Methods of Mathematical Physics (PH 301)

The Review of Vector Analysis

Elementary definition of scalars and vectors

Scalar = number + proper - units; distance = 5 m; speed = 5 m / sec

mass = 5 kg, time = 10 sec, temperature = 300 K

Vector = number + proper - units + direction;

displacement = 5 m due North; Velocity = 5 m / sec due East

Other examples : acceleration, force, momentum, and angular momentum


Notation : A or 
A or A

Magnitude = number + proper - units ⟶ scalar


Notation : A  or A

Vectors in 3 - dimensions
Ax
   
A = Ax x + Ay y + Az z = (Ax , Ay , Az ) ⟺ A = Ay [Matrix representation]
Az

A1
   
A = A1 e1 + A2 e2 + A3 e3 = (A1 , A2 , A3 ) ⟺ A = A2
A3
Bx
   
B = Bx x + By y + Bz z = (Bx , By , Bz ) ⟺ B = By
Bz
B1
   
B = B1 e1 + B2 e2 + B3 e3 = (B1 , B2 , B3 ) ⟺ B = B2
B3

Einstein Summation Convention

    3  
A = A 1 e 1 + A2 e 2 + A3 e 3 = Σ A i e i = A i e i
i=1

[Einstein ' s summation convention : Summation over repeated indices]

2 Lecture_A.nb

 3  3  3    
A = Σ Ai e i = Σ Aj e j = Σ Ak e k ⇒ Ai e i = A j e j = A k e k
i=1 j=1 k=1

repeated indices i, j, k are dummy indices

 
A · B = Ai Bi = Aj Bj = Ak Bk repeated indices i, j, k are dummy indices

More than two repeated indices are not allowed according to this convention

How to combine two vectors?

Vector Addition
Vector addition is both commutative and associative
         
A + B = B + A [commutative] & A + B + C = A + B + C [associative]

Vector Multiplication
How to express the product of two vectors ?

A1 A1 B1 A1 B2 A1 B3
Ai Bj ⇔ A2 ( B1 B2 B3 ) = A2 B1 A2 B2 A2 B3 = ? or 3 × 3 = 9 =? 1 + 3 + 5
A3 A3 B1 A3 B2 A3 B3

Scalar/Dot Product

   
A · B = A B cosθ = B · A [ cos (- θ) = cosθ]
           
Example : x · x = 1 = y · y = z · z; x · y = 0 = y · z = z · x;
 
A · B = A x B x + Ay B y + Az B z
           
Example : e 1 · e1 = 1 = e2 · e2 = e3 · e3 ; e1 · e2 = 0 = e2 · e3 = e3 · e1 ;
 
In general, e i · ej = 1 if i = j otherwise zero
  3
A · B = A 1 B 1 + A2 B 2 + A3 B 3 = Σ A i B i = A i B i
i=1
[Einstein ' s summation convention : Summation over repeated indices]
 
A · B = Ai Bi = Aj Bj = Ak Bk repeated indices i, j, k are dummy indices

More than two repeated indices are not allowed according to this convention

B1
 
A · B = A i Bi ⟺ A T B = ( A1 A2 A3 ) B2
B3
   
Example : A · A = A2 = Ax Ax + Ay Ay + Az Az ⇒ A = A2x + A2y + A2z
 Lecture_A.nb 3

  
A+B = C ⇒ Ai + Bi = Ci or Aj + Bj = Cj i, j are free indices [which appear once]

Free indices are not summed over but they can take any value, 1, 2, 3

Kronecker Delta [ δ ]
A square matrix has equal # of rows and columns : M ij ith, jth row, column of matrix M

1 0 0
Identity matrix = I = 0 1 0 ⟹ I11 = I22 = I33 = 1 & Ii≠j = 0
0 0 1
1 0 0
Kronecker Delta = δ = 0 1 0 ⟹ δij =  1 for i = j
0 for i ≠ j
0 0 1
Tr[δ] = 3 or Tr[δ] = δ11 + δ22 + δ33 = δii

