Movatterモバイル変換


[0]ホーム

URL:


Jump to content
WikipediaThe Free Encyclopedia
Search

Čech cohomology

From Wikipedia, the free encyclopedia
Theory in algebraic topology
APenrose triangle depicts a nontrivial element of the first cohomology of anannulus with values in the group of distances from the observer.[1]

Inmathematics, specificallyalgebraic topology,Čech cohomology is acohomology theory based on the intersection properties ofopencovers of atopological space. It is named for the mathematicianEduard Čech.

Motivation

[edit]

LetX be a topological space, and letU{\displaystyle {\mathcal {U}}} be an open cover ofX. LetN(U){\displaystyle N({\mathcal {U}})} denote thenerve of the covering. The idea of Čech cohomology is that, for an open coverU{\displaystyle {\mathcal {U}}} consisting of sufficiently small open sets, the resulting simplicial complexN(U){\displaystyle N({\mathcal {U}})} should be a good combinatorial model for the spaceX. For such a cover, the Čech cohomology ofX is defined to be thesimplicialcohomology of the nerve. This idea can be formalized by the notion of agood cover. However, a more general approach is to take thedirect limit of the cohomology groups of the nerve over the system of all possible open covers ofX, ordered byrefinement. This is the approach adopted below.

Construction

[edit]

LetX be atopological space, and letF{\displaystyle {\mathcal {F}}} be apresheaf ofabelian groups onX. LetU{\displaystyle {\mathcal {U}}} be anopen cover ofX.

Simplex

[edit]

Aq-simplex σ ofU{\displaystyle {\mathcal {U}}} is an ordered collection ofq+1 sets chosen fromU{\displaystyle {\mathcal {U}}}, such that the intersection of all these sets is non-empty. This intersection is called thesupport of σ and is denoted |σ|.

Now letσ=(Ui)i{0,,q}{\displaystyle \sigma =(U_{i})_{i\in \{0,\ldots ,q\}}} be such aq-simplex. Thej-th partial boundary of σ is defined to be the (q−1)-simplex obtained by removing thej-th set from σ, that is:

jσ:=(Ui)i{0,,q}{j}.{\displaystyle \partial _{j}\sigma :=(U_{i})_{i\in \{0,\ldots ,q\}\setminus \{j\}}.}

Theboundary of σ is defined as the alternating sum of the partial boundaries:

σ:=j=0q(1)j+1jσ{\displaystyle \partial \sigma :=\sum _{j=0}^{q}(-1)^{j+1}\partial _{j}\sigma }

viewed as an element of thefree abelian group spanned by the simplices ofU{\displaystyle {\mathcal {U}}}.

Cochain

[edit]

Aq-cochain ofU{\displaystyle {\mathcal {U}}} with coefficients inF{\displaystyle {\mathcal {F}}} is a map which associates with eachq-simplex σ an element ofF(|σ|){\displaystyle {\mathcal {F}}(|\sigma |)}, and we denote the set of allq-cochains ofU{\displaystyle {\mathcal {U}}} with coefficients inF{\displaystyle {\mathcal {F}}} byCq(U,F){\displaystyle C^{q}({\mathcal {U}},{\mathcal {F}})}.Cq(U,F){\displaystyle C^{q}({\mathcal {U}},{\mathcal {F}})} is an abelian group by pointwise addition.

Differential

[edit]

The cochain groups can be made into acochain complex(C(U,F),δ){\displaystyle (C^{\bullet }({\mathcal {U}},{\mathcal {F}}),\delta )} by defining thecoboundary operatorδq:Cq(U,F)Cq+1(U,F){\displaystyle \delta _{q}:C^{q}({\mathcal {U}},{\mathcal {F}})\to C^{q+1}({\mathcal {U}},{\mathcal {F}})} by:

