|
original poster
hbghlyj
posted 2025-7-18 22:09
Why the semidirect product \(\mathbb{Z} \rtimes \mathbb{Z}\) in the short exact sequence \(1 \to \pi_1(X) \to \pi_1(X/\mathbb{Z}) \to \mathbb{Z} \to 1\) is actually a direct product \(\mathbb{Z} \oplus \mathbb{Z}\)? This happens precisely when the action of the deck transformation group \(\mathbb{Z}\) on \(\pi_1(X) \cong \mathbb{Z}\) is trivial, which implies that the two generators of \(\pi_1(X/\mathbb{Z})\) commute.
To see this explicitly, consider the following representatives for the generators:
One generator, call it \(\alpha\), is the image under the quotient map \(p: X \to X/\mathbb{Z}\) of a loop around the origin in \(X\), such as the unit circle \(\gamma(t) = (\cos(2\pi t), \sin(2\pi t))\) for \(t \in [0,1]\). This generates the subgroup \(p_*(\pi_1(X)) \cong \mathbb{Z}\) in the exact sequence.
The other generator, call it \(\beta\), comes from a choice of section to the exact sequence (lifting the generator of the deck group \(\mathbb{Z}\)). Fix a basepoint in \(X\), say \((1,1)\). The deck transformation \(\phi\) sends this to \((2,1/2)\). Connect these by a path in \(X\) along the hyperbola \(xy=1\), such as \(\sigma(t) = (e^{t \log 2}, e^{-t \log 2})\) for \(t \in [0,1]\) (which starts at \((1,1)\) and ends at \((2,1/2)\)). In the quotient \(X/\mathbb{Z}\), the endpoints are identified, so \(p(\sigma)\) is a loop \(\beta\).
In \(\pi_1(X/\mathbb{Z})\), the relation is \(\beta \alpha \beta^{-1} = \phi_*(\alpha)\), where \(\phi_*\) is the induced map on \(\pi_1(X)\). Here, \(\phi\) is linear with positive determinant 1, so it is orientation-preserving and \(\phi_*\) is the identity on \(\pi_1(X) \cong \mathbb{Z}\). Thus, \(\beta \alpha \beta^{-1} = \alpha\), or equivalently, \(\alpha\) and \(\beta\) commute: \(\alpha \beta = \beta \alpha\). |
|