Unit 19 Continuous Groups

UNIT 19
CONTINUOUS
GROUPS
Structure
19.1 Introduction 19.4 Minkowski Space and the
Expected Learning Outcomes Lorentz Group
19.2 Continuous Groups: An Introduction Minkowski Space
Active and Passive Transformations The Lorentz Group
Simplest Continuous Groups †
19.5 SL(2,C) and the Lorentz Group L
Orthogonal Group O(3) †
Homomorphism SL(2,C)  L is 2:1
The Subgroup SO(3) or Rotation Group
Lie Groups or Continuous Groups Six Parameters of SL(2,C)
One Parameter Subgroups of SO(3) Boosts and Hermitian SL(2,C) Matrices
19.3 Generators of One-parameter Rotations and SU(2)
Subgroups Form of a General SL(2,C) Matrix
Commutator of Generators The Matrix  and Other Useful Formulas
Lie Algebra of Generators 19.6 Summary
19.7 Terminal Questions
19.8 Solutions and Answers

19.1 INTRODUCTION
You have so far discussed groups that are finite groups which contain a
number of discrete elements. There can be groups with infinitely many
elements, like the set of all integers with respect to addition. This group has
infinitely many elements, but the elements are denumerable.
Now think about the set of rotations through an angle, say , about a fixed
axis. The rotations form a group, but the group is labelled by the continuous
parameter . There are infinitely many elements in this group and they are not
denumerable either. This is an example of a continuous group.
In this unit we discuss continuous groups in physics which are related to
symmetry transformations. In Sec. 19.2 we introduce continuous groups and
discuss some simple continuous groups. In Sec. 19.3 we discuss about the
generators of one-parameter subgroups. In Sec. 19.4 we will talk about
Minkowski space and the Lorentz group. In Sec. 19.5 we discuss the group of
2  2 complex matrices SL(2,C) and its connection to the Lorentz group. 69
Block 5 Group Theory
Expected Learning Outcomes
After studying this unit, you should be able to:
 define and identify active and passive transformations;

 define continuous groups and give examples;

 determine the generators of one parameter subgroups;

 solve problems based on continuous groups and their generators;

 define Minkowski space and Lorentz group;

 solve problems related to SL (2,C); and

 define the matrix  and solve problems related to it.

19.2 CONTINUOUS GROUPS: AN

INTRODUCTION
In this section we will introduce you to continuous groups, which in physics are
transformation groups. So, first we explain transformations – active and
passive.

19.2.1 Active and Passive Transformations

All concepts of symmetry, and symmetry groups, of physics deal with
transformations. It is important to remember the distinction between active
and passive transformations.

When we think of a rigid body being rotated about an axis by a certain angle, it
is an example of an active transformation. The position coordinates of various
points of the body change after the rotation. This change of the position
coordinates is an active transformation of rotation.

But we can think of rotation in another way.

We can keep the body fixed, but we move the coordinate system by a rotation.
The coordinates of the points of the body in the old coordinate system are
transformed to the coordinates of the same points in the new coordinate
system. This is a passive transformation, where the rigid body remains
stationary and it is the coordinate system describing it which rotates.

In the active case, there is one coordinate system, and infinitely many different
rotations can be imparted to the body.

In the passive case, there are an infinite number of coordinate systems related
by rotations.

Mathematically, both are symmetries and they are equivalent, but the
parameters have opposite signs.

If a body is actively rotated by an angle /4 about the z-axis, a point of the
body on the xy-plane with coordinates (1, 0, 0) will acquire coordinates

70
 
1/ 2,1/ 2, 0 (see Fig. 19.1a).
Unit 19 Continuous Groups
On the other hand if the rigid body is kept fixed, and the coordinate system is
rotated by the same amount, the same point will have coordinates
 
1/ 2,  1/ 2, 0 in the new coordinate system (Fig. 19.1b).

y y

y

x

 x  1/ 2, y   1/ 2 

x  1/ 2, y   1/ 2 
(x = 1, y = 0) (x = 1, y = 0)
x 
Passive
x
Active

(a) (b)

Fig. 19.1: a) Active transformation; b) passive transformation.

Passive and active transformations are both used frequently in physics.

19.2.2 Simplest Continuous Groups

Let us now define continuous groups. Continuous groups or Lie groups are
groups such that each element of the group is specified by one or several (but
finite number of) real parameters. Such groups necessarily have an infinite
number of elements. The group law is the rule which tells us how to calculate
the parameters of the product element given two sets of parameters specifying
the two elements.

Given below are some of the simplest examples of continuous groups.

1. Let  be the set of all real numbers. Then x   specifies the group
element. The group law is the ordinary addition of real numbers. The
identity element is the number 0 and the inverse of x is  x.

2. Let   be the subset of non-zero positive real numbers. The group law is
multiplication of real numbers. The identity is the number 1, and the
inverse of x    is 1/x.

3. The group U(1): Let U(1) be the set of complex numbers of modulus unity,
that is, complex numbers z with z  1 . The group elements can be
pictured as belonging to the unit circle in the complex plane. The product
of two elements of U(1) is also of modulus unity z1z 2  z1 z 2  1 . The
71
Block 5 Group Theory
identity is the number 1 and the inverse of z is z* = 1/z. The group is so
named because complex numbers of U(1) can be considered as 1  1
unitary matrices.

19.2.3 Orthogonal Group O(3)

As a simple and important example, let us consider rotations in three-
dimensional space. Rotations have the property that (1) there is one special
point which remains fixed, and (2) distances between any pair of points of the
rotated body remain unchanged.

We can take the fixed point, O, as the origin of Cartesian coordinates. Let a
general point P of the body have the coordinates ( x1, x 2 , x 3 ) before the
rotation, and ( x1 , x 2 , x 3 ) after the rotation. Then

( x1 )2  ( x 2 )2  ( x 3 )2  ( x1)2  ( x 2 )2  ( x 3 )2

We can write it as a matrix equation:

 x1   x 1 
   
If X   x 2  , X    x 2 
   
x 3   x 3 
   

then

X T X   X T X

If the relationship between the coordinates X and X is linear, then we can

write
X = RX,

where R is a 3  3 real matrix. Written fully,

 x 1  R11 R12 R13   x1 

     
 2    2
 x   R 21 R 22 R 23  x 
     
 x 3  R 31 R 32 R 33  x 3 
   

The condition X T X   X T X implies (see Sec. 8.4 in Unit 8).

RT R  1 or RT  R 1 therefore RR T  1 as well.

Thus, R is a 3  3 real orthogonal matrix.

