Suppose that $(e_1, \ldots, e_k)$ is an oriented basis for $T_pM$ where $M$ is an oriented Riemannian manifold. In general, we know that we can extend to a smooth local frame $(X_1, \ldots, X_k)$ on $U\ni p$ such that $X_i\rvert_p = e_i$ for each $i$.

But can we further stipulate that $(X_1\rvert_q, \ldots, X_k\rvert_q)$ is oriented for each $q\in U$?

I have tried thinking of ways to shrink $U$ in a suitable way. I have played around with using the orientation form $\omega$ on $M$ and with first applying Gram-Schmidt to $(X_1, \ldots, X_k)$, but no progress.

Any help is much appreciated.


The local frame $(X_i)$ will agree with the chosen orientation form: if $\omega(e_1,\dots,e_k)>0$, then $\omega(X_1,\dots,X_k)>0$ in $U$, because $\omega(X_1,\dots,X_k)|_q=0$ would imply that the $X_i's$ do not form a basis of $T_qM$.

| cite | improve this answer | |
  • $\begingroup$ This solution shows how to restrict $U$ appropriately, but I don't see why vanishing at $\omega \rvert_q$ would imply that the $X_i$ don't form a basis. It's possible that an alternating tensor vanishes at a linearly independent tuple, right? $\endgroup$ – CuriousKid7 Dec 5 '18 at 5:01
  • $\begingroup$ You don't need to restrict $U$ in any way. $M$ is orientable and you have fixed an orientation form $\omega$. What I'm saying is that if $(X_i)$ is a frame inside $U$, then $\omega(X_1,\dots,X_k)$ doesn't change sign inside $U$, because if it does there is a point at which it must vanish, and at that point $(X_i)$ would not be a frame $\endgroup$ – Federico Dec 5 '18 at 13:51
  • 1
    $\begingroup$ "It's possible that an alternating tensor vanishes at a linearly independent tuple, right?" No, unless $\omega=0$ at that point. If it vanishes at a linearly independent tuple, it vanishes on any tuple. $\endgroup$ – Federico Dec 5 '18 at 13:52
  • 1
    $\begingroup$ Assume $\omega(e_1,\dots,e_k)=0$ for some basis $(e_i)$. Take any $k$-uple $v_j=a^i_je_i$. Then $\omega(v_1,\dots,v_k)=\det(A)\omega(e_1,\dots,e_k)=0$, where $A$ is the matrix with entries $a^i_j$. $\endgroup$ – Federico Dec 5 '18 at 13:54
  • $\begingroup$ Got it, thanks for clarifying! $\endgroup$ – CuriousKid7 Dec 5 '18 at 16:27

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.