MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

The $kth$ derivative of a function $f:\mathbb{R}^n \to \mathbb{R}$ can be thought of as a symmetric $k$-tensor. (Well, almost. It is not invariant under coordinate changes, and should really be thought of as part of a jet bundle. But we have nice coordinates on $\mathbb{R}^n$, so lets roll with it)

The exterior derivative maps alternating $k$-forms to alternating $k+1$-forms.

These two "types" of derivative are really the only ones I have met.

Is there a unified notion of "derivative" which contains both of these concepts? Assuming there is, are there other examples besides these two?

share|cite|improve this question
up vote 12 down vote accepted

This is a big subject. There are many, many different notions of derivative in different contexts, all of which derive in some way from jet bundles. This question was considered in great generality in the foundational works on so-called 'differential invariants' by Lie, Cartan, and their followers and was reconsidered in modern times by many people. Some of the basic ideas in modern form can be found in R. Palais' Natural operations on differential forms and there is a more comprehensive modern treatment in the book Natural operations in differential geometry by Kolar, Michor, and Slovak.

Basically, there are methods for defining and computing the invariants of a given geometric structure under the action of some specified (pseudo-)group of transformations, and they yield, as special cases, the exterior derivative of exterior forms, the covariant derivatives of functions with respect to an affine connection, the Schwarzian derivative of a function with respect to linear fractional transformations, and a host of others. Even the Levi-Civita connection of a Riemannian metric can be regarded as the natural first derivative of the metric, which gives a simple example of a (nonlinear) derivative that is not a tensor. Using Cartan's method of equivalence, for example, one can define the invariants to all orders of most $G$-structures and test them for equivalence under various pseudo-groups.

share|cite|improve this answer
The Kolar, Michor and Slovak book is publicly available here: – arsmath Feb 14 '14 at 20:43
Thanks, this looks great! – Steven Gubkin Feb 14 '14 at 21:04

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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