Taking the determinant, we see that

RT R  RT R   R 2  1, or R  1

We call O(3) the set of all 3  3 real orthogonal matrices. They form a group
because the product of two such matrices is again one such matrix:

72 (R1R 2 )T (R1R 2 )  R T RT R R  RT
2 1 1 2
R 1
2 2
Unit 19 Continuous Groups
The identity matrix trivially satisfies orthogonality condition, and the inverse
R 1  R T is also in the set because

(R 1)T R 1  (RT )T RT  RRT  1

The space inversion (or parity) matrix which reverses the sign of each
coordinate

 1 0 0
 
I s   0 1 0   1
 
 0 0  1

belongs to O(3).

The space inversion Is is actually not a rotation, because it cannot be

physically achieved by rotating a body. The actual rotation group is a
subgroup of O(3) which we discuss next.

19.2.4 The Subgroup SO(3) or Rotation Group

A general element of O(3) can have a determinant either 1 or 1. The

product of two matrices with determinant 1 is again a matrix with determinant
1. Therefore, the subset of O(3) which has matrices with determinant 1 is a
subgroup denoted by SO(3) called the special orthogonal group.

Thus, the group O(3) has two cosets, the subgroup SO(3) and the set of
matrices with determinant –1. Since the space inversion matrix has
determinant equal to –1, any O(3) matrix T with determinant equal to –1 can
be written as:

T = Is R

Note that if T  O(3) then both T and T 1 have the same determinant
because T 1  T T .

SAQ 1

Is SO(3) an invariant subgroup of O(3)? If yes, find the factor group.

19.2.5 Lie Groups or Continuous Groups

O(3) and SO(3) are more elaborate examples of continuous or Lie groups.
As mentioned above these groups have a continuous infinity of group
elements. Each group element is labeled by a number of real parameters.

In our case O(3) needs three real parameters to specify a group element. A
3  3 matrix R has 9 elements. The orthogonality condition R T R  1 is
equivalent to six conditions, ( R T R is a symmetric matrix with six elements
actually independent: three elements on the diagonal and three elements on
one side of the diagonal. Elements on one side of the diagonal are equal to 73
Block 5 Group Theory
corresponding elements on the other side in a symmetric matrix.) Therefore,
there are only three free parameters.

The three independent parameters can be chosen in infinitely many ways.

19.2.6 One Parameter Subgroups of SO(3)

A rotation in which not just a single point, but a whole line remains fixed
is called rotation about an axis.

These are especially simple rotations because if the axis is fixed, a single
parameter, the angle of rotation is enough to specify the rotation completely.

It was proved by Euler (in 1776) that every rotation can be seen as a rotation
about an axis by a definite angle.

What it means is this: although one can arrive at the same final position of a
rigid body in infinitely many possible ways through all kinds of rotations, the
final position can always be obtained from the initial position by a single
rotation about an axis by a certain angle.

P

Q

Fig. 19.2: Rotation about an axis.

Refer to Fig. 19.2. Let the fixed point be O and P and Q two points on the
body. Let the final positions of the two points be P and Q.

The axis can be found as follows: draw a plane perpendicular to POP and
bisecting the angle POP. Similarly, draw a plane perpendicular to QOQ and
bisecting the angle QOQ. If these planes are not coincident, then the
intersecting line of these planes is the axis. If the two planes are coincident,
then the axis is the intersecting line of planes POQ and POQ.

Rotations about a fixed axis can be labeled by giving the direction of the axis,
which requires two parameters, and the angle by which the rotation is made,
which is a number lying between 0 and 2 when the angle is measured in
radians. At angle 2 the rotation becomes equal to identity. One should avoid
the identity element to be given by two different parameters, therefore rotation
by 2 is equated to identity.
74
Unit 19 Continuous Groups
If we fix the axis, then there is only one parameter to specify the rotation.
These rotations about a fixed axis define a subgroup because two successive
rotations by angles 1 and  2 give a rotation by angle 1 +  2 . It is possible
that 1 +  2 exceeds 2. In that case the added angle is the excess over 2.

This one parameter group of rotations with a fixed axis is called the group
SO(2) as rotations in the two dimensional plane perpendicular to the axis. It is
a subgroup of SO(3).

SAQ 2

Find the matrices corresponding to rotations about the z-axis and show that
they are in one-one correspondence with orthogonal 2  2 matrices with
determinant 1.

19.3 GENERATORS OF ONE-PARAMETER

SUBGROUPS
Since rotations about a fixed axis add up, it is possible to build finite rotations
by successive application of a very large number of very small rotations. This
suggests that we should look at the structure of infinitesimal rotations.

An active rotation by an angle  about the 3-axis (in a right-handed Cartesian

coordinate system) is the SO(3) matrix:

cos   sin  0 1   2 / 2!...     3 / 3!... 0

   
 
R3 ()   sin  cos  0     3 / 3!... 1   2 / 2!... 0  ...
   
 0 0 1  0 0 1

where  > 0 means that the rotation is in the right-handed screw sense (that is
anti-clockwise in the plane parallel to 1-2 plane).

For infinitesimal  this matrix is:

0 1 0 1 0 0
   
2 
lim R 3 ()  1    1 0 
0  0 1 0
0 2! 
   
0 0 0 0 0 0

2 2
 1   J3  J  ...
2! 3

The matrix J 3 is:

0 1 0
 
J 3   1 0 0
 
0 0 0 75
Block 5 Group Theory
We can see that

J 32  1, J 33  J 3 , J 34  1, ...

therefore even for finite ,

2 2
R 3 ()  exp( J 3 )  1  J 3  J  ...
2 3

J 3 is called the generator of rotations about the 3-axis because all finite
angle rotations about 3-axis can be built out of it.

Similarly, an infinitesimal rotation about the 1-axis,

1 0 0  1 0 0 
   
R1()  0 cos   sin   0 1   2 / 2!...     / 3!...  ...
3

   
0 sin  cos   0    3 / 3!... 1   2 / 2!... 

can be written as:

2 2
R1()  1  J1  J  ...  exp (J1 )
2 1

with

0 0 0
 
J1  0 0  1
 
0 1 0 

And an infinitesimal rotation about the 2-axis,

 cos  0 sin  
 
R 2 ()   0 1 0 
 
 sin  0 cos 

gives, in a similar manner,

2 2
R 2 ()  1  J 2  J  ...  exp (J 2 )
2 2

with

0 0 1
 
J 2   0 0 0
 
 1 0 0
76
Unit 19 Continuous Groups

19.3.1 Commutator of Generators

We have discussed the generators of rotations about the three coordinate

axes. There will be a similar generator about every axis.

Rotations about axes in different directions do not commute.

We study the simple case of two infinitesimal rotations: R1  R1() about

coordinate axis 1, and R2  R2 () about axis 2,  and  being small.

Let us see how much mismatch is there when these two infinitesimal rotations
are applied one after the other in two different ways.

