Relativistic Kinematics

Raghunath Sahoo∗

Indian Institute of Technology Indore, India-452020

arXiv:1604.02651v1 [nucl-ex] 10 Apr 2016

∗
Email: [email protected], (Lecture delivered in IX SERC School on Experimental High Energy
Physics, IIT Madras, India,12-14 Dec. 2013)

1
Relativistic Kinematics Raghunath Sahoo 2

I. WHY HIGH ENERGIES?

In Particle Physics, we deal with elementary constituents of matter. By elementary

we mean the particles have no substructure or they are point-like objects. However, the
elementariness depends on the spatial resolution of the probe used to investigate the possible
structure/sub-structure [1]. The resolution is ∆r if two points in an object can just be
resolved as separate when they are a distance ∆r apart. Assuming that the probing beams
themselves consist of point-like particles like electrons or positrons, the resolution is limited
by the de Broglie wavelength of these beam particles, which is given by λ = h/p , where p
is the beam momentum and h is Planck’s constant. Hence beams of high momentum have
short de Broglie wavelengths and can have high resolution. For example if we need to probe
a dimension of 1 F ermi (10−15 meter) (let’s say the inner structure of proton, the charge
radius of proton being ∼ 0.840 f m), we need to use a beam of momentum 1.47 GeV which
is given by de Broglie’s above expression. Figure 1 shows a schematic picture where a low
energy probe fails to probe the inner structure of an object unlike a high energy probe with
higher resolution. In addition to the above consideration, Einstein’s formula E = mc2
helps us to produce particles of higher masses (like the massive gauge bosons, Higgs etc.) in
nature’s way.

object

probe

FIG. 1: Left: A low-energy probe probing an object, Right: a high-energy probe able to probe
deep inside of the object because of higher resolution.

II. SPECIAL THEORY OF RELATIVITY & INVARIANTS

In particle physics, the particles are treated relativistically, meaning E ≈ pc  mc2 and
thus special theory of relativity becomes an mathematical tool in describing the particle
kinematics.

• Space and time can not be treated independently as is done in Newtonian mechanics.
• Physical objects that were treated as an independent three component vector and a
scalar in non-relativistic physics mix in high-energy phenomena.
• Combined to form a four-component Lorentz vector that transforms like a time and
space coordinate.
For their consistent and unified treatment, one relies on Einstein’s theory of special
relativity (STR), having the following two underlying principles.
Relativistic Kinematics Raghunath Sahoo 3

• Invariance of velocity of light: velocity of light always remains as the constant c in any
inertial frame.

• Relativity Principle: This requires covariance of the equations, namely the physical
law should keep its form invariant in any inertial frame of reference. In mathematical
language this amounts to the fact that physical laws have to be expressed in Lorentz
tensors.

Note that the principle of relativity applies to Galilei transformation and is valid in New-
tonian mechanics as well. But the invariance of the velocity of light necessitates Lorentz
transformation in changing from one inertial system to another that are moving relative to
each other with constant speed.

A. Lorentz Transformation

Consider a Lorentz boost in x-direction. Here, a particle at (t, x, y, z, ) in a coordinate

frame L is boosted to (t0 , x0 , y 0 , z 0 ) with velocity v. This statement is equivalent to changing
to another coordinate frame L0 which is moving in the x-direction at velocity −v. L0 is
assumed to coincide with L at t = t0 = 0. Then the two coordinates are related by the
following equations:

t + (v/c2 ) x 0
t → t0 = q ⇒ x0 = γ x0 + βx ,

1 − (v/c)2
x + vt 0
x → x0 = q ⇒ x1 = γ βx0 + x ,


(1)
1 − (v/c)2
0
x2 = x2 ,
0
x3 = x3

where, β = v/c, γ = √ 1 . The above equations can be written in matrix form as:
1−β 2
 00     0
x γ βγ 0 0 x
 x1 0  βγ γ 0 0  x1 
 0= ×  (2)
 x2   0 0 1 0   x2 
0
x3 0 0 0 1 x3

1. The Proper Time (τ )

It is the time an observer feels in the observer’s rest frame.

p
Proper time dτ ≡ dt 1 − β 2 is a Lorentz invariant scalar.
Proof:
Relativistic Kinematics Raghunath Sahoo 4

ds2 = (cdt)2 − dx2 − dy 2 − dz 2

dx 2 dy 2 dz 2
 
2 2
= c dt 1 − − −
dt dt dt
2 2 2


= c dt 1 − β
= (cdτ )2
is Lorentz invariant by definition.

B. What is need of the variable called ”Rapidity”?

Successive Lorentz boost in the same direction is represented by a single boost, where
the transformation velocity is given by
β+β 0
β 00 = |v/c|00 = 1+ββ 0

Proof:
Assume velocity v 0 in frame L is observed as v 00 in frame L00 , where the frame L0 is travelling
0
in the x-direction with −v in frame L. The coordinates (t0 , x1 ) are expressed in terms of
(t, x1 ) using the usual Lorentz transformation equations given by Eqn. 1. Omitting other
coordinates for simplicity, one obtains:

0
x0 = γ x0 + βx1


0
x1 = γ βx0 + x1

v0 dx1
β0 = = 0
c dx
then
0
00dx1 γ (βdx0 + dx1 )
β = 00 =
dx γ (dx0 + βdx1 )
β + β0
⇒ β 00 = (3)
1 + ββ 0
The velocity is not an additive quantity. i.e. non-linear in successive transforma-
tion. Here comes the need of ”Rapidity” to circumvent this drawback, by defining
 
β = tanh y or y = 12 ln 1+β 1−β
. One can show (we will be showing this in subsequent
sections) that rapidity is an additive quantity i.e.

y 00 = y + y 0

Using the rapidity, a Lorentz transformation with finite η, can be decomposed into N
successive transformations with rapidity
Relativistic Kinematics Raghunath Sahoo 5

∆y = y/N

Solving β, γ in terms of y, we have

β = tanh y, γ = cosh y, βγ = sinh y

Lorentz boost given by Eqn. 1 can be rewritten as

0
x0 = (cosh y) x0 + (sinh y) x1 ,
0 (4)
x1 = (sinh y) x0 + (cosh y) x1 ,
Comparing this with rotation in the x − y plane:

x0 = xcos θ − ysin θ,
(5)
y 0 = xsin θ + ycos θ,
Eqn. 4 can be obtained from Eqn. 5 by substituting

• θ → −iy (this y is rapidity variable )

• x → ix0

• y → x1 (this y is the Cartesian coordinate )

Lorentz boost (in the x-direction) is formally a rotation by an angle (−iy) in the x and
imaginary time (ix0 ) plane.

Experimental Consideration: In high-energy collider experiments, the secondary par-

ticles which are produced from the interaction, are boosted in the z-direction (along the
beam axis). The boosted angular distribution is better expressed as rapidity distribution.
At high-energies, each particle has E ∼ pc, pII = pcos θ, and its rapidity is approximated
by so-called pseudo-rapidity:
   
0 1 1 + βII 1 E + pII c
η = ln = ln ∼ −ln tan θ/2. (6)
2 1 − βII 2 E − pII c

This fact is taken into account in designing detectors, which are divided into modules that
span the same solid angle in the η − φ (azimuthal angle) plane.

C. Four Vectors

Th position-time 4-vector: xµ , µ = 0, 1, 2, 3; with x0 = ct, x1 = x, x2 = y, x3 = z.

I ≡ (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2

0 0 0 0
= (x0 )2 − (x1 )2 − (x2 )2 − (x3 )2 (7)

I is called the 4-dimensional length element, which is Lorentz Invariant (LI). A quantity
having same value in all inertial frames is called an ”invariant”. This is like r2 = x2 +y 2 +z 2
Relativistic Kinematics Raghunath Sahoo 6

being invariant under spatial rotation. I could be written in the form of a sum:

3
X
I = xµ xµ (8)
µ=0

To take care of the negative signs, let’s define a ”metric” gµν such that
 
1 0 0 0
0 −1 0 0
g =   (9)
0 0 −1 0 
0 0 0 −1
Now
I = gµν xµ xν (10)
Define covariant 4-vector xµ (index down):

xµ = gµν xν (11)
xµ (index up) is called ”contravariant 4-vector”. With the above definitions, now

I = xµ xµ = x µ xµ (12)

To each contravariant 4-vector aµ , a covariant 4-vector could be assigned and vice-versa.

aµ = g µν aν (13)
aµ = gµν aν (14)

g µν are the elements in g −1 . Since g −1 = g, g µν = gµν .