A1 1 0 0 A1
As A2 = 0 1 0 A2 ⟺ Ai = δij Aj
A3 0 0 1 A3

i = 1 ⇒ A1 = δ1 j Aj = δ11 A1 + δ12 A2 + δ13 A3 = A1

i = 2 ⇒ A2 = δ2 j Aj = δ21 A1 + δ22 A2 + δ23 A3 = A2

i = 3 ⇒ A3 = δ3 j Aj = δ31 A1 + δ32 A2 + δ33 A3 = A3

1 0 0 1 0 0 1 0 0
0 1 0 = 0 1 0 0 1 0 ⟺ δ = δδ ⟺ δij = δik δkj
0 0 1 0 0 1 0 0 1
Tr[δ] = Tr[δ δ] ⟺ δii = δik δki = δkk

1 0 0 1 0 0 1 0 0 1 0 0
0 1 0 = 0 1 0 0 1 0 0 1 0 ⟺ δ = δδδ ⟺ δij = δik δkm δmj
0 0 1 0 0 1 0 0 1 0 0 1
B1 1 0 0 B1
( A1 A2 A3 ) B2 = ( A1 A2 A3 ) 0 1 0 B2 ⟺ Ai Bi = Ai δij Bj
B3 0 0 1 B3
       
Note that e i · ej = δij ⇒ A · B = Ai ei · Bj ej = Ai ei · ej  Bj = Ai δij Bj = Ai Bi

A1 A1 B1 A1 B2 A1 B3
A2 ( B1 B2 B3 ) = A2 B1 A2 B2 A2 B3 ⟺ A i Bj
A3 A3 B1 A3 B2 A3 B3
4 Lecture_A.nb

Vector/Cross Product
    
A × B = A B sinθ n = - B × A [ sin (- θ) = - sinθ]

y x
+
z
              
Example : x × x = 0 = y × y = z × z; x × y = z; y × z = x; z × x = y;
              
Example : e 1 × e1 = 0 = e2 × e2 = e3 × e3 ; e1 × e2 = e3 ; e2 × e3 = e1 ; e3 × e1 = e2 ;

The area of a parallelogram as the magnitude of a cross product

    
The cross product A × B is defined as a vector C that is perpendicular orthogonal to both A and B,
with a direction given by the right - hand rule
and a magnitude equal to the area of the parallelogram that the vectors span.

The definition of a cross product can be used to define a vector area ?

 
a =da

Levi - Civita
    
A × B = (Ay Bz - Az By ) x + (Az Bx - Ax Bz ) y + (Ax By - Ay Bx ) z
 
A × B = (A2 B3 - A3 B2 , A3 B1 - A1 B3 , A1 B2 - A2 B1 )
     
A × B = A2 B3 - A3 B2 ; A × B = A3 B1 - A1 B3 ; A × B = A1 B2 - A2 B1 ;
1 2 3

 
A × B = ϵijk Aj Bk ; ϵijk = Levi - Civita Symbol = completely antisymmetric
i

 
A × B = ϵ1 jk Aj Bk = ϵ11 k A1 Bk + ϵ12 k A2 Bk + ϵ13 k A3 Bk
1

 
A × B = (ϵ111 A1 B1 + ϵ112 A1 B2 + ϵ113 A1 B3 ) +
1
(ϵ121 A2 B1 + ϵ122 A2 B2 + ϵ123 A2 B3 ) + (ϵ131 A3 B1 + ϵ132 A3 B2 + ϵ133 A3 B3 ) = A2 B3 - A3 B2

⇒ ϵ111 = 0 = ϵ112 = ϵ113 = ϵ121 = ϵ122 = ϵ131 = ϵ133 & ϵ123 = 1; ϵ132 = - 1

 
A × B = ϵijk Aj Bk
i

     
Example : A × B = Aj Bk ej × ek i = ϵijk Aj Bk ⇒ ej × ek i = ϵijk
i
 Lecture_A.nb 5

Another Method
     
A × B = A2 B3 - A3 B2 ; A × B = A3 B1 - A1 B3 ; A × B = A1 B2 - A2 B1 ; Eq .1
1 2 3

 
A × B = (Aj Bk - Ak Bj ) with i ≠ j ≠ k in cyclic order
i

But free indices i, j, k on both sides of the above expression do not match
 
Better representation is ϵijk A × B = (Aj Bk - Ak Bj ) ⇒ ϵijk = - ϵikj
i

Now we need to explore ϵijk further :

ϵijk = - ϵikj ⇒ ϵijk = 0 if j = k

     
ϵ1 jk A × B + ϵ2 jk A × B + ϵ3 jk A × B = 1 (Aj Bk - Ak Bj )
1 2 3 2
Comparing with Eq .1 we get