(δqf)(σ):=j=0q+1(1)jres|σ||jσ|f(jσ),{\displaystyle \quad (\delta _{q}f)(\sigma ):=\sum _{j=0}^{q+1}(-1)^{j}\mathrm {res} _{|\sigma |}^{|\partial _{j}\sigma |}f(\partial _{j}\sigma ),}

whereres|σ||jσ|{\displaystyle \mathrm {res} _{|\sigma |}^{|\partial _{j}\sigma |}} is therestriction morphism fromF(|jσ|){\displaystyle {\mathcal {F}}(|\partial _{j}\sigma |)} toF(|σ|).{\displaystyle {\mathcal {F}}(|\sigma |).} (Notice that ∂jσ ⊆ σ, but |σ| ⊆ |∂jσ|.)

A calculation shows thatδq+1δq=0.{\displaystyle \delta _{q+1}\circ \delta _{q}=0.}

Thecoboundary operator is analogous to theexterior derivative ofDe Rham cohomology, so it sometimes called the differential of thecochain complex.

Cocycle

[edit]

Aq-cochain is called aq-cocycle if it is in the kernel ofδ{\displaystyle \delta }, henceZq(U,F):=ker(δq)Cq(U,F){\displaystyle Z^{q}({\mathcal {U}},{\mathcal {F}}):=\ker(\delta _{q})\subseteq C^{q}({\mathcal {U}},{\mathcal {F}})} is the set of allq-cocycles.

Thus a (q−1)-cochainf{\displaystyle f} is a cocycle if for allq-simplicesσ{\displaystyle \sigma } the cocycle condition

j=0q(1)jres|σ||jσ|f(jσ)=0{\displaystyle \sum _{j=0}^{q}(-1)^{j}\mathrm {res} _{|\sigma |}^{|\partial _{j}\sigma |}f(\partial _{j}\sigma )=0}

holds.

A 0-cocyclef{\displaystyle f} is a collection of local sections ofF{\displaystyle {\mathcal {F}}} satisfying a compatibility relation on every intersectingA,BU{\displaystyle A,B\in {\mathcal {U}}}

f(A)|AB=f(B)|AB{\displaystyle f(A)|_{A\cap B}=f(B)|_{A\cap B}}

A 1-cocyclef{\displaystyle f} satisfies for every non-emptyU=ABC{\displaystyle U=A\cap B\cap C} withA,B,CU{\displaystyle A,B,C\in {\mathcal {U}}}

f(BC)|Uf(AC)|U+f(AB)|U=0{\displaystyle f(B\cap C)|_{U}-f(A\cap C)|_{U}+f(A\cap B)|_{U}=0}

Coboundary

[edit]

Aq-cochain is called aq-coboundary if it is in the image ofδ{\displaystyle \delta } andBq(U,F):=Im(δq1)Cq(U,F){\displaystyle B^{q}({\mathcal {U}},{\mathcal {F}}):=\mathrm {Im} (\delta _{q-1})\subseteq C^{q}({\mathcal {U}},{\mathcal {F}})} is the set of allq-coboundaries.

For example, a 1-cochainf{\displaystyle f} is a 1-coboundary if there exists a 0-cochainh{\displaystyle h} such that for every intersectingA,BU{\displaystyle A,B\in {\mathcal {U}}}

f(AB)=h(A)|ABh(B)|AB{\displaystyle f(A\cap B)=h(A)|_{A\cap B}-h(B)|_{A\cap B}}

Cohomology

[edit]

TheČech cohomology ofU{\displaystyle {\mathcal {U}}} with values inF{\displaystyle {\mathcal {F}}} is defined to be the cohomology of the cochain complex(C(U,F),δ){\displaystyle (C^{\bullet }({\mathcal {U}},{\mathcal {F}}),\delta )}. Thus theqth Čech cohomology is given by

Hˇq(U,F):=Hq((C(U,F),δ))=Zq(U,F)/Bq(U,F){\displaystyle {\check {H}}^{q}({\mathcal {U}},{\mathcal {F}}):=H^{q}((C^{\bullet }({\mathcal {U}},{\mathcal {F}}),\delta ))=Z^{q}({\mathcal {U}},{\mathcal {F}})/B^{q}({\mathcal {U}},{\mathcal {F}})}.