Given any two 4-vectors, aµ and bµ ,

aµ bµ = aµ bµ = a0 b0 − a1 b1 − a2 b2 − a3 b3 (15)
is L.I.. The above operation is called ”4-vector scalar product”. Remember Einstein’s
summation convention (repeated Greek indices are to be summed).

a.b ≡ aµ bµ
= a0 b0 − ~a.~b (16)

a2 ≡ a.a = (a0 )2 − ~a2 (17)

The 1st term is called ”temporal” and the 2nd is called spatial component.

• If a2 > 0: aµ is called time-like. Events are in the forward light-cone. They appear
later than the origin, O. Events in the backward light-cone appear earlier to O. Only
events in the backward light-cone can influence O. And O can have an influence only
on the events in the forward cone.
Relativistic Kinematics Raghunath Sahoo 7

• If a2 < 0: aµ is called space-like. Events are called space-like events and there is no
interaction with O. This is related to ”causality”.
• If a2 = 0: aµ is called light-like. Connects all those events with the origin which can
be reached by a light signal.
This is shown pictorially in Figure 2.

ct
x timelike future ne
=
-c htco
t lig
lig ct
ht
co x=
ne

spacelike spacelike x

timelike past

FIG. 2: A schematic of lightcone diagram.

D. Energy-Momentum Four-Vector

The velocity of a particle is given by

d~x
~v = (18)
dt
where d~x is the distance travelled in the laboratory frame and dt is the time measured in
the same frame. Proper velocity of the particle is given by
d~x
~η = (19)
dτ
where d~x is the distance travelled in the laboratory frame and dτ is the proper time. Now
d~x d~x dt
~η = =
dτ dt dτ
= ~v γ
⇒ ~η = γ~v (20)

It is easy to work with the proper velocity, ~η , as only d~x transforms under Lorentz transfor-
mation. Furthermore,
dxµ
ηµ = (21)
dτ
Relativistic Kinematics Raghunath Sahoo 8

so

dx0 d(ct)
η0 = = 1 = γc (22)
dτ γ
dt
Hence
η µ = γ (c, vx , vy , vz ) (23)
This is called the proper velocity 4-vector. Remember that the spatial component brings up
the negative sign for covariant tensor. Now

η µ ηµ = γ 2 c2 − vx2 − vy2 − vz2

v2
 
2 2
= γ c 1− 2
c
= γ c 1 − β2
2 2

= c2 , (24)

which is Lorentz Invariant. This also proves that 4-vector scalar product is L.I.

We know momentum = mass × velocity. And velocity can be ”ordinary

velocity” or ”proper velocity”. Classically, both are equal (non-relativistic limit).
If p~ = m~v , the conservation of momentum is inconsistent with the principle of relativity.
In relativity, momentum is the product of mass and proper velocity.

p~ ≡ m~η (25)

pµ = mη µ (26)
The spatial component of pµ constitutes the (relativistic) momentum 4-vector:

m~v
p~ = γm~v = p (27)
1 − v 2 /c2

p0 = γmc (28)
Relativistic energy, E:
mc2
E ≡ γmc2 = p (29)
1 − v 2 /c2
Hence,
E
p0 = (30)
c
and the energy-momentum 4-vector:
 
µ E
p = , p x , py , pz (31)
c
Relativistic Kinematics Raghunath Sahoo 9

Now

µ E2
p pµ = 2
− p~ 2 = m2 c2 : L.I.
c
= (mη µ ) (mηµ )
= m2 (η µ ηµ )
= m2 c2
⇒ pµ pµ = m2 (N atural U nits). (32)

⇒ E 2 = p~ 2 c2 + m2 c4
In natural units,
E 2 = p~ 2 + m2 (33)

• p2 = m2 > 0: Ordinary massive particle

• p2 = m2 = 0: Massless particles like photons, gravitons etc.
• p2 < 0: Tachyon or virtual particles
• pµ = 0: Vacuum

Remember that the relativistic equations p~ = γm~v and E = γm do not hold good for
massless particles and m = 0 is allowed only if the particle travels with the speed of light.
For massless particles,

v = c and E = |~p|c.

E. The Choice of Units

We know

x2 = c2 t2 − x21 − x22 − x23 (34)

p2 = m20 c2 (35)

The velocity of light ”c” appears directly in these and many other formulas. Furthermore,
de Broglie relation between 4-momentum and wave vector of a particle is

E = h̄ω (obtained f rom Einstein0 s equation) (36)

In 4-vector notation,
P = h̄K (37)
E ω
where P = c
, p ,K = c
, k . If we choose a system of unit in which

h̄ = c = 1 ,

where h̄ = h
2π
= 1.055 × 10−34 Joule sec: unit of action/angular momentum (M L2 /T ).
Relativistic Kinematics Raghunath Sahoo 10

c = 2.998 × 108 meter sec−1 : unit of velocity, the velocity of light in vacuum (L/T ).

Now the relativistic formula for energy,

E 2 = p2 c2 + m20 c4 (38)
in this new system of unit (called natural unit and more popularly used in high-energy
(particle) physics) becomes

E 2 = p2 + m20 (39)
We can define system of units completely, if we specify the unit of energy (M L2 /T 2 ). In
particle physics, unit of energy is GeV (1 GeV = 109 eV ). This choice is motivated by
the rest mass of proton ∼ 1 GeV . This gives rise to mass (m), momentum (mc), energy
h̄
(mc2 ) in GeV . Length ( mc ) and time ( mch̄ 2 ) in GeV −1 .

Taking the values of h̄ = c = 1, one obtains,

1 sec = 1.52 × 1024 GeV −1

1 meter = 5.07 × 1015 GeV −1

1 f ermi ≡ 1 f m = 10−13 cm = 10−15 m

⇒ 1 f m = 5.07 GeV −1

1 f m = 3.33 × 10−23 sec

197 M eV = 1 f m−1

Note: 1 T eV = 103 GeV = 106 M eV = 109 KeV = 1012 eV .

The additional advantage of using natural unit in high energy particle physics is that we
deal with strong interaction, whose life time ∼ 10−23 sec, the decay length of particle can
be better expressed in terms of fermi.

F. Collider Vs Fixed Target Experiment

1. For Symmetric Collisions (A + A)

Consider the collision of two particles. In LS, the projectile with momentum p1 , energy
E1 and mass m1 collides with a particle of mass m2 at rest. The 4-momenta of the particles
are
p1 = (E1 , p1 ), p2 = (m2 , 0)
In CMS, the momenta of both the particles are equal and opposite, the 4-momenta are
Relativistic Kinematics Raghunath Sahoo 11

p∗1 = (E1∗ , p∗1 ), p∗2 = (E2∗ , −p∗1 )

The total 4-momentum of the system is a conserved quantity in the collision.

In CMS,

pµ pµ = (p1 + p2 )2 = (E1 + E2 )2 − (p1 + p2 )2

= (E1 + E2 )2
2
= Ecm ≡ s (40)
√
s is the total energy in the CMS which is the invariant mass of the CMS.

In LS,
pµ pµ = (p1 + p2 )2 = m21 + m22 + 2E1 m2 (41)
Hence
√ q
Ecm = s = m21 + m22 + 2Eproj m2 (42)

where E1 = Eproj , the projectile

√ energy in LS. Hence it is evident here that the CM frame
with an invariant mass s moves in the laboratory in the direction of p1 with a velocity
corresponding to:
Lorentz factor,
E1 + m2
γcm =√ (43)
s
√ Elab
⇒ s = , (44)
γcm
this is because E = γm and
ycm = cosh−1 γcm . (45)
The center of mass or center of momentum frame (CM/CMS) is at rest and the total
momentum is zero. This makes it a suitable choice for solving kinematics problems.

