t_{i}),\ \gamma(t_{i+1})\in X_{exc}. If UΒ―i\bar{U}_{i} is homeomorphic to an interval of 𝐑{\mathbf{R}}, then Uiβ‰ Ui+1U_{i}\neq U_{i+1} and Uiβ‰ Uiβˆ’1U_{i}\neq U_{i-1}. Otherwise, UΒ―i\bar{U}_{i} is homeomorphic to S1S^{1} and if Ui=Ui+1U_{i}=U_{i+1}, then the paths Ξ³|Ui\gamma_{|U_{i}} and Ξ³|Ui+1\gamma_{|U_{i+1}} have the same orientation.

Let Ξ³β€²\gamma^{\prime} be a minimal path homotopic to Ξ³\gamma. We will show the existence of Ο•\phi as in the lemma by induction on nn. Since Ξ³βˆ˜Ξ³β€²βˆ’1\gamma\circ\gamma^{\prime-1} is not minimal, there is Ξ΅>0\varepsilon>0 such that γ′​([0,Ξ΅])βŠ‚UΒ―1\gamma^{\prime}([0,\varepsilon])\subset\bar{U}_{1}. Consider Ξ΅\varepsilon maximal with this property.

Assume γ′​(Ξ΅)βˆ‰Xe​x​c\gamma^{\prime}(\varepsilon){\not\in}X_{exc}. We have Ξ΅=1\varepsilon=1. Let Ξ΅β€²βˆˆ(t0,t1]\varepsilon^{\prime}\in(t_{0},t_{1}] such that γ⁑(Ξ΅β€²)=γ′​(Ξ΅)\gamma(\varepsilon^{\prime})=\gamma^{\prime}(\varepsilon). The path Ξ³|[Ξ΅β€²,1]\gamma_{|[\varepsilon^{\prime},1]} is homotopic to the identity, hence n=1n=1, Ξ΅β€²=1\varepsilon^{\prime}=1 and γ⁑(t1)=γ′​(Ξ΅)\gamma(t_{1})=\gamma^{\prime}(\varepsilon).

If γ′​(Ξ΅)∈Xe​x​c\gamma^{\prime}(\varepsilon)\in X_{exc}, then γ′​(Ξ΅)=γ⁑(t1)\gamma^{\prime}(\varepsilon)=\gamma(t_{1}) as well. In both cases, the paths Ξ³|[0,t1]\gamma_{|[0,t_{1}]} and Ξ³β€²|[0,Ξ΅]\gamma^{\prime}_{|[0,\varepsilon]} are injective and have the same image. So, there is a homeomorphism ψ:[0,Ξ΅]β†’βˆΌ[0,t1]\psi:[0,\varepsilon]\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}[0,t_{1}] such that γ′​(t)=γ⁑(ψ⁑(t))\gamma^{\prime}(t)=\gamma(\psi(t)) for t∈[0,Ξ΅]t\in[0,\varepsilon] and the existence of ψ\psi follows by induction.

∎

0P8U

Definition 7.1.19. Let ΢\zeta be a non-identity homotopy class of paths in a 11-dimensional space XX. We define the support supp⁑(΢)\operatorname{supp}\nolimits(\zeta) of ΢\zeta to be the subspace γ⁑([0,1])\gamma([0,1]) of XX, where γ\gamma is a minimal path in ΢\zeta.

Lemma 7.1.18 ensures that the support is well defined. Note that supp⁑(ΞΆ)=⋂γγ⁑([0,1])\operatorname{supp}\nolimits(\zeta)=\bigcap_{\gamma}\gamma([0,1]), where Ξ³\gamma runs over paths with [Ξ³]=ΞΆ[\gamma]=\zeta.

Since a minimal path [0,1]β†’X[0,1]\to X is a morphism of 11-dimensional manifolds, it follows that the support of ΞΆ\zeta is a compact connected 11-dimensional subspace of XX.

We define the support of the identity homotopy class idx\operatorname{id}\nolimits_{x} at a point xx to be {x}\{x\}.

0P8V

Lemma 7.1.20. Let f:X→X′f:X\to X^{\prime} be a morphism of 11-dimensional spaces and let γ\gamma, γ′\gamma^{\prime} be two paths in XX.