If R1, R2   R1R2  R2R1 were zero then the result of R1R2 and R2R1 would
have been the same acting on any vector X.

But, in general

R1R2 X  R2R1 X or, multiplying by inverses, R 21R11R 2 R1  1.

The failure of commuting can therefore be estimated by calculating

R 21R11R 2 R1  R 2 ( )R1( )R 2 ()R1()

Keeping quantities up to second order,

R 2 ()R1()  (1  J 2   2 J 22 / 2  ...) (1  J1   2 J12 / 2  ...)

 1  J 2  J1  J 2 J1   2 J 22 / 2   2 J12 / 2  ...

Similarly, changing signs of  and 

R 2 ( )R1( )  1  J 2  J1  J 2 J1   2 J 22 / 2   2 J12 / 2  ...

Therefore,

R2 ()R1()R2 ()R1()  (1  J 2  J1  J 2 J1

  2 J 22 / 2   2 J12 / 2  ...)

 (1  J 2  J1  J 2 J1   2 J 22 / 2   2 J12 / 2  ...)

Keeping again up to second order, we see that

R2 ()R1() R2 ()R1()  1  (J 2 J1  J1J 2 )  ...

This shows that the effect of applying infinitesimal rotation by  about axis 1,
followed by an infinitesimal rotation by  about axis 2, and then an inverse
infinitesimal rotation about axis 1 by –  followed by an infinitesimal rotation
about axis 2 by angle –  is equivalent to an infinitesimal rotation by an angle
 about an axis whose generator is
77
Block 5 Group Theory
0 0 1 0 0 0  0 0 0 0 0 1
      
J2J1  J1J2   0 0 0 0 0  1  0 0  1. 0
 0 0
      
 1 0 0 0 1 0  0 1 0   1 0 0

 0 1 0
 
  1 0 0
 
 0 0 0

 J 3

Thus, the resulting rotation axis is the negative 3-axis!

We have taken rotations about the coordinate axes, but one could take any
arbitrary directions and obtain similar relations between generators.

The relation above shows that the commutator of the generators of rotations
in 1- and 2-directions is the generator in the 3-direction:

J1, J 2   J1, J 2  J 2 , J1  J 3

We can show similarly that

J 2 , J 3   J1 and J 3 , J1  J 2

19.3.2 Lie Algebra of Generators

We calculated the generators of rotations by looking at the infinitesimal

rotations about the coordinate axes. We now discuss a general infinitesimal
rotation.

A general rotation that is a member of SO(3) is an orthogonal 3  3 matrix R

with determinant equal to one. The identity matrix corresponds to no rotation.
Therefore, an infinitesimal rotation is a matrix infinitesimally different from the
identity matrix. We write

R=1+M

where elements of M are all small quantities.

The condition of orthogonality requires that

1  RT R  (1  M )T (1  M )

 1  M T  M  second order of smallness

Therefore, in the limit of infinitesimal rotation,

M T  M  0, or M T  M.

The infinitesimal rotation matrix differs from the identity matrix by an

antisymmetric matrix.
78
Unit 19 Continuous Groups
We have already seen that the generators J1, J 2 , J 3 in the last section are
antisymmetric matrices.

There are as many generators as the number of possible rotation axes, which
are infinite in number. These generators are all possible antisymmetric
matrices.

But, the set of all generators, that is, the set of all anti-symmetric 3  3 real
matrices forms a real vector space (of dimension 3 as we see below) because
the sum of two antisymmetric matrices is also antisymmetric and if we multiply
an antisymmetric matrix by a real number it remains antisymmetric.

Moreover, the commutator of two antisymmetric matrices is also

antisymmetric. Let M and N be two antisymmetric matrices, then, using
MT  M and NT  N , we have

[M, N]T  (MN  NM)T  NT MT  MT NT  NM  MN  [M, N]

What is the dimension of this space?

A general 3  3 real matrix has 9 independent elements. If it is antisymmetric,

the diagonal elements are zero, and the remaining 6 elements are pairwise
negative to each other. Thus there are only three independent matrices; all
others can be written as linear combination of these three.

0 a b
 
M    a 0 c 
 
 b c 0 

0 1 0 0 0 1 0 0 0
     
 a  1 0 0  b  0
 0 0  c 0
 0  1
     
0 0 0  1 0 0 0 1 0 

 cJ1  bJ 2  aJ3

Therefore every generator of SO(3) is a member of the vector space of 3  3

antisymmetric matrices. Moreover this vector space has the commutator of
two elements again belonging to the same space. The commutator acts like a
bilinear product,
[M, N  P ]  [M, N ]  [M, P ], [aM , N ]  a[M, N ],

[M  N, P ]  [M, P ]  [N, P ], [M, aN ]  a[M, N ]

In addition, it is antisymmetric

[M, N] =  [N, M]

and satisfies the Jacobi identity:

[L, [M, N]] + [M, N], L]] + [N, L], M]] = 0 79

Block 5 Group Theory
Definition

Any vector space which has a bilinear relation defined on it is an algebra. A

vector space (like the one above) with an antisymmetric bilinear product
satisfying Jacobi identity is called a Lie algebra.

The bilinear ‘product’ here is the commutator of two matrices. Most

products encountered in physics are associative products, for example,
the matrix multiplication, (AB)C = A(BC). But the Jacobi identity shows
that the product in a Lie algebra is non-associative: [[M, N], P]  [M, [N, P]]
in general. The only other place a non-associative product is
encountered in elementary physics is the Poisson bracket in classical
mechanics. There too, there is a Lie algebra of phase space functions.

Since every element of the Lie algebra can be written as a linear combination
of its basis elements, it is enough to specify the commutator only for the basis
elements.

In a Lie algebra of dimension n with basis E1,E2,..., En we can write [E i , E j ]

which again belongs to the vector space of the Lie algebra as a linear
combination of the basis vectors:

[E i , E j ]   c ij k E k ,
k

where the n 2 (n  1) / 2 real numbers c ij k  c ji k are called the structure

constants of the Lie algebra.

SAQ 3

Why are there n 2 (n  1) / 2 structure constants for a Lie algebra of dimension

n?

Important note on notation:

The Lie algebra of SO(3) is a real algebra with real structure constants.

But when the rotation group is used as a symmetry in Quantum Mechanics,

the rotations are represented by unitary operators, and infinitesimal rotations
can be written as

Uˆ (R)  1  Tˆ  ...

where Tˆ is an infinitesimal operator. The condition for unitarity means

1  U(R )† U(R )  1  Tˆ †  Tˆ  ...

Therefore the generators Tˆ of rotation in Quantum Mechanics are

anti-hermitian operators (Tˆ †  Tˆ ).

