Register to reply

Dual basis and differential forms...

by Buri
Tags: basis, differential, dual, forms
Share this thread:
Buri
#1
Dec28-10, 03:25 PM
P: 273
I was reading about dual spaces and dual bases in the book Linear Algebra by Friedberg, Spence and Insel (FSI) and they give an example of a linear functional, f_i (x) = a_i where [x]_β = [a_1 a_2 ... a_n] denotes the matrix representation of x in terms of the basis β = {x_1, x_2, ..., x_n} of V. Now they go on to prove that {f_1, f_2, ..., f_n} is in fact the dual basis of β for V* by actually never using the fact that f_i (x) = a_i, but rather that f_i (x_j) = δ_ij, where δ_ij denotes the Kronecker delta function. I also happened to have read other references by Halmos and Lang and they did not go about actually finding the linear functionals, but rather use the fact that a linear functional that satisfies φ_i (x_j) = δ_ij exists. Now I suppose my question is, doesn't it so happen that the dual basis of β ALWAYS equals {f_1, f_2, ..., f_n} as defined by FSI? Because since φ and f agree on the basis elements in turns out that φ_i = f_i for all x.

Now here is where differential forms come in. I was reading do Carmo's book on Differential forms and he says that the basis of R is {dx_i; i = 1,2,3} where x_i represents the i-th coordinate function. So I guess its true that dx_i = f_i(x) from above then right? I, to be honest, find this a bit unexpected as x_i is precisely f_i when we're using the canonical basis, yet differentiating doesn't mess it up. So I actually went on to check it on some values that dx_i really does equal f_i(x). And so it seems to also verify what I was asking in the first paragraph.

Is this all right? I'd appreciate someone clarifying some of this stuff to me.

Thanks!

EDIT: I guess another way of putting my question is, is the dual basis unique, in the sense that the linear functionals that form the basis are always identically equal to f_i(x) as given by FSI?
Phys.Org News Partner Science news on Phys.org
Scientists develop 'electronic nose' for rapid detection of C. diff infection
Why plants in the office make us more productive
Tesla Motors dealing as states play factory poker
Hurkyl
#2
Dec28-10, 04:07 PM
Emeritus
Sci Advisor
PF Gold
Hurkyl's Avatar
P: 16,091
Another way of putting your question is:

If you have two lists f and g of linear functionals satisfying:
fi(xj) = δ_ij
gi(xj) = δ_ij
Can you prove that, for each i,
fi = gi
?
Buri
#3
Dec28-10, 04:12 PM
P: 273
I guess I could since they agree on basis elements they should be equal - that's a result from linear algebra. I guess that answers my question, but the differential form example sorta throws me off...

Buri
#4
Dec28-10, 04:15 PM
P: 273
Dual basis and differential forms...

I suppose the purpose of 'rewriting' the f_i (x) as dx_i is probably just more convenient as it is "differential" forms.
JSuarez
#5
Dec28-10, 08:33 PM
P: 402
...purpose of 'rewriting' the f_i (x) as dx_i is probably just more convenient as it is "differential" forms.
The purpose is that do Carmo's is interested in differential forms defined on a manifold (or in an open set of it), not just on a single point. Therefore, given a manifold M and [itex]p \in M[/itex] and R3 is its tangent space at p, then the dual basis to the canonical one is indeed dx_i(p) and is exactly the same as the dual basis as defined in linear algebra.

This notation is useful when, for example, you have a function f defined on an open set U of M and you need df, which is then written (usually supressing p):

[tex]df(p) = \sum_{i = 1}^{n}a_{i}\left(p\right)dx_{i}\left(p\right)[/tex]

Where the a_i's are (usually) smooth functions on U.
Buri
#6
Dec29-10, 08:55 AM
P: 273
Quote Quote by JSuarez View Post
The purpose is that do Carmo's is interested in differential forms defined on a manifold (or in an open set of it), not just on a single point. Therefore, given a manifold M and [itex]p \in M[/itex] and R3 is its tangent space at p, then the dual basis to the canonical one is indeed dx_i(p) and is exactly the same as the dual basis as defined in linear algebra.

This notation is useful when, for example, you have a function f defined on an open set U of M and you need df, which is then written (usually supressing p):

[tex]df(p) = \sum_{i = 1}^{n}a_{i}\left(p\right)dx_{i}\left(p\right)[/tex]

Where the a_i's are (usually) smooth functions on U.
Ahh that makes sense to me now. Thanks a lot JSuarez! Also thanks to Hurkyl for helping me out too!


Register to reply

Related Discussions
Dual basis problem. (Linear Algebra) Calculus & Beyond Homework 11
Abstract Linear Algebra: Dual Basis Calculus & Beyond Homework 3
Dual basis in Minkowski space Special & General Relativity 2
Dual Basis Linear & Abstract Algebra 5