Note: We know that for a collider with head-on collision (θ = 1800 )

2
s = Ecm = m21 + m22 + 2(E1 .E2 + |p1 ||p2 |) (46)

For relativistic collisions, m1 , m2  E1 , E2

2
Ecm ' 4E1 E2 (47)

For two beams crossing at an angle θ,

2
Ecm = 2E1 E2 (1 + cos θ) (48)

The CM energy available in a collider with equal energies (E) for new particle production
rises linearly with E i.e.
Ecm ' 2E (49)
Relativistic Kinematics Raghunath Sahoo 12

For a fixed-target experiment the CM energy rises as the square root of the incident energy:
p
Ecm ' 2m2 E1 (50)
Hence the highest energy available for new particle production is achieved at collider
experiments. For example, at SPS fixed-target experiment to achieve a CM energy of 17.3
AGeV the required incident beam energy is 158 AGeV.

Problem: Suppose two identical particles, each with mass m and kinetic energy T ,
collide head-on. What is their relative kinetic energy, T 0 (i.e. K.E. of one in the rest frame
of the other). Apply this to an electron-positron collider, where K.E. of electron (positron)
is 1 GeV. Find the K.E. of electron if positron is at rest (fixed target). Which experiment
is preferred, a collider or a fixed-target experiment?

Note: Most of the times the energy of the collision is expressed in terms of nucleon-
nucleon center of mass energy. In the nucleon-nucleon CM frame, two nuclei approach each
√
other with the same boost factor γ. The nucleon-nucleon CM is denoted by sN N and is
related to the total CM energy by
√ √
s = A sN N (51)
This is for a symmetric collision with number of nucleons in each nuclei as A. The colliding
√
nucleons approach each other with energy sN N /2 and with equal and opposite momenta.
The rapidity of the nucleon-nucleon center of mass is
yN N = 0 and taking m1 = m2 = mp , the projectile and target nucleons are at equal and
opposite rapidities.
√
−1 sN N
yproj = − ytarget = cosh = ybeam . (52)
2mp
Note: Lorentz Factor

√
E s
γ = =
M 2A mp
√ √
A sN N sN N
= =
2A mp 2 mp
CM S
E
= beam (53)
mp
where E and M are Energy and Mass in CMS respectively. Assuming mass of a proton,
mp ∼ 1 GeV, the Lorentz factor is of the order of beam energy in CMS for a symmetric
collision.

2. For Asymmetric Collisions (A + B)

During the early phase of relativistic nuclear collision research, the projectile mass was
limited by accelerator-technical conditions (38 Ar at the Bevalac, 28 Si at the AGS, 32 S at the
Relativistic Kinematics Raghunath Sahoo 13

Elab/A (GeV)
3 5 6
102 10 104 10 10 107

Lorentz Factor (γ )
103

102

10

10 102 103
√ sNN (GeV)

FIG. 3: A plot to demonstrate how the Lorentz Factor increases with collision energy for a sym-
metric collision. Center of mass system is compared with fixed target experiment.

SPS). Nevertheless, collisions with mass ≈ 200 nuclear targets were investigated. Analysis of
such collisions is faced with the problem of determining an ”effective” center of mass frame,
to be evaluated from the numbers of projectile and target participant nucleons, respectively.
Their ratio - an thus the effective CM rapidity - depends on impact parameter. Moreover,
this effective CM frame refers to soft hadron production only, whereas hard processes are
still referred to the frame of nucleon-nucleon collisions. The light projectile on heavy target
kinematics are described in [17].
The center of mass energy of a collision of two different systems with charge Z1 , Z2 and
atomic numbers A1 , A2 with Z = A = 1, for a proton is
r
√ √ √ Z1 Z2
sN N ' 2 spp + spp (54)
A1 A2
√
where sub-index N N refers to the energy per nucleon inside the colliding nucleus and spp
is the corresponding energy in pp collisions. The rapidity shift in non-symmetric systems is
given by
 
1 Z1 A2
∆y ' ln (55)
2 Z2 A1
This is due to the fact that the center-of-mass frame of the pA collision doesn’t coincide
with the laboratory center-of-mass frame. The rapidity shift in p + P b collision is
 
1 Z1 A2
∆y ' ln
2 ZA
 2 1 
1 82 × 1
= ln
2 1 × 208
= −0.465
Hence ∆y = ±0.465 for p + P b collisions, flipping the beams. This rapidity shift need to be
taken into account for the comparison with Pb+Pb data.
Relativistic Kinematics Raghunath Sahoo 14


p1 = (E1, p1 )


p2 = (E2 , p2 )
θ

FIG. 4: The Laboratory system: two particle kinematics.

At LHC, the maximum√ proton beam energy is 7 TeV, while the maximum Pb beam
energy is 2.75 TeV, s = 8.775 TeV. The difference in available energy is due to the charge-
to-mass ratio, Z/A. More is the number of neutrons in the nucleus, difficult it is to accelerate
to higher energies. Because of different energies, the two beams will also not have the same
rapidity. For the proton beam yp = 9.61 and for the Pb beam it is yP b = 8.67. Thus the
center of the collision is shifted away from ycm = 0 by ∆ycm = (yp − yP b )/2 = 0.47.

G. The Energy and Velocity of the Center of Momentum

As discussed in Ref. [2], lets consider a Lorentz system- call it the laboratory system
(frame), with two particles with masses, m1 and m2 and four momenta p1 and p2 , respec-
tively as shown in Fig. 4.

What is the centre of mass energy, E ?

1. It is independent of the Lorentz system, where p1 and p2 are defined.

2. It must be possible to have an answer in terms of the three invariants, namely
p21 = m21 and p22 = m22
and [p1 p2 or (p1 + p2 )2 or (p1 − p2 )2 ].
The answer would be trivial in the center of momentum (mass) frame itself. We use asterisk
(*) in the CM frame to distinguish that from the laboratory frame.

In CM frame,

(p~1 ∗ + p~2 ∗ ) = 0
=⇒ p1 + p2 = (E1∗ + E2∗ , ~0)
and E ∗ = E1∗ + E2∗ (56)
Relativistic Kinematics Raghunath Sahoo 15

Hence,

(E ∗ )2 = (E1∗ + E2∗ )2
= (p∗1 + p∗2 )2
= (p1 + p2 )2 . (57)
This is because (p1 + p2 )2 is invariant.

Let us define the total mass M of the system as the square of the total 4-momentum as:

M 2 = (p1 + p2 )2 = P 2 = (E ∗ )2 = (E1 + E2 )2 − (p1 + p2 )2 = Invariant

Kinematically, for the given system, the two particles p1 and p2 are equivalent to one
single particle with 4-momentum P and mass M = ECM . Generalizing this, one can
consider these individual particles to represent a system of particles.

Further, we know

p~ = m~v γ, and E = mγ , where γ = √ 1 .

1−β 2

Hence, the 4-momentum of the two-particle system is given by:

~
P~ = M βγ

E = M γ.
Using the above, one obtains:

P~
βCM = (58)
E
(p~1 + p~2 )
= , (59)
(E1 + E2 )
is the velocity of the CM seen from the laboratory system.

1
γCM = p
1 − β2
E
= (60)
M
E1 + E2
= p
(E1 + E2 )2 − (p~1 + p~2 )2
E1 + E2
= , (61)
ECM
is the Lorentz factor or the Lorentz boost of the CM. In general, the Lorentz factor of
the CM is the ratio of the sum of the energies of the particles in the laboratory system and
the energy of the CM.
Relativistic Kinematics Raghunath Sahoo 16

H. The Energy, Momentum and Velocity of one particle as seen from the rest
system of another one

Lets assume that we sit on particle 1, which is in motion. What will be the energy of
particle 2, for us?

The answer to this question must always be the same, irrespective of the Lorentz system
we start with. It is thus expressible by the invariants discussed in the preceding subsection
(point-2), where the last two variables are the well-known Mandelstam’s s and t variables.

Let E21 : is the energy of particle 2, if we look at it sitting on particle 1, which then
appears to be at rest for us.
E21 = E2 , in the system where p~1 = 0. One needs to write E21 in terms of the invariants
available in this problem.

Now, p1 p2 = E1 E2 − p~1 .p~2 = m1 E2 (since p~1 = 0).

p1 p2
Hence, E21 = E2 = m1
.

Note that the RHS is already in an invariant form.

