Orientation of a real vector bundle

A choice of consistent orientation in each fiber of a real vector bundle, varying continuously across the base.
Orientation of a real vector bundle

Let π:EM\pi:E\to M be a smooth real vector bundle of rank rr over a .

An orientation of EE can be defined in any of the following equivalent ways:

  1. Atlas definition (transition determinants). Choose a vector bundle atlas with local trivializations over an open cover {Uα}\{U_\alpha\} such that the transition functions

    gαβ:UαUβGL(r,R) g_{\alpha\beta}:U_\alpha\cap U_\beta\to \mathrm{GL}(r,\mathbb R)

    all have positive determinant. Two such atlases are equivalent if their union still has positive-determinant transition functions. An orientation is an equivalence class of such atlases.

  2. Frame definition. An orientation is a choice, for each xMx\in M, of one of the two connected components of the set of bases of ExE_x, varying locally so that on overlaps it is preserved; equivalently, it is a choice of the “positive” in each fiber.

  3. Determinant line bundle definition. Let det(E):=ΛrE\det(E):=\Lambda^rE be the top . Then det(E)\det(E) is a real line bundle, and an orientation of EE is the choice of a connected component of det(E){0}\det(E)\setminus\{0\} in each fiber; equivalently, it is the choice of a nowhere-vanishing section of det(E)\det(E) up to multiplication by a positive function.

If MM is connected, an orientation (if it exists) is a global structure; if it does not exist, EE is called non-orientable.

Examples

  1. Tangent bundle of an oriented manifold. An orientation of MM determines an orientation of the TMTM by declaring coordinate frames with positively oriented Jacobian to be positive.

  2. Trivial bundle. The bundle M×RrM\times\mathbb R^r has a canonical orientation given by the standard basis of Rr\mathbb R^r in each fiber.

  3. Möbius line bundle. The Möbius real line bundle over S1S^1 is non-orientable: its transition function on an overlap has negative determinant (it is the constant 1-1 in GL(1,R)\mathrm{GL}(1,\mathbb R)).