  • β€’

    γ\gamma is minimal if and only if f⁑(γ)f(\gamma) is minimal. In particular, supp([f(γ)])=f(supp([γ)])\operatorname{supp}\nolimits([f(\gamma)])=f(\operatorname{supp}\nolimits([\gamma)]).

  • β€’

    If f⁑(Ξ³)=f⁑(Ξ³β€²)f(\gamma)=f(\gamma^{\prime}), then Ξ³=Ξ³β€²\gamma=\gamma^{\prime} or Ξ³\gamma and Ξ³β€²\gamma^{\prime} are constant paths at two distinct points of XX having the same image under ff.

  • β€’

    If [f⁑(Ξ³)]=[f⁑(Ξ³β€²)][f(\gamma)]=[f(\gamma^{\prime})], then [Ξ³]=[Ξ³β€²][\gamma]=[\gamma^{\prime}] or [Ξ³]=idx1[\gamma]=\operatorname{id}\nolimits_{x_{1}} and [Ξ³β€²]=idx2[\gamma^{\prime}]=\operatorname{id}\nolimits_{x_{2}} for some x1β‰ x2∈Xx_{1}\neq x_{2}\in X with f⁑(x1)=f⁑(x2)f(x_{1})=f(x_{2}).

0P8W

Proof. A minimal path is a locally injective path. Since every point of XX has an open neighbourhood on which ff is injective (cf Lemma 7.1.8), the image by ff of a minimal path is a minimal path.

Consider the set Ξ©={t∈[0,1]|γ⁑(t)≠γ′​(t)}\Omega=\{t\in[0,1]\ |\ \gamma(t)\neq\gamma^{\prime}(t)\}, an open subset of [0,1][0,1]. Let II be a connected component of Ξ©\Omega. If I=[0,1]I=[0,1], then Ξ³\gamma and Ξ³β€²\gamma^{\prime} are constant paths at distinct points of XX with the same image under ff. Otherwise, let s∈IΒ―βˆ’Is\in\overline{I}-I. There is an open neighbourhood UU of γ​(s)=γ′​(s)\gamma(s)=\gamma^{\prime}(s) such that f|Uf_{|U} is injective. There is t∈It\in I such that γ⁑(t)\gamma(t) and γ′​(t)\gamma^{\prime}(t) are in UU, hence γ​(t)=γ′​(t)\gamma(t)=\gamma^{\prime}(t), a contradiction. This shows the second assertion of the lemma.

Assume Ξ³\gamma and Ξ³β€²\gamma^{\prime} are minimal. Since f⁑(Ξ³)f(\gamma) and f⁑(Ξ³β€²)f(\gamma^{\prime}) are minimal and homotopic, it follows from Lemma 7.1.18 that there is Ο•:[0,1]β†’βˆΌ[0,1]\phi:[0,1]\mathrel{\mathop{\kern 0.0pt\to}\limits^{\sim}}[0,1] with ϕ⁑(0)=0\phi(0)=0 and ϕ⁑(1)=1\phi(1)=1 such that f⁑(Ξ³β€²)=f⁑(Ξ³)βˆ˜Ο•=f⁑(Ξ³βˆ˜Ο•)f(\gamma^{\prime})=f(\gamma)\circ\phi=f(\gamma\circ\phi). It follows from the previous assertion of the lemma that Ξ³β€²=Ξ³βˆ˜Ο•\gamma^{\prime}=\gamma\circ\phi.

Assume now Ξ³\gamma is minimal. Since f⁑(Ξ³)f(\gamma) is minimal, it follows that [f⁑(Ξ³β€²)][f(\gamma^{\prime})] is not the identity, hence [Ξ³β€²][\gamma^{\prime}] is not the identity. We deduce that the third assertion of the lemma holds when [Ξ³][\gamma] and [Ξ³β€²][\gamma^{\prime}] are not both identities. The case where they are both identities is clear. ∎

7.1.5. Tangential multiplicity

Let XX be a 11-dimensional space. Let x∈Xx\in X and UU be a small open neighborhood of xx.