The Čech cohomology ofX is defined by consideringrefinements of open covers. IfV{\displaystyle {\mathcal {V}}} is a refinement ofU{\displaystyle {\mathcal {U}}} then there is a map in cohomologyHˇ(U,F)Hˇ(V,F).{\displaystyle {\check {H}}^{*}({\mathcal {U}},{\mathcal {F}})\to {\check {H}}^{*}({\mathcal {V}},{\mathcal {F}}).} The open covers ofX form adirected set under refinement, so the above map leads to adirect system of abelian groups. TheČech cohomology ofX with values inF{\displaystyle {\mathcal {F}}} is defined as thedirect limitHˇ(X,F):=limUHˇ(U,F){\displaystyle {\check {H}}(X,{\mathcal {F}}):=\varinjlim _{\mathcal {U}}{\check {H}}({\mathcal {U}},{\mathcal {F}})} of this system.

The Čech cohomology ofX with coefficients in a fixed abelian groupA, denotedHˇ(X;A){\displaystyle {\check {H}}(X;A)}, is defined asHˇ(X,FA){\displaystyle {\check {H}}(X,{\mathcal {F}}_{A})} whereFA{\displaystyle {\mathcal {F}}_{A}} is theconstant sheaf onX determined byA.

A variant of Čech cohomology, callednumerable Čech cohomology, is defined as above, except that all open covers considered are required to benumerable: that is, there is apartition of unityi} such that each support{xρi(x)>0}{\displaystyle \{x\mid \rho _{i}(x)>0\}} is contained in some element of the cover. IfX isparacompact andHausdorff, then numerable Čech cohomology agrees with the usual Čech cohomology.

Relation to other cohomology theories

[edit]

IfX ishomotopy equivalent to aCW complex, then the Čech cohomologyHˇ(X;A){\displaystyle {\check {H}}^{*}(X;A)} isnaturally isomorphic to thesingular cohomologyH(X;A){\displaystyle H^{*}(X;A)\,}. IfX is adifferentiable manifold, thenHˇ(X;R){\displaystyle {\check {H}}^{*}(X;\mathbb {R} )} is also naturally isomorphic to thede Rham cohomology; the article on de Rham cohomology provides a brief review of this isomorphism. For less well-behaved spaces, Čech cohomology differs from singular cohomology. For example ifX is theclosed topologist's sine curve, thenHˇ1(X;Z)=Z,{\displaystyle {\check {H}}^{1}(X;\mathbb {Z} )=\mathbb {Z} ,} whereasH1(X;Z)=0.{\displaystyle H^{1}(X;\mathbb {Z} )=0.}

IfX is a differentiable manifold and the coverU{\displaystyle {\mathcal {U}}} ofX is a "good cover" (i.e. all the setsUα arecontractible to a point, and all finite intersections of sets inU{\displaystyle {\mathcal {U}}} are either empty or contractible to a point), thenHˇ(U;R){\displaystyle {\check {H}}^{*}({\mathcal {U}};\mathbb {R} )} is isomorphic to the de Rham cohomology.

IfX is compact Hausdorff, then Čech cohomology (with coefficients in a discrete group) is isomorphic toAlexander-Spanier cohomology.

For a presheafF{\displaystyle {\mathcal {F}}} onX, letF+{\displaystyle {\mathcal {F}}^{+}} denote itssheafification. Then we have a natural comparison map

χ:Hˇ(X,F)H(X,F+){\displaystyle \chi :{\check {H}}^{*}(X,{\mathcal {F}})\to H^{*}(X,{\mathcal {F}}^{+})}

from Čech cohomology tosheaf cohomology. IfX is paracompact Hausdorff, thenχ{\textstyle \chi } is an isomorphism. More generally,χ{\textstyle \chi } is an isomorphism whenever the Čech cohomology of all presheaves onX with zero sheafification vanishes.[2]