(p1 p2 )2 −m21 m22
|~p21 |2 = E21
2
− m22 = m21
.

If p1 and p2 are the momentum 4-vectors of any two particles in any Lorentz system, then

p1 p2
E21 =
m1
(p1 p2 )2 − m21 m22
|~p21 |2 = (p1 p2 ≡ E1 E2 − p~1 p~2 )
m21
2 |~p21 |2 (p1 p2 )2 − m21 m22
v21 = 2
= (62)
E21 (p1 p2 )2

The above three equations give the energy, momentum and velocity of particle 2, as seen
from particle 1. The velocity v21 is the relative velocity, which is symmetric in 1 and 2. Note
here that all these expressions are invariant and can be evaluated in any Lorentz system.

I. The Energy, Momentum, and Velocity of a Particle as seen from the CM System

This problem is now like all the above quantities are as seen from a fictitious particle M ,
called the “center-of-momentum-particle”, whose 4-momentum is

P = p1 + p 2 (63)
We need to apply formulae (Eq. 62) with p1 replaced by P and p2 by the 4-momentum
Relativistic Kinematics Raghunath Sahoo 17

of that particle whose energy, momentum and velocity, we like to determine. From Eq. 62,
P p1
E1∗ =
M
2 2 2
(P p 1 ) − M m1
|~p∗1 |2 = 2
M
2
∗2 (P p 1 ) − M 2 m21
v 1= (64)
(P p1 )2
This is by using Eq. 63 and by using

1
p1 p2 = [(p1 + p2 )2 − p21 − p22 ]
2
1
= (M 2 − m21 − m22 ), (65)
2
one obtains
M 2 + (m21 − m22 )
E1∗ = (66)
2M
M − (m21 − m22 )
2
E2∗ = (67)
2M
E1∗ + E2∗ = M (68)

|~p ∗ |2 = |~p1∗ |2 = |~p2∗ |2

M 4 − 2M 2 (m21 + m22 ) + (m21 − m22 )
=
4M 2
[M − (m1 + m2 )2 ] [M 2 − (m1 − m2 )2 ]
2
= (69)
4M 2
2
|~p ∗ |

∗2
v1 = . (70)
E1∗
Here, E1∗ , v1∗ are energy and velocity of particle 1, as seen from their common CM system
and M 2 = P 2 = (p1 + p2 )2 is the square of the total mass. The above equations give the
energy, momenta and the velocities of two particles m1 and m2 for which

M → m1 + m2

J. Description of Nucleus-Nucleus Collisions in terms of Light-Cone Variables

In relativistic nucleus-nucleus collisions, it is convenient to use kinematic variables which

take simple forms under Lorentz transformations for the change of frame of reference. A few
of them are the light cone variables x+ and x− , the rapidity and pseudorapidity variables,
y and η. A particle is characterized by its 4-momentum, pµ = (E, p). In fixed target
and collider experiments where the beam(s) define reference frames, boosted along their
Relativistic Kinematics Raghunath Sahoo 18

FIG. 5: Description of heavy-ion collisions in one space (z) and one time (t) dimension.

direction, it is important to express the 4-momentum in terms of more practical kinematic

variables.
Figure 5 shows the collision of two Lorentz contracted nuclei approaching each other with
velocities nearly equal to velocity of light. The vertical axis represents the time direction
with the lower half representing time before the collision and the upper half, time after the
collision. The horizontal axis represents the spatial direction. Both the nuclei collide at
(t, z) = (0, 0) and then the created fireball expands in time going through various processes
2 2
√ particles freeze-out and reach the detectors. The lines where t − z = 0
till the created
(note that t2 − z 2 ≡ τ , τ being the proper time of the particle) along the path of the
colliding nuclei define the light cone. The upper part of the light-cone, where t2 − z 2 > 0,
is the time-like region. In nucleus-nucleus collisions, particle production occurs in the upper
half of the (t, z)-plane within the light-cone. The region outside the light cone for which
t2 − z 2 < 0 is called space-like region. The space-time rapidity is defined as
 
1 t+z
ηs = ln (71)
2 t−z
It could be seen that ηs is not defined in the space-like region. It takes the value of positive
and negative infinity along the beam directions for which t = ±z respectively. A particle is
”light-like” along the beam direction. Inside the light-cone which is time-like, ηs is properly
defined.
For a particle with 4-momentum p (p0 , pT , pz ), the light-cone momenta are defined by

p+ = p 0 + pz (72)
p− = p 0 − pz (73)

p+ is called “forward light-cone momentum” and p− is called “backward light-cone momen-

tum”.
For a particle travelling along the beam direction, has higher value of forward light-cone
Relativistic Kinematics Raghunath Sahoo 19

momentum and travelling opposite to the beam direction has lower value of forward light-
cone momentum. The advantages of using light-cone variables to study particle production
are the following.
1. The forward light-cone momentum of any particle in one frame is related to the forward
light-cone momentum of the same particle in another boosted Lorentz frame by a constant
factor.
2. Hence, if a daughter particle c is fragmenting from a parent particle b, then the ratio of
the forward light-cone momentum of c relative to that of b is independent of the Lorentz
frame.
Define
pc0 + pcz
x+ = b (74)
p0 + pbz
c+
= .
b+
The forward light-cone variable x+ is always positive because c+ can’t be greater than b+ .
Hence the upper limit of x+ is 1. x+ is Lorentz invariant.
3. The Lorentz invariance of x+ provides a tool to measure the momentum of any particle
in the scale of the momentum of any reference particle.

K. Pictorial Representation of Detector System

FIG. 6: A schematic decomposition of particle momentum p~ (in CM frame) into parallel and
longitudinal components. Note the angle of inclination θ of p~ and the azimuthal angle φ of p⊥ [3].

In an collider experimental environment, a particle is emitted from the collision point

making a polar angle θ with the collision point. When the momenta of the particles are
determined by a tracking detector, the (pseudo)-rapidity and azimuthal angles are given by
y = tanh−1 vz = tanh−1 pEz = tanh−1 √ 2 p2 z 2 2
px +py +pz +m0

py !
and φ = tan−1 .
px
Relativistic Kinematics Raghunath Sahoo 20

The polar angle θ is given by

pz |p~T |
θ = cos−1 = tan−1 .
|~p| pz
Pictorially, these are shown in the Figure 6. A detector plane is spanned by (η, φ), with
η decreasing while going away from the beam axis in annular rings and φ is scanned making
an angle with the beam axis and increasing it anti-clock-wise, as is shown in the picture.

L. The Rapidity Variable

One of the important tasks in hand is to introduce key kinematic variables that relate
particle momentum to the dynamics that is occurring in heavy-ion collisions. It is essen-
tially the convenience of working in center-of-momentum system, we need to introduce the
observable rapidity. To do so, let’s proceed as follows: The co-ordinates along the beam
line (conventionally along the z-axis) is called longitudinal and perpendicular to it is called
transverse (x-y). The 3-momentum can be decomposed into the longitudinal (pz ) and the
transverse (pT ), pT being a vector quantity which is invariant under a Lorentz boost along
the longitudinal direction. The variable rapidity “y” is defined by
 
1 E + pz
y = ln (75)
2 E − pz
 
E + pz
= ln (76)
mT

It is a dimensionless quantity related to the ratio of forward light-cone to backward light-

cone momentum. The rapidity changes by an additive constant under longitudinal Lorentz
boosts.
For a free particle which is on the mass shell (for which E 2 = p2 + m2 ), the 4-momentum
has only three degrees of freedom and can be represented by (y, pT ). (E, pT ) could be
expressed in terms of (y, pT ) as

E = mT cosh y (77)
pz = mT sinh y (78)

mT being the transverse mass which is defined as

m2T = m2 + p2T . (79)

The advantage of rapidity variable is that the shape of the rapidity distribution remains
unchanged under a longitudinal Lorentz boost. When we go from CMS to LS, the rapidity
distribution is the same, with the y-scale shifted by an amount equal to ycm . This is shown
below.

1. Rapidity of Center of Mass in the Laboratory System

√
The total energy in √
the CMS system
√ is Ecm = s. The energy and momentum of the
CMS in the LS are γcm s and βcm γcm s respectively. The rapidity of the CMS in the LS
Relativistic Kinematics Raghunath Sahoo 21

is
 √ √ 
1 γcm s + βcm γcm s
ycm = ln √ √
2 γcm s − βcm γcm s
 
1 1 + βcm
= ln (80)
2 1 − βcm

It is a constant for a particular Lorentz transformation.

