There is no metric that makes the integral curves of the field $X=(y,-x)$ geodesics?

96 Views Asked by At

I used the fact that $\nabla_X X=0$, which implies that the integral curves of $X$ are geodesics, to find the Christoffel symbols(the non-zero ones are $\Gamma_{22}^1=1/x,\Gamma_{11}^2=1/y$). I then used $\nabla g=0$ to find the metric from the connection, but I get a singular metric, which cannot happen.

But I hoped I'd get the metric of a cylinder as the solution, as if the metric is that of a cylinder, this vector field corresponds to revolutions around the cylinder, which are geodesics. What is wrong with what I'm doing or with the interpretation?

1

There are 1 best solutions below

4
On

The simplest approach is to use polar coordinates.

The geodesics of the metric $$\tag{1} ds^2=dr^2+d\theta^2 $$ are straight lines in $(r,\theta)$ because the Christoffel symbols all vanish. Among those geodesics are

  • rays going through the origin $r\mapsto (r,\theta)\,,$ where the angle $\theta$ is fixed, and

  • circles $\theta\mapsto (r,\theta)\,$ where the radius $r$ is fixed.

To see how this metric looks like in Cartesian coordinates we start from \begin{align}\tag{2} dx&=\cos\theta\,dr-r\sin\theta\,d\theta\,,&dy=\sin\theta\,dr+r\cos\theta\,d\theta \end{align} and get \begin{align}\tag{3} \begin{pmatrix}dr\\r\,d\theta\end{pmatrix}=\begin{pmatrix}\cos\theta&\sin\theta\\-\sin\theta&\cos\theta\end{pmatrix}\begin{pmatrix}dx\\dy\end{pmatrix}\,, \end{align} which can be written as \begin{align}\tag{4} dr&=\cos\theta\,dx+\sin\theta\,dy\,,&d\theta=\frac{-\sin\theta\,dx+\cos\theta\,dy}{r}\,. \end{align} Therefore, the metric in Cartesian coordinates is \begin{align} ds^2&=\Big(\cos^2\theta+\frac{\sin^2\theta}{r^2}\Big)\,dx^2+\Big(\sin^2\theta+\frac{\cos^2\theta}{r^2}\Big)\,dy^2 +2\sin\theta\cos\theta\Big(1-\frac{1}{r^2}\Big)\,dx\,dy\nonumber\\ &=\frac{x^2r^2+y^2}{r^4}\,dx^2+\frac{y^2r^2+x^2}{r^4}\,dy^2+2xy\frac{r^2-1}{r^4}\,dx\,dy\tag{5} \end{align} where the abbreviation $r^2=x^2+y^2$ is used.

With sympy I calculated the Christoffel symbols: \begin{align*} {\Gamma^x}_{xx} &= - \frac{x y^2}{r^4}\,,&{\Gamma^x}_{xy} &={\Gamma^x}_{yx}= - \frac{y^3}{r^4}\,,&{\Gamma^x}_{yy} &= \frac{(x^2 + 2 y^2) x}{r^4}\,,\\ {\Gamma^y}_{xx} &= \frac{(2 x^2 + y^2) y}{r^4}\,,&{\Gamma^y}_{xy} &={\Gamma^y}_{yx}= - \frac{x^3}{r^4}\,,&{\Gamma^y}_{yy} &= - \frac{x^2 y}{r^4}\,. \end{align*} To check that circles are geodesics lets parametrize them by $$\tag{6} x(\theta)=r\cos(\theta)\,,\quad y(\theta)=r\sin(\theta) $$ and observe that the equations of motion are $$\tag{7} \ddot x=-x\,,\quad \ddot y=-y\,. $$ The geodesic equation are, as we know: \begin{align*} 0&=\ddot x+{\Gamma^x}_{xx}\,\dot x^2+{\Gamma^x}_{yy}\,\dot y^2+{\Gamma^x}_{xy}\,\dot x\dot y+{\Gamma^x}_{yx}\,\dot x\dot y\,,\tag{8}\\[2mm] 0&=\ddot y+{\Gamma^y}_{xx}\,\dot x^2+{\Gamma^y}_{yy}\,\dot y^2+{\Gamma^y}_{xy}\,\dot x\dot y+{\Gamma^y}_{yx}\,\dot x\dot y\,.\tag{9} \end{align*} Since \begin{align*} \dot x=-y\,,\quad \dot y =x\tag{10} \end{align*} we get \begin{align} 0&=\ddot x+{\Gamma^x}_{xx}\,y^2+{\Gamma^x}_{yy}\,x^2-{\Gamma^x}_{xy}\,xy-{\Gamma^x}_{yx}\,xy\,,\tag{11}\\[2mm] 0&=\ddot y+{\Gamma^y}_{xx}\,y^2+{\Gamma^y}_{yy}\,x^2-{\Gamma^y}_{xy}\,xy-{\Gamma^y}_{yx}\,xy\,.\tag{12} \end{align}

Plugging the Christoffel symbols into (11) and (12) gives \begin{align*} 0&=\ddot x+\frac{-xy^4+x^5+2x^3y^2+2xy^4}{r^4}=\ddot x+x\frac{(x^2+y^2)^2}{r^4}=\ddot x+x\,,\tag{13}\\ 0&=\ddot y+\frac{2x^2y^3+y^5-x^4y+2x^4y}{r^4}=\ddot y+y\frac{(x^2+y^2)^2}{r^4}=\ddot y+y\,.\tag{14} \end{align*} which are the equations of motion (7) as it should be.

Again with sympy I also checked that the Riemann tensor of the metric (5) vanishes as it must.

A pseudo Riemannian Metric

The pseudo Riemannian metric $$\tag{15} g=\begin{pmatrix}xy^{-1}&0\\0&-yx^{-1}\end{pmatrix} $$ leads to \begin{align*} {\Gamma^x}_{xx}&=\frac{1}{2x}\,, &{\Gamma^x}_{yy}&=-\frac{y^2}{2x^3}\,,&{\Gamma^x}_{xy}&=-\frac{1}{2y}\,,&{\Gamma^x}_{yx}&=-\frac{1}{2y}\,,\\[2mm] {\Gamma^y}_{xx}&=-\frac{x^2}{2y^3}\,, &{\Gamma^y}_{yy}&=\frac{1}{2y}\,,&{\Gamma^y}_{xy}&=-\frac{1}{2x}\,,&{\Gamma^x}_{yx}&=-\frac{1}{2x}\,. \end{align*} Then the geodesic equations collapse again to the equations of motion (7) that the circles satisfy.