TENSORS AND ONE-FORMS

Link to: physicspages home page.

1. T ENSORS
Tensors feature prominently in relativity, so it’s a good idea to get a feel
for just what they are. First, it’s important to realize that a tensor, like a
vector, is an object that is independent of the coordinate system being used.
What do depend on the coordinate system are the components of the tensor.
0

A tensor (written as N by Schutz in A First Course in General Rela-
tivity) is actually a linear function of N vectors. The result of this function
is to produce an ordinary number after operating on the N vectors. Thus
the metric tensor g is actually a function of 2 vectors, with this function
providing the scalar product of the two vectors. That is,
 
g A,~ B
~ =A ~ ·B~ (1)
We’re usually used to seeing the scalar product written using the compo-
nents of the metric tensor in a given coordinate system. In a Lorentz frame,
for example, we have

A~ ·B
~ = ηµν Aµ B ν (2)
The scalar product is independent of the coordinate system (it’s the same in
every Lorentz frame), and is a function of two vectors, so the metric tensor
0

defined this way is a 2 tensor.

2. O NE - FORMS
The simplest tensor is a 00 tensor, which is a function of no vectors, and

just returns a scalar. Next up the scale we have a 01 tensor, called a one-


form or 1-form. According to the definition above, a one-form is a function
of a single vector and returns a scalar as its result.
Consider the momentum p of a particle. This is a vector (we’re dealing
with relativity, so all vectors are four-vectors). According to quantum me-
chanics, however, a free particle with momentum p is represented by a plane
wave. This wave has evenly spaced peaks and troughs which are described
1
 TENSORS AND ONE-FORMS 2

by the phase φ of the wave at any point. For a given value of the phase φ,
we can draw a surface in 4-space of all locations where the phase has that
value. The pattern of such surfaces is an example of a one-form.
Now consider two events P0 and P and draw the vector between these
two points as

v = P − P0 (3)
This vector will pierce some number of these constant surfaces represented
by the one-form. If v connects two points where the phase is equal, then it
pierces zero surfaces; if it connects two points with different phases, then
the number of surfaces pierced is the difference in phase between these two
points.
When we’re considering a particle’s momentum, the one-form represent-
ing the phase is often represented by the symbol k̃k . The number of surfaces
of k̃k pierced by v is then written

k̃k , v (4)
Schutz uses the notation

k̃ (vv ) (5)
for the same thing. Schutz’s notation is perhaps clearer, since it emphasizes
the fact that a one-form (a tensor, remember) is a function of a single vector.

3. C OMPONENTS
When dealing with vectors, we usually have to resort to some component
representation in a particular coordinate system. Schutz gives us a method
for determining the components of a one-form.
In a given coordinate system O, we have a set of basis vectors ~eα . The
components of a one-form p̃ are obtained by letting it act on each basis
vector. That is

pα ≡ p̃ (~eα ) (6)
From this, we find the expression for the one-form acting on a general vector
~ Since A
A. ~ is written in terms of the basis vectors as

~ = Aα~eα
A (7)
and the one-form is a linear function, we have, since the components Aα
are just numbers:
 TENSORS AND ONE-FORMS 3

 
~ = p̃ (Aα~eα )
p̃ A (8)
= Aα p̃ (~eα ) (9)
= A α pα (10)
We can use this result to obtain a set of basis one-forms in a given frame
O. We would like to be able to write a one-form in terms of a set ω̃ α of
basis one-forms as

p̃ = pα ω̃ α (11)
As it stands, this is an abstract equation, since we must remember that a
one-form is a function of a vector. Thus in order for 11 to make sense, we
~ above. We
have to use it to act on a vector. Let’s applying to the vector A
get
   
~
p̃ A = pα ω̃ Aα ~
(12)
 
= pα ω̃ α Aβ ~eβ (13)
= pα Aβ ω̃ α ~eβ


(14)
Notice that this formula involves two sums; one over α and one over β.
Also note that we have a set of 4 ω̃ α s, each of which acts on each of the 4
basis vectors ~eβ . Each one-form has 4 components, as given by 6. Now in
~ we must have
order for 14 to be consistent with 10 for any vector A,

ω̃ α ~eβ = δ αβ


(15)
That is, the components of ω̃ α in the coordinate system O must be

ω̃ 0 → (1, 0, 0, 0)
ω̃ 1 → (0, 1, 0, 0)
(16)
ω̃ 2 → (0, 0, 1, 0)
ω̃ 3 → (0, 0, 0, 1)
These are the basis one-forms relative to the basis vectors ~eα . If we used
different basis vectors, we’d get different basis one-forms.
What is the relation between a vector p and a corresponding one-form
p̃p? In quantum mechanics, the phase becomes the momentum when it is
multiplied by h̄ (in relativistic quantum theory, we usually take h̄ = 1, so
then momentum and phase become the same). If we multiply k̃k above by h̄,
 TENSORS AND ONE-FORMS 4

we then get the momentum one-form p̃p. If we now use this one-form to act
on some vector v , we get

hp̃p, v i (17)
which is the number of surfaces of p̃p pierced by v . This is just another way
of saying that we are finding the projection of v onto the vector p , which is
given by the scalar product. In other words,

p · v = hp̃p, v i (18)
P INGBACKS
Pingback: Components of one-forms and vectors
Pingback: Gradient as a one-form
Pingback: Normal vectors and unit normal one-forms
Pingback: Covariant derivative of a general tensor
Pingback: Metric tensor - inverse and raising and lowering indices
Pingback: Tensors in Maple - definition and index actions
Pingback: Representation of rotation group with rank-2 tensors
Pingback: One-form basis