0
$\begingroup$

If one defines a smooth $(r,s)$ tensor field on a differentiable manifold as a $C^{\infty}(M)$ multilinear map $$T : (\Gamma(T^*M))^r \times (\Gamma(TM))^s \xrightarrow{\sim} C^{\infty}(M)$$ from the modules of smooth sections of the cotangent and tangent bundles respectively, how does one define contractions (in a coordinate free way)? Must an approach with tensor products be employed?

I know the idea of contraction is to take a vector field input and covector field input and let the second act on the first to multiply the tensor by a smooth function, but I'm not sure how to implement it in this picture.

Approach via tensor products: Coordinate-free notation for tensor contraction?

The answer to that post gives a coordinate free way to get a contraction over a single vector space in terms of the universal property of the tensor product, similar to the idea I quoted above. So this is only applicable if we consider tensors as sections of the tensor bundle and hence are able to write in some coordinate patch on $M$ in a form like $$T = T_{\mu\nu} dx^\mu \otimes dx^\nu$$

So my question becomes: can one define tensor field contraction without going to a pointwise construction?

$\endgroup$
2
  • 1
    $\begingroup$ Not sure what you mean. Tensor contraction is a pointwise operation. I think of a tensor bundle as a parameterized family of vector spaces where the parameter space is the manifold. Any linear algebra or tensor operation that can be applied to a single vector space can be applied to each fiber of vector bundle. For me that’s easier to work with than the infinite dimensional space of sections. $\endgroup$ Commented Mar 24, 2023 at 23:48
  • $\begingroup$ Okay, this is what I was beginning to suspect. Thank you for making it clear. $\endgroup$ Commented Mar 25, 2023 at 0:40

0

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.