home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky sci.math:10491 sci.physics:13264
- Newsgroups: sci.math,sci.physics
- Path: sparky!uunet!cs.utexas.edu!torn!watserv2.uwaterloo.ca!watserv1!nuntius
- From: Deane Yang <yang@fields.waterloo.edu>
- Subject: Re: tensors
- Message-ID: <BtCLry.EBC@watserv1.uwaterloo.ca>
- Sender: news@watserv1.uwaterloo.ca
- Organization: The Fields Institute for Research in Mathematical Sciences
- X-Useragent: Nuntius v1.1
- References: <1992Aug20.190041.6215@pellns.alleg.edu>
- <5130@tuegate.tue.nl> <5134@tuegate.tue.nl> <3djygrk@rpi.edu>
- Date: Fri, 21 Aug 1992 19:21:33 GMT
- Lines: 55
-
- In article <3djygrk@rpi.edu> Tom Pierce, pierct@rpi.edu writes:
- >I know all about 1st and second order tensor, but what's a third order
- >tensor look like? Also, aren't these the same mathematical critters as
- >"matrices", witha the added "square" criterion?
- >What would you use a third or highr order tensor for? Any PHYSICAL
- >applications?
-
- Perhaps it's worth distinguishing between tensors and tensor fields.
- A tensor is an algebraic object, living inside a tensor product of
- vector spaces. A tensor field is analogous to a vector field.
- A tensor field is a function with values inside the tensor product,
- just like a vector field is a function taking values in a vector space.
- However, we're usually sloppy about this, calling a tensor field a tensor.
-
- So a second-order tensor is very much like a matrix,
- but a 2nd order tensor field is like a matrix-valued function.
-
- But all of this is a gross oversimplification. It turns out in physics
- or differential geometry, when you work with vector fields or
- tensor fields, you don't want the vector space or the tensor products
- to be a single, fixed object. Instead, each point in the domain
- has a vector space of its own. The vector spaces have an abstract
- intuitive meaning, e.g. "velocity vectors", a.k.a. tangent vectors,
- and when coordinates given on the domain, they induce a natural
- basis of each vector space. In any case, working all of this out carefully
- leads to the notion of tangent bundles, vector bundles, and tensor
- bundles.
-
- The crucial point of tensor fields is
- that the components of the tensor field depend on the
- coordinates you use. Moreover, when the coordinate change, the components
- change linearly, indicating that they represent objects lying in some
- abstract vector space. Understanding the behavior of tensor fields under
- coordinate changes is important, because one of the guiding principles of
- differential geometry and modern physics is that all invariants or
- observable phemomena are invariant under coordinate changes. This
- is usually called "invariance under diffeomorphisms" and considered
- a special case of "gauge invariance".
-
- It turns out that higher order tensors are important. The easiest way
- to see this is that the "derivative" of a k-th order tensor field
- is a (k+1)-th order tensor. So the moment you have the desire to
- differentiate a second order tensor, you've got a third order tensor
- on your hands.
-
- Also, in differential geometry, general relativity, and in gauge theory,
- you work with the curvature tensor, which is a fourth order
- tensor.
-
- Still another natural class of higher order tensors are differential
- forms, which are the correct objects to use for multiple integrals,
- since they encode the change of variable formula, as well as Stoke's
- theorem in a both elegant and easily computed form.
- Deane Yang
- Polytechnic University
-