III- INDEX NOTATION

The index notation for vectors is far more useful than the notation that we have used
before. Index notation has the dual advantages of being more concise and more transparent.
Proofs are shorter and simpler. It becomes easier to visualize what the different terms in
equations mean.

III-1 Index notation and the Einstein summation convention

We begin with a change of notation, instead of writing 𝑎⃗ = 𝑎 𝚤̂ + 𝑎 𝚥̂ + 𝑎 𝑘
we write 𝑎⃗ = 𝑎 𝑒̂ + 𝑎 𝑒̂ + 𝑎 𝑒̂ = ∑ 𝑎 𝑒̂
We simplify this further by introducing the Einstein summation convention: if an index
appears twice in a term, then it is understood that the indices are to be summed from 1 to 3.
Thus we write
𝑎⃗ = 𝑎 𝑒̂
Examples:
𝑎 = 𝑏  𝑎 = 𝑏 ,𝑎 = 𝑏 , 𝑎 = 𝑏
𝑎 𝑏 = 𝑎 𝑏 +𝑎 𝑏 +𝑎 𝑏

III-2 Summation convention and dummy indices

Let consider: 𝑎⃗. 𝑏⃗ = 𝑎 𝑏 + 𝑎 𝑏 + 𝑎 𝑏
This indicate that we can write a scalar product as : 𝑎⃗. 𝑏⃗ = 𝑎 𝑏
In the term aibi, an index like i that is summed over is called a dummy index (or more cavalierly
as a dummy variable). The index used is irrelevant - just as the integration variable is irrelevant
in an integral (though in this case the term dummy variable is entirely appropriate).
This means: 𝑎 𝑏 = 𝑎 𝑏 = 𝑎 𝑏 = 𝑎 𝑏 = 𝑎 𝑏 + 𝑎 𝑏 + 𝑎 𝑏
NB : A term cannot contain an index more than twice, if a compound calculation would lead to
such a situation, the dummy index should be changed. An index that appears only once in
a term is called a free or floating index.
For example: 𝑎 𝑏 𝑐 = (𝑎 𝑏 + 𝑎 𝑏 + 𝑎 𝑏 )𝑐

𝑎 𝑏 = 𝑎 𝑏 =𝑎 𝑏 +𝑎 𝑏 +𝑎 𝑏

j: repeated subscripts = dummy subscripts

i: free or distinct subscripts

In an equation, all terms must contain the same free indices, in particular you should note
that : 𝑎 𝑏 𝑐 ≠ 𝑎 𝑏 𝑐
If we have s = aibi and we want sci, as noted above we change the subscripts on a and b because
they are the dummy indices. Do not change the free indices because you risk changing the
equation. Thus s𝑐 = 𝑎 𝑏 𝑐
Contracted product
Contraction is the process of setting two free indices equal
10
  The contracted product of two 2nd order tensors is a 2nd order tensor defined by :
𝐶 = 𝐴. 𝐵 𝐶 = 𝐴 𝐵
 The double contrated product of two 2nd order tensors est un scalar
𝑠 = 𝐴. 𝐵 𝐴 𝐵 = 𝐴 𝐵 = 𝑇𝑟(𝐴. 𝐵 )
 The contracted product of a 2nd order tensors by a vector b is a vector :
 𝐶 = 𝐴. 𝐵 𝐶 = 𝐴 𝐵
𝐴. 𝑏⃗ = 𝑐⃗ 𝐴 𝑏 =𝑐
𝑏⃗. 𝐴 = 𝑑⃗ 𝑏𝐴 =𝑑