In algebraic geometry

[edit]

Čech cohomology can be defined more generally for objects in asiteC endowed with a topology. This applies, for example, to the Zariski site or the etale site of aschemeX. The Čech cohomology with values in somesheafF{\displaystyle {\mathcal {F}}} is defined as

Hˇn(X,F):=limUHˇn(U,F).{\displaystyle {\check {H}}^{n}(X,{\mathcal {F}}):=\varinjlim _{\mathcal {U}}{\check {H}}^{n}({\mathcal {U}},{\mathcal {F}}).}

where thecolimit runs over all coverings (with respect to the chosen topology) ofX. HereHˇn(U,F){\displaystyle {\check {H}}^{n}({\mathcal {U}},{\mathcal {F}})} is defined as above, except that ther-fold intersections of open subsets inside the ambient topological space are replaced by ther-foldfiber product

U×Xr:=U×X×XU.{\displaystyle {\mathcal {U}}^{\times _{X}^{r}}:={\mathcal {U}}\times _{X}\dots \times _{X}{\mathcal {U}}.}

As in the classical situation of topological spaces, there is always a map

Hˇn(X,F)Hn(X,F){\displaystyle {\check {H}}^{n}(X,{\mathcal {F}})\rightarrow H^{n}(X,{\mathcal {F}})}

from Čech cohomology to sheaf cohomology. It is always an isomorphism in degreesn = 0 and 1, but may fail to be so in general. For theZariski topology on aNoetherianseparated scheme, Čech and sheaf cohomology agree for anyquasi-coherent sheaf. For theétale topology, the two cohomologies agree for any étale sheaf onX, provided that any finite set of points ofX are contained in some open affine subscheme. This is satisfied, for example, ifX isquasi-projective over anaffine scheme.[3]

The possible difference between Čech cohomology and sheaf cohomology is a motivation for the use ofhypercoverings: these are more general objects than the Čechnerve

NXU:U×XU×XUU×XUU.{\displaystyle N_{X}{\mathcal {U}}:\dots \to {\mathcal {U}}\times _{X}{\mathcal {U}}\times _{X}{\mathcal {U}}\to {\mathcal {U}}\times _{X}{\mathcal {U}}\to {\mathcal {U}}.}

A hypercoveringK ofX is a certainsimplicial object inC, i.e., a collection of objectsKn together with boundary and degeneracy maps. Applying a sheafF{\displaystyle {\mathcal {F}}} toK yields asimplicial abelian groupF(K){\textstyle {\mathcal {F}}(K_{\ast })} whosen-th cohomology group is denotedHn(F(K)){\textstyle H^{n}({\mathcal {F}}(K_{\ast }))}. (This group is the same asHˇn(U,F){\displaystyle {\check {H}}^{n}({\mathcal {U}},{\mathcal {F}})} in caseK equalsNXU{\displaystyle N_{X}{\mathcal {U}}}.) Then, it can be shown that there is a canonical isomorphism

Hn(X,F)limKHn(F(K)),{\displaystyle H^{n}(X,{\mathcal {F}})\cong \varinjlim _{K_{*}}H^{n}({\mathcal {F}}(K_{*})),}

where the colimit now runs over all hypercoverings.[4]

Examples

[edit]

The most basic example of Čech cohomology is given by the case where the presheafF{\displaystyle {\mathcal {F}}} is aconstant sheaf, e.g.F=R{\displaystyle {\mathcal {F}}=\mathbb {R} }. In such cases, eachq{\displaystyle q}-cochainf{\displaystyle f} is simply a function which maps everyq{\displaystyle q}-simplex toR{\displaystyle \mathbb {R} }. For example, we calculate the first Čech cohomology with values inR{\displaystyle \mathbb {R} } of the unit circleX=S1{\displaystyle X=S^{1}}. DividingX{\displaystyle X} into three arcs and choosing sufficiently small open neighborhoods, we obtain an open coverU={U0,U1,U2}{\displaystyle {\mathcal {U}}=\{U_{0},U_{1},U_{2}\}} whereUiUj{\displaystyle U_{i}\cap U_{j}\neq \emptyset } butU0U1U2={\displaystyle U_{0}\cap U_{1}\cap U_{2}=\emptyset }.