Let c∈C⁑(x)c\in C(x) and let UcU_{c} be the connected component of Uβˆ’{x}U-\{x\} corresponding to cc. Given Ξ³\gamma a path in XX, let Ic+​(Ξ³)I_{c}^{+}(\gamma) (resp. Icβˆ’β€‹(Ξ³)I_{c}^{-}(\gamma)) be the set of elements t∈[0,1]t\in[0,1] such that γ⁑(t)=x\gamma(t)=x and there is Ξ΅>0\varepsilon>0 with t+Ξ΅<1t+\varepsilon<1 and γ⁑((,,,))βŠ‚Uc\gamma((t,t+\varepsilon))\subset U_{c} (resp. tβˆ’Ξ΅>0t-\varepsilon>0 and γ⁑((,,,))βŠ‚Uc\gamma((t-\varepsilon,t))\subset U_{c}).

When Ξ³\gamma is minimal, the set Ic±​(Ξ³)I_{c}^{\pm}(\gamma) is finite and it follows from Lemma 7.1.18 that its cardinality depends only on the homotopy class [Ξ³][\gamma]. We put mc±​([Ξ³])=|Ic±​(Ξ³)|βˆˆπ™β‰₯0m_{c}^{\pm}([\gamma])=|I_{c}^{\pm}(\gamma)|\in{\mathbf{Z}}_{\geq 0} for Ξ³\gamma minimal and mc​([Ξ³])=mc+​([Ξ³])βˆ’mcβˆ’β€‹([Ξ³])m_{c}([\gamma])=m_{c}^{+}([\gamma])-m_{c}^{-}([\gamma]). Similarly, whether or not 0∈Ic+0\in I_{c}^{+} depends only on the homotopy class [Ξ³][\gamma] (for Ξ³\gamma minimal).

0P8X

Lemma 7.1.21. Let Ξ³\gamma be a path in XX such that Ξ³βˆ’1​(x)\gamma^{-1}(x) has finitely many connected components, none of which contain 00 or 11 in the closure of their interior.

We have βˆ‚(Ξ³βˆ’1​(x))=⋃c∈C⁑(x)(Ic+​(Ξ³)βˆͺIcβˆ’β€‹(Ξ³))\partial(\gamma^{-1}(x))=\bigcup_{c\in C(x)}(I_{c}^{+}(\gamma)\cup I_{c}^{-}(\gamma)) and |Ic+​(Ξ³)|βˆ’|Icβˆ’β€‹(Ξ³)|=mc​([Ξ³])|I_{c}^{+}(\gamma)|-|I_{c}^{-}(\gamma)|=m_{c}([\gamma]) for all c∈C⁑(x)c\in C(x).

0P8Y

Proof. The first statement is clear. Let us now prove the second statement. That statement is clear if γ⁑((0,1))∩(Xe​x​cβˆͺ{x})=βˆ…\gamma((0,1))\cap(X_{exc}\cup\{x\})=\emptyset.

The left side of the equality is additive under compositions of paths, and so is the right side by Lemma 7.1.22 below.

Assume now Ξ³βˆ’1​(Xe​x​cβˆͺ{x})\gamma^{-1}(X_{exc}\cup\{x\}) is finite. The path Ξ³\gamma is a (finite) composition of paths mapping (0,1)(0,1) into the complement of Xe​x​cβˆͺ{x}X_{exc}\cup\{x\}, hence the statement holds for Ξ³\gamma.

Consider now the general case. The proof of Lemma 7.1.16 for E=Xe​x​cβˆͺ{x}E=X_{exc}\cup\{x\} produces a path Ξ³β€²\gamma^{\prime} homotopic to Ξ³\gamma such that Ξ³β€²βˆ’1​(E)\gamma^{\prime-1}(E) is finite and such that |Ic+​(Ξ³)|βˆ’|Icβˆ’β€‹(Ξ³)|=|Ic+​(Ξ³β€²)|βˆ’|Icβˆ’β€‹(Ξ³β€²)||I_{c}^{+}(\gamma)|-|I_{c}^{-}(\gamma)|=|I_{c}^{+}(\gamma^{\prime})|-|I_{c}^{-}(\gamma^{\prime})|. Since the statement holds for Ξ³β€²\gamma^{\prime}, it follows that it holds for Ξ³\gamma. ∎

Let ΞΆ\zeta be the homotopy class of a minimal path Ξ³\gamma. Let x=΢⁑(0)x=\zeta(0). There is a unique c∈C⁑(x)c\in C(x) such that 0∈Ic​(Ξ³)+0\in I_{c}(\gamma)^{+} and we define ΢⁑(0+)={c}\zeta(0+)=\{c\}. Similarly, we define ΢⁑(1βˆ’)={cβ€²}\zeta(1-)=\{c^{\prime}\}, where cβ€²βˆˆC⁑(΢⁑(1))c^{\prime}\in C(\zeta(1)) is unique such that 1∈Ic​(Ξ³)βˆ’1\in I_{c}(\gamma)^{-}.

