Five-term exact sequence

In mathematics, five-term exact sequence or exact sequence of low-degree terms is a sequence of terms related to the first step of a spectral sequence.

More precisely, let

be a first quadrant spectral sequence, meaning that vanishes except when p and q are both non-negative. Then there is an exact sequence

0 E21,0 H 1(A) E20,1 E22,0 H 2(A).

Here, the map is the differential of the -term of the spectral sequence.

Example

0 H 1(G/N, AN) H 1(G, A) H 1(N, A)G/N H 2(G/N, AN) H 2(G, A)
in group cohomology arises as the five-term exact sequence associated to the Lyndon–Hochschild–Serre spectral sequence
H p(G/N, H q(N, A)) H p+q(G, A)
where G is a profinite group, N is a closed normal subgroup, and A is a G-module.

Construction

The sequence is a consequence of the definition of convergence of a spectral sequence. The second page differential with codomain E21,0 originates from E21,1, which is zero by assumption. The differential with domain E21,0 has codomain E23,1, which is also zero by assumption. Similarly, the incoming and outgoing differentials of Er1,0 are zero for all r ≥ 2. Therefore the (1,0) term of the spectral sequence has converged, meaning that it is isomorphic to the degree one graded piece of the abutment H 1(A). Because the spectral sequence lies in the first quadrant, the degree one graded piece is equal to the first subgroup in the filtration defining the graded pieces. The inclusion of this subgroup yields the injection E21,0 H 1(A) which begins the five-term exact sequence. This injection is called an edge map.

The E20,1 term of the spectral sequence has not converged. It has a potentially non-trivial differential leading to E22,0. However, the differential landing at E20,1 begins at E22,2, which is zero, and therefore E30,1 is the kernel of the differential E20,1 E22,0. At the third page, the (0, 1) term of the spectral sequence has converged, because all the differentials into and out of Er0,1 either begin or end outside the first quadrant when r ≥ 3. Consequently E30,1 is the degree zero graded piece of H 1(A). This graded piece is the quotient of H 1(A) by the first subgroup in the filtration, and hence it is the cokernel of the edge map from E21,0. This yields a short exact sequence

0 E21,0 H 1(A) E30,1 0.

Because E30,1 is the kernel of the differential E20,1 E22,0, the last term in the short exact sequence can be replaced with the differential. This produces a four-term exact sequence. The map H 1(A) E20,1 is also called an edge map.

The outgoing differential of E22,0 is zero, so E32,0 is the cokernel of the differential E20,1 E22,0. The incoming and outgoing differentials of Er2,0 are zero if r ≥ 3, again because the spectral sequence lies in the first quadrant, and hence the spectral sequence has converged. Consequently E32,0 is isomorphic to the degree two graded piece of H 2(A). In particular, it is a subgroup of H 2(A). The composite E22,0 E32,0 H2(A), which is another edge map, therefore has kernel equal to the differential landing at E22,0. This completes the construction of the sequence.

Variations

The five-term exact sequence can be extended at the cost of making one of the terms less explicit. The seven-term exact sequence is

0 E21,0 H 1(A) E20,1 E22,0 Ker(H 2(A) E20,2) E21,1 E23,0.

This sequence does not immediately extend with a map to H3(A). While there is an edge map E23,0 H3(A), its kernel is not the previous term in the seven-term exact sequence.

For spectral sequences whose first interesting page is E1, there is a three-term exact sequence analogous to the five-term exact sequence:

There are also low degree exact sequences for homological spectral sequences, as well as for spectral sequences in the third quadrant. When additional terms of the spectral sequence are known to vanish, the exact sequences can sometimes be extended further. For example, the long exact sequence associated to a short exact sequence of complexes can be derived in this manner.

gollark: AE2 can do multiple recipes per item, multiple-output recipes, and to some extent loops IIRC.
gollark: I mean "simple" as in "one recipe for each item, no muultiple-output recipes, no loops", which is quite limiting.
gollark: Very simple autocrafting *is* doable without huge problems - Dragon had an implementation - but that's not very good.
gollark: Are you suggesting we should cover anything but the maximally general case? HERESY!
gollark: In fact, not only is it computationally NP-hard, apparently (Squid did some insanity here: https://squiddev.cc/2018/01/28/ae-sat.html), but to do it *well* you also have to somehow make highly subjective decisions!

References

  • Neukirch, Jürgen; Schmidt, Alexander; Wingberg, Kay (2000), Cohomology of Number Fields, Grundlehren der Mathematischen Wissenschaften, 323, Berlin: Springer-Verlag, ISBN 978-3-540-66671-4, MR 1737196, Zbl 0948.11001
  • Weibel, Charles A. (1994). An introduction to homological algebra. Cambridge Studies in Advanced Mathematics. 38. Cambridge University Press. ISBN 978-0-521-55987-4. MR 1269324. OCLC 36131259.
This article is issued from Wikipedia. The text is licensed under Creative Commons - Attribution - Sharealike. Additional terms may apply for the media files.