Given any 1-cocyclef{\displaystyle f},δf{\displaystyle \delta f} is a 2-cochain which takes inputs of the form(Ui,Ui,Ui),(Ui,Ui,Uj),(Uj,Ui,Ui),(Ui,Uj,Ui){\displaystyle (U_{i},U_{i},U_{i}),(U_{i},U_{i},U_{j}),(U_{j},U_{i},U_{i}),(U_{i},U_{j},U_{i})} whereij{\displaystyle i\neq j} (sinceU0U1U2={\displaystyle U_{0}\cap U_{1}\cap U_{2}=\emptyset } and hence(Ui,Uj,Uk){\displaystyle (U_{i},U_{j},U_{k})} is not a 2-simplex for any permutation{i,j,k}={1,2,3}{\displaystyle \{i,j,k\}=\{1,2,3\}}). The first three inputs givef(Ui,Ui)=0{\displaystyle f(U_{i},U_{i})=0}; the fourth gives

δf(Ui,Uj,Ui)=f(Uj,Ui)f(Ui,Ui)+f(Ui,Uj)=0f(Uj,Ui)=f(Ui,Uj).{\displaystyle \delta f(U_{i},U_{j},U_{i})=f(U_{j},U_{i})-f(U_{i},U_{i})+f(U_{i},U_{j})=0\implies f(U_{j},U_{i})=-f(U_{i},U_{j}).}

Such a function is fully determined by the values off(U0,U1),f(U0,U2),f(U1,U2){\displaystyle f(U_{0},U_{1}),f(U_{0},U_{2}),f(U_{1},U_{2})}. Thus,

Z1(U,R)={fC1(U,R):f(Ui,Ui)=0,f(Uj,Ui)=f(Ui,Uj)}R3.{\displaystyle Z^{1}({\mathcal {U}},\mathbb {R} )=\{f\in C^{1}({\mathcal {U}},\mathbb {R} ):f(U_{i},U_{i})=0,f(U_{j},U_{i})=-f(U_{i},U_{j})\}\cong \mathbb {R} ^{3}.}

On the other hand, given any 1-coboundaryf=δg{\displaystyle f=\delta g}, we have

{f(Ui,Ui)=g(Ui)g(Ui)=0(i=0,1,2);f(Ui,Uj)=g(Uj)g(Ui)=f(Uj,Ui)(ij){\displaystyle {\begin{cases}f(U_{i},U_{i})=g(U_{i})-g(U_{i})=0&(i=0,1,2);\\f(U_{i},U_{j})=g(U_{j})-g(U_{i})=-f(U_{j},U_{i})&(i\neq j)\end{cases}}}

However, upon closer inspection we see thatf(U0,U1)+f(U1,U2)=f(U0,U2){\displaystyle f(U_{0},U_{1})+f(U_{1},U_{2})=f(U_{0},U_{2})} and hence each 1-coboundaryf{\displaystyle f} is uniquely determined byf(U0,U1){\displaystyle f(U_{0},U_{1})} andf(U1,U2){\displaystyle f(U_{1},U_{2})}. This gives the set of 1-coboundaries:

B1(U,R)={fC1(U,R): f(Ui,Ui)=0,f(Uj,Ui)=f(Ui,Uj),f(U0,U2)=f(U0,U1)+f(U1,U2)}R2.{\displaystyle {\begin{aligned}B^{1}({\mathcal {U}},\mathbb {R} )=\{f\in C^{1}({\mathcal {U}},\mathbb {R} ):\ &f(U_{i},U_{i})=0,f(U_{j},U_{i})=-f(U_{i},U_{j}),\\&f(U_{0},U_{2})=f(U_{0},U_{1})+f(U_{1},U_{2})\}\cong \mathbb {R} ^{2}.\end{aligned}}}