When ΞΆ\zeta is the homotopy class of a constant path we put ΢⁑(0+)=C⁑(΢⁑(0))\zeta(0+)=C(\zeta(0)), ΢⁑(1βˆ’)=C⁑(΢⁑(1))\zeta(1-)=C(\zeta(1)) and mc±​(ΞΆ)=mc​(ΞΆ)=0m_{c}^{\pm}(\zeta)=m_{c}(\zeta)=0.

Given a category π’ž{\mathcal{C}}, we denote by H0​(π’ž)H_{0}({\mathcal{C}}) the abelian group generated by maps in π’ž{\mathcal{C}} modulo the relation f+g=f∘gf+g=f\circ g for any two composable maps ff and gg. We denote by ⟦f⟧\llbracket f\rrbracket the class in H0​(π’ž)H_{0}({\mathcal{C}}) of a map ff of π’ž{\mathcal{C}}. Note that if ff is an identity map, then ⟦f⟧=0\llbracket f\rrbracket=0.

Note that H0H_{0} is left adjoint to the functor sending an abelian group to the category with one object with endomorphism monoid that abelian group.

Let R⁑(X)=H0​(Π⁑(X))R(X)=H_{0}(\Pi(X)). Note that R⁑(X)R(X) is generated by the set II of homotopy classes of paths Ξ³\gamma such that Ξ³\gamma is injective. It follows from the description of the composition of two minimal paths in Β§7.1.4 that R⁑(X)R(X) has a presentation with generating set the non-identity homotopy classes of paths and relations [Ξ³βˆ˜Ξ³β€²]=[Ξ³]+[Ξ³β€²][\gamma\circ\gamma^{\prime}]=[\gamma]+[\gamma^{\prime}] if Ξ³\gamma, Ξ³β€²\gamma^{\prime} and Ξ³βˆ˜Ξ³β€²\gamma\circ\gamma^{\prime} are minimal and [Ξ³]+[Ξ³βˆ’1]=0[\gamma]+[\gamma^{-1}]=0 for Ξ³\gamma minimal. Note finally that every element of R⁑(X)R(X) is a linear combination of non-identity homotopy classes of paths such that the intersection between the supports of two distinct homotopy classes is finite.

0P8Z

Lemma 7.1.22. Given c∈T⁑(X)c\in T(X), the map mcm_{c} induces a morphism of groups R⁑(X)→𝐙R(X)\to{\mathbf{Z}}.

0P90

Proof. Consider Ξ³\gamma and Ξ³β€²\gamma^{\prime} two injective composable paths such that Ξ³βˆ˜Ξ³β€²\gamma\circ\gamma^{\prime} is injective. We have mc±​([γ​γ′])=mc±​([Ξ³])+mc±​([Ξ³β€²])m_{c}^{\pm}([\gamma\gamma^{\prime}])=m_{c}^{\pm}([\gamma])+m_{c}^{\pm}([\gamma^{\prime}]).

Consider now Ξ³\gamma a minimal path. We have mc±​([Ξ³])=mcβˆ“β€‹([Ξ³βˆ’1])m_{c}^{\pm}([\gamma])=m_{c}^{\mp}([\gamma^{-1}]), hence mc​([Ξ³])+mc​([Ξ³βˆ’1])=0=mc​([Ξ³βˆ’1∘γ])m_{c}([\gamma])+m_{c}([\gamma^{-1}])=0=m_{c}([\gamma^{-1}\circ\gamma]). The lemma follows. ∎

The next lemma shows how to realize R⁑(X)R(X) as a subgroup of the group of maps U→𝐙U\to{\mathbf{Z}}, where UU is a dense subset of XX.