The commutator of two anti-hermitian operators is again anti-hermitian, so the

Lie algebra is a real Lie algebra, with real structure constants as in the Lie
algebra of rotation matrices.
80
Unit 19 Continuous Groups
However, the generators of all symmetry transformations are
proportional to physical quantities or observables, which are
represented by hermitian operators.

It is a time honoured practice to redefine generators with a factor proportional

to imaginary unit and convert the anti-hermitian operator into a hermitian
operator.

Thus, usually the operator version of the generators are defined as:

U (1  J1  ...)  1  iJˆ1  ...

U (1  J 2  ...)  1  iJˆ 2  ...

U (1  J 3  ...)  1  iJˆ3  ...

with the result that

[Jˆ1, Jˆ2 ]  iJˆ3 , and so on.

This may give the impression that the structure constants are imaginary, but
that is not so. As is well known, the generators Ĵ are called components of
angular momentum.

We have given the generators starting from the active rotation matrices. In
many books, as well as in this unit, passive transformations are used as well.
There the relations given above will appear as:

U(1  J1  ...)  1  iJˆ1  ...

and so on. You must be careful about these signs. Of course, just as bases
can be chosen in infinitely many ways, the generators too can be chosen in
every possible way. But the basis with [J1, J 2 ]  iJ3 , and its cyclic versions,
are universally accepted norm for angular momentum.

19.4 MINKOWSKI SPACE AND THE LORENTZ

GROUP
We now discuss Minkowski space and the Lorentz group. For this, let us first
revise Lorentz transformation. A word about notation.
Notation: We use units such that time is measured in metres with the
help of the universal constant c which represents velocity of light in
vacuum. This, effectively makes c = 1 in relativity formulas, and
velocities become dimensionless numbers.

Recall the familiar Lorentz transformation from your UG courses:

 t    1 1  v 2 v 1 v 2 0 0  t 
   
     
 x   v 1 v 2 1 1 v 2 0 0 x
    
 y    y 
   0 0 1 0  
     
 z   z
   0 0 0 1    81
Block 5 Group Theory
It relates the same event P which is specified by (t, x, y, z) and (t , x , y , z ) ,
respectively in the two frames of reference S and S (which means we are
using passive transformation). The two frames are such that

1. their axes are parallel,

2. the x and x-axes are collinear,
3. S moves with velocity v with respect to S along the x-axis,
4. the clocks are set so that when the origins of the two frames coincide,
t  0  t .

This is written as the matrix equation:

x = Lx

where x is the single column matrix

x0   t 
   
   
 x1   x 
  
 2  y 
x   
   
x3   z 
   

and similarly for x. The matrix L is a 4  4 matrix.

The example given above is a special Lorentz transformation called a “boost”

in the 1-direction.

Definition

A Lorentz transformation is a real 4  4 matrix L which transforms x into

x = Lx such that

 ( x 0 )2  ( x 1 )2  ( x 2 )2  ( x 3 ) 2  ( x  0 )2  ( x 1 ) 2  ( x  2 ) 2  ( x  3 ) 2

This condition can be written more compactly using a matrix,

1 
 
 
 1 
 
 1 
 
 
 1


as

x T x  x T x 

Therefore, if x   Lx then the condition for L to be a Lorentz matrix is:

x T x   xT LT Lx  xT x
82
Unit 19 Continuous Groups
Since this is true for any arbitrary x, it follows that

LT L  

Thus, any real 4  4 matrix that satisfies the above equation is a Lorentz
transformation.

In the next two sub-sections we discuss Minkowski space and Lorentz group.

19.4.1 Minkowski Space

We can think of a 4-dimensional vector space (in analogy to the 3-dimensional

space) where ‘position vectors’ are:

 x0 
 
 
 x1 
x  
 2
x 
 
 x3 
 

and where the inner product or ‘dot product’ between two vectors x and y is
given by

x, y  x . y  x T y

This four-dimensional space is called the Minkowski space.

Just as in three-dimensions, rotations do not change the dot product of two

vectors, similarly, the Lorentz transformations do not change the above dot
product.

But unlike the three-dimensional space, the dot product of a vector with itself
is not necessarily positive. It can be positive, negative or even zero. We still
call

x . y  xT x  ( x 0 )2  ( x1)2  ( x 2 )2  ( x 3 )2

as the “length squared” of a vector.

The Minkowski space gets divided into three regions with respect to the origin.

It is conventional to imagine the x 0 -axis or the time axis as a vertical line and
the plane perpendicular to it carrying the spatial axes corresponding to
x 1, x 2 , x 3 .

The set of vectors corresponding to x . x = 0 are called light-like or null

vectors.

Null vectors form a double cone. This is the surface in Minkowski space
satisfying

( x 0 )2  ( x 1)2  ( x 2 )2  ( x 3 )2
83
Block 5 Group Theory
or

( x 0 )   ( x 1)2  ( x 2 )2  ( x 3 )2

with the vertex at the origin (0,0,0,0)T and vertex angle of 45. This cone is
called the light cone. The time axis is the axis of the cone. The positive side
of the cone (that is, the half-cone through which the positive half of the time
axis passes) is called the forward light cone and the negative side of the
cone the backward light cone.

All vectors with x.x  0 lie within the light cone and they are called time-like
vectors.

The remaining vectors with x.x  0 lie in the wedge-like region and they are
called space-like vectors.

19.4.2 The Lorentz Group

Let L be the set of all 4  4 Lorentz transformation matrices. These form a
group, called the Lorentz group with matrix multiplication as the group law.
The inverse of each matrix exists because det L  0. We know that the
determinant of any Lorentz matrix can be either +1 or 1. This follows from
LT L   and taking determinant on both sides to obtain (det L)2  1.

You can check that

L1  LT 

SAQ 4

a) Prove that

 L00  L10  L20  L30 

 
 
  L01 L11 L21 L31 
L1  LT    ,
 L L12 L22 L32 
 02 
 
 L L33 
 03 L13 L23

so that the inverse Lorentz matrix is just the transposed matrix with 0i and
i0 elements occurring with a minus sign.

b) Show that if L00  1 then not only

L00   L10
2  L2  L2
20 30

(which follows from the definition) but also

L00   L201  L202  L203

84
Unit 19 Continuous Groups

19.5 SL(2,C) AND THE LORENTZ GROUP L

The proper orthochronous Lorentz group L has a deep relation with the
group SL(2, C) of 2  2 complex matrices.

SL(2, C) stands for Special Linear Group of 2  2 complex matrices. The

only special thing about elements of SL(2, C) is that their determinant is equal
to one. The fundamental fact about these matrices is this:

For every A  SL(2, C) there is a proper orthochronous Lorentz

transformation L(A)  L such that

L( A1 ) L( A2 )  L( A1 A2 )

To see this, let us consider the set of all 2  2 hermitian matrices.

