Torsion and linking number for a surface diffeomorphism (Q2633133)

Let \(\mathrm{Diff}^1(S)\) denote the group of \(C^1\) diffeomorphisms on a parallelizable Riemannian surface \(S\). An isotopy in \(\mathrm{Diff}^1(S)\) is an arc \((f_t)_{t\in[0,1]}\) in \(\mathrm{Diff}^1(S)\) which is continuous with respect to the weak or compact-open \(C^1\) topology on \(\mathrm{Diff}^1(S)\). Suppose \(f_0=id_S\) and \(f_1=f\). Then the isotopy \((f_t)_{t\in[0,1]}\) can be extended to any positive time as follows: for \(t\in\mathbb{R}_+\) the \(C^1\) diffeomorphism \(f_t:S\to S\) is defined as \(f_t:=f_{\{t\}}\circ f^{[t]}\), where \(\{t\}\) and \([t]\) denote the fractionary and integer part of \(t\), respectively. The extended isotopy is also denoted as \((f_t)_t\). Fix an orientation and endow \(S\) with a Riemannian metric so that the notion of oriented angle between two non zero vectors of the same tangent space is well-defined. Then fix a reference continuous unit vector field \(X\) over \(S\) and define the function \(v((f_t)_t):TS_\ast\times\mathbb{R}\to\mathbb{R}/(2\pi\mathbb{Z})\) by \[ (x,\xi,t)\mapsto\theta(X(f_t(x)), Df_t(x)\xi) \] where the notation \(\theta(u,v)\) denotes the oriented angle between two non zero vectors \(v\) and \(u\). For \(x\in S\) and \(\xi\in T_xS\setminus\{0\}\) the angle function \(v((f_t)_t )(x,\xi,\cdot)\) is continuous and thus has a continuous determination \(\tilde{v}((f_t)_t)(x,\xi,\cdot):\mathbb{R}\to\mathbb{R}\), i.e. a continuous function \(\tilde{v}((f_t)_t )(x,\xi,\cdot):\mathbb{R}\to\mathbb{R}\) such that \(\tilde{v}((f_t)_t )(x,\xi, s)\) is a measure of the oriented angle \({v}((f_t)_t )(x,\xi, s)\) for any \(s\in\mathbb{R}\). The torsion at finite time \(n\in\mathbb{N}\setminus\{0\}\) is defined by \[ \mathrm{Torsion}_n((f_t)_t , x, \xi) := \frac{1}{n}(\tilde{v}((f_t)_t )(x,\xi,n)-\tilde{v}((f_t)_t )(x,\xi,0)). \] When for some \(\xi\in T_xS\setminus\{0\}\) the quantity \(\mathrm{Torsion}_n((f_t)_t,x,\xi)\) converges as \(n\to\infty\), the torsion of the orbit of \(x\) is defined by \[ \mathrm{Torsion}((f_t)_t , x)=\lim_{n\to\infty}\mathrm{Torsion}_n((f_t)_t , x, \xi). \] Let \(S=\mathbb{R}^2\) be endowed with the counterclockwise orientation, and then fix the constant vector field \(X = (1, 0)\). For an isotopy \((F_t)_t\) in \(\mathrm{Diff}^1(\mathbb{R}^2)\) joining the identity to \(F_1=F\) and \(\triangle:= \{(z_1, z_2)\in\mathbb{R}^4\,:\, z_1=z_2\}\) and define the function \[ u((F_t)_t) : (\mathbb{R}^4\setminus\triangle)\times\mathbb{R}\to\mathbb{R}/(2\pi\mathbb{Z}),\; (z_1,z_2,t)\mapsto \theta((1,0), F_t(z_2)-F_t(z_1)). \] Then for each \((z_1,z_2)\in\mathbb{R}^4\setminus\triangle\) the angle function \(u((F_t)_t)(z_1, z_2,\cdot)\) has a continuous determination \(\tilde{u}((F_t)_t)(z_1, z_2,\cdot):\mathbb{R}\to\mathbb{R}\). The linking number of \(z_1\) and \(z_2\) at finite time \(n\) is defined by \[ \mathrm{Linking}_n((F_t)_t , z_1, z_2) := \frac{1}{2}(\tilde{u}((F_t)_t)(z_1, z_2, n)-\tilde{u}((F_t)_t)(z_1, z_2, 0)), \] and the linking number of the orbits of \(z_1\) and \(z_2\) is defined by \[ \mathrm{Linking}((F_t)_t , z_1, z_2) := \lim_{n\to+\infty}\mathrm{Linking}_n((F_t)_t , z_1, z_2) \] whenever the limit exists. As an positive answer to a question raised by [\textit{F. Béguin} and \textit{Z. R. Boubaker}, J. Math. Soc. Japan 65, No. 1, 137--168 (2013; Zbl 1268.37064)], Theorem~1.1 in the paper under review stated that for an isotopy \((F_t)_{t\in [0,1]}\) in \(\mathrm{Diff}^1(\mathbb{R}^2)\) there exists a point \(z\in [x,y]\) so that \(\mathrm{Torsion}_1((F_t)_t, z, y-x) = l\) provided that there exist two distinct points \(x,y\in\mathbb{R}^2\), \(x\ne y\) such that \(\mathrm{Linking}_1((F_t)_t, x, y)=l\in\mathbb{R}\). For the annulus \(\mathbb{A} =\mathbb{R}/(2\pi\mathbb{Z})\times\mathbb{R}\) and an isotopy \((f_t)_{t\in [0,1]}\) in \(\mathrm{Diff}^1(\mathbb{A})\) joining \(Id_{\mathbb{A}}\) to \(f_1 = f\) which is a \(C^1\) positive twist map, Theorem 1.2 showed that \(\mathrm{Torsion}_n((f_t)_t, z, (0,1))\in (-\pi,0)\) for any \(z\in\mathbb{A}\), \((0, 1)\in T_z\mathbb{A}\setminus\{0\}\), and any \(n\in\mathbb{N}\setminus\{0\}\). As a consequence of this and Theorem~1.1, \(\mathrm{Torsion}((f_t)_t, z)\in [-\pi,0]\) for any point \(z\in\mathbb{A}\) for which the limit of the torsion exists.