I'm building a vector bundle structure for the space $L_{alt}^k(TM) = \bigcup_{p \in M}L_{alt}^k(T_pM)$, the bundle of alternating k-multilinear maps in $T_pM$.

$L_{alt}^k(T_pM)$ is a vector space of dimension $n \choose k$ and has $\{dx^{i_1} \wedge \dotsc \wedge dx^{i_k}:1 \leq i_1 < \dotsc < i_k \leq n \}$ as a basis.

So, if $\alpha \in L_{alt}^k(T_pM)$, then $ \alpha = \sum a_{i_1\dotsc i_k} dx^{i_1} \wedge \dotsc \wedge dx^{i_k}$, the coefficients uniquely determined by $\alpha$ and $\{dx^{i_1} \wedge \dotsc \wedge dx^{i_k}:1 \leq i_1 < \dotsc < i_k \leq n \}$.

This defines a linear isomorphism $\Phi_p:L_{alt}^k(T_pM) \to \mathbb{R}^d$ where $ d = {n \choose k} $ for each $p \in M$.

Given a chart $(U_\alpha, x_\alpha)$ of $M$, the collection of maps $\{\Phi_p\}_{p \in U_\alpha}$ is my candidate for the principal part $\Phi_\alpha$ of the local trivialisation $\phi_\alpha=(\pi,\Phi_\alpha)$ for that chart.

How can I show that $\Phi_\alpha$ is smooth and $\Phi_p$ is a diffeomorphism for each $p \in U_\alpha$??

  • $\begingroup$ By brute force, using the definitions. Only perserverance is required. $\endgroup$ Oct 7 '15 at 22:30
  • $\begingroup$ @Mariano Could you elaborate on that? I only know smoothness for maps between smooth manifolds... $\endgroup$
    – Luis Vera
    Oct 7 '15 at 22:55
  • $\begingroup$ You know the definition of manifolds and you have candidate for charts. Compute the transition functions and observe that they are smooth. $\endgroup$ Oct 7 '15 at 23:02
  • $\begingroup$ @MarianoSuárez-Alvarez edited, is this argument correct? $\endgroup$
    – Luis Vera
    Oct 8 '15 at 0:20
  • $\begingroup$ I suggest you review the construction of the tangent bundle itself and its dual. $\endgroup$ Oct 8 '15 at 0:29

Finally solved it by revisiting the construction of the cotangent bundle. We have that given two charts that overlap $(U,x), (V,y)$
$$dx^{i_k} = \sum_{j_k}\frac{\partial x^{i_k}}{\partial y^{j_k}}\cdot dy^{j_k} $$ so after some tedious but straightforward calculations we get that $$a_{i_1\cdots i_k} = \sum_{\overset{\to}{L}}a_{i_1\cdots i_k}\frac {\partial(x^{i_1},\ldots,x^{i_k})}{\partial(y^{l_1},\ldots,y^{l_k})}:= \sum_{\overset{\to}{L}}a_{i_1\cdots i_k}\frac{\partial x^I}{\partial x^L} $$ gives the transformation of local coordinates for each coefficient, so given $a \in \mathbb{R}^d$ we get that each $a_i$ goes to $$a_i \overset{proy_i \circ \Phi_x^{-1}}{\longmapsto} \alpha_i = a_{i_1\dotsc i_k} dx^{i_1} \wedge \dotsc \wedge dx^{i_k} \overset{proy_i \circ \Phi_y}{\longmapsto}\sum_{\overset{\to}{L}}a_{i_1\cdots i_k}\frac{\partial x^I}{\partial x^L} $$ which is clearly smooth. A similar procedure gives the inverse transformation and since $\Phi$ is biyective we get the result I was looking for.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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