Doubt about affine connections: what is a covariant derivative?

216 Views Asked by At

And the Connection fellow enters the stage:

So, an (affine) connection is the map:

$$\begin{array}{rl} \nabla: \chi(\mathcal{M}) \times \chi(\mathcal{M}) &\to \chi(\mathcal{M}) \\ (V,W) &\mapsto \nabla_{V}W \end{array} $$

with the axioms:

$1. \nabla_{V}(W+Z) = \nabla_{V}(W) + \nabla_{V}(Z)$

$2.\nabla_{V+Z}(W) = \nabla_{V}(W) + \nabla_{Z}(W)$

$3.\nabla_{fV}(W) = f\nabla_{V}(W)$

$4.\nabla_{V}(fW) = V[f]W+f \nabla_{V}(W)$

Now, given a connection we can take the tangent vector field $\dot{\gamma} = \dot{t}^{i}\frac{\partial}{\partial x^{i}}$ to a given curve $\gamma(\lambda)$ of the manifold, and ask ourselfs about the change of the vector field $V$ along the curve $\gamma(\lambda)$: $\nabla_{\dot{\gamma}} V$ .

Now, in a given chart, we whole thing is written as

\begin{equation} \nabla _{\dot{\gamma}} V = \Bigg[\dot{t}^{i}\Big(\partial_{i}v^{k} + \Gamma^{k}\hspace{0.1mm}_{ij} v^{j}\Big)\Bigg]\partial_{k} \tag{1} \end{equation}

But in general case we have:

\begin{equation} \nabla _{X} V = \Bigg[X^{i}\Big(\partial_{i}v^{k} + \Gamma^{k}\hspace{0.1mm}_{ij} v^{j}\Big)\Bigg]\partial_{k} \tag{2} \end{equation}

Furthermore, since we can define (at this point is actually just a mere change of symbology):

$$ \nabla_{i}v^{k}= \partial_{i}v^{k} + \Gamma^{k}\hspace{0.1mm}_{ij} v^{j} \tag{3}$$

we note that we have covariant and a contravariant indexes in a new object:

$$T^{k}_{i} := \nabla_{i}v^{k} = \partial_{i}v^{k} + \Gamma^{k}\hspace{0.1mm}_{ij} v^{j} \tag{4}$$

After we change the coordinates we realize that:

$$ \nabla_{a'}v^{b'} = \frac{\partial x^{b'}}{\partial x^{k}}\frac{\partial x^{i}}{\partial x^{a'}}\nabla_{i}v^{k} \tag{5}$$

Therefore we can define:

$$\nabla V := (\nabla_{i}v^{k}) dx^{i} \otimes \frac{\partial}{\partial x^{k}} \tag{6}$$

So, we have then at least three objects: $\nabla _{\dot{\gamma}} V$, $\nabla _{X} V$, $\nabla V$. And, another one that we use to define geodesics: $\nabla _{\dot{\gamma}} \dot{\gamma}$:

\begin{equation} \nabla_{\dot{\gamma}}\dot{\gamma} = \Bigg( \frac{d^{2}x^{k}}{d\lambda^{2}} + \Gamma^{k}\hspace{0.1mm}_{ij}\frac{dx^{i}}{d\lambda} \frac{dx^{j}}{d\lambda} \Bigg)\partial_{k} \tag{7} \end{equation}

My question is: is it legal to construct the tensor $(6)$ the way that I did?(*)

(*)I'm asking this because, in General Relativity text books, we call a covariant derivative precisely the components: $\nabla_{i}v^{k} = \partial_{i}v^{k} + \Gamma^{k}\hspace{0.1mm}_{ij} v^{j}$ and not the $X^{i}(\nabla_{i}v^{k}) = X^{i}(\partial_{i}v^{k} + \Gamma^{k}\hspace{0.1mm}_{ij} v^{j})$ (which are precisely the components of $\nabla_{X}V$ a $(1,0)-$tensor)