III-3 The Kronecker delta or the substitution operator

The Kronecker delta, 𝛿 = 1 𝑖𝑓 𝑖 = 𝑗 𝑎𝑛𝑑 𝛿 = 0 𝑖𝑓 𝑖 ≠ 𝑗
So 𝛿 = 𝛿 = 𝛿 = 1 𝑎𝑛𝑑 𝛿 = 𝛿 = 𝛿 = 0
We will sometimes find it convenient this result in an array
𝛿 𝛿 𝛿 1 0 0
𝛿 = 𝛿 𝛿 𝛿 = 0 1 0
𝛿 𝛿 𝛿 0 0 1
Why is the Kronecker delta also known as the substitution operator? We can figure this out by
making a calculation.
Consider 𝛿 𝑎 = 𝛿 𝑎 let j take on the values 1, 2, and 3. Then we have:
𝑗 =1∶ 𝛿 𝑎 +𝛿 𝑎 +𝛿 𝑎 = 𝑎
𝑗 =2∶ 𝛿 𝑎 +𝛿 𝑎 +𝛿 𝑎 = 𝑎
𝑗 =3∶ 𝛿 𝑎 +𝛿 𝑎 +𝛿 𝑎 = 𝑎
From this we can see that 𝛿 𝑎 = 𝑎
Thus applying Kronecker delta allows us to drop a repeated index and changes one index
into another.
Examples:
𝐴 𝐵𝐶𝛿 =𝐴 𝐵𝐶 =𝐴 𝐵𝐶
So we see that if two indices are repeated, only one is dropped. We should note the following
obvious results:
𝛿 = 1 + 1 + 1 = 3 And 𝛿 𝛿 = 𝛿

III-4 The permutation symbol or Levi-Civita tensor

The number 1, 2, 3 are in cyclic order if they occur in the order 1, 2, 3 on a
counterclockwise path starting from 1.
⃗ ⃗

+𝟏
𝒌⃗ −𝟏
𝒌⃗
⃗
⃗

Cyclic Permutations Non Cyclic Permutations

11
 Cyclic permutations are sometimes called even, non-cyclic permutations are sometimes
called odd. This idea can be used in the evaluation of vector products. The idea is introduced
through the permutation symbol 𝜀
𝜀 = +1 if 𝑖𝑗𝑘 is a cyclic permutation of 1,2,3.
𝜀 = −1 if 𝑖𝑗𝑘 is a non − cyclic permutation of 1,2,3.
𝜀 = 0 if 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒, 𝑖. 𝑒 𝑎𝑛 𝑖𝑛𝑑𝑒𝑥 𝑖𝑠 𝑟𝑒𝑝𝑒𝑎𝑡𝑒𝑑.
So we find
𝜀 =𝜀 =𝜀 = +1 and 𝜀 =𝜀 =𝜀 = −1
While
𝜀 =𝜀 =𝜀 =0
We should also note the following properties:
𝜀 =𝜀 =𝜀
But when we swap indices
𝜀 = −𝜀 𝑎𝑛𝑑 𝜀 = −𝜀

III-5 Vector Product in index notation

Recall
𝑖 𝑗 𝑘
𝑎⃗ 𝑏 =
⃗ 𝑎 𝑎 𝑎
𝑏 𝑏 𝑏
Now consider
𝐶 =𝜀 𝑎𝑏
This is a vector characterized by a single free index i. The indices j and k are dummy
indices and are summed out. We get the three values of 𝐶 by letting i = 1, 2, 3 independently.
This is useful but the method is made more powerful by the methods of the next section.
𝑈⃗  𝑉⃗ = 𝜀 𝑈 𝑉
Scalar Triple Product
𝑎 𝑎 𝑎
𝑎⃗ 𝑏⃗ 𝑐⃗ = 𝑏 𝑏 𝑏 =𝜀 𝑎𝑏𝑐
𝑐 𝑐 𝑐

III-6 The  - identity

𝜀 𝜀 =𝛿 𝛿 −𝛿 𝛿
This identity can be used to generate all the identities of vector analysis, it has four free indices.
To prove it by exhaustion, we would need to show that all 81 cases hold. Note that the ε’s have
the repeated index first, and that in the δ’s, the free indices are take in this order:
1. both second
2. both third
3. one second, one third
4. the other second, the other third

