research34 min studyLesson 568 in Mathematics

Čech cohomology and sheaf cohomology

Compute global gluing obstructions with an acyclic affine cover and justify when the Čech complex computes derived cohomology.

On this page

What you’ll understand

  • Construct the alternating Čech differential with a consistent trivialization.
  • State the affine acyclicity hypotheses that identify Čech and derived sheaf cohomology.
  • Compute H⁰ and H¹ of selected twists on the projective line using Laurent exponents.

Sheaf gluing tells us when local sections agree, but global geometry also needs a language for failures of solving compatibility equations. Cohomology supplies that language. Čech cochains organize sections on overlaps of a chosen cover; derived sheaf cohomology is an invariant of the sheaf and space independent of that cover. These constructions agree under useful acyclicity hypotheses, not by definition for every arbitrary cover. The projective line allows an explicit calculation in which a few missing Laurent exponents measure a genuine global obstruction even though each affine chart has no higher quasi-coherent cohomology.

The complex measures compatibility and its obstructions

For an ordered open cover, a degree-p Čech cochain assigns a section to every increasing intersection of p+1 opens. The differential is the alternating sum of restrictions, omitting one index at a time. Each twice-omitted restriction appears with opposite signs, so the differential squares to zero. Degree-zero cocycles are exactly compatible families and hence global sections. A degree-one cocycle represents overlap data, while a coboundary is overlap data obtained by changing local choices on the individual opens.

If the sheaf is a line bundle, choose local frames and translate all overlap sections into the same frame before subtracting them. Forgetting a transition function can turn a correct gluing problem into the cohomology of the wrong sheaf. Signs do not change the span of coboundaries in our two-chart calculation, but powers of the coordinate do. Naming the frame convention explicitly makes the quotient calculation reproducible and prevents a positive twist from accidentally being treated as a negative one.

(δc)i0⋯ip+1=∑j=0p+1(−1)jci0⋯ij^⋯ip+1∣Ui0∩⋯∩Uip+1,δ2=0(\delta c)_{i_0\cdots i_{p+1}}=\sum_{j=0}^{p+1}(-1)^j c_{i_0\cdots\widehat{i_j}\cdots i_{p+1}}\big|_{U_{i_0}\cap\cdots\cap U_{i_{p+1}}},\qquad\delta^2=0
All sections are restricted to the common intersection before summation. The equation is for an abelian sheaf or a sheaf of modules, where subtraction is defined.↔ Scroll wide equations horizontally.

Affine acyclicity justifies the comparison theorem

Derived sheaf cohomology is obtained by applying global sections to an injective resolution in the category of sheaves of modules. It is independent of the chosen resolution. For a quasi-coherent sheaf on an affine scheme, all positive-degree cohomology groups vanish. Consequently, if a cover has affine finite intersections, its Čech complex computes the derived sheaf cohomology of a quasi-coherent sheaf in every degree. This is the comparison theorem used here, and it requires both the sheaf condition and the acyclicity input.

A separated scheme has affine intersections of affine opens, so a quasi-compact separated scheme admits a finite affine cover suitable for this calculation. More general acyclic covers are possible; separatedness is a convenient sufficient condition rather than the definition of Čech cohomology. For an arbitrary sheaf or an arbitrary cover, higher cohomology on the intersections can contribute through a spectral sequence, and the naive Čech groups need not equal the derived groups. Local acyclicity must therefore be checked before interpreting a calculated quotient as a global invariant.

Hq(Ui0∩⋯∩Uip,F)=0 (q>0)⟹Hˇp(U,F)≅Hp(X,F)H^q(U_{i_0}\cap\cdots\cap U_{i_p},\mathcal F)=0\ (q>0)\quad\Longrightarrow\quad\check H^p(\mathcal U,\mathcal F)\cong H^p(X,\mathcal F)
Affine finite intersections and quasi-coherence imply the displayed acyclicity. Without it, Čech-to-derived comparison is a theorem requiring further analysis, not an automatic identity.↔ Scroll wide equations horizontally.

On the projective line, Laurent exponents expose the obstruction

Cover the projective line by the affine t-chart and the affine u-chart with u=1/t. For O(n), take frames e₀ and e∞ with e∞=tⁿe₀ on the overlap. Expressed in e₀, chart sections contribute k[t] and tⁿk[t⁻¹] inside the Laurent ring. Their intersection gives global sections, while the quotient of the Laurent ring by their sum gives first cohomology. The two-chart ordered complex has no higher terms.