ϵ123 = 1; ϵ223 = 0; ϵ323 = 0;

ϵ131 = 0; ϵ231 = 1; ϵ331 = 0;
ϵ112 = 0; ϵ212 = 1; ϵ312 = 1;

⇒ ϵijk = 0 if i = j or j = k

ϵ123 = ϵ231 = ϵ312 = 1 ⇒ ϵ123 = - ϵ213 = - ϵ321 ⇒ ϵijk = - ϵjik = - ϵkji

Therefore, Levi - Citiva ϵijk is totally antisymmetric in its indices i, j, k

Exercise : 3 × 3 = 9 = 1 + 3 + 5
   
Given that Ai Bj = Cij A · B + Cijk A × B + Sij ; Find Cij = ? Cijk = ? Sij = ?
k

Scalar-triple Product
   
A = Ax x + Ay y + Az z = (A1 , A2 , A3 );
   
B = Bx x + By y + Bz z = (B1 , B2 , B3 );
   
C = Cx x + Cy y + Cz z = C1 , C2 , C3 ;

A1 A2 A3
  
A · B × C = B1 B2 B3
C1 C2 C3
        
A · B × C = 0 if either A = B or A = C or B = C

    
A · B × C = Ai B × C = Ai ϵijk Bj Ck = ϵijk Ai Bj Ck
i

        
A · B × C = B · C × A = C · A × B

 A
B

+
C
6 Lecture_A.nb

The scalar triple product gives the volume of a parallelepiped,

where the three vectors represent the adjacent sides of the parallelepiped

  
A · B × C = 0 if all three vectors are coplanar

        
Example : A · B × C = Ai ei · Bj ej × Ck ek  = ei · ej × ek  Ai Bj Ck = ϵijk Ai Bj Ck
  
⟹ ϵijk = ei · ej × ek 

Vector-triple Product
           
A × B × C =? a B · C A + b A · C B + c A · B C

     
Since A × B × C is perpendicular to both A & B × C ⟹ a = 0

         
Or assume A = B × C and take the dot product with B × C ⇒ a B · C A · B × C = 0 ⇒ a = 0

        
A × B × C = b A · C B + c A · B C

   
As B × C = - C × B ⟹ c = - b
        
A × B × C = bA · C B - A · B C

     
Let ' s take, A = y; B = z; C = y
            
y × z × y = y × - x = z = by · y z - y · z y = b z ⟹ b = 1

        
A × B × C = B A · C - C A · B [BAC - CAB rule]

        
A × B × Ci = B A · C - C A · Bi

     
ϵijk Aj B × C = Bi A · C - Ci A · B
k

ϵijk Aj ϵkmn Bm Cn = Bi Al Cl - Ci Al Bl

[ϵijk ϵkmn ] Aj Bm Cn = (δim Bm ) (δlj Aj ) δln Cn  - δin Cn  (δlj Aj ) (δlm Bm )

[ϵijk ϵkmn ] Aj Bm Cn = [δim δlj δln - δin δlj δlm ] Aj Bm Cn

δlj δln = δjn ; δlj δlm = δjm ; [I I = I]

ϵijk ϵkmn = δim δjn - δin δjm

 Lecture_A.nb 7

ϵijk ϵkjn = δij δjn - δin δjj ;

δjj = δ11 + δ22 + δ33 = 1 + 1 + 1 = 3 [Tr[I] = δjj = 3]

ϵijk ϵkjn = δin - 3 δin = - 2 δin

ϵijk ϵkjn = - 2 δin

ϵijk ϵkji = - 2 δii = - 2 3 = - 6

ϵijk ϵijk = 3 !

Consider two square matrices A & B

[C = A * B]ij ⟹ [C]ij = [A * B]ij = [A]ik [B]kj

δlj δln = δjn ; [I I = I]jn ⟹ [I I]jn = [I]jl [I]ln = [I]jn

[I I]jn = [I]jl [I]ln = [I]jn ⟹ δjl δln = δjn

ϵijk ϵlmn = δil δjm δkn + ...

six - permutations, [l m n, l n m, m l n, m n l, n l m, n m l ]

Understand, I vs Tr[I] ⟺ δij vs δii

 
Exercise : Show that ϵijk r × pk = ri pj - rj pi = r[i pj]

Jacobi' s identity for vector products

Excercise : Show that
        
A × B × C + B × C × A + C × A × B = 0

A[i B[j Ck]] + C[i A[j Bk]] + B[i C[j Ak]] =? 0

Rotation Matrix
What is a vector ?

