It is well-known that a linear differential equation in a finite-dimensional vector space depends continuously on some external parameters (for details see below). I search for an explicit reference where this result is extended to the case of a differential equation in a finite-dimensional Lie group. In particular, I need the parameter space be merely a metrisable topological vector space.

(If this question and its answer are folklore and thus not appropriate for Mathoverflow, I apologize in advance)

In more detail, one finds the following result (slightly paraphrased) in Dieudonne's Treatise on Analysis Vol.1 (10.7.2):

Let $ I \subseteq \mathbb R $ be an open interval and $P $ be a metric space (the parameter space). Consider the linear differential equation in the finite-dimensional vector space $ E $ \begin{equation} \dot{x}(t) = A(t, p) \cdot x(t) + b(t, p), \end{equation}
where $ A: I \times P \to L(E, E) $ and $ b: I \times P \to E $ are continuous. If $ t \mapsto u(t,p ) $ is a solution to the initial value problem $ u(t_0, \cdot ) = x_0 $, then $ u $ is continuous in $I \times P $.

I'm interested in the case where $E$ is not a vector space but a finite-dimensional Lie group and the differential equation is adapted accordingly.


See appendix B of the book Lie groups by Duistermaat and Kolk, 2000.

  • 1
    $\begingroup$ I think you refer to Theorem B3, which however only handles the case of a scalar parameter $\epsilon \in \mathbb{R}$. Thanks nonetheless! $\endgroup$ – Tobias Diez Dec 7 '13 at 17:21
  • $\begingroup$ Indeed, but this result extends straigthforwardly at least to normed vector spaces (not necessarily complete), see for example Theorem A6 in my book (or also PhD thesis). Note that there I prove a bit more abstractly that the flow depends smoothly on the vector field in a uniform context, but the same argument can be used to prove that it depends $C^k$ on parameters in a normed space. I don't know if this implies metrisable TVS's. $\endgroup$ – Jaap Eldering Dec 8 '13 at 12:47

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.