Let us show that all 2  2 hermitian matrices form a four-dimensional real

vector space.

The sum of two hermitian matrices is hermitian, as well as multiplication by a

real number.

A 2  2 hermitian matrix has the general form:

 1  2  i 3 
 
 
  2  i 3 4 

where the ’s are four real numbers.

We can choose the following four matrices as a basis in this vector space.
These four standard hermitian matrices are called Pauli matrices:

1 0 0 1
0    1   
   
0 1 1 0

0 i 1 0
2    3   
   
i 0 0  1

Any four linearly independent hermitian matrices would have been equally
good, but these are the ones traditionally used and they have the properties
listed below:

i) The squares of all the Pauli matrices are equal to the identity

02  1, 12  1, 22  1, 32  1

ii) Matrices i  1, 2, 3 have trace zero, and

iii) 1 2  2 1  i 3 and other similar relations are obtained by rotating

1,2,3 cyclically. 85
Block 5 Group Theory
These properties imply that

Tr ( )  2

where  is the Kronecker delta, equal to zero when    and 1 when the
two indices are the same.

Caution about notation: Although the placing of subscripts on the

constant Pauli matrices is like that of a covariant vector, it is not a
covariant 4-vector. We will always use Pauli matrices with a lower index.

With a given space-time point with coordinates x   we form a hermitian

matrix,

3  x0  x3 x 1  ix 2 
 
X   x   
 x 1  ix 2 
 0  x 0  x 3 

whose determinant is

det X  x 0   x 1   x 2   x 3    x . x.
2 2 2 2

Let A  SL(2, C) be an arbitrary 2  2 complex matrix with determinant 1 and

X as above. We define

Y  AX A†

where A† is the hermitian conjugate of A. The matrix Y is also hermitian and,

therefore, can be expanded in the basis of   matrices:

3
Y   y  
 0

Moreover,

y . y  y T y   detY   det X  xT x  x . x

because the determinant of A as well as A† is equal to one.

What is the relation between coordinates y  and x  ? Certainly y depends

linearly on the x. Therefore, these coordinates are linearly related. We write
3
y   L( A) x 
 0

L(A) has to be a Lorentz matrix because y . y = x . x.

To obtain how L(A) depends on A, we proceed as follows.

We know that the defining relation for the Lorentz transformation

corresponding to A  SL(2, C) is:

AX A †  A( x    )A †  y     (   x  )  ,   L( A)
86
Unit 19 Continuous Groups
It holds good for any point x. Therefore, we must have:

A  A †     

Multiplying on the left by  and taking the trace, we find that

1
   L( A)   Tr (  A  A † )
2

By applying two successive transformations, we get:

X  Y  A1 X A1† , y   L( A1 )  x 

and

Y  Z  A2Y A2† , z   L( A2 )  y 

We deduce that

Z  A2 ( A1X A1† ) A2†  ( A2 A1) X ( A2 A1)† , z   L( A2 A1) x 

so that

z   L( A2 )  L( A1 )   L( A2 A1 )  x 

Therefore,

LA2 LA1   LA2 A1 

We can work out the relationship between A and L(A) in the following detailed
and very useful form:

a b 
A , ad  bc  1
c d 
 


L00  a 2  b 2  c 2  d 2 2 
L01   ab * cd *

L02   ab * cd *


L03  a 2  b 2  c 2  d 2 2 
L10   a * c  b * d  L11   ad *  b * c 

L12   (ad * b * c ) L13   ac *  bd *

L20   (a* c  b * d ) L21   (a* d  b * c )

L22   (a * d  bc*) L23   ( a * c  bd *) 87

Block 5 Group Theory
L30  a  2
b 2
c 2
d 2
2
L31   ab *  cd * L32   ab *  cd *


L33  a 2  b 2  c 2  d 2 2 
Here the real and imaginary parts of a complex number z are denoted by (z )
and (z ) , respectively.

We just calculate two of the elements. You can calculate the remaining
L (A) yourself.

1 1
L00  Tr ( 0 A 0 A † )  Tr AA†
2 2

1  a b

a *

c *

 Tr
2  c d  b *
 d * 

1  a  b ca * db * 
2 2
 Tr
2  ac *  bd * c2  d2




1 2
2

a  b2  c2  d 2 
Similarly,

1
L12  Tr (1A 2 A † )
2

1  0 1 a b  0  i  a * c *
 Tr     
2  1 0  c
 d   i 0   b * d * 

1 c d    ib *  id * 
 Tr   
2 a b   ia * ic * 


1   icb * ida *  

 Tr
2    iad * ibc * 

 ad * b * c 

Proof of L( A)  L†

From the expressions above, we see that L00 ( A)  0 for any A. As L(A) is a
Lorentz transformation, it maps a time-like vector to a time-like vector. This
shows that the L(A) preserves the time direction.

To show that det L(A) = +1 we must prove that there is no A which can give
space inversion.
88
Unit 19 Continuous Groups
The formulas above show that for L( A)  I s , to be true, L00  L33  0 implies
that a 2  d 2  0 or a = 0 and d = 0.

As det A = ad – bc = 1, this means bc = –1.

But L11  L22  0 implies that (b * c )  0, and L12  L21  0 implies that
(b * c )  0, giving b * c  0. This is a contradiction because bc = –1.

19.5.1 Homomorphism SL(2,C)  L† is 2 : 1

The formula relating A  L(A) also shows that the correspondence is a 2 : 1

mapping, with A and – A giving the same L(A)  L† .

But, how do we know that there are just two, and not more A’s for a given
L(A)?

We try to find its kernel.

We ask in the following example: If L(A) = 1 (the 4  4 unit matrix), then what
should A be?

Example 19.1

Using the formulas above, show that the only possibilities are A = 1, or A = – 1
(the 2  2 unit matrices).

Solution : Given the formulas above for L (A) we find that if L(A) = 1, then /

i) L03  0, and L30  0 imply that L03  L30  0 which gives a 2  d 2 and
c2  b2.

ii) L00  1 and L33  1 imply that L11  L33 which gives b 2  c 2  0
meaning that b = 0 and c = 0.

iii) L01, L10, L02, L20, L13, L31, L23, L32 become zero because of b = 0, c = 0.

iv) L11  L22  (ad *)  1. And L12  (ad *)  0 as well as (a * d )  L21  0.
This means that ad* is a real number. Therefore, ad* = a*d = 1. The
determinant of an SL(2, c) matrix is equal to one, that is ad – bc = ad = 1.
Therefore, a = 1/d = 1/d* or d = d* is real. So is a = 1/d*.

v) a 2  d 2 , and therefore, d   a. But ad = 1, so that ad  a2  1. As a is

real, only d  a is allowed. Therefore, a2  1 and both a and d can be +1
or –1, but both must have the same sign.

vi) Therefore A can be 1 or A = – 1, proving the mapping SL(2, C)  L† to be

2 : 1.
89
Block 5 Group Theory
19.5.2 Six Parameters of SL(2,C)

A typical matrix in SL(2, C) is:

a b
A , ad  bc  1
 
c d

where a, b, c, d are complex numbers with one complex condition among

them. This means that there are eight real parameters with two real conditions
among them. Such a matrix is parametrized by six parameters equal to the
number of independent parameters in the Lorentz group (three for rotations
and three for boosts).