Consider the position vector :

       
r = x x + y y + z z = x ' x ' + y ' y ' + z ' z ' = r ' [under passive rotation]

Anti - clockwise passive rotation about the z - axis :

8 Lecture_A.nb

x' cosφ sinφ 0 x

y' = - sinφ cosφ 0 y
z' 0 0 1 z
Roation about the y - axis :

x' cosφ 0 - sinφ x

y' = 0 1 0 y
z' sinφ 0 cosφ z
Rotation about the x - axis :

x' 1 0 0 x
y' = 0 cosφ sinφ y
z' 0 - sinφ cosφ z

This is also true for an arbitrary vector, A = (Ax , Ay , Az )

Rotation about the z - axis :

Ax ' cosφ sinφ 0 Ax

Ay ' = - sinφ cosφ 0 Ay
Az ' 0 0 1 Az

In matrix form : A' = R A

In component form : Ai ' = Rij Aj

cosφ - sinφ 0
 Ax ' A y ' A z '  =  A x Ay Az  sinφ cosφ 0
0 0 1
Ax ' cosφ - sinφ 0 cosφ sinφ 0 Ax
 Ax ' A y ' A z '  Ay ' =  A x Ay Az  sinφ cosφ 0 - sinφ cosφ 0 Ay
Az ' 0 0 1 0 0 1 Az

A2 = Ax '2 + Ay '2 + Az '2 = (Ax )2 + (Ay )2 + (Az )2

A 'T A ' = AT RT R A = AT A as RT R = I = R RT

RT R = Iij ⟹ RT Rij = [RT ]ik [R]kj = δij ⟹ Rki Rkj = δij = Rik Rjk

Considering, R T R = I, to be more general, we see that

DetRT R = I ⟹ DetRT R = Det[RT ] Det[R] = Det[I] = 1

Det[R]2 = 1 ⟹ Det[R] = ± 1 ⟹ Det[R] = + 1 ⟶ proper - rotation

 Lecture_A.nb 9

cosφ sinφ 0
Det - sinφ cosφ 0  = 1
0 0 1
cosφ sinφ 0
Rr = - sinφ cosφ 0 ; [proper - rotation]
0 0 1
x' -1 0 0 x
y' = 0 -1 0 y [space - inversion [improper rotation] or Parity]
z' 0 0 -1 z
-1 0 0
RP = 0 -1 0 ; Det[RP ] = - 1
0 0 -1
-1 0 0
Or RPx = 0 1 0 ; Det[RPx ] = - 1
0 0 1
RT R = I = R RT [Orthogonality Condition]
   
Polar vector : r , p, F, E
   
Axial  Pseudovector; L = r × p, B

 d          
F = p; Fm = q v × B; Fe = q E; dW = F.d r; τ = r × F;
dt
 
Scalar : r · r, W
  
Pseudoscalar : A .B × C

Pseudos : S ' = J S; Ai ' = J Rij Aj ; Tij ' = J Rim Rjn Tmn ; [ J = Det[R]]

Under Parity : proper tensor ⟶ - 1rank proper tensor

Under Parity : pseudo tensor ⟶ - - 1rank pseudo tensor [Jparity = - 1]

rank scalar = 0; rank vector = 1;

rank of a tensor = number of indices

Tensors
  
Tij ~ Ai Bj [tensor of rank 2] T = Tij ei ⊗ ej

Ai ' = Rim Am & Bj ' = Rjn Bn ;

Tij ' ~ Ai ' Bj ' = Rim Am Rjn Bn = Rim Rjn (Am Bn )

Tij ' = Rim Rjn Tmn or Tij ' = Rim Tmn RT nj

Tij ' = R T RT ij ⟹ T ' = R T RT

Isotropic tensors [T ' = R T RT = T]

If T = I; then I ' = R I RT = R RT = I ⟹ δij is an isotropic tensor

In component form : δij = Rim Rjn δmn

10 Lecture_A.nb

What about the Levi - Civita ϵijk ?