2. Relationship between Rapidity of a particle in LS and rapidity in CMS

The
 rapidities
 of a particle
 ∗ in∗ the LS and CMS of the collision are respectively, y =
1
2
ln E−pz and y = 2 ln E
E+pz ∗ 1 +pz
E ∗ −p∗z
. For a particle travelling in longitudinal direction, the
Lorentz transformation of its energy and momentum components give
 ∗    
E γ −γβ E
∗ = · , PT∗ = PT (81)
PL −γβ γ PL

where PL and PT are the longitudinal and transverse components of P~ , which are parallel
and perpendicular to β, respectively. Hence, the inverse Lorentz transformations on E and
pz give

γ(E ∗ + βp∗z ) + γ(βE ∗ + p∗z )

 
1
y = ln
2 γ(E ∗ + βp∗z ) − γ(βE ∗ + p∗z )
 ∗
E + p∗z
  
1 1 1+β
= ln + ln (82)
2 E ∗ − p∗z 2 1−β
⇒ y = y ∗ + ycm . (83)

Hence the rapidity of a particle in the laboratory system is equal to the sum of the ra-
pidity of the particle in the center of mass system and the rapidity of the center of mass
in the laboratory system. It can also be state that the rapidity of a particle in a moving
(boosted) frame is equal to the rapidity in its own rest frame minus the rapidity of the
moving frame. In the non-relativistic limit, this is like the subtraction of velocity of the
moving frame. However, this is not surprising because, non-relativistically, the rapidity y
is equal to longitudinal velocity β. Rapidity is a relativistic measure of the velocity. This
simple property of the rapidity variable under Lorentz transformation makes it a suitable
choice to describe the dynamics of relativistic particles. The simple shape invariance nature
of rapidity spectra brings its importance in the analysis of particle production in nuclear
collisions. For instance, in fixed-target experiments, we can study particle spectra using y as
a variable without making an explicit transformation to the CM frame of reference and from
the rapidity spectra, we deduce the point of symmetry corresponding to the CM rapidity. In
symmetric collisions with fixed targets, the CM frame is located in the middle between the
rapidities of the projectile and target i.e. ycm = yproj /2. In this case, the particle rapidity
spectrum must be symmetric around ycm . This allows for complementing measured particle
spectra: if these are available for, e.g. y ≥ ycm , a reflection at the symmetry point ycm gives
us the part of the spectrum with y ≤ ycm , for which an experimental measurement is absent.
Relativistic Kinematics Raghunath Sahoo 22

Beam Rapidity
9

3
10 102 10
√ sNN (GeV)

FIG. 7: A plot of beam rapidity as a function of center of mass energy.

3. Relationship between Rapidity and Velocity

Consider a particle travelling in z-direction with a longitudinal velocity β. The energy E

and the longitudinal momentum pz of the particle are

E = γm (84)
pz = γβm (85)
pz = βE (86)

where m is the rest mass of the particle. Hence the rapidity of the particle travelling in
z-direction with velocity β is
   
1 E + pz 1 γm + γβm
yβ = ln = ln
2 E − pz 2 γm − γβm
 
1 1+β
= ln (87)
2 1−β
Note here that yβ is independent of particle mass. A particle is said to be relativistic in
nature if γ >> 1, β ≈ 1 and E ≈ p. In other words, the energy of the particle is much
higher than its rest mass i.e. E >> m0 and hence the energy and total momentum of the
particle are comparable. In the non-relativistic limit when β is small, expanding yβ in terms
of β leads to
yβ = β + O(β 3 ) (88)
Thus the rapidity of the particle is the relativistic realization of its velocity.

4. Beam Rapidity

We know,
E = mT cosh y, pz = mT sinh y and m2T = m2 + p2T .
Relativistic Kinematics Raghunath Sahoo 23

For the beam particles, pT = 0.

Hence, E = mb cosh yb and pz = mb sinh yb ,
where mb and yb are the rest mass and rapidity of the beam particles.

yb = cosh−1 (E/mb ) (89)

Note that Eq. 89 is used for fixed target experiment with E as the energy of individual
beam particles, neglecting the rest mass. For example, for SPS beam energy of 158 AGeV,
E = 158 GeV. This leads to yb = 5.82 in fixed target framework. Now when we convert Eq.
89 for a collider system, we get

yb = cosh−1 (E/mb )
√ 
−1 sN N
= cosh (90)
2 mp
√
⇒ yb = ln( sN N /mp ) (91)

Now, for the above SPS beam energy of 158 AGeV, the equivalent center of mass energy
√
per nucleon is, sNN = 17.3 GeV. By using Eq. 91, we get yb = 2.91. This makes sense to
us, as in a collider experiment both the beams come in opposite directions with equal beam
rapidities.
Further,
yb = sinh−1 (pz /mb ) (92)
In terms of the beam velocity, beam rapidity can be expressed as

yb = tanh−1 (pz /E) = tanh−1 β (93)

√
Here mp is the mass of a proton. Note that the beam energy E = sN N /2 for a symmetric
collider.
√
Furthermore, it could be shown that: yb = ∓ ln( sN N /mp ) = ∓ ymax
√
Example For the nucleon-nucleon center of mass energy sN N = 9.1 GeV, the beam
rapidity yb = cosh−1 2×0.938
9.1


= 2.26
For p+p collisions
 with lab momentum 100 GeV/c,
−1 pz −1 100


yb = sinh mb
= sinh 0.938
= 5.36
and for Pb+Pb collisions at SPS with lab energy 158 AGeV, yb = 2.91.

√
TABLE I: Table of collision energy, sN N Vs the beam rapidity, yb .
√
sN N (GeV) 7.7 9.1 11.5 17.3 19.6 27.0 39.0 62.4 130 200 900 2360 2760 5520 7000 14000
yb 2.10 2.27 2.50 2.91 3.03 3.35 3.72 4.19 4.93 5.36 6.86 7.83 7.98 8.68 8.91 9.61
Relativistic Kinematics Raghunath Sahoo 24

5. Rapidity of the CMS in terms of Projectile and Target Rapidities

Let us consider the beam particle “b” and the target particle “a”.
bz = mT sinh yb = mb sinh yb . This is because pT of beam particles is zero. Hence
yb = sinh−1 (bz /mb ). (94)
The energy of the beam particle in the laboratory frame is
b0 = mT cosh yb = mb cosh yb .
Assuming target particle a has longitudinal momentum az , its rapidity in the laboratory
frame is given by
ya = sinh−1 (az /ma ) (95)
and its energy
a0 = ma cosh ya . (96)
The CMS is obtained by boosting the LS by a velocity of the center-of-mass frame βcm such
that the longitudinal momentum of the beam particle b∗z and of the target particle a∗z are
equal and opposite. Hence βcm satisfies the condition,
a∗z = γcm (az − βcm a0 ) = − b∗z = − γcm (bz − βcm b0 ), where γcm = √ 1 2 . Hence,
1−βcm

az + b z
βcm = . (97)
a0 + b 0
We know the rapidity of the center of mass is
 
1 1 + βcm
ycm = ln (98)
2 1 − βcm
Using equations 97 and 98, we get
 
1 a0 + az + b 0 + b z
ycm = ln . (99)
2 a0 − az + b 0 − b z
Writing energies and momenta in terms of rapidity variables in the LS,
 
1 ma cosh ya + ma sinh ya + mb cosh yb + mb sinh yb
ycm = ln
2 ma cosh ya − ma sinh ya + mb cosh yb − mb sinh yb
ma eya + mb eyb
 
1 1
= (ya + yb ) + ln (100)
2 2 ma eyb + mb eya
For a symmetric collision (for ma = mb ),
1
ycm = (ya + yb ) (101)
2
Rapidities of a and b in the CMS are
1
ya∗ = ya − ycm = − (yb − ya ) (102)
2
1
yb∗ = yb − ycm = (yb − ya ). (103)
2
Given the incident energy, the rapidity of projectile particles and the rapidity of the target
particles can thus be determined. The greater is the incident energy, the greater is the
separation between the projectile and target rapidity.
Relativistic Kinematics Raghunath Sahoo 25

Central Rapidity The region of rapidity mid-way between the projectile and target
rapidities is called central rapidity.
Example In p+p collisions at a laboratory momentum of 100 GeV /c, beam rapidity yb =
5.36, target rapidity ya = 0 and the central rapidity ≈ 2.7.

