[go: up one dir, main page]
Scribd
Upload
27 views3 pages

Exercise (12.16) (amended) : α α = 0 ⇔ Α = γ δ (i.e. α is simple)

This document proves two assertions about dual tensors: 1. A tensor αrs is simple (i.e. αrs is a product of Kronecker deltas) if and only if α[rsαu]v = 0. 2. If tensors ψ and α are dual, then ψr..tu αuv...w = 0 if and only if ψ and α are simple. The proof uses properties of the Levi-Civita symbol to relate the dual tensors to their simple forms.

Uploaded by

Jaco Greeff
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
27 views3 pages

Exercise (12.16) (amended) : α α = 0 ⇔ Α = γ δ (i.e. α is simple)

This document proves two assertions about dual tensors: 1. A tensor αrs is simple (i.e. αrs is a product of Kronecker deltas) if and only if α[rsαu]v = 0. 2. If tensors ψ and α are dual, then ψr..tu αuv...w = 0 if and only if ψ and α are simple. The proof uses properties of the Levi-Civita symbol to relate the dual tensors to their simple forms.

Uploaded by

Jaco Greeff
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Exercise [12.

16] (amended)

[12.16] First assertion:

α[rsαu]v = 0 ⇔ αrs = γ[rδs] (i.e. αrs is simple)

Proof "⇒"

Choose vectors xν and zµ such that

z μ xν α μν = 2 (1)

(This is always possible if α≠0, because for the choice xν = δνk and zµ = δµm the sum reduces
to a single component αmk. Choosing m, k such that a nonzero αmk ≠ 0 is selected, and
normalizing xν and/or zµ appropriately yields eq.(1).)

We can then define:

γ r = xν α rν δ s = z μα μs = − z μα sμ (2)

(the last equation holds because αrs can be assumed to be antisymmetric – cf. Penrose,
Chapter 12.4). We then get:

2 ⋅ γ [ rδ s ] = (γ rδ s − γ sδ r ) = xν α rν ⋅ z μα μs − xν α sν ⋅ z μα μr = xν z μ (α rν ⋅ α μs − α sν ⋅ α μr )
( 2)
(3)

Next we exploit the "⇒" prerequisite, i.e.

0 = α[ rsα μ ]v =
1
3!
(
α rs ⋅ α μν + α sμ ⋅ α rν + α μr ⋅ α sν − α sr ⋅ α μν − α μs ⋅ α rν − α rμ ⋅ α sν ) (4)

to replace the bracket in eq. (3) (the terms of the bracket are underlined in (4)):

2 ⋅ γ [ r δ s ] = ... = xν z μ (α rs ⋅ α μν + α sμ ⋅ α rν − α sr ⋅ α μν − α rμ ⋅ α sν )
( 4)

= α rs ⋅ xν z μα μν + z μα sμ ⋅ xν α rν − α sr ⋅ xν z μα μν − z μα rμ ⋅ xν α sν

(1)( 2 )
= α rs ⋅ 2 + (−δ s ) ⋅ γ r − α sr ⋅ 2 − (−δ r ) ⋅ γ s = 2α rs − 2α sr + δ r ⋅ γ s − δ s ⋅ γ r

= 2 ⋅ (α rs − α sr ) − (γ r ⋅ δ s − γ s ⋅ δ r ) = 4 ⋅ α rs − 2 ⋅ γ [ r ⋅ δ s ]

⇒ γ [ r δ s ] = α rs q.e.d.

Juergen Beckmann
Proof "⇐"

Let αrs be simple, i.e.

α rs = γ [ rδ s ] =
1
(γ rδ s − γ sδ r ) = 1 (r , s − s, r ) (5)
2 2

The last term of this equation uses the short form

γ rδ s = r , s (6)

We then have:

2 ⋅ 3! α[ rsα u ]v = 2(α rs − α sr ) ⋅ α uv + 2(α su − α us ) ⋅ α rv + 2(α ur − α ru ) ⋅ α sv

= 4α rs ⋅ α uv + 4α su ⋅ α rv + 4α ur ⋅ α sv (αrs is antisymmetric!)

( 5)
= (r , s − s, r ) ∗ (u, v − v, u ) + ( s, u − u, s ) ∗ (r , v − v, r ) + (u, r − r , u ) ∗ ( s, v − v, s )

= r , s ∗ u , v − r , s ∗ v, u − s, r ∗ u , v + s, r ∗ v, u + ...

... + s, u ∗ r , v − s, u ∗ v, r − u , s ∗ r , v + u , s ∗ v, r + ...

... + u, r ∗ s, v − u , r ∗ v, s − r , u ∗ s, v + r , u ∗ v, s = 0

In the last line, terms that are underlined in the same colour cancel, taking into account that,
for example:

r , s ∗ u , v = γ r δ s ∗ γ u δ v = γ r δ v ∗ γ uδ s = r , v ∗ u , s

(i.e. corresponding indices can be swapped in the products).

The aforementioned swapping of indices is possible because they belong to single factors,
i.e. because αrs is assumed to be simple. This is surely the secret why the assertion holds
also for higher numbers of indices, i.e. αr…s, but I did not succeed to find a formal proof.

Juergen Beckmann
[12.16] Second assertion:

If ψr..tu and αuv…w are dual, then

ψr..tu αuv…w = 0 ⇔ ψr..tu and αuv…w are simple

Proof:

Because ψ and α are dual, we have:

ψ r ...tuα uv...w = ψ r ...tuε uv...wx... zψ x... z (7)

Two things can be observed with respect to eq.(7):

1. The right sum (summation convention!) comprises only addends in which all n indices
uv…wx…z are different from each other (otherwise εuv…wx…z would be zero). The
indices v…w are given in advance. Due to the summation, the residual numbers from
the set {1,2,…n} are distributed to the indices u,x,…z in all possible permutations.

2. Hence each addend is proportional to C⋅ε12…n ⋅ sign(π) with C being a nonzero


constant and π being the permutation of (u,x…z) that corresponds to this addend.

Accordingly, eq. (7) can be continued as follows:

ψ r ...tuα uv...w = Cε12...n


π
∑ sign(π ) ⋅ψ
( u , x ... z )
r ...t π ( u )
ψ π ( x )...π ( z ) = Cε12...n (k + 1)!ψ r ...t [uψ x... z ] (8)

(with k being the number of indices of ψr…tu).

From eq.(8) it follows that

ψ r ...tuα uv...w = 0 ⇔ ψ r ...t [uψ x... z ] = 0 ⇔ ψ r ...tu is simple (due to the first assertion).

An analogous reasoning holds for αuv…w (using ∈uv...wx... z in eq.(7) to express ψ by α).

Juergen Beckmann

You might also like