R11 R12 R13

Consider Det[R] = R21 R22 R23
R31 R32 R33

The interchange of any two rows or columns generates a minus sign, therefore we can write

Ri1 Ri2 Ri3 R1 l R1 m R1 n

ϵijk Det[R] = Rj1 Rj2 Rj3 or ϵlmn Det[R] = R2 l R2 m R2 n or
Rk1 Rk2 Rk3 R3 l R3 m R3 n

Ril Rim Rin

ϵijk ϵlmn Det[R] = Rjl Rjm Rjn [holds true for any arbitrary matrix R]
Rkl Rkm Rkn

If R = I then

δil δim δin

ϵijk ϵlmn = δjl δjm δjn
δkl δkm δkn

Excercise : Show that ϵijk = Det[R] Ril Rjm Rkn ϵlmn [isotropic pseudo - tensor]

ϵijk ' = J Ril Rjm Rkn ϵlmn = ϵijk

Tiso
ijkl ∝ (c1 δij δkl + c2 δik δjl + c3 δil δjk )

Tiso
ijkl = < Ai Bj Ck Dl >
∝ [c1 δij δkl + c2 δik δjl + c3 δil δjk ]
           
Find ci where it could be a combination of A · B C · D, A · C B · D & A · D C · B

    
Tiso
ijkmn = < Ai Bj Ck Dm En > ∝ A · B × C  D · E  ϵijk δmn + ...

Tiso
ijklmn = < Ai Bj Ck Dl Em Fn > ∝ [? ]

  
Excercise : Show that L = r × p is a pseudovector, i.e., Li ' = J Rij Lj
  
Excercise : Show that S = A · B × C is a pseudoscalar,
  
i.e., S ' = J S if A , B & C are all vectors.

Contraction
When we set two indices on a tensor equal and sum we say that we contract the two indices.

Active Rotation
Anti - clockwise rotation of a vector about the z - axis [active rotation] :

x' cosφ - sinφ 0 x

y' = sinφ cosφ 0 y
z' 0 0 1 z
Roation about the y - axis :
 Lecture_A.nb 11

x' cosφ 0 sinφ x

y' = 0 1 0 y
z' - sinφ 0 cosφ z
Rotation about the x - axis :

x' 1 0 0 x
y' = 0 cosφ - sinφ y
z' 0 sinφ cosφ z
0
 
Example : Consider a unit vector : r = z ⇔ 0
1
First rotate the vector counterclockwise by angle θ about the y -
axis then counterclockwise by φ about the z - axis
cosφ - sinφ 0 cosθ 0 sinθ 0 cosφ - sinφ 0 sinθ cosφsinθ
sinφ cosφ 0 0 1 0 0 = sinφ cosφ 0 0 = sinφsinθ
0 0 1 - sinθ 0 cosθ 1 0 0 1 cosθ cosθ
   
r = sinθcosφ x + sinθsinφ y + cosθ z [position unit vector in spherical coordinate]

[Active] Rotation about an arbitrary direction

A formula which transforms a given coordinate


system by rotating it through a angle Φ about an axis n .
         
r ' = ON + NV + VQ = n n · r + r - n n · r cosΦ + r × n sinΦ
      
r ' = r cosΦ + n n · r[1 - cosΦ] + r × n sinΦ
      
r ' = r cosΦ - n × r sinΦ + n n · r[1 - cosΦ]
[active clockwise - rotation]

Φ=φ [for passive anticlockwise - rotation]

xi ' = xi cosφ - ϵikj nk xj sinφ + ni (nj xj )[1 - cosφ] = Rij xj

Rij = δij cosφ + ϵijk nk sinφ + ni nj [1 - cosφ]

Excercise : show that Rik Rjk = δij for the above expression of Rij

xi ' = xi cosφ + ϵijk nj xk sinφ + ni (nj xj )[1 - cosφ] = Rij xj

 
Anti - clockwise rotation about the z - axis : n = z; ni = 0, 0, 1

Rij = δij cosφ + ϵijk nk sinφ + ni nj [1 - cosφ]

