Documentation

TauCeti.AlgebraicTopology.Singular.MappingTorus.Torsion

The torsion of the homology of a mapping torus #

Let φ : F ≃ₜ F be a homeomorphism and T_φ its mapping torus, with singular homology taken with coefficients in a commutative ring k. The Wang sequence

Hₙ₊₁(F) ⟶ Hₙ₊₁(F) ⟶ Hₙ₊₁(T_φ) ⟶ Hₙ(F) ⟶ Hₙ(F),

with maps 𝟙 - φ_*, the fibre inclusion and the Wang connecting morphism, identifies the image of the fibre in Hₙ₊₁(T_φ) with the cokernel of 𝟙 - φ_* on Hₙ₊₁(F) (TauCeti.MappingTorus.wangCokernelEquiv).

Suppose k is a domain, Hₙ₊₁(F) is free of finite rank, det (𝟙 - φ_*) ≠ 0 on it, and Hₙ(F) is torsion-free. The cokernel of 𝟙 - φ_* is then killed by det (𝟙 - φ_*), while the quotient of Hₙ₊₁(T_φ) by the image of the fibre embeds in Hₙ(F). Hence the image of the fibre is exactly the torsion submodule of Hₙ₊₁(T_φ). Over a principal ideal domain its Nat.card equals that of k ⧸ (det (𝟙 - φ_*)), and over ℤ it is finite of order |det (𝟙 - φ_*)|.

In degree one the torsion-freeness hypothesis always holds, since H₀(F) is free on the path components of F. When H₁(F; ℤ) is free of finite rank, as for a closed orientable surface, and φ_* has no eigenvalue one on it, this computes the torsion subgroup of H₁(T_φ; ℤ) as the cokernel of 1 - φ_*, and its order as |det (1 - φ_*)|.

Main results #

See A. Hatcher, Algebraic Topology, Section 2.2, for the Wang sequence of a mapping torus.

If Hₙ(F) is torsion-free, every torsion class of Hₙ₊₁(T_φ) comes from the fibre: the Wang connecting morphism sends it to a torsion class of Hₙ(F), hence to zero.

The torsion of the homology of a mapping torus. If Hₙ₊₁(F) is free of finite rank, det (𝟙 - φ_*) ≠ 0 on it and Hₙ(F) is torsion-free, the torsion submodule of Hₙ₊₁(T_φ) is the image of the fibre, which wangCokernelEquiv identifies with the cokernel of 𝟙 - φ_*.

In degree one, the torsion of H₁(T_φ) is the image of the fibre whenever H₁(F) is free of finite rank and det (𝟙 - φ_*) ≠ 0 on it; no hypothesis on H₀(F) is needed, since it is free.