矢量沿无穷小正交回路平行移动的不变性及变化量探究

Alright, let's walk through exactly how to calculate the vector change $\Delta v^\beta = v^\beta_{|| P} - v_P^\beta$ when parallel transporting a vector around an infinitesimal closed parallelogram in curved spacetime. This is one of the most intuitive ways to connect parallel transport to the Riemann curvature tensor, so let's break it down step by step.

We start at point $P$ with coordinates $x^\mu_P$, and our parallelogram is spanned by two infinitesimal vectors $a^\mu$ and $b^\mu$. The path we'll take is:

• $P \to P+a$: move along $a^\mu$
• $P+a \to P+a+b$: move along $b^\mu$
• $P+a+b \to P+b$: move back along $-a^\mu$
• $P+b \to P$: move back along $-b^\mu$

Since everything is infinitesimal, we can use Taylor expansions for the connection coefficients $\Gamma^\alpha_{\mu \nu}$ at points away from $P$.

Recall the parallel transport equation:
$$ \frac{d v^\beta}{d\sigma} + \Gamma^\beta_{\mu \nu} v^\mu \frac{d x^\nu}{d \sigma} = 0. $$
For infinitesimal displacements, we can approximate this as a finite change (ignoring higher-order small terms):
$$ v^\beta_{\text{final}} = v^\beta_{\text{initial}} - \Gamma^\beta_{\mu \nu}(\text{start point}) v^\mu_{\text{initial}} \Delta x^\nu. $$

First path: $P \to P+a \to P+a+b$ #

1. From $P$ to $P+a$:
   $$ v^\beta_{P+a} = v^\beta_P - \Gamma^\beta_{\mu \nu}(P) v^\mu_P a^\nu. $$
2. From $P+a$ to $P+a+b$: we need the connection at $P+a$, which we Taylor-expand to first order:
   $$ \Gamma^\beta_{\mu \nu}(P+a) = \Gamma^\beta_{\mu \nu}(P) + \partial_\alpha \Gamma^\beta_{\mu \nu}(P) a^\alpha. $$
   Plugging this into the parallel transport formula (and substituting $v^\beta_{P+a}$), we get:
   $$
   \begin{align*}
   v^\beta_{P+a+b, 1} &= v^\beta_{P+a} - \Gamma^\beta_{\mu \nu}(P+a) v^\mu_{P+a} b^\nu \
   &= v^\beta_P - \Gamma^\beta_{\mu \nu}(P)v^\mu_P a^\nu - \left(\Gamma^\beta_{\mu \nu}(P) + \partial_\alpha \Gamma^\beta_{\mu \nu}(P)a^{\alpha\right)\left(v}\mu_P - \Gamma^\mu_{\rho \sigma}(P)v^\rho_P a^{\sigma\right)b}\nu \
   &= v^\beta_P - \Gamma^\beta_{\mu \nu}(P)v^\mu_P(a\nu + b^\nu) + \Gamma^\beta_{\rho \nu}(P)\Gamma^\rho_{\mu \alpha}(P)v^\mu_P a^\alpha b^\nu - \partial_\alpha \Gamma^\beta_{\mu \nu}(P)v^\mu_P a^\alpha b^\nu.
   \end{align*}
   $$
   (We dropped all terms higher than second order in $a$ or $b$ since they're negligible for infinitesimal displacements.)

Second path: $P \to P+b \to P+a+b$ #

1. From $P$ to $P+b$:
   $$ v^\beta_{P+b} = v^\beta_P - \Gamma^\beta_{\mu \nu}(P) v^\mu_P b^\nu. $$
2. From $P+b$ to $P+a+b$: Taylor-expand the connection at $P+b$:
   $$ \Gamma^\beta_{\mu \nu}(P+b) = \Gamma^\beta_{\mu \nu}(P) + \partial_\alpha \Gamma^\beta_{\mu \nu}(P) b^\alpha. $$
   Parallel transporting gives:
   $$
   \begin{align*}
   v^\beta_{P+a+b, 2} &= v^\beta_{P+b} - \Gamma^\beta_{\mu \nu}(P+b) v^\mu_{P+b} a^\nu \
   &= v^\beta_P - \Gamma^\beta_{\mu \nu}(P)v^\mu_P(a\nu + b^\nu) + \Gamma^\beta_{\rho \nu}(P)\Gamma^\rho_{\mu \alpha}(P)v^\mu_P b^\alpha a^\nu - \partial_\alpha \Gamma^\beta_{\mu \nu}(P)v^\mu_P b^\alpha a^\nu.
   \end{align*}
   $$

Now we find the difference between the two vectors at $P+a+b$:
$$ \Delta v^\beta_{P+a+b} = v^\beta_{P+a+b,1} - v^\beta_{P+a+b,2}. $$
Most terms cancel out, leaving:
$$
\Delta v^\beta_{P+a+b} = \left( -\partial_\alpha \Gamma^\beta_{\mu \nu} + \partial_\nu \Gamma^\beta_{\mu \alpha} + \Gamma^\beta_{\rho \nu}\Gamma^\rho_{\mu \alpha} - \Gamma^\beta_{\rho \alpha}\Gamma^\rho_{\mu \nu} \right) v^\mu_P a^\alpha b^\nu.
$$

Since we're dealing with infinitesimal displacements, parallel transporting this difference back to $P$ doesn't change its value (the correction would be higher-order and negligible). So this is exactly the change $\Delta v^\beta$ we're looking for.

Notice that the expression in parentheses is the negative of the standard Riemann curvature tensor (definitions can vary slightly by sign convention, but this is the core structure):
$$ R^\beta_{\ \mu \alpha \nu} = \partial_\alpha \Gamma^\beta_{\mu \nu} - \partial_\nu \Gamma^\beta_{\mu \alpha} + \Gamma^\beta_{\rho \alpha}\Gamma^\rho_{\mu \nu} - \Gamma^\beta_{\rho \nu}\Gamma^\rho_{\mu \alpha}. $$

Substituting this in, we get:
$$ \Delta v^\beta = -R^\beta_{\ \mu \alpha \nu} v^\mu_P a^\alpha b^\nu. $$

We can also rewrite this using the antisymmetry of the parallelogram's area element (since swapping $a$ and $b$ reverses the path direction):
$$ \Delta v^\beta = \frac{1}{2} R^\beta_{\ \mu \alpha \nu} v^\mu_P \left(a^\alpha b^\nu - a^\nu b^\alpha\right). $$

This result shows that the change in the vector after parallel transport around a closed loop is directly proportional to the Riemann curvature tensor, the initial vector, and the "area" of the loop (encoded in the antisymmetric product of $a$ and $b$). This is the geometric heart of the Riemann tensor: it quantifies how much parallel transport fails to return a vector to its original value—i.e., how "curved" the spacetime is.

内容的提问来源于stack exchange，提问作者Aegon