6. Mid-rapidity in Fixed target and Collider Experiments

In fixed-target experiments (LS), ytarget = 0.

ylab = ytarget + yprojectile = ybeam Hence mid-rapidity in fixed-target experiment is given
by,
LS
ymid = ybeam /2. (104)
In collider experiments (center of mass system),
yprojectile = − ytarget = yCM S = ybeam /2.
Hence, mid-rapidity in CMS system is given by
CM S
ymid = (yprojectile + ytarget )/2 = 0. (105)
This is valid for a symmetric energy collider. The rapidity difference is given by yprojectile −
ytarget = 2yCM S and this increases with energy for a collider as y increases with energy.

7. Light-cone variables and Rapidity

Consider a particle having rapidity y and the beam rapidity is yb . The particle has
forward light-cone variable x+ with respect to the beam particle
pc + pcz
x+ = b0
p0 + pbz
mc
= Tb ey−yb (106)
m
where mcT is the transverse mass of c. Note that the transverse momentum of the beam
particle is zero. Hence,
 
mb
y = yb + ln x+ + ln (107)
mcT
Similarly, relative to the target particle a with a target rapidity ya , the backward light-cone
variable of the detected particle c is x− . x− is related to y by
mcT ya −y
x− = b e (108)
m
and conversely,  
ma
y = ya − ln x− − ln . (109)
mcT
In general, the rapidity of a particle is related to its light-cone momenta by
 
1 p+
y = ln (110)
2 p−
Note that in situations where there is a frequent need to work with boosts along z-direction,
it’s better to use (y, pT ) for a particle rather than using it’s 3-momentum, because of the
simple transformation rules for y and pT under Lorentz boosts.
Relativistic Kinematics Raghunath Sahoo 26

8. The Maximum Accessible Rapidity in an Interaction

 
√
FIG. 8: sN N (vertical axis) of various accelerators as a function of the projectile and target
rapidities seen from the CM frame. Shaded areas: energy ranges accessible at various accelerators
[3].

We know the variable rapidity “y” of a particle is defined by Eqn. 76:

 
1 E + pz
y = ln . (111)
2 E − pz

This corresponds to
pz
tanh(y) = (112)
E
where, pz is the longitudinal momentum along the direction of the incident particle, E is
the energy, both defined for a given particle. The accessible range of rapidities for a given
interaction is determined by the available centre-of-mass energy and all participating parti-
cles’ rest masses. One usually gives the limit for the incident particle, elastically scattered
at zero angle:

|y|max = ln[(E + pz )/m]

= ln[γ + γβ]
p
= ln[γ + γ 2 − 1]
' cosh−1 γ, if γ  1 (113)

with all variables referring to the through-going particle given in the desired frame of
reference (e.g. in the centre of mass). A Lorentz boost β along the direction of the incident
particle adds a constant, ln[γ + γβ], to the rapidity. Rapidity differences, therefore, are
invariant to a Lorentz boost. Statistical particle distributions are flat in y for many physics
∂y
production models. Note here that ∂p z
= 1/E.
Relativistic Kinematics Raghunath Sahoo 27

Furthermore,
Ebeam
γ = (114)
mp
and for symmetric collision,
√ √
sN N sN N
Ebeam = ⇒ γ = (115)
2 2mp
Hence,
√ 
−1 sN N
ymax = cosh
2mp
 
Elab
= cosh−1
Amp
√ 
sN N
⇒ ymax = yb = ln (116)
mp

Note that the maximum accessible rapidity is independent of the collision species for a
symmetric collider and only depends on the center of mass energy.

Examples:
√
(a) For RHIC top energy, sN N = 200 GeV, γ = Ebeam /mp = 106.609. Hence y|max = 5.36.
For LHC, center of mass energy of 5.5 TeV (γ = 2931.768), y|max = 8.67.
lab
(b) ybeam = 5.8 for Elab = 158 AGeV and 4.4 for Elab = 40 AGeV.

M. The Pseudorapidity Variable

Let us assume that a particle is emitted at an angle θ relative to the beam axis. Then
its rapidity can be written as
√ 
 
E+PL m2 +p2 +p cos θ
1
y = 2 ln E−PL = 2 ln √ 2 2
1
. At very high energy, p  m and hence
m +p −p cos θ

 
1 p + p cos θ
y = ln
2 p − p cos θ
= −ln tan θ/2 ≡ η (117)

η is called the pseudorapidity. Hence at very high energy,

y ≈ η = − ln tan θ/2. (118)

In terms of the momentum, η can be re-written as

 
1 |p| + pz
η = ln . (119)
2 |p| − pz
θ is the only quantity to be measured for the determination of pseudorapidity, independent
of any particle identification mechanism. Pseudorapidity is defined for any value of mass,
Relativistic Kinematics Raghunath Sahoo 28

-2

-4
0 30 60 90 120 150 180

[deg]

FIG. 9: A plot of pseudorapidity variable, η as a function of the polar angle, θ.

FIG. 10: As angle increases from zero, pseudorapidity decreases from infinity.

momentum and energy of the collision. This also could be measured with or without mo-
mentum information which needs a magnetic field. A plot of pseudorapidity as a function
of the polar angle, θ is shown in Fig. 9. Table II shows the values of η corresponding to
the polar angle of emission a particle. One speaks of the ”forward” direction in a collider
experiment, which refers to regions of the detector that are close to the beam axis, at high
|η|. The difference in the rapidity of two particles is independent of Lorentz boosts along the
beam axis. Pseudorapidity is odd about θ = 90 degrees. In other words, η(θ) = −η(180−θ).

TABLE II: Table of pseudorapidity, η Vs the polar angle, θ.

θ 0 5 10 20 30 45 60 80 90 100 110 ... 175 180
η ∞ 3.13 2.44 1.74 1.32 0.88 0.55 0.175 0 −0.175 −0.55 ... −3.13 -∞
Relativistic Kinematics Raghunath Sahoo 29

√
FIG. 11: The mid-rapidity dNch /dη for Au+Au collisions at sN N = 200 GeV [6].

1. Change of variables from (y, pT ) to (η, pT )

By equation 119,
s
|p| + pz
eη = (120)
|p| − pz
s
|p| − pz
e−η = (121)
|p| + pz

Adding both of the equations, we get

|p| = pT cosh η (122)

p
pT = |p|2 − p2z . By subtracting the above equations, we get

pz = pT sinh η (123)

Using these equations in the definition of rapidity, we get

"p #
1 p2T cosh2 η + m2 + pT sinh η
y = ln p 2 (124)
2 pT cosh2 η + m2 − pT sinh η

Similarly η could be expressed in terms of y as,

"p #
1 m2T cosh2 y − m2 + mT sinh y
η = ln p 2 (125)
2 mT cosh2 y − m2 − mT sinh y

The distribution of particles as a function of rapidity is related to the distribution as a

function of pseudorapidity by the formula
s
dN m2 dN
= 1− 2 2
. (126)
dηdpT mT cosh y dydpT
Relativistic Kinematics Raghunath Sahoo 30