R11 = δ11 cosφ + ϵ11 k nk sinφ + n1 n1 [1 - cosφ] = cosφ

12 Lecture_A.nb

R22 = δ22 cosφ + ϵ22 k nk sinφ + n2 n2 [1 - cosφ] = cosφ

R33 = δ33 cosφ + ϵ33 k nk sinφ + n3 n3 [1 - cosφ] = cosφ + [1 - cosφ] = 1

R12 = δ12 cosφ + ϵ12 k nk sinφ + n1 n2 [1 - cosφ] = ϵ123 n3 sinφ = sinφ

R21 = δ21 cosφ + ϵ21 k nk sinφ + n2 n1 [1 - cosφ] = ϵ213 n3 sinφ = - sinφ

R13 = δ13 cosφ + ϵ13 k nk sinφ + n1 n3 [1 - cosφ] = ϵ132 n2 sinφ = 0

cosφ sinφ 0
Rz = - sinφ cosφ 0 [passive anti - clockwise rotation]
0 0 1

General Tensors
Consider the following general transformations :

x ' = x ' (x, y, z), y ' = y ' (x, y, z), z ' = z ' (x, y, z) or xi ' = xi ' (xk )

Using chain rule we can write

dx ' = ∂ x ' dx + ∂ x ' dy + ∂ x ' dz

∂x ∂y ∂z

dy ' = ∂ y ' dx + ∂ y ' dy + ∂ y ' dz

∂x ∂y ∂z

dz ' = ∂ z ' dx + ∂ z ' dy + ∂ z ' dz

∂x ∂y ∂z

Or in short, dxi ' = ∂ xi ' dxj

∂ xj

Now consider
∂ = ∂ x dx + ∂ y dy + ∂ z dz
∂x' ∂x' ∂x' ∂x'
∂ = ∂ x dx + ∂ y dy + ∂ z dz
∂y' ∂y' ∂y' ∂y'
∂ = ∂ x dx + ∂ y dy + ∂ z dz
∂z' ∂z' ∂z' ∂z'

Or in short, ∂ =  ∂ xk  dx
k
∂ xi ' ∂ xi '
   
Work done : dW = F · d r = Fi dxi ; F = - ∇ Φ [conservative force]

Let ' s assume Φ ' = Φ [a scalar]

Displacement Vector; dxi ' = ∂ xi ' dxj ; Ai ' = ∂ xi ' Aj [contravariant]
∂ xj ∂ xj

∂ xj
Force Vector; Fi ' = -  ∂ Φ  ' = -  ∂ Φ '  = -  ∂ Φ  = - ∂Φ
∂ xi ∂ xi ' ∂ xi ' ∂ xi ' ∂ xj

Force Vector; Fi ' = 

∂ xj
 Fj Ai ' = ∂ xj Aj [covariant]
∂ xi ' ∂ xi '
∂ xj ∂x
Work done : dW = Fi ' dxi ' =   Fj  ∂ xi '  dxk = Fj  j  dxk = Fj (δjk ) dxk
∂ xi ' ∂ xk ∂ xk
 Lecture_A.nb 13

Work done : dW = Fi ' dxi ' = Fi dxi

    
Note that r = xi ei = xi ωi such that e j · ωi = δij

∂ xi ' ∂ xj
In general, ≠ 
∂ xj ∂ xi '

∂ xj ∂ xj
If ∂ xi ' = [Cartesian Tensors] otherwise general tensors  ∂ xi ' ≠ 
∂ xj ∂ xi ' ∂ xj ∂ xi '

xi ' = Rik xk ⟹ ∂ xi ' = Rik ∂ xk = Rik δjk = Rij  ∂ xk = δjk 

∂ xj ∂ xj ∂ xj

x' = R x ⟹ x = RT x ' ⟹ xi = RT x 'i = [RT ]ik xk ' = Rki xk '

∂ xj
xj = Rkj xk ' ⇒ = Rij = ∂ xi '
∂ xi ' ∂ xj
∂ xj
A general relation between x ' & x that satifies ∂ xi ' = :
∂ xj ∂ xi '
∂ xj
xi ' = Rik xk + ai ⟹ ∂ xi ' = [where Rik Rjk = δij ]
∂ xj ∂ xi '
∂ xj
Note that an arbitrary linear transformation does not imply the condition ∂ xi ' =
∂ xj ∂ xi '

Exercise : For xi ' =

∂ xj
Aijk xj xk + Bij xj + Ci find the conditions on A, B, C if any such that ∂ xi ' =
∂ xj ∂ xi '
2
Exercise : Show that m d x2i = Fi is invariant under the Galileon transformations,
dt
xi ' = Rij xj + vi t + di & t ' = t + t0 where Fi = Σ fij (xi - xj )
j