When n is nonnegative, the intersection has basis 1,t,…,tⁿ, and every Laurent exponent lies in the sum, so first cohomology vanishes. When n is at most minus two, no exponent is in both chart ranges, but the exponents strictly between n and zero are missing from their sum. They give a basis of first cohomology. Thus dimensions can be computed without mistaking global sections for all functions on a single affine chart; the transition power controls the obstruction.

Three obstruction classes for a negative twist

Compute H⁰(P¹,O(−4)) and H¹(P¹,O(−4)) over a field k using the two standard affine charts and the frame convention e∞=t^(−4)e₀.

  1. In the e₀ frame the degree-zero cochain images are k[t] and t^(−4)k[t⁻¹] inside k[t,t⁻¹]. Their monomials have exponents respectively at least 0 and at most −4.

  2. The two exponent ranges do not intersect, so the only globally compatible section is zero. Therefore H⁰(P¹,O(−4))=0.

  3. In the quotient by the sum, precisely the exponents −1,−2,−3 remain. Their Laurent monomials are independent modulo the two chart images and span every remaining class. The affine acyclicity comparison identifies this quotient with H¹.

The result

H¹(P¹,O(−4))≅k³, with representatives t⁻¹,t⁻²,t⁻³ in the chosen frame. Its dimension records a global obstruction invisible on either affine chart separately.

Explore the model

Interactive lab

Interactive lab · spectral

Count independent directions after a quotient calculation

Pause and Reset; set Mode count N=3,Coefficient decay p=0. At the zero clock predict sum 0 at x=π/2, smallest wavelength 2π/3, and highest-mode coefficient 1. For a fresh comparison Reset and set N=1,p=0; predict sum 1 and smallest wavelength 2π. The display shows the finite sum and highest included mode. Its distinct sine frequencies are independent, but they are not the Laurent representatives above. The tool does not compute a coboundary quotient; the three cohomology classes come from the checked Čech calculation.

Count independent directions after a quotient calculationA finite sum of sin(nx − t)/nᵖ. Playback changes t; increasing the cutoff includes finer spatial scales. This is not a proof about a continuum limit or renormalization. Current readings: Sum at x = π/2: 0.867; Smallest wavelength: 2π / 5; Highest-mode coefficient: 0.2. All parameters are adjustable with the labeled controls below.01.573.144.716.28−2.56−1.2801.282.56x · radiansFinite sine sum
Finite sumHighest included mode
Sum at x = π/2
0.867
Smallest wavelength
2π / 5
Highest-mode coefficient
0.2
t = 0 s

A finite sum of sin(nx − t)/nᵖ. Playback changes t; increasing the cutoff includes finer spatial scales. This is not a proof about a continuum limit or renormalization. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.

Look forA cohomology dimension counts independent classes after removing coboundaries; choosing a display with that many modes does not prove the count.

Check your understanding

Why does the two-affine-chart Čech calculation compute derived cohomology for O(n) on P¹?

Try it yourself

With the same frame convention, compute H⁰(P¹,O(3)) and H¹(P¹,O(3)). Identify the relevant intersection and quotient of Laurent polynomial spaces.

Take a moment to make a prediction or work it out on paper. Open a hint when you need one.

Give me a hint

Compare the exponent ranges m≥0 and m≤3 inside k[t,t⁻¹].

Compare with the solution

The intersection k[t]∩t³k[t⁻¹] has basis 1,t,t²,t³, so H⁰ has dimension four. The sum k[t]+t³k[t⁻¹] contains every Laurent monomial, since any integer is either at least zero or at most three. Its quotient is zero, so H¹=0. The affine cover comparison applies exactly as in the worked negative-twist calculation.

Take it with you

  • Čech coboundaries depend on restrictions and the chosen bundle transitions.
  • Affine finite intersections and quasi-coherence justify the comparison with derived cohomology.
  • On P¹, missing Laurent exponents compute the higher cohomology of negative twists.

Read further

Original sources and deeper explanations to follow at your own pace.

Optional progress, saved in this browser when available. Revisit any lesson at any time.

Lesson not marked complete.