1
$\begingroup$

I'm trying to show the curvature of a one dimensional vector bundle with a Riemannian metric vanishes, no matter what the connection is. I found this can be done for orientable bundles, because an orientable Riemannian line bundle is trivial.

My questions are,

  1. Are Riemannian line bundles trivial?
  2. If not, how to prove their curvature vanish?
$\endgroup$
4
  • 3
    $\begingroup$ 1) You can put a bundle metric on any vector bundle using a partition of unity. 2) The curvature of a metric-compatible connection is skew-symmetric and a skew-symmetric map of a one-dimensional space is zero. $\endgroup$ Commented Dec 11, 2021 at 3:17
  • $\begingroup$ I'm not familiar with metric compatibility on bundles. Is what you said translate to [X,Y] <s_i,s_j> = <R(X,Y)s_i, s_j> + <s_i, R(X,Y)s_j> = 0? Here s_i, s_j are smooth sections and X,Y are tangent vectors. R is the curvature. $\endgroup$
    – chan
    Commented Dec 11, 2021 at 3:53
  • 1
    $\begingroup$ Your formula cannot be correct since the RHS is tensorial in X,Y and the LHS is not. Metric compatibility of connection D means X<s,t> = <D_Xs,t> + <s,D_Xt>. Under this condition one has the skew-symmetry <R(X,Y)s,t>+<s,R(X,Y)t> = 0. Under the rank-one condition, this skew-symmetry implies R(X,Y)=0. (When I was learning this material the most useful textbook I found was Morita "Geometry of Differential Forms.") $\endgroup$ Commented Dec 11, 2021 at 4:13
  • 1
    $\begingroup$ I just started to read the book, and already like his explanation style. Thank you for the answer, it answers some other questions I had too. $\endgroup$
    – chan
    Commented Dec 11, 2021 at 5:30

1 Answer 1

2
$\begingroup$

Given a connection $A_{i\alpha}^\beta$ the curvature is $$F_{ij\alpha}^\beta=\frac{\partial A_{j\alpha}^\beta}{\partial x^i}-\frac{\partial A_{i\alpha}^\beta}{\partial x^j}+A_{i\gamma}^\beta A_{j\alpha}^\gamma-A_{j\gamma}^\beta A_{i\alpha}^\gamma.$$ In the case of a line bundle one can only have $\alpha=\beta=1$ so denoting $A_i=A_{i1}^1$ you get $$F_{ij}=\frac{\partial A_j}{\partial x^i}-\frac{\partial A_i}{\partial x^j}.$$ In general this is nonzero.

Now, given a bundle metric $g_{\alpha\beta}$, metric-compatibility of the connection means that $A_{i\alpha}^\gamma g_{\beta\gamma}+A_{i\beta}^\gamma g_{\alpha\gamma}=\partial_i g_{\alpha\beta}.$ Differentiating relative to $x^j$, and subtracting off the same equation with $i$ and $j$ swapped, you get the skew-symmetry $$F_{ij\alpha}^\gamma g_{\beta\gamma}+F_{ij\beta}^\gamma g_{\alpha\gamma}=0.$$ Returning again to the case of a line bundle, this says $2F_{ij}=0$ i.e. $F=0.$

$\endgroup$
4
  • $\begingroup$ Not sure why you need this messy coordinate computation when (as you say in the comments) all you use is that $R(X,X)=0$ and that the curvature is a tensor. $\endgroup$ Commented Dec 11, 2021 at 18:47
  • $\begingroup$ In my view it's exactly the same as what I said in the comments, just put in a different language. I think it's very clean, but I know opinions vary on use of coordinates. $\endgroup$ Commented Dec 11, 2021 at 19:23
  • $\begingroup$ Coordinates are fine if they give a shorter explanation, which is not the case here. $\endgroup$ Commented Dec 11, 2021 at 19:46
  • 1
    $\begingroup$ Ok. I think the two explanations have identical length. Anyway it's good to have facility and to know how the argument looks both ways $\endgroup$ Commented Dec 11, 2021 at 20:24

Not the answer you're looking for? Browse other questions tagged .