Therefore,Hˇ1(U,R)=Z1(U,R)/B1(U,R)R{\displaystyle {\check {H}}^{1}({\mathcal {U}},\mathbb {R} )=Z^{1}({\mathcal {U}},\mathbb {R} )/B^{1}({\mathcal {U}},\mathbb {R} )\cong \mathbb {R} }. SinceU{\displaystyle {\mathcal {U}}} is agood cover ofX{\displaystyle X}, we haveHˇ1(X,R)R{\displaystyle {\check {H}}^{1}(X,\mathbb {R} )\cong \mathbb {R} } byLeray's theorem.

We may also compute the coherent sheaf cohomology ofΩ1{\displaystyle \Omega ^{1}} on the projective linePC1{\displaystyle \mathbb {P} _{\mathbb {C} }^{1}} using the Čech complex. Using the cover

U={U1=Spec(C[y]),U2=Spec(C[y1])}{\displaystyle {\mathcal {U}}=\{U_{1}={\text{Spec}}(\mathbb {C} [y]),U_{2}={\text{Spec}}(\mathbb {C} [y^{-1}])\}}

we have the following modules from the cotangent sheaf

Ω1(U1)=C[y]dyΩ1(U2)=C[y1]dy1{\displaystyle {\begin{aligned}&\Omega ^{1}(U_{1})=\mathbb {C} [y]dy\\&\Omega ^{1}(U_{2})=\mathbb {C} \left[y^{-1}\right]dy^{-1}\end{aligned}}}

If we take the conventions thatdy1=(1/y2)dy{\displaystyle dy^{-1}=-(1/y^{2})dy} then we get the Čech complex

0C[y]dyC[y1]dy1d0C[y,y1]dy0{\displaystyle 0\to \mathbb {C} [y]dy\oplus \mathbb {C} \left[y^{-1}\right]dy^{-1}{\xrightarrow {d^{0}}}\mathbb {C} \left[y,y^{-1}\right]dy\to 0}

Sinced0{\displaystyle d^{0}} is injective and the only element not in the image ofd0{\displaystyle d^{0}} isy1dy{\displaystyle y^{-1}dy} we get that

H1(PC1,Ω1)CHk(PC1,Ω1)0 for k1{\displaystyle {\begin{aligned}&H^{1}(\mathbb {P} _{\mathbb {C} }^{1},\Omega ^{1})\cong \mathbb {C} \\&H^{k}(\mathbb {P} _{\mathbb {C} }^{1},\Omega ^{1})\cong 0{\text{ for }}k\neq 1\end{aligned}}}

References

[edit]

Citation footnotes

[edit]
  1. ^Penrose, Roger (1992), "On the Cohomology of Impossible Figures",Leonardo,25 (3/4):245–247,doi:10.2307/1575844,JSTOR 1575844,S2CID 125905129. Reprinted fromPenrose, Roger (1991),"On the Cohomology of Impossible Figures / La Cohomologie des Figures Impossibles",Structural Topology,17:11–16, retrievedJanuary 16, 2014
  2. ^Brady, Zarathustra."Notes on sheaf cohomology"(PDF). p. 11.Archived(PDF) from the original on 2022-06-17.
  3. ^Milne, James S. (1980), "Section III.2, Theorem 2.17",Étale cohomology, Princeton Mathematical Series, vol. 33,Princeton University Press,ISBN 978-0-691-08238-7,MR 0559531
  4. ^Artin, Michael;Mazur, Barry (1969),"Theorem 8.16",Etale homotopy, Lecture Notes in Mathematics, vol. 100, Springer, p. 98,ISBN 978-3-540-36142-8

General references

[edit]
Retrieved from "https://en.wikipedia.org/w/index.php?title=Čech_cohomology&oldid=1272406991"
Category:
Hidden categories:

[8]ページ先頭

©2009-2025 Movatter.jp