next up previous contents
Next: How to calculate. Up: The tangent space. Previous: The derivative of a   Contents

Co-ordinate tangent vectors and one-forms.

Let $ (U, \psi)$ be a set of co-ordinates on $ M$ where $ \psi = (\psi^1,\dots, \psi^n)$. Then each of the component functions $ \psi^i$ is a real function so we can define $ n$ one-forms $ d\psi^i(x) \in T_x^{*}M$ called the co-ordinate one-forms.

We have seen that

$\displaystyle d\psi^{-1} \colon T_xM \to \mathbb{R}^n
$

is a linear isomorphism. We denote by

$\displaystyle \frac{\partial\phantom{\psi^i}}{\psi^i}(x) = d\psi^{-1}(\psi(x))(e^i)
$

the image under this map of the standard basis vector $ e^i$ in $ \mathbb{R}^n$. We call the set of these the basis of co-ordinate tangent vectors. Consider what happens when we apply $ d\psi^i(x)$ to $ \partial /\partial \psi^j(x)$. We have

$\displaystyle d\psi^i(x)(\frac{\partial\phantom{\psi^j}}{\psi^j}(x)) = d\psi^i(x) (d\psi^{-1}(x)(e^j))
= d(\psi^i\circ \psi^{-1})(\psi(x))(e^j)).
$

Notice that if $ y=(y^1,\dots, y^n)$ is a point in $ \mathbb{R}^n$ then $ \psi^i\circ \psi^{-1}(y) = y^i$ is a linear map so equal to its own derivative. Hence $ d(\psi^i\circ \psi^{-1})(\psi(x))(e^j)$ is the $ i$th component of the vector $ e^j$ or just $ \Delta_{ij}$. It follows from linear algebra that $ d\psi^1(x),\dots, d\psi^n(x)$ is a basis of $ T_xM$ and, in fact, the dual basis to the basis $ \partial /\partial \psi^i$.



Michael Murray
1998-09-16