A flow on a manifold $M$ is a one-parameter group of diffeomorphism $\psi_t: M \to M$. In particular, this induces a smooth group action on $M$, $$ \mathbb{R} \times M \to M $$ $$ (t,m) \mapsto \psi_t(m)$$ So that we have a group homomorphism $\phi: \mathbb{R} \to Diff(M)$.

My question is the following: Given any diffeomorphism $g \in Diff(M)$, does there exist a flow (and thus a vector field) on $M$ such that $g$ is in the image of $\phi$?

  • $\begingroup$ I added the "differential-topology" tag to your post. Cheers! $\endgroup$ – Robert Lewis Apr 22 '18 at 5:04

The answer is no, and the reason may be concisely stated: for any flow

$\phi: \Bbb R \times M \to M \tag 1$


$\phi_t: M \to M, \; \forall x \in M \; \phi_t(x) = \phi(t, x) \tag 2$

is homotopic to the identity map, via a homotopy

$H(s, t, x) = \phi_{st}(x), \; s \in [0, 1]; \tag 3$

it is easily seen that

$H(0, t, x) = \phi_0(x) = x, \; H(1, t, x) = \phi_t(x). \tag 4$

For many $M$ there are, however, diffeomorphisms $\psi:M \to M$ which are not homomtopic to the identity; an example is provided by the antipodal map $\psi(x) = -x$ on $S^{2n}$ for any positive integer $n$.


No, the group of diffeomorphism of a manifold is not always connected, for example for the torus, the connected components are in bijection with $SL(2,\mathbb{Z})$.


  • $\begingroup$ Ok. So we definitely would need a manifold to be connected for this to hold. Does it necessarily hold for, let us say, compact, connected manifolds? $\endgroup$ – Jadwiga Apr 22 '18 at 5:02
  • $\begingroup$ The two dimensional torus is connected. $\endgroup$ – Tsemo Aristide Apr 22 '18 at 5:03

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.