12
 Let’s put this to use by proving what would be a tough identity using ordinary vector
methods.
We’ll prove the bac-cab rule.
Example : Proof that :
𝐴⃗  𝐵⃗  𝐶⃗ = 𝐴⃗ . 𝐶⃗ 𝐵⃗ − (𝐴⃗. 𝐵⃗ ) 𝐶⃗
To prove this, let :
𝐴⃗  𝐵⃗  𝐶⃗ = 𝐴⃗  𝐷⃗ = 𝐸⃗
We convert to index notation as follows :
𝐵⃗  𝐶⃗ = 𝜀 𝐵 𝐶 = 𝐷
Then
𝐴⃗  𝐷⃗ = 𝜀 𝐴 𝐷 = 𝜀 𝐴 𝜀 𝐵 𝐶 = 𝐸
Rearranging terms, we have
𝐸 =𝜀 𝜀 𝐴 𝐵𝐶 =𝜀 𝜀 𝐴 𝐵𝐶
And using the - identity
𝐸 = (𝛿 𝛿 − 𝛿 𝛿 )𝐴 𝐵 𝐶
Then
𝐸 =𝛿 𝛿 𝐴 𝐵 𝐶 −𝛿 𝛿 𝐴 𝐵 𝐶
Then using the substitution properties of the Kronecker deltas, this becomes
𝐸 =𝐴 𝐵 𝐶 −𝐴 𝐵 𝐶
= 𝐵 (𝐴 𝐶 ) − 𝐶 (𝐴 𝐵 )
= 𝐵⃗ 𝐴⃗ . 𝐶⃗ − 𝐶⃗(𝐴⃗. 𝐵⃗ ) QED

III-7 Matrix Product

Product between a matrix and a vector
𝐴 𝑎⃗ = 𝐴 𝑎 = 𝑎 𝐴
𝑎⃗ 𝐴 = 𝑎 𝐴 = 𝐴 𝑎
𝑎⃗ : Transpose (3X1) (1x3)
Product between two matrices A and B
𝐴𝐵 = 𝐴 𝐵
𝐴𝐵 = 𝐴 𝐵
𝐴 𝐵= 𝐴 𝐵
𝑡𝑟(𝐴 𝐵) = 𝐴 𝐵
𝑡𝑟(𝐴 𝐵 ) = 𝑡𝑟(𝐴 𝐵) = 𝐴 𝐵
With
𝐴 :Transpose matrix of A
𝑡𝑟𝐴 : Trace of the matrix A defined as : 𝑡𝑟𝐴 = 𝐴 = 𝐴 +𝐴 +𝐴