Quotient Rule
"The quotient rule is a substitue for illegal division of tensors"

K*A = B ⟹ K = B [Algebraic Equation]

A
1 Ki Ai = B [if A is a vector and B is a pseudo - scalar]

2 Kkl Al = Bk [If both A and B are tensors]

3 Kij Ajk = Bik [If both A and B are tensors]

4) Kijkl Aij = Bkl [If both A and B are tensors]

5) Kij Ak = Bijk [A is an axial vector and B is a pseudo tensor]

where A & B are known tensors and K is an unknown quantity. Find the transformation properties of K
such that above equations hold in all (rotated ) Cartesian coordinate systems.
xi ' = Rik xk

Let ' s consider 2 :

Kkl Al = Bk ⟶ 2
14 Lecture_A.nb

Given that Bi ' = Rik Bk & Aj ' = Rjl Al [Al = Rjl Aj '] & Kij ' Aj ' = Bi '

Kij ' Aj ' = Bi ' = Rik Bk (2)

= Rik (Kkl Al ) = Rik Kkl Rjl Aj ' = Rik Kkl Rjl Aj '

Kij ' - Rik Kkl Rjl  Aj ' = 0

As Aj is an arbitrary vector,

Kij ' = Rik Rjl Kkl ⟹ K is a second rank tensor.

Number of independent elements/components of a tensor

T : 1; Ti : 3; Tij : 9; Tijk : 27; , ....

N N + 1
Symmetric tensor : Sij = Sji : in N - dimensions
2
[S12 = S21 ; S13 = S31 ; S23 = S32 ; S11 ; S22 ; S33 ]
3 3 + 1
= 6 # of elements
2
Sij ' = Rik Rjl Skl = Sji

N N - 1
Anti - symmetric tensor : Aij = - Aji : in N - dimensions
2
[A12 = - A21 ; A13 = - A31 ; A23 = - A32 ; A11 = 0; A22 = 0; A33 = 0]
3 3 - 1
= 3 # of elements
2
Aij ' = Rik Rjl Akl = - Aji ⟹ Aii = 0 [Traceless]

Tij : T ' = R T RT ; Tr[T '] = TrR T RT  = TrRT R T = Tr[I T] = Tr[T]

Tii : Trace [contraction of indices] is invariant under rotation

Irreducible decomposition of tensor products

Tij = 1 [Tij + Tji ] + 1 [Tij - Tji ] = Sij + Aij
2 2

Sij = 1 [Tij + Tji ] = Sji ; Aij = 1 [Tij - Tji ] = - Aji

2 2
Note that Sii = Tii = Tr[T] but Aii = 0

Tij = 1 Tr[T] δij + Sij + Aij

3

Sij = Sij - 1 Tr[T] δij = Trace - less symmetric tensor

3

Sii = Sii - 1 Tr[T] δii = Tr[T] - 1 Tr[T] 3 = Tr[T] - Tr[T] = 0

3 3

Tij = 1 Tr[T] δij + Sij + Aij

3
3 × 3 = 9 = 1 + 5 + 3 ≠ [1 + 2 + 2 + 2, 3 + 3 + 3, ....]
   
Ai Bj = 1 A · B δij + Sij + ϵijk A × B
3 k
 Lecture_A.nb 15

 
where 〈Ai Bj 〉 = 1 A · B δij = isotropic tensor
3
Exercise : For Ai Bj = C δij + ϵijk Ck + Cij such that Cij = Cji & Cii = 0, find {C, Ci , Cij }

⇒ Tij = Tr[T] δij

?
Exercise : Show that T ij = Ril Rjm Tlm = Tij
3
Exercise :
  
Ai Bj Ck = 1 A · B × C ϵijk + ...
6
  
where 〈Ai Bj Ck 〉 = 1 A · B × C ϵijk = isotropic tensor
6
3 × 3 × 3 = 27 = ?

⇒ Tijk = (ϵlmn Tlmn ) ϵijk

?
Exercise : T ijk = Ril Rjm Rkn Tlmn = Tijk
6

Addition of Spins
j 1 × j2 = j 1 + j2 , j 1 + j2 - 1 , , , j 1 - j2

J : 1 × 1 = 0 + 1; or 2j+1: 2×2 = 1+3

2 2
J : 1 × 1 = 0 + 1 + 2; or 2j+1: 3×3 = 1+3+5

J : 1 × 1 × 1 = 0 + 1 + 2 × 1 = 1 + 1 × 1 + 2 × 1 = 1 + 0 + 1 + 2 + 1 + 2 + 3;
2 j + 1 : 3 × 3 × 3 = 3 + 1 + 3 + 5 + 3 + 5 + 7

3 × 3 × 3 = 27 = 3 + 1 + 3 + 5 + 3 + 5 + 7

Exercise : How many singlets  scalars appear in the product of N - vectors ?

Exercise : How many triplets  vectors appear in the product of N - vectors ?

Irreducible tensors [under rotation in N-space]

T : 1; Ti : 3; Tij : 9 = 1 + 3 + 5; Tijk : 27 = 3 + 1 + 3 + 5 + 3 + 5 + 7; ....

Tij : 9 = 1 + 3 + 5 ⇒  1 Tll δij , T[ij] = 1 ϵijk Tk , , T(ij) - 1 Tii δij 

3 2! 3

where T(ij) ≡ 1 (Tij + Tji ) & T[ij] ≡ 1 (Tij - Tji )

2! 2!
Tijk : 27 = 3 + 1 + 3 + 5 + 3 + 5 + 7 ⇒
Till δjk + per,  1 Tlll ϵijk , Ttraceless trace traceless traceless
([ij] k) , ?, T(ijk) , T[(ij) k] , T(ijk) 
3!
Cij = ϵimn T[mn] j , now reduce Cij and then get T[ij] k = ?

T(ij) k = 1 [(T(ij) k + T(ik) j + T(kj) i ) + (T(ij) k - T(ik) j ) + (T(ij) k - T(kj) i )]

3

T(ij) k = T(ijk) + 1 [(T(ij) k - T(ik) j ) + (T(ij) k - T(kj) i )]

3
(T(jm) n - T(jn) m ) (T(ij) k - T(ik) j )
Bij = ϵimn  , now reduce Bij and then get   = ?
3 3
16 Lecture_A.nb

Ttraceless
(ijk) = T(ijk) - Ttrace
(ijk)

where T(ijk) ≡ 1 (Tijk + Tikj + Tjki + Tjik + Tkji + Tkji ) & Ttrace
(ijk) =
1 (T
(ill) δjk + T(ljl) δik + T(llk) δij )
3! 5
N N + 1
S(ij) : Sij = + Sji ⇒ # of independent elements
2!
Example : Inertia Tensor, Iij = Iij

Example : Energy Momentum Tensor, Tμν = Tνμ

Example : Metric Tensor, gμν = gνμ

Example : Einstein Tensor, Gμν = Gνμ

N N + 1 N + 2
S(ijk) : # of independent elements
3!
∂ gjk N N + 1
If gjk = gkj then : N× # of independent elements
∂ xi 2!

∂2 gij N N + 1 N N + 1
: × # of independent elements
∂ xm ∂ xn 2! 2!

∂3 gmn :
N N + 1 N N + 1 N + 3
× # of independent elements
∂ xi ∂ xj ∂ xk 2! 3!

Dual Tensors
N N - 1
A[ij] : A[ij] = - A[ji] ⇒ # of independent elements
2!

A[ij] ⟺ Ai = 1 ϵijk A[jk] [Ai & Ajk are dual to each other]
2!
N N - 1 N - 2
A[ijk] : # of independent elements
3!

A[ijk] ⟺ A = 1 ϵijk A[ijk] [A & Aijk are dual to each other]

3!
  4 4 - 1
E, B ⟺ Fμν = - Fνμ  = 6 = 3 + 3 in N = 1 + 3 dimension
2
Fμν = electromegnetic field tensor

Riemann Tensor
1 Rμνρσ = - Rνμρσ

2 Rμνρσ = - Rμνσρ

3 Rμνρσ = + Rρσμν

4) Rμνρσ + Rμρσν + Rμσνρ = 0

Excercise. Find the number of the independent components of Rμνρσ satisfying the above conditions
Excercise : Find the number of the independent
components of the rotation matrix Rij in N dimensions satisfying the conditions :
 Lecture_A.nb 17

Rik Rjk = δij