19.5.3 Boosts and Hermitian SL(2,C) Matrices

Every hermitian SL(2, C) matrix can be written in the standard form:

A  cosh(  / 2)  n.  sinh(  / 2),

because if A hermitian, it can be written as A   a with real a  , and

the determinant det A = 1 means that a02  a12  a22  a32   1. Such matrices
represent boosts in the direction of the unit vector n with velocity given by
tanh .

As an example
e / 2 0 
 
B3 ( )  cosh(  / 2)   3 sinh(  / 2)   
 0 e  / 2 


corresponds to
 cosh  0 0 sinh  
 
 
 0 1 0 0 
L(B3 ( ))   
 0 0 1 0 
 
 
 sinh  cosh  
 0 0

Note that if  > 0 then this is an active Lorentz boost in the direction of
coordinate 3-axis, with velocity v = tanh . This boost transforms the
momentum four vector (m, 0, 0, 0) of a particle of mass m at rest, to that of the
particle moving along the 3-axis with momentum 4-vector
p   ( p 0  m cosh , 0, 0, p 3  m sinh ). The speed of the particle is:
p
v  tanh .
p0

SAQ 5

Show that if A is hermitian, then L(A) is symmetric.

90
Unit 19 Continuous Groups
Let us take up another example.

Example 19.2

The SL(2, C) matrix Bp for a boost in p direction that transforms a four vector
(m, 0, 0, 0) representing the four-momentum of a particle of rest mass m in its
 
rest frame into p 0  p 2  m 2 , p , that is

 p0  m
   
   
 p1  0
   L(Bp )  
 2 0
p   
   
 p3  0
   

is given by

B p  cosh(  / 2)  n .  sinh(  / 2).

Show that we can write it as

m 0  p   
Bp 
2mp 0  m 

where n  p / p and cosh( )  p 0 / m, sinh( )  p / m.

Solution : We know that

p0
2 cosh2 ( / 2)  cosh   1  1
m

Therefore,

cosh   1 p0  m p0  m
cosh(  / 2)   
2 2m 2m( p 0  m )

and, similarly,

cosh   1 ( p 0  m)
sinh(  / 2)  
2 2m

( p 0 )2  m 2 p
 
2m( p 0  m ) 2m( p 0  m )

It is worth remembering that boosts do not form a subgroup because the

product of two Hermitian matrices is not Hermitian unless they commute with
each other.

91
Block 5 Group Theory

SAQ 6
A particle of rest mass zero moves with four-momentum
p (0)   ( , 0, 0, ),   0. Find the SL(2, C) matrix which represents the
boost along the 3-direction and which takes (, 0, 0, ) to ( p 0 , 0, 0, p 0 ) with
p 0  0.

19.5.4 Rotations and SU(2)

Unitary SL(2,C) matrices form a subgroup of SL(2,C) and they correspond to
rotations. Let us demonstrate this in Example 19.3.

Example 19.3

a) Show that unitary matrices of SL(2, C) form a subgroup.

b) Show that for A  SU(2), the Lorentz matrix L(R) is of the form:

1 0 0 0
 
0 
 
L( A)   
0 Rij 
 
 
0 

where R ij is a 3  3 rotation matrix RT R  1, det R  1.

Solution : a) The product of two unitary matrices is unitary:

U 2U1 † U 2U1   U1† U 2† U 2U1  U1† U1  1

As the inverse U 1 is equal to hermitian adjoint U † and

U † † U †  UU †  1

the inverse matrix is unitary. The identity is trivially a unitary matrix.

This subgroup of SL(2, C) is denoted by SU(2) (for special unitary matrices

of dimension 2).

b) If A is unitary A †  A 1 , then

1 1
L00 ( A)  Tr ( 0 A 0 A † )  Tr ( AA 1 )  1
2 2

Therefore, (see SAQ 4b in Sec. 19.4.1).

L10  L20  L30  L01  L02  L03  0

92
Unit 19 Continuous Groups
The Lorentz matrix has the form:

1 0 0 0 
 
0 R11 R12 R13 

 
0 R 21 R 22 R 23 
 
 
0 R 31 R 23 R 33 

and the fact that column vectors of a Lorentz matrix form an orthonormal
basis, shows that R is an orthogonal matrix.

Any SU(2) matrix can be written as

  
A   0 cos  in .  sin
2 2

where

n .   n11  n 22  n 3 3

and n  (n1, n 2 , n 3 ) is a unit vector. This follows from the fact that for such a
matrix A †  A 1 , so that if

a b
A  , ad  bc  1
c d 


then

d  b  a * c *
A 1     A†
 c a  b * d * 


Comparing the two sides, we get two conditions

d  a*, c  b *

Therefore, a general SU(2) matrix has the form

 a b
A , a2  b2 1
 b * a *


If a  a1  ia2 and b  b1  ib2 , (the real and imaginary parts), then

 a1  ia2 b1  ib2 
A   a1 0  ia2  3  ib2 1  ib1 2
 b  ib a1  ia2 
 1 2

with the condition that

a 2  b 2  a12  a 22  b12  b22  1

93
Block 5 Group Theory
We can see that the coefficients of 1,  2 ,  3 are i times b2 , b1, a2 . We
choose a1  cos  / 2. Then

b22  b12  a 22  1  cos 2  / 2  sin 2  / 2

We define the unit vector n by

(b2 , b1, a2 )  n sin( / 2)

so that A can be written as:


A  cos( / 2)0  i sin( / 2)n. 

The nature of rotation R can be found by looking at simple examples. If

e i / 2 0   
A    0 cos  i 3 sin
 0 e i / 2  2 2

then n = (0, 0, 1) and

1 0 0 0
 
 
0 cos  sin  0
L(A)   
0  sin  cos  0
 
 
0 1 
 0 0

which is a rotation about the axis in the direction of n.