13
IV- COORDINATE TRANSFORMATIONS
Units basics vectors
𝑥
𝑥 {𝑒̂ } = {𝑒̂ , 𝑒̂ , 𝑒̂ }
𝒘⃗ {𝑒̂ } = {𝑒̂ , 𝑒̂ , 𝑒̂ }

𝑒̂ 𝑥
𝑒̂ Two Cartesian frames : (𝑥 , 𝑥 , 𝑥 ) and
𝑒̂ (𝑥 , 𝑥 , 𝑥 )
𝑒̂ 𝑥
𝑒̂ O
𝑄 = 𝐶𝑜𝑠𝑖𝑛𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑎𝑛𝑔𝑙𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑡ℎ𝑒 𝑥
𝑒̂
− 𝑎𝑥𝑖𝑠 𝑎𝑛𝑑 𝑡ℎ𝑒 𝑥 − 𝑎𝑥𝑖𝑠 ∶
𝑥 𝑄 = 𝐶𝑜𝑠(𝑥 , 𝑥 )
𝑥
The basic vectors in the primed coordinated frame is :
𝑒̂ = 𝑄 𝑒̂ + 𝑄 𝑒̂ + 𝑄 𝑒̂
𝑒̂ = 𝑄 𝑒̂ + 𝑄 𝑒̂ + 𝑄 𝑒̂
𝑒̂ = 𝑄 𝑒̂ + 𝑄 𝑒̂ + 𝑄 𝑒̂
In index notation : 𝑒 = 𝑄 𝑒
The opposite transformation is : 𝑒 = 𝑄 𝑒
Let consider arbitrary a vector 𝑤⃗
𝑤⃗ = 𝑤 𝑒̂ + 𝑤 𝑒̂ + 𝑤 𝑒̂ = 𝑤 𝑒̂
= 𝑤 𝑒̂ + 𝑤 𝑒̂ + 𝑤 𝑒̂ = 𝑤 𝑒̂
By substituting 𝑒 ….
𝑤⃗ = 𝑤 𝑒̂ = 𝑤 𝑄 𝑒̂
Since 𝑤⃗ = 𝑤 𝑒̂  𝑤 𝑄 𝑒̂ = 𝑤 𝑒̂  𝑤 𝑄 𝑒̂ = 𝑤 𝑒̂  𝑤 𝑄 = 𝑤
 𝒘𝒊 = 𝑸𝒊𝒋 𝒘𝒋
Analogously,
𝑒̂ = 𝑄 𝑒̂ and 𝑤⃗ = 𝑤 𝑒̂ 
𝑤⃗ = 𝑤 𝑄 𝑒̂
 𝒘𝒊 = 𝑸𝒋𝒊 𝒘𝒋
𝑤⃗ = 𝑤 𝑒̂
We have transformation between orthogonal coordinate system:
𝑤 =𝑄 𝑤 = 𝑄 𝑄 𝑤
Remember Kronecker delta 𝛿
𝛿 𝑤 = 𝑄 𝑄 𝑤  (𝑄 𝑄 − 𝛿 )𝑤 = 0
Since it is true for all vectors wk  𝑄 𝑄 −𝛿 =0𝑄 𝑄 =𝛿
Analogously :
𝑤 =𝑄 𝑤 = 𝑄 𝑄 𝑤
𝛿 𝑤 = 𝑄 𝑄 𝑤  (𝑄 𝑄 − 𝛿 )𝑤 = 0
Since it is true for all vectors 𝑤  𝑄 𝑄 −𝛿 =0𝑄 𝑄 =𝛿
The two relations:

14
𝑄 𝑄 =𝛿
 Orthogonality condition: 𝑑𝑒𝑡 (𝑄 ) = 1
𝑄 𝑄 =𝛿

Example 1 : Transformation
Let consider the components of a first and second order tensor in a particular coordinate
frame as :
1 1 0 3 𝑥
𝑎 = 4 ; 𝑎 = 0 2 2
𝑥
2 3 2 4
Determine the components of each tensor in a 𝑥
new coordinate system (x’, y’, z’) obtained 𝑥
60°
through a rotation of 60° as shown in the
𝑥 𝑥
figure aside. Choose a counterclockwise
rotation.

Solution
𝑐𝑜𝑠60° 𝑐𝑜𝑠30° 𝑐𝑜𝑠90° 1 ⁄2 √3⁄2 0
𝑄 = 𝑐𝑜𝑠150° 𝑐𝑜𝑠60° 𝐶𝑜𝑠90° = −√3⁄2 1⁄2 0
𝑐𝑜𝑠90° 𝐶𝑜𝑠90° 𝐶𝑜𝑠0° 0 0 1
The transformation of the vector quantity is given by :
1
1 ⁄2 √3⁄2 0 1 + 2√3
𝑎 = 𝑄 𝑎 = −√3⁄2 1⁄2 0 4 = 2
2 − √3/2
0 0 1 2 2
1 ⁄2 √3⁄2 0 1 0 3 1 ⁄2 √3⁄2 0
𝑎 = 𝑄 𝑄 𝑎 = −√3⁄2 1⁄2 0 0 2 2 −√3⁄2 1⁄2 0
0 0 1 3 2 4 0 0 1
3
⎡ 7 ⁄4 √3⁄4 + √3 ⎤
⎢ 2 ⎥
= ⎢ √3⁄4 5 ⁄4 1 − 3 √3⁄2⎥
⎢3 ⎥
+ √3 1 − 3 √3 ⁄2 4
⎣2 ⎦

15