Consider an open embedding $\varphi:\Bbb R^{1,1}_- \hookrightarrow (0,1)^2$ with $\varphi(x_1,x_2)=(e^{x_1},e^{x_2})$

I need to put a metric on $(0,1)^2$.

Here’s what I know so far. I need to transport the Lorentz metric: $ds^2=dxdy. $

Since my map is a diffeomorphism I can push forward the Lorentz metric to $(0,1)^2$.

I know that $n=\log(x)\log(y)$ is a preserved metric of a transformation on $(0,1)^2$ that is related to the classical Lorentz transformation.

However I don't think I can obtain a metric purely from this fact.

How can I push the Lorentz metric forward?

My wild guess is: $ds^2=\frac{1}{x}dx\frac{1}{y}dy.$

Also how can one generalize this metric to dimension 3?

I think it has to do with adding in a positive definite matrix of dimension $n-2$ somewhere into the metric. This makes sense because that matrix will do the job of rotating the coordinate system.


1 Answer 1


I figured out the first question. The new metric in $(0,1)^2$ is $ds^2=\frac{1}{u}du\frac{1}{v}dv.$

This is done by pulling back the Lorentz metric $ds^2=dxdy.$

$\begin{align} (\varphi^{-1})^* (dxdy) &=( d((\varphi^{-1})^*x))( d((\varphi^{-1})^*y)) \\ &= (d (x\circ \varphi^{-1}))(d (y\circ \varphi^{-1})) \\ &= (d \log u)(d\log v) \\ &= \left( \frac{1}{u} du\right)\left( \frac{1}{v} dv\right) \\ &= \frac{1}{u} du\frac{1}{v} dv. \end{align}$

Still not sure about the dimension 3 case.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.