Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Let $V \rightarrow M$ be an oriented vector bundle over a compact oriented manifold $M$ equipped with a metric $h$ (the metric $h$ is a metric on the Vector bundle $V$, not on the manifold $M$). Is there some ``natural'' differential $\omega_T$ form representing the Thom Class of $V$? In particular I want the following properties:

1) If $(M,g)$ is a compact two dimensional Riemannian manifold,
and $V = TM$, the tangent bundle of $M$ and $X_0 : M \rightarrow TM$ the zero vector field, then $$ X_0^{*} (\omega_T) = \frac{K}{2 \pi} dA $$ equality holding on the level of forms, where $K$ is the Gaussian curvature and $dA$ is the area form.

2) If $V\rightarrow M$ is a complex vector bundle with a hermitian metric $h$ and $s_o : M \rightarrow V$ the zero section then $s_0^*(\omega_T)$ is the differential form for the top Chern class obtained by Chern Weil theory (again equality holds on the level of forms).

Notice that on the level of cohomology, the pull back via the zero section of the Thom class gives us the Euler class of $V$. My basic question is that what should one take the Thom class to be, to obtain equality on ``the level of forms'' when there is a natural form representing the Euler class.

share|improve this question
add comment

2 Answers

up vote 4 down vote accepted

Here is another construction which goes back to Chern's proof of the Gauss-Bonnet theorem.

Suppose that $\pi: E\to M$ is an oriented rank $2k$ real vector bundle over the manifold $M$. Assume additionally that $E$ is equipped with a metric $g$, and a connection $\nabla$ compatible with the metric.

In your case $2k=\dim M =2$, $E= TM$, $g$ is a Riemann metric and the connection $\nabla$ is the Levi-Civita connection.

Denote by $S(E)$ the unit sphere bundle of $E$. Using the connection $\nabla$ one can explicitly construct a form

$$\Psi(\nabla) \in \Omega^{2k-1}(S(E)) $$

with the property that the integral over each fiber of $S(E)$ is equal to $-1$. The form $\Psi(\nabla)$ is known as the global angular form determined by $g$ and $\nabla$.

The connection $\nabla$ also explicitly defines via the Chern-Weil construction a closed $2k$-form $\newcommand{\be}{\boldsymbol{e}}$

$$\be(\nabla)\in \Omega^{2k}(M). $$

This the the so-called Euler form which in your case is $\frac{1}{2\pi} K dA$.

$\newcommand{\bR}{\mathbb{R}}$ Now choose a smooth function $\rho :[0,\infty)\to [0,\infty)$ which ir equal to $-1$ near $0$ and equal to $0$ on $[1, \infty)$. Let $r: E\to \bR$ denote the radial distance along the fibers of the vector bundle. Define

$$\omega(\nabla)=-\rho'(r) dr\wedge \Psi(\nabla) + \rho(r)\pi^*\be(\nabla). $$

One can then show the following.

1. $\omega(\nabla)$ is a closed form representing the Thom class of $E$.

2. If $\zeta_0: M\to E$ denotes the zero section, then

$$\zeta_0^* \omega(\nabla)=\be(\nabla). $$

3. Suppose that $\zeta:M\to E$ is a section with transversal zero set. For $t>0$ we set $\zeta_t:=t\zeta$. Note that $\zeta$ and $\zeta_t$ have the same zero sets. Let $Z=\zeta^{-1}(0)$ so that $Z$ is a codimension $2k$-submanifold of $M$ whose normal bundle carries a natural orientation. Then the following hold.

a. For any neighborhood $\newcommand{\eN}{\mathscr{N}}$ $\eN$ of $Z$ in $M$ there exists $T=T(\eN)$ such that the support of $\zeta_t^* \omega(\nabla)$ is contained in $\eN$ for any $t>T$.

b. The form $\zeta_t^*\omega(\nabla)$ is Poincare dual to the cohomology class determined by $Z$.

c. As $t\to \infty$ the forms $\zeta_t^*\omega(\nabla)$ converge in the sense of currents to the current of integration determined by $SZ$.

In your case, the zero set consists of finitely many points, the $2$-form $\zeta_t^*\omega(\nabla)$ has the description

$$\zeta_t^*\omega(\nabla)=\rho_t dA $$

where the function $\rho_t$ is concentrated near the zeros of the section, and forms high (positive or negative) peaks around these points.

The statement 3.c. above contains as a special case the Poincare-Hopf theorem.

For more details see Section 8.3.2 of these notes.

share|improve this answer
add comment

Have you looked at Mathai-Quillen's paper: "Superconnections, Thom classes, and equivariant differential forms." Topology 25: 85-100 (1986)? Where they build a form representing the Thom class (together with some very nice properties). You can also look at: "Mathai-Quillen Formalism" Siye Wu available on arXiv(hep-th) (section 2.2).

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.