If I have a vector $X^{\mu}(x)$, and then I consider an infinitesimal coordinate transformation of the form $x^{\mu} \to x^{\mu} + v^{\mu}(x)$, then how does my vector $X^{\mu}(x)$ transform?

From some reading online, it seems like the answer is along the lines of: $$ X^{\mu}(x) \ \to \ X^{\mu}(x) + v^{\sigma}(x) \partial_{\sigma} X^{\mu}(x) - X^{\sigma}(x) \partial_{\sigma} v^{\mu}(x) $$

I'm not really understanding where this is coming from though...is it because we're taking a Taylor expansion of $X^{\mu}(x+v)$? The minus sign is particularly unsettling for me.

  • $\begingroup$ Would Mathematics be a better home for this question? $\endgroup$
    – Qmechanic
    Feb 7, 2017 at 21:02

1 Answer 1


If $X^\mu(x)$ is a vector field, in terms of coordinates $\{x^\mu\}$ and we consider an infinitesimal shift,

$$x^\mu \to x^\mu + v^\mu(x)$$

then the vector field $X^\mu$ changes according to the Lie derivative w.r.t. to the vector $v^\mu$, that is, we have that,

$$\delta X^\mu = \mathcal L_v X^\mu = v^\nu \nabla_\nu X^\mu - X^\nu \nabla_\nu v^\mu$$

by simplying applying the rules of Lie differentiation of a tensor. If the manifold is entirely flat, then covariant derivatives are demoted to partial derivatives,

$$\delta X^\mu = v^\nu \partial_\nu X^\mu - X^\nu \partial_\nu v^\mu$$

recovering the expression given by the OP. Notice that $v^\nu \partial_\nu$ is the same as the vector $v^\mu$ expressed as a derivation.$^\dagger$ As such, we can write,

$$\mathcal L_v X = [v,X]$$

with the Lie bracket, where $v$ and $X$ are the fields expressed as derivations, i.e. operators.

$\dagger$ A vector as a derivation can be thought of as a directional derivative. Specifically, for a vector $v$ and a map $f : \mathbb R^n \to \mathbb R$, we have that,

$$D_v f(x) = \frac{d}{d\lambda} f(x + \lambda v) \bigg\rvert_{\lambda = 0} = v^\mu \partial_\mu f (x).$$

A case of interest is when $v$ is a vector along a curve, that is, it is the vector tangent along a path, in which case one may define differentiation of a map along a path on the manifold.

  • $\begingroup$ This is exactly what I was hoping for - thank you. As a follow up question, if I build a tensor out of vectors $X^{\mu}$ and $Y^{\mu}$ and consider the same kind of coordinate shift, for example building $H^{\mu\nu} := X^{\mu}Y^{\mu}$, then is it true that the transformed tensor is simply of the form $\left( X^{\mu} + \mathcal{L}_{v}(X^{\mu}) \right)\left( Y^{\mu} + \mathcal{L}_{v}(Y^{\mu}) \right)$? $\endgroup$ Feb 6, 2017 at 20:58

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.