coordinate basis

From Mathematics wiki

Jump to: navigation, search

Also called a holonomic basis, a coordinate basis is a set of basis vectors that are derivatives along coordinate curves:

\vec{e}_i \equiv \frac{\partial}{\partial x^i}\,.

In writing a general vector as

\vec{V} = \frac{d}{d\lambda} = V^i\frac{\partial}{\partial x^i}\,

we have implicitly made use of the coordinate basis vectors \vec{e}_i\,, so that we can write

\vec{V} = V^i \vec{e}_i\,.

Note that acting \vec{V} = \frac{d}{d\lambda}\, on an arbitrary function f\,, we obtain a new function \vec{V}(f)\,. If f\, is suitably differentiable, then we can act upon this function with another vector field \vec{W} = \frac{d}{d\mu}\,:

\vec{W}\left(\vec{V}(f)\right)\, = \frac{d}{d\mu}\left( \frac{df}{d\lambda} \right)\,,
= \frac{d}{d\mu}\left( V^i \frac{\partial f}{\partial x^i} \right)\,,
= W^j \frac{\partial}{\partial x^j}\left( V^i \frac{\partial f}{\partial x^i} \right)\,,
= W^j \frac{\partial V^i}{\partial x^j}\frac{\partial f}{\partial x^i} + W^j V^i \frac{\partial^2f}{\partial x^j \partial x^i}\,.

Notice that this is no longer a first order differential operator (not that we expected it to be), but consider the fact that second partial derivatives commute. Then,

[\vec{W},\vec{V}] f\, \equiv  \vec{W}\left(\vec{V}(f)\right) - \vec{V}\left(\vec{W}(f)\right)\,,
= W^j \frac{\partial V^i}{\partial x^j}\frac{\partial f}{\partial x^i} + W^j V^i \frac{\partial^2f}{\partial x^j \partial x^i} - V^j \frac{\partial W^i}{\partial x^j}\frac{\partial f}{\partial x^i} - V^j W^i \frac{\partial^2f}{\partial x^j \partial x^i}\,,
= \left(W^j \frac{\partial V^i}{\partial x^j} - V^j \frac{\partial W^i}{\partial x^j}\right) \frac{\partial f}{\partial x^i} \,.

Since f\, is arbitrary, we obtain a new first order differential operator (and vector field):

[\vec{W},\vec{V}] = \left(W^j \frac{\partial V^i}{\partial x^j} - V^j \frac{\partial W^i}{\partial x^j}\right) \frac{\partial}{\partial x^i}\,.

The vector field [\vec{W},\vec{V}]\, is known as the Lie bracket of \vec{W}\, and \vec{V}\,. It is the Lie derivative of \vec{V}\, in the direction \vec{W}\,, and can be written as \mathcal{L}_\vec{W}\vec{V}\,.


Finally, if \lambda\, and \mu\, are two coordinates of a coordinate chart, e.g. y^1 =\lambda\, and y^2 =\mu\,, then

[\vec{W},\vec{V}] f = \frac{\partial^2 f}{\partial y^2 \partial y^1} - \frac{\partial^2 f}{\partial y^1 \partial y^2} = 0\,.

Therefore a coordinate basis has the property that

[\vec{e}_i, \vec{e}_j] = 0\,,

for all i\, and j\,. Although it would appear that this property should hold for general tangent vectors \frac{d}{d\lambda}\, and \frac{d}{d\mu}\,, the fact that second partial derivatives commute is the statement that an infinitesimal displacement \Delta x^i\, along the coordinate curve x^i\,, holding x^j\, for some j\neq i\, fixed, followed by an infinitesimal displacement \Delta x^j\, along the coordinate curve x^j\,, holding x^i\, for i\neq j\, fixed, is equivalent to the infinitesimal displacement in the reverse order. In general, it is always not possible to vary \lambda\, while keeping \mu\, fixed, so that the total derivatives \frac{d}{d\lambda}\, and \frac{d}{d\mu}\, do not commute. In such a case \lambda\, and \mu\, are not good coordinates.

A non-coordinate basis, also called an anholonomic basis, is one for which

[\vec{e}_i, \vec{e}_j] \neq 0\,.
Personal tools