In the region y  0, the pseudorapidity distribution (dN/dη) and the rapidity distribution
dN dN
(dN/dy) which are essentially the pT -integrated values of dηdp T
and dydpT
respectively, are
approximately the same. In the region y ≈ 0, there is a small “depression” in dN/dη dis-
tribution compared to dN/dy distribution due to the above transformation. At very high
energies where dN/dy has a mid-rapidity plateau, this transformation gives a small dip in
dN/dη around η ≈ 0 (see Figure 11). However, for a massless particle like photon, the
dip in dN/dη is not expected (which is clear from the above equation). Independent of the
frame of reference where η is measured, the difference in the maximum magnitude of dN/dη
appears due to the above transformation. In the CMS, the maximum ofpthe distribution is
located at y ≈ η ≈ 0 and the η-distribution is suppressed by a factor 1 − m2 / < m2T >
with reference to the rapidity distribution. In the laboratory frame, however the maxi-
mum
p is located around half of the beam rapidity η ≈ yb /2 and the suppression factor is
1 − m2 / < m2T > cosh2 (yb /2), which is about unity. Given the fact that the shape of the
rapidity distribution is independent of frame of reference, the peak value of the pseudora-
pidity distribution in the CMS frame is lower than its value in LS. This suppression factor
at SPS energies is ∼ 0.8 − 0.9.
It’s to note here again that the conversion of rapidity to pseudorapidity phase space is
associated with a Jacobian J(y, η), which is given by the right hand side multiplier of Eqn.
126. This depends on the momentum distribution of the produced particles. In the limit
of rest mass of the particles being much smaller than their momenta, J(y, η) = 1. The
value of the Jacobian is smaller at LHC energies, compared to that at RHIC energies, as
the average transverse momentum of particles increases with beam energy. As measured by
√
the PHENIX experiment, for central Au+Au collisions at sN N = 200 GeV, J(y, η) = 1.25.
Whereas, the corresponding measurement of J(y, η) = 1.09 for Pb+Pb central collisions at
√
sN N = 2.76 TeV by the CMS experiment at LHC. Rewriting Eqn. 126 after taking an
integration over pT , one obtains:
dN dN
= v(y) (127)
dη dy
where v is the velocity of the particle. For a hadron of mass m and momentum p, which
emerges at an angle 900 with respect to the beam direction, y = η = 0, the above relationship
becomes,
dN dN
|η=0 = v |y=0 (128)
dη dy
As most of the particles are pions in the final state, with an average momentum of three times
the pion mass, we have v = 0.95. At mid-rapidity, the rapidity-pseudorapidity conversion
hence involves with almost a constant factor. Hence the shape is not affected to a greater
extent. However, when one considers the whole rapidity range, where the particle velocity in
fact becomes a function of the rapidity, the shape of the pseudorapidity distribution (which
is characterized by the width) is different from the rapidity distribution.
Usually, the rapidity spectra are parametrized by the sum of two Gaussian distributions
positioned symmetrically with respect to mid-rapidity [7, 8],
(   2     2 )
dN hN i 1 y − y0 1 y + y0
= √ exp − + exp − (129)
dy 2 2πσ 2 2 σ 2 σ

where hN i , σ and y0 are fit parameters. σ is the width of the rapidity distribution.
Relativistic Kinematics Raghunath Sahoo 31

Landau’s energy dependent Gaussian rapidity distribution is given by [9, 10]

1 dσ dN
=
σin dy dy
N
exp. −y 2 /2L


= 1/2
(130)
(2πL)

with parameters
1
ln s/4m2p


L =
2 √
= ln γ = ln ( sN N /2mp ) (131)

where s ≡ squared total center of mass energy. Comparing Eqns. 129 and 130, we get

√
q
σy = ln( sN N /2mp ) (132)

σy is the width of the rapidity distribution, with mp being the mass of proton. Interest-
ingly, the width of the rapidity distribution is related to the longitudinal flow and velocity
of sound in the medium and hence can probe the equation of state of the produced matter.
However, this is affected by the final state rescattering, which could be understood through
a pT dependent study of σy to disentangle the initial hard scattering from the final state
rescattering. With the assumption that the velocity of sound cs is independent of temper-
ature, the rapidity density in the framework of Landau hydrodynamic model is given by
[11, 12]
1/4
y2
 
dN sN N
= K p exp. − 2 (133)
dy 2π σy2 2σy
where
8 c2s √
σy2 = 4
ln ( sN N /2mp )
3 1 − cs
8 c2s
⇒ σy2 = ln γ (134)
3 1 − c4s

and K ≡ normalization factor. Inverting the above equation for σy2 , one can get [13]
v"
√

u 4 ln √sN N /2mp 2
u
#
−4 ln s N N /2mp
c2s = + t + 1 (135)
3 σy2 3 σy2

For an ideal gas (Landau model prediction) the velocity of sound is, c2s = 1/3.
However, for a gas of hadrons, c2s = 1/5. This implies, the expansion of the system
is slower compared to an ideal gas scenario. The equation of state is given by
Relativistic Kinematics Raghunath Sahoo 32

FIG. 12: Left: The width of rapidity distributions of π − in central Pb+Pb (Au+Au) collisions as
a function the beam rapidity. The dotted line indicates Landau model predictions with c2s = 1/3,
while the full line shows a linear fit to data points. Right: Speed of sound as a function of beam
energy showing a softest point at Ebeam = 30A GeV [13].

∂P
= c2s (136)
∂
where, P is the pressure and  is the energy of the system under consideration. It should be
noted here that when the expansion of the matter proceeds as longitudinal and superimposed
transverse expansions, a rarefaction wave moves radially inwards with the velocity of sound.
The velocity of sound in the medium formed in heavy-ion collisions, when studied as a
function of collision (beam) energy, shows the softest point occurring around Ebeam = 30
AGeV. This could be a signature of deconfinement transition [13].
Relativistic Kinematics Raghunath Sahoo 33

Notation [14]
Coordinate Space Momentum Space
(t, x, y, z) → (τ, r, φ, η) (E, px , py , pz ) → (mT , p~T , y)
√ p
τ = t2 − z 2 = proper time mT = E 2 − p2z = transverse mass

η = tanh−1 zt = space-time rapidity y = tanh−1 pEz = tanh−1 vz = rapidity

Conversely, Conversely,

t = τ cosh η, z = τ sinh η E = mT cosh y, pz = mT sinh y

N. The Invariant Yield

The rapidity variable has the useful property that it transforms linearly under a Lorentz
transformation so that the invariant differential single particle inclusive cross section be-
comes:

Ed3 σ Ed3 σ d3 σ
= = , (137)
dp3 pT dpT dpL dφ pT dpT dydφ
where
dpL
dy = (138)
E
3
First we proceed to show dEp is Lorentz invariant. The differential of Lorentz boost in
longitudinal direction is given by
dp∗z = γ(dpz − βdE). (139)
Taking the derivative of the equation E 2 = p2 + m2 , we get
EdE = pz dpz . (140)
Using equations 139 and 140 we get
pz dpz
dp∗z = γ(dpz − β )
E
dpz ∗
= E . (141)
E
As pT is Lorentz invariant, multiplying pT on both the sides and re-arranging gives
d3 p∗ d2 pT dpz d3 p
= = . (142)
E∗ E E
d3 p
In terms of experimentally measurable quantities, E
could be expressed as
d3 p
= dpT dy
E
= pT dpT dφdy (143)
= mT dmT dφdy. (144)
Relativistic Kinematics Raghunath Sahoo 34

3 3
The Lorentz invariant differential cross-section Ed
dp3
σ
= Ed N
dp3
is the invariant yield. In terms
of experimentally measurable quantities this could be expressed as
Ed3 σ 1 d3 N
=
dp3 mT dmT dφdy
1 d2 N
=
2π mT dmT dy
1 d2 N
= . (145)
2π pT dpT dy
To measure the invariant yields of identified particles equation 145 is used experimentally.

O. Inclusive Production of Particles and the Feynman Scaling variable xF

A reaction of type
beam + target −→ A + anything
where A is known is called an “inclusive reaction”. The cross-section for particle production
could be written separately as functions of pT and pL :

σ = f (pT )g(pL ). (146)

This factorization is empirical and convenient because each of these factors has simple
parametrizations which fit well to experimental data.

Similarly the differential cross-section could be expressed by

d3 σ d2 σ dσ
= (147)
dp3 p2T dpL
Define the variable
p∗L
xF =
p∗L (max)
2p∗
⇒ xF = √ L (148)
s

xF is called the Feynman scaling variable: longitudinal component of the cross-section

when measured in CMS of the collision, would scale i.e. would not depend on the energy
√
s. This is the fraction of maximum allowed longitudinal momentum (−1 ≤ xF ≤ 1)
carried by the particle in the CMS. This is used in comparing the shapes of particle
distributions at different collision energies near the projectile or target√rapidity. Instead of
dσ
, dσ is measured which wouldn’t depend on energy of the reaction, s. This Feynman’s
dp∗L dxF
assumption is valid approximately. This variable is usually used to compare particle
distribution at different collision energies.