0P91

Lemma 7.1.23. Let UU be a dense subset of Xβˆ’(βˆ‚XβˆͺXe​x​c)X-(\partial X\cup X_{exc}). Given x∈Ux\in U, fix a group morphism lx:𝐙C⁑(x)→𝐙l_{x}:{\mathbf{Z}}^{C(x)}\to{\mathbf{Z}} that does not factor through the sum map.

The morphism (lx∘(mc)c∈C⁑(x))x∈U:R⁑(X)→𝐙U(l_{x}\circ(m_{c})_{c\in C(x)})_{x\in U}:R(X)\to{\mathbf{Z}}^{U} is injective.

0P92

Proof. Let LL be a non-empty finite subset of II such that supp⁑(ΞΆ)∩supp⁑(ΞΆβ€²)\operatorname{supp}\nolimits(\zeta)\cap\operatorname{supp}\nolimits(\zeta^{\prime}) is finite for any two distinct elements ΞΆ\zeta and ΞΆβ€²\zeta^{\prime} in LL. Let r=βˆ‘ΞΆβˆˆLaΞΆβ€‹βŸ¦ΞΆβŸ§r=\sum_{\zeta\in L}a_{\zeta}\llbracket\zeta\rrbracket where aΞΆβˆˆπ™βˆ’{0}a_{\zeta}\in{\mathbf{Z}}-\{0\} for ΢∈L\zeta\in L. Let ΞΆ0∈L\zeta_{0}\in L. There is x∈supp⁑(ΞΆ0)∩Ux\in\operatorname{supp}\nolimits(\zeta_{0})\cap U with xβˆ‰{ΞΆ0​(0),ΞΆ0​(1)}x{\not\in}\{\zeta_{0}(0),\zeta_{0}(1)\} and xβˆ‰β‹ƒΞΆβˆˆLβˆ’{ΞΆ0}supp⁑(ΞΆ)x{\not\in}\bigcup_{\zeta\in L-\{\zeta_{0}\}}\operatorname{supp}\nolimits(\zeta). Let c∈C⁑(x)c\in C(x) and ι⁑(c)\iota(c) be the other element of C⁑(x)C(x). We have mc​(ΞΆ0)=βˆ’mι⁑(c)​(ΞΆ0)=Β±1m_{c}(\zeta_{0})=-m_{\iota(c)}(\zeta_{0})=\pm 1, while mc​(ΞΆβ€²)=mι⁑(c)​(ΞΆβ€²)=0m_{c}(\zeta^{\prime})=m_{\iota(c)}(\zeta^{\prime})=0 for ΞΆβ€²βˆˆLβˆ’{ΞΆ0}\zeta^{\prime}\in L-\{\zeta_{0}\}. It follows that mc​(r)=βˆ’mι⁑(c)​(r)=Β±aΞ³m_{c}(r)=-m_{\iota(c)}(r)=\pm a_{\gamma}. Consequently, (lx∘(mc,mι⁑(c)))​(r)=Β±lx​(aΞ³,βˆ’aΞ³)β‰ 0\bigl(l_{x}\circ(m_{c},m_{\iota(c)})\bigr)(r)=\pm l_{x}(a_{\gamma},-a_{\gamma})\neq 0. Since every non-zero element of R⁑(X)R(X) is of the form rr as above, the lemma follows. ∎

Let f:X→X′f:X\to X^{\prime} be a morphism of 11-dimensional spaces. The next lemma follows from the injectivity statement of Lemma 7.1.8.

0P93

Lemma 7.1.24. Given x∈Xx\in X, c∈C⁑(X)c\in C(X) and ΞΆ\zeta a homotopy class of paths in XX, we have mf⁑(c)±​(f⁑(ΞΆ))=mc±​(ΞΆ)m_{f(c)}^{\pm}(f(\zeta))=m_{c}^{\pm}(\zeta) and mf⁑(c)​(f⁑(ΞΆ))=mc​(ΞΆ)m_{f(c)}(f(\zeta))=m_{c}(\zeta).

Note that ff induces a morphism of groups f:R⁑(X)β†’R⁑(Xβ€²)f:R(X)\to R(X^{\prime}).

