I have been watching this video, which explains in fair detail what each of the terms in Einstein’s field equation represents. The portion I linked explains the tensor object ${\Gamma^\alpha}_{\beta\gamma}$, which I believe is also called a Christoffel symbol.

To summarize, if spacetime is curved or if a coordinate system is curved, then the basis vector in a particular dimension is not necessarily constant throughout space. The change in the basis vector along a dimension can be described as a linear combination of all the coordinate system’s basis vectors.

image 1image 2

The video indicates that for the tensor object ${\Gamma^\alpha}_{\beta\gamma}$,

  • $\alpha$ indicates which basis vector it is multiplying;
  • $\beta$ indicates which basis vector is ‘moving’ or being examined; and
  • $\gamma$ indicates the direction in which the basis vector (of the $\beta$-dimension) is moving.

I would like to understand how the difference in a basis (covariant) vector $\tilde e_\mu$ between two points in space, $\vec a = \begin{pmatrix} x_a & y_a & z_a \end{pmatrix}^\mathsf{T}$ and $\vec w = \begin{pmatrix} x_w & y_w & z_w \end{pmatrix}^\mathsf{T}$, is mathematically described. (Note that $\tilde e_\mu$ is not the component of $e$ in the $\mu$-dimension but rather the basis vector $\tilde e$ of the $\mu$-dimension.)

I figured that one could define $$\tilde e_\mu(\vec a) + \Delta\tilde e_\mu = \tilde e_\mu(\vec w)$$

These are my particular thoughts:

  • For three-dimensional space, one must sum over $\alpha\in\{1,2,3\}$.
  • If one is investigating the change in $\tilde e_\mu$, then one should set $\beta=\mu$.
  • Since $\vec w - \vec a$ could span any number of dimensions, then I suppose that one ought to sum over $\gamma\in\{1,2,3\}$.

This leads me to my best guess that $$\Delta\tilde e_\mu = \sum_{\gamma=1}^{3} \sum_{\alpha=1}^{3} {\Gamma^\alpha}_{\mu\gamma} \, \tilde e_\alpha$$

Naturally, one would also have to take into account that ${\Gamma^\alpha}_{\beta\gamma}$ could vary across space, so this should probably look more like $$\Delta\tilde e_\mu = \sum_{\gamma=1}^{3} \sum_{\alpha=1}^{3} {\Gamma^\alpha}_{\mu\gamma}(\vec a,\vec w) \, \tilde e_\alpha$$

I have the presentiment that one does not define actually ${\Gamma^\alpha}_{\beta\gamma}$ as a function of two vectors but rather resolves the issue of ${\Gamma^\alpha}_{\beta\gamma}$ varying across space by integrating along the path from $\vec a$ to $\vec w$.

How close am I to correct? How is this actually done in practice?

  • $\begingroup$ Just to clarify, are you asking how basis vectors "transform"? $\endgroup$ – Dr. Ikjyot Singh Kohli Dec 24 '17 at 21:56
  • $\begingroup$ @Dr.IkjyotSinghKohli Ehm, no, I don’t believe so. I am familiar with how basis vectors transform with a coordinate system in that they are covariant. What I am curious about is how the change in a basis vector across a coordinate system that is curved (either by definition or by a warping of spacetime) is described mathematically using this $\Gamma$ tensor. $\endgroup$ – gen-ℤ ready to perish Dec 24 '17 at 21:59
  • $\begingroup$ Christoffel symbols are not tensors: they don't transform the "right" way. See Zee's "Einstein Gravity" for a pretty clear explanation. $\endgroup$ – NickD Dec 24 '17 at 22:15
  • $\begingroup$ @Nick My apologies—I’ve reflected that now. $\endgroup$ – gen-ℤ ready to perish Dec 24 '17 at 22:18

So, assume that you have a set of basis vectors, call them $\mathbf{e}_{\mu}$. One actually defines the connection coefficients as the components of the directional derivative of the basis vectors:

$\nabla_{e_v} e_u \equiv \Gamma^{\alpha}_{\mu v} e_{e_\alpha}$

So, how you read this is: the connection coefficients $\Gamma^{\alpha}_{\mu v}$ is the $\alpha$-component of the change of $e_{\mu}$ by a change in the $e_{v}$ direction, so I would say your formulation is correct. One then uses this definition along with properties of the Koszul connection to define the covariant derivative in a coordinate-independent manner.

(Also, just a comment: the general definition of a basis vector: $e_{\mu} = \frac{\partial}{\partial x^{\mu}}$ is valid in general (flat/curved) spacetimes)


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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