19.5.5 Form of a General SL(2,C) Matrix

Any SL(2, C) matrix can be written as a product of a boost and an SU(2)

matrix. Specifically,

A  V H, H  ( A† A)1/ 2 , V  A( A† A) 1/ 2

The square-root (and square root inverse) of ( A † A) is well defined because it

is a positive definite hermitian matrix, that is, a matrix with strictly positive
eigenvalues. The proof is especially easy because A † A it is a 2  2 hermitian
matrix. It has determinant (equal to product of the two eigenvalues) equal to 1
and the trace (equal to the sum of the eigenvalues) is
 a 2  c 2  b 2  d 2   0 where a, b, c, d are elements of A.
 

To calculate the square root of a positive definite hermitian matrix, we first

diagonalize it by a similarity transformation. Then take the square-root of the
positive eigenvalues along the diagonal. Then go back to the original basis by
the inverse of the same similarity transformation.
94
Unit 19 Continuous Groups

19.5.6 The Matrix  and Other Useful Formulas

A very useful and interesting matrix is the SU(2) element

0 1
  .
 
 1 0

which corresponds to a rotation by  about the y-axis. It is antisymmetric real

matrix whose negative is its inverse

(  )   1

It converts any SL(2, C) matrix to its inverse-transpose as follows:

 A  1  ( A 1 )T ,

as can be immediately verified.

Parity inverted Pauli matrices  P

 are defined as:

1 0 0  1
 0P  0   , 1P  1   ,
   
0 1 1 0

0 i 1 0
P      , P      .
2 2   3 3  
 i 0 0 1

These matrices are related to the standard Pauli matrices by

 
    1   P
 *

which can be verified directly.

†
The basic formula connecting SL(2, C) to L is:

A  A †     

Multiplying by   and summing over , we get:

  A  A †      ( T ) 

   ( T )      

  
1 P


Taking the complex conjugate of the above equation and multiplying by  and
1 on two sides, we get

( A 1 )   A 1  
1    ( 1T )
B B  95
Block 5 Group Theory
Therefore, the SL(2, C) matrix which generates 1T is the inverse-dagger of
the matrix which generates .

Let us now summarise what we have studied in this unit.

19.6 SUMMARY
In this unit we have discussed the following concepts:

 continuous groups: an introduction;

 active and passive transformations;

 simplest continuous groups;

 orthogonal group O(3);

 the subgroup SO(3) or rotation group;

 Lie groups or continuous groups;

 one parameter subgroups of SO(3);

 generators of one-parameter subgroups;

 commutators of generators;

 Lie algebra of generators;

 Minkowski space and the Lorentz group;

 Minkowski space;

 Lorentz group;
†
 SL(2,C) and the Lorentz group L ;

†
 homomorphism SL(2,C)  L is 2:1;

 six parameters of SL(2,C);

 boosts and hermitian SL(2,C) matrices;

 rotations and SU(2);

 form of a general SL(2,C) matrix; and

 the matrix  and other useful formulas.

19.7 TERMINAL QUESTIONS

1. Find the rotation matrix (of active transformation) which corresponds to a
rotation about an axis in the direction of a unit vector n by an angle .

2. Calculate the structure constants for the Lie algebra of SO(3).

†
3. Prove that if L  L then for a positive time-like vector x  (x 0 , x ), x 0  x ,
96 the vector x   Lx , is also positive time like: x  0  x  .
Unit 19 Continuous Groups
4. Show that the four column vectors of a Loretnz matrix form an orthonormal
basis in Minkowski space. (This result is similar to that for the orthogonal
matrices.)

5. Calculate L(A) for A = cosh (/2) + n .  sinh(  / 2).

6. Show that

  
A   0 cos  in .  sin
2 2

corresponds to a passive rotation by an angle  about the axis in the

direction n. Compare the result with the SO(3) matrix of TQ 1 after changing
the sign of .

19.8 SOLUTIONS AND ANSWERS

Self-Assessment Questions
1. Yes, because if R  SO(3) and T  O(3) then

TRT 1  T R T 1  R  1

There are two cosets. Apart from SO(3), the other is obtained by
multiplying elements of SO(3) by the matrix I s  1. The factor group is
isomorphic to the multiplicative group of two elements (1, 1).

2. An active rotation about the z-axis in the right-handed screw sense

changes only the x-and y-coordinates (see Fig. 19.3).

x   R cos(    )  x cos   y sin ,


y   R sin(    )  y cos   x cos 


  x = R cos, y = R sin 

Active x

Fig. 19.3: Diagram for solution of SAQ 2.

x   x cos   y sin 

y   y cos   x sin 

z  z 97
Block 5 Group Theory
Therefore, the SO(2) matrices are:

cos   sin 
 
 sin  cos  


and they generate a one-parameter additive group:

cos   sin  cos   sin   cos(   )  sin(   )

    
 sin  cos    sin  cos    sin(   ) cos(   ) 
 

3. The structure constants are zero for i = j. Because of antisymmetry,

c ij k  c ji k , so there are n(n – 1)/2 independent pairs for each k which
can take n values.

4. a) (LT )L  (LT L)    1.

The Lorentz group matrices L can be divided into two mutually

exclusive subsets:

i) those for which det L = +1, and

ii) those for which det L = 1.

They are called, respectively, the proper and improper Lorentz

groups.

The former, written as L with det = +1 form a subgroup, but those

with det = 1 (called L ) do not form a subgroup because the product
of two matrices with determinant –1 is +1, not –1.

These two subsets are further divided into subsets, as can be seen
below:

The equation LT L  , when written for 00 element, is

00  1   (LT )0 () L0

, 

 (L00 )2  (L10 )2  (L20 )2  (L30 )2

In other words

(L00 )2  1  (L10 )2  (L20 )2  (L30 )2 .

so that (L00 )2  1 and therefore, either L00  1 or L00  1.

Thus, denoting matrices with L00  1 by an ‘up arrow’ and those with
L00  1 by a ‘down arrow’,

L  L  L , L  L  L

98
Unit 19 Continuous Groups

Members of L , that is, Lorentz matrices with determinant +1 and

L00  1 are called proper orthochronus transformations. They form a
subgroup within L .

b) We look at the inverse matrix L1. Then L001  L00 and Li 01  L0i .
Therefore,

L00  L001   (L10

1 )  (L1 ) 2  (L1 ) 2   L2  L2  L2
20 30 01 02 03

5. As A †  A, the expression for L(A) is

1 1
L ( A)  Tr (  A  A)  Tr (  A  A)  L ( A)
2 2

using the property of the trace Tr(AB) = Tr(BA).

6. The boost in 3-direction is of the forms

e / 2 0 
B3   ,
   / 2 
 0 e 

leading to Lorentz transformation which must change  to p 0 .

 p 0   cosh  0 0 sinh    
     
     
 0   0 1 0 0  0
    
   0 0 1 0  0
 0     
     
 p 0   sinh  cosh    
   0 0  

Therefore,

p 0  (cosh   sinh )  e  , that is e  / 2  p 0 / 

and

 p0 /  0 
 
B3   
 0  / p 0 


Terminal Questions

1. Let the fixed point of rotation be at the origin, and P = r = (x, y, z) be the
point which gets rotated about axis n by  to a new point
P = r = (x, y, z). We have to find P = r. (See Fig. (19.4a).

