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

I'm looking for a definition of pseudo differential forms in ordinary differential geometry. However searching the web gave me nothing. There are definitions in supergeometry but that is not what I'm after.

Recently I read, that pseudo-differentialforms are the natural structure to integrate, since integration works on any kind of submanifold (orientation not required) for them, but those texts don't gave a 'clean' definition of these kind of forms.

What are pseudo-differentialforms?

Can pseudo differentialforms be defined as sections of some kind of fiber bundle? If yes that's a definition I would prefer.

share|cite|improve this question
I suspect you are looking for a definition of 1-densities. – Eugene Lerman Feb 7 '13 at 15:13
They're not densities. They're forms with coefficients in a (flat) real line bundle. – Liviu Nicolaescu Feb 7 '13 at 15:15
I'm guessing you're looking for the notion of k-densities as explanined in my answer to this MO question: If you insist on a complicated definition they are sections of a determinant line bundle over the grassmannian bundle on manifold, but they're simple objects that we use every day like $\sqrt{dx^2 + dy^2}$ – alvarezpaiva Feb 7 '13 at 15:33
By the way, here is another MO question on this topic – alvarezpaiva Feb 7 '13 at 15:35
Is your definition of a density functorial? – Nevermind Feb 7 '13 at 16:01

There at least two sources I am aware of.

  1. Theodore Frankel, The Geometry of Physics, Section 2.8 and 3.4.

  2. Georges De Rham, Varietes Differentiables. Formes, courants, formes harmoniques, Chap. II.

share|cite|improve this answer
thanks, I will look on these references tomorrow, since I don't have access right now. – Nevermind Feb 7 '13 at 16:22
up vote 2 down vote accepted


Let $M$ be a topological manifold and $PM$ the frame bundle of $M$. If $dim(M)=n$ then $PM$ is a $Gl(n)$-principal bundle.

Let $\tau: Gl(n) \to \mathbb{R} \; ; \; A \mapsto abs(det(A))$ the map, that maps any linear isomorphism $f \in Gl(n)$ to the absolute value of its determinant. This defines a left action of $Gl(n)$ on $\mathbb{R}$ by

$$\cdot: Gl(n) \times \mathbb{R} \to \mathbb{R} \; ; \; (A,x) \mapsto \tau(A)x$$

The bundle of pseudo-forms then is the associated (line) bundle


of this action and pseudo-forms are sections of this bundle. If $M$ is smooth, this is a smooth bundle,since the action is smooth. ($\tau$ is smooth since $det(A)\neq0$ for $A\in Gl(n)$)

But this gives only pseudo-forms that behaves right in respect to integration on $dim(M)$-dimensional submanifolds. Remains the question, ow to generalize this to submanifolds of arbitrary dimension.

share|cite|improve this answer
Just using the absolute value here sounds a bit arbitrary to me. What's the reason for that and how to generalize? – Nevermind Feb 12 '13 at 0:02
Just use the bundle of $k$-frames ($k$ linearly independent tangent vectors), which is a principal $GL(k)$ bundle. The rest is the same. – Deane Yang Feb 12 '13 at 0:03
You want to use the absolute value, so the integral does not depend on orientation and the thing you're integrating looks more like a measure on the manifold or submanifold, rather than a differential form. in particular, it allows you to define integration on a non-orientable manifold. If you use forms without the absolute value, there are no global sections to integrate. – Deane Yang Feb 12 '13 at 0:05
@Nevermind. Remember the co-ordinate change formula for the integral ? What comes out is the absolute value of the determinant of the jacobian of the co-ordinate change function. If you don't have an orientation this is how things have to transform to get a sensible definition of integral. – Michael Murray Feb 12 '13 at 11:09
From a top-down perspective, there are only so many linear actions of [the multiplicative group of R] on a 1-dimensional vector space, so that maybe it's not surprising that this one crops up here. – Tim Campion Sep 22 '13 at 22:01

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.