The differential cross-section for the inclusive production of a particle is then written as
d3 σ
= F (s, xF , pT ) (149)
dxF d2 pT
Relativistic Kinematics Raghunath Sahoo 35

Feynman’s assumption that at high energies the function F (s, xF , pT ) becomes asymptot-
ically independent of the energy means:

lims→∞ F (s, xF , pT ) = F (xF , pT ) = f (pT ) g(xF )

According to Feynman, the mean number of particles rises logarithmically in the
asymptotic limit of large energies. This in fact applies to any kind of particles:
√
< N > ∝ ln W ∝ ln s,
√
where W = s/2, is the beam energy of a symmetric collider. However, these conclusions
are based on phenomenological arguments about the exchange of quantum numbers between
the colliding particles. It was assumed that the number of particles with a given mass and
transverse momentum (pT ) in a longitudinal interval pz depends on the energy
E = E(pZ ) as
dN 1
∼ (150)
dpz E
This means the among all the produced particles, the probability of finding a particle of
kind i with transverse momentum pT , mass m and longitudinal momentum pz is of the form:

dpz 2
fi (pT , xF = pz /W ) d pT (151)
p E
with E = m2 + p2T + p2z being the total energy of the particle under discussion. The
function fi (pT , xF ) denotes the particle distribution. Let’s derive Eqn. 151 to have a hand
on experience.

Rewriting Eqn. 151 in the form of invariant cross section

1 d3 σ
E = fi (pT , xF ) (152)
σ dpz d2 pT
fi factorizes approximately (found experimentally) and a normalization gi is chosen such
that
Z Z
2
fi (pT , xF ) d pT = fi (xF ) gi (pT ) d2 pT
= f (xF ) (153)
Here Z
gi (pT ) d2 pT = 1 (154)
Now, integrating Eqn. 152 and applying Eqn. 153, we get
d3 σ d3 p
Z
1
E = <N >
σ dpz d2 pT E
d3 p
Z
= fi (pT , xF )
E
Z
dpz
= fi (xF ) p (155)
W x2F + m2T
2
Relativistic Kinematics Raghunath Sahoo 36

On the left hand side we have used the definition of invariant cross section with the average
particle multiplicity < N >, and for mT an effective average pT is used with xF = pz /W .
Hence Z +1
dxF
<N > = fi (xF ) q , (156)
−1 2 m2T
xF + W 2
dpz
with dxF = W
.

The integral is symmetric because fi (xF ) is symmetric for collisions of identified particles.
For asymmetric collision systems, the integration could be performed separately for negative
and positive xF and yields the same result.

fi (xF ) ≤ B is finite and bounded due to energy conservation. Feynman assumed that
for xF = 0 a finite limit is reached.

pz pz
xF → 0 ⇒ = → 0
W Elab
⇒ Elab → ∞

Hence

Z 1 Z 1
dxF dxF
<N > = 2 fi (xF ) q ≤ 2 Bq
0 m2T 0 m2T
x2F + W2
x2F + W2
" r #
m2T 1
= 2B ln xF + x2F +

W2

0
" r #
m2T m 
T
= 2B ln 1 + 1 + 2 − 2B ln (157)
W W

In the limit W → ∞, the first term of the above equation could be shown to be constant
and the second term is proportional to ln W . hence, Feynman scaling tells that the average
total multiplicity scales as:

√
< N > ∝ ln W ∝ ln s (158)
√
If one considers
√ the maximum reachable rapidity in a collision to increase like ln s
(ymax ∼ ln s) and √ in addition, particles are evenly distributed in rapidity, it follows that
dN
dy
is independent of s:

dN
= Constant (159)
dy
Feynman’s Assumption: fi (pT , xF ) which denotes the particle distribution, becomes
independent of W at high energies. This assumption is known as Feynman Scaling and fi
Relativistic Kinematics Raghunath Sahoo 37

is called the scaling function or Feynman function. The variable

√
xF = pz /W = 2pz / s (160)
is called Feynman-x.

Feynman-x is the ratio of the longitudinal momentum of the particle to the total energy
of the incident particle. Note:

1. dNdy
= Constant ⇒ Height of the rapidity distribution around mid-rapidity (the
√
so-called plateau) is independent of collision energy, s.

2. Equivalently, if Feynman scaling holds good, the pseudorapidity density of charged

particles at mid-rapidity i.e. dN
dη
(η = 0) is approximately constant.

Scaling properties of various distributions can also be studied in terms of the scaled
rapidity:

z = y ∗ /ymax
∗
. (161)
The two scaled variables: xF and z emphasize different kinematic regions: the detailed
structure of the central part of the distribution (i.e. large emission angles) can be better
seen using xF , while the far ”wings” (i.e. small angles) using z.

P. What is a pT or mT Spectrum?

1 d2 N
pT spectrum: 2πpT dy dpT
Vs pT
1 d2 N
mT spectrum: 2πmT dy dmT
Vs mT or (mT − m0 )

The invariant yield when plotted as a function of pT or mT or (mT − m0 ) is called pT or

mT spectrum, respectively. Experimentally at kinetic freeze-out when the elastic collisions
between the final state particles almost cease to happen (in other words the particle mean
free path becomes higher than the system size: size of the produced fireball) then the pT or
mT -spectrum is frozen, which carries the kinetic freeze-out properties of the system. Recall
here that pT dpT = mT dmT .
The distribution of particles as a function of pT is called pT -distribution. Mathematically,
dN dN
= (162)
dpT 2π |pT |d|pT |

where dN is the number of particles in a particular pT -bin. People usually plot pTdN
dpT
as a
function of pT taking out the factor 1/2π which is a constant. Here pT is a scalar quantity.
The low-pT part of the pT -spectrum is well described by an exponential function having
thermal origin given by Eqn. (163). However, a QCD-inspired power-law function (given by
Eqn. (164)) seems to provide a better description of the high pT (∼ > 3 GeV/c) region. To
describe the whole range of the pT -spectrum, one uses the Levy function given by Eqn.(165)
Relativistic Kinematics Raghunath Sahoo 38

which has an exponential part to describe low-pT and a power-law function to describe the
hight-pT part which is dominated by hard scatterings (high momentum transfer at early
times of the collision).
1 d2 N −mT
= Ae T , (163)
2πpT dy dpT
−n
d2 N

1 pT
= B 1+ , (164)
2πpT dy dpT p0

p !−n
1 d2 N dN (n − 1)(n − 2) p2T + m20 − m0
= × 1+ , (165)
2πpT dy dpT dy 2πnC[nC + m0 (n − 2)] nC

where A, T, B, p0 , n, dN
dy
, C, and m0 are fit parameters [15]. The inverse slope parameter
of pT -spectra is called the effective temperature (Tef f ), which has a thermal contribution
because of the random kinetic motion of the produced particles and a contribution from
the collective motion of the particles. This will be described in details in the section of
freeze-out properties and how to determine the chemical and kinetic freeze-out temperatures
experimentally.

If we look deeper into Eqn. 163, we expect this for a 2-dimensional classical thermalized
fluid at rest. We recall here that Boltzmann distribution ∝ exp.(−βE), where β ≡ 1/T
is the inverse temperature. We use a semi-logarithmic plot (y-axis logarithmic scale) of
N exp. (−βmT ) vs (mT − m0 ) which happens to be a straight line with slope −β, from
which the effective temperature is extracted. A semi-log plot of N exp. (−βmT ) vs pT is an
approximate straight line if pT  m0 .
The most important parameter is then the mean pT which carries the information of the
effective temperature of the system. Experimentally, hpT i is studied as a function of dNdηch
which is the measure of the entropy density of the system. This is like studying the temper-
ature as a function of entropy to see the signal of phase transition. The phase transition is
of 1st order if a plateau is observed in the spectrum signalling the existence of latent heat of
the system (like liquid-vapour phase transition). This was first proposed by L. Van Hove [16].

The average of any quantity A following a particular probability distribution f (A) can
be written as R
A f (A) dA
hAi = R . (166)
f (A) dA
Similarly,
R∞
pT ( dN ) dpT
hpT i =
0
R ∞