0P94

Lemma 7.1.25. Let HH be the subgroup of R⁑(Xβ€²)R(X^{\prime}) generated by classes [Ξ³][\gamma] with supp⁑(Ξ³)βŠ‚Xβ€²βˆ’f⁑(X)Β―\operatorname{supp}\nolimits(\gamma)\subset\overline{X^{\prime}-f(X)}.

The composition R⁑(X)→𝑓R⁑(Xβ€²)β†’canR⁑(Xβ€²)/HR(X)\xrightarrow{f}R(X^{\prime})\xrightarrow{{\mathrm{can}}}R(X^{\prime})/H is injective.

0P95

Proof. Let Uβ€²=Xβ€²βˆ’(Xfβ€²βˆͺXe​x​cβ€²βˆͺβˆ‚Xβ€²)U^{\prime}=X^{\prime}-(X^{\prime}_{f}\cup X^{\prime}_{exc}\cup\partial X^{\prime}), a dense subset of Xβ€²X^{\prime}. Note that U=fβˆ’1​(Uβ€²)U=f^{-1}(U^{\prime}) is a dense subset of Xβˆ’(Xe​x​cβˆͺβˆ‚X)X-(X_{exc}\cup\partial X). Given xβ€²βˆˆUβ€²x^{\prime}\in U^{\prime}, fix a morphism lxβ€²:𝐙C⁑(xβ€²)→𝐙l_{x^{\prime}}:{\mathbf{Z}}^{C(x^{\prime})}\to{\mathbf{Z}} that does not factor through the sum map. Given x∈Ux\in U, let lx=lxβ€²βˆ˜f:𝐙C⁑(x)→𝐙l_{x}=l_{x^{\prime}}\circ f:{\mathbf{Z}}^{C(x)}\to{\mathbf{Z}}. Lemma 7.1.23 shows that (lx∘(mc)c∈C⁑(x))x∈U:R⁑(X)→𝐙U(l_{x}\circ(m_{c})_{c\in C(x)})_{x\in U}:R(X)\to{\mathbf{Z}}^{U} is injective. This map is equal to the composition

R⁑(X)→𝑓R⁑(Xβ€²)β†’(lxβ€²βˆ˜(mcβ€²)cβ€²βˆˆC⁑(xβ€²))xβ€²βˆˆU′𝐙Uβ€²β†’fβˆ—π™UR(X)\xrightarrow{f}R(X^{\prime})\xrightarrow{(l_{x^{\prime}}\circ(m_{c^{\prime}})_{c^{\prime}\in C(x^{\prime})})_{x^{\prime}\in U^{\prime}}}{\mathbf{Z}}^{U^{\prime}}\xrightarrow{f^{*}}{\mathbf{Z}}^{U}

since mf⁑(c)±​(f⁑(ΞΆ))=mc±​(ΞΆ)m_{f(c)}^{\pm}(f(\zeta))=m_{c}^{\pm}(\zeta) and mf⁑(c)​(f⁑(ΞΆ))=mc​(ΞΆ)m_{f(c)}(f(\zeta))=m_{c}(\zeta) for all x∈Xx\in X, c∈C⁑(X)c\in C(X) and all homotopy classes of paths ΞΆ\zeta in XX (Lemma 7.1.24). Since HH is contained in the kernel of the composition

R⁑(Xβ€²)β†’(lxβ€²βˆ˜(mcβ€²)cβ€²βˆˆC⁑(xβ€²))xβ€²βˆˆU′𝐙Uβ€²β†’fβˆ—π™U,R(X^{\prime})\xrightarrow{(l_{x^{\prime}}\circ(m_{c^{\prime}})_{c^{\prime}\in C(x^{\prime})})_{x^{\prime}\in U^{\prime}}}{\mathbf{Z}}^{U^{\prime}}\xrightarrow{f^{*}}{\mathbf{Z}}^{U},

it follows that the composite map of the lemma is injective. ∎

Given MM a subset of XX, we denote by RM​(X)R_{M}(X) the subgroup of R⁑(X)R(X) generated by classes of paths Ξ³\gamma with endpoints in MM.

Original mathematics by the credited authors. Source collection and HTML conversion remain in progress.

Andrew Manion, Raphael Rouquier

Original source: arXiv:2009.09627v2