From Fig. 19.4b, we can arrive at the vector OP as the following sum:
   
OP   OA  AB  BP 

OA is in the direction of the axis n. 99

Block 5 Group Theory
AB is perpendicular to n and in the plane OAP.

BP is perpendicular to the plane OAP.

The length OA is equal to n . r = r cos where  is the angle AOP

between n and r and r  r .

The length AP = AP is equal to n  r  r sin 

Length AB = AP cos = AP cos = n  r cos 

Length BP = AP sin = n  r sin 

The unit vector in the direction of BP is m = (n  r)/ n  r

The unit vector in the direction of AB is 1 = m  n = (n  r)  n/ n  r

   
Thus, r   OP   OA  AB  BP 

 n . r n  n  r cos  1  n  r m

n  r   n n  r 
 n . r  n  n  r cos   n  r sin 
nr nr

 n . r n  cos  n  r   n  sin  n  r

 (1  cos ) n . r  n  (cos ) r  (sin ) n  r

P
P
A

P


A B P

(a) (b)

Fig. 19.4: a) Diagram for solution of TQ 1; b) Enlarged part of a) showing the

100 region ABPP.
Unit 19 Continuous Groups
2. If we take the basis J1, J2, J3 for the Lie algebra of SO(3), then in
[J i , J j ]   c ij k J k , the nine structure constants are:
k

c12 k  0, 0,1 for k  1, 2, 3, respectively

c 23 k  1, 0, 0 for k  1, 2, 3, respectively

c 31k  0,1, 0 for k  1, 2, 3, respectively

3. We already know that x 0   x 2  x  0   x  2  0. We only need to

2 2

prove that x 0  0 .

x  0  L00 x 0  L01x 1  L02 x 2  L03 x 3

On the right-hand side the terms L01x 1  L02 x 2  L03 x 3 are like the dot
product of two three-dimensional vectors a  (L01, L02 , L03 ) and
x  ( x 1, x 2 , x 3 ). Therefore,

L01x 1  L02 x 2  L03 x 3  x 0   a . x  a x

2

The lowest value that L01x 1  L02 x 2  L03 x 3 can have is:

 a x   L201  L202  L203 x

But

L00  L201  L202  L203 and x0  x

Therefore,

( x )0  L00 x 0  L01x1  L02 x 2  L03 x 3

 L00 x 0  L201  L202  L203 x

0

On the other hand, members of L flip the time sense. That is so because
every member of L is equal to a matrix of L multiplied by the time
inversion matrix I t   . The former does not flip the time sense but the
latter does. The product of two L matrices is, of course, in L .

4. The defining equation for the Lorentz matrix is:

LT L   101
Block 5 Group Theory
We call the column vectors of the matrix as:

L00  L01  L02  L03 

       
L  L  L  L 
 10   11   12   13 
E0   , E1   , E2   , E3   ,
L20  L21  L22  L23 
       
       
L30  L31  L32  L33 

Then the defining equations can be written as:

E , E  E E   L  L  (LT L)  

5. Let n = (n1, n2 , n3 ) be the unit vector with n12  n 22  n 32  1. We write

n  n1  in2 , c  cosh(  / 2) and s  sinh(  / 2) . Then

 c  n3 s ns 
A   
 n s c  n 3 s 

Note that a and d are real and

n   n1, n   n 2 , n 2  n12  n 22 , n 2  2n1n 2 .

To calculate the various products in the expressions for the matrix

elements L(A), it is helpful to first calculate quantities like ab*, etc.

a* b* c* d*

a (c  n 3 s ) 2 (c  n3 s )n  s (c  n3 s )n  s (c 2  n 32 s 2 )

b (c  n 3 s ) n  s (n12  n 22 ) s 2 n 2 s 2 (c  n 3 s ) n s

c (c  n3 s )n  s n 2 s 2 (n12  n 22 )s 2 (c  n 3 s )n  s

d (c 2  n 32 s 2 ) (e  n3 s )n  s (c  n3 s )n  s (c  n 3 s ) 2

You can show that

L00  cosh 

L11  1  n12 (cosh   1)

L22  1  n 22 (cosh   1)

L33  1  n 32 (cosh   1)

L01  n1 sinh   L10

102
Unit 19 Continuous Groups
L02  n2 sinh   L20

L03  n3 sinh   L30

L12  n1n2 (cosh   1)  L21

L13  n1n3 (cosh   1)  L31

L23  n2n3 (cosh   1)  L32

This kind of boost is called direct boost like the B3 above. It is used to
connect the state of a particle (of proper mass m) at rest when it is boosted
in the direction of p to a state when its three momentum becomes p. The
Lorentz matrix in the direction of momentum given by the unit vector
n  p / p , so that it acquires a speed

p
v  tanh  
p 2  m2

6. With notation

 
n  (n1, n 2 , n 3 ), n12  n 22  n 32  1, c  cos , s  sin ,
2 2

our matrix is,

a b   c  in3 s n 2  in1  s 
A   d  a*, c  b*
   
c d    n 2  in1  s c  in3 s 

We use the formulas connecting A to L(A).

We have already proved that a unitary matrix like this will be a pure
rotation for which L00  1, L0i  Li 0  0. For the remaining:

L11  (ad * b * c )  (a 2  c 2 )  cos   (1  cos )n12

L12  (ad * b * c )  (a 2  c 2 )  u 3 sin   (1  cos )n1n2

L13  2(ac*)  n2 sin   (1  cos )n1n3

L21  (a * 2 c 2 )  n3 sin   (1  cos )n1n2

L22  (a * 2 c * 2 )  cos   (1  cos )n 22

L23  2(ac*)  n1 sin   (1  cos )n2n3

L31  2(ac )  n2 sin   (1  cos )n1n3

L32  2(ac )  n1 sin   (1  cos )n2n3

L33  a 2  c 2  cos   (1  cos )n 23

103
Block 5 Group Theory
Comparison with the result of TQ 1 with     is

r   (1  cos ) (n.r ) n  (cos ) r  (sin ) n  r

or, written in detail, it is:

x 1  (1  cos ) n1x 1  n2 x 2  n3 x 3 n1

 x 1 cos   n2 x 3  n3 x 2 sin 

 
 cos   (1  cos ) n12 x 1  n 3 sin   (1  cos ) n1n 2  x 2

  n2 sin   (1  cos )n1n3 

 L11x 1  L12 x 2  L13 x 3 .

Similarly, you can determine for x  2 and x  3 .

104