# The Schwartz Space on a Manifold

I asked this question a couple of days ago on math.stackexchange, but have yet to receive a response, so I have decided to post this here.

This question is also vaguely related (both questions arose from the same thing I was working on) to this question I just asked last night.

The question is simple: on a general manifold $M$, can one generalize the space of Schwartz functions on $\mathbb{R}^n$ to a space of smooth functions $\mathcal{S}(M)$ on $M$ that obeys similar properties? I would like to be able to define the convolution of two Schwartz functions, so I guess I better require that $M$ at least be a (unimodular) Lie group. Is it then possible to define $\mathcal{S}(M)$? What about the Fourier transform? Is there a natural definition of the Fourier Transform on $\mathcal{S}(M)$?

• +1 because I have been wondering about getting some kind of algebra of Schwartz functions on a unimodular Lie group -- there are notions for certain Lie groups, due originally (I think) to Harish-Chandra and elaborated by others, but I am a complete novice in that area and don't know if those definitions only make sense for e.g. radial functions. – Yemon Choi Nov 5 '11 at 0:00
• Regarding your last question: if your group is nonabelian then the Fourier transform ought to naturally take values in some topological algebra of operators on some direct sum of "well-behaved irreducible representations". E.g. if your group is compact then there is a well-developed notion of the nonabelian Fourier transform from $L^1(G)$ into $\bigoplus_\pi {\rm End}(H_\pi)$. – Yemon Choi Nov 5 '11 at 0:03
• @Yemon Choi: For closed manifolds, can we use the Nash embedding theorem to treat it some kind of submanifolds in Euclidean space? – Bombyx mori Dec 11 '14 at 3:25

To define a Schwartz space, you need a notion of decay at infinity, so you need a norm'', i.e. a distance to some origin. So the convenient framework is a complete Riemannian manifold. However, even on a Lie group, it is not enough to choose an invariant Riemannian structure to get a Schwartz space having the properties that you require (convolution algebra, good Fourier transformation...). See e.g. the subtlety in the definition of Harish-Chandra's Schwartz space on a semi-simple Lie group, where you have to throw in the $$\Xi$$-function.
For Lie groups, at least for those that embed into $GL_n(R)$ for some $n$, my favorite treatment of the Schwartz space is in Casselman's paper "Introduction to the Schwartz Space of $\Gamma \backslash G$", Can. J. Math. XL, No 2, 1989. There Casselman defines an appropriate Schwarz space on $\Gamma \backslash G$ whenever $G$ is the Lie group obtained by taking the $R$-points of an affine algebraic group over $R$, and $\Gamma$ is any discrete subgroup of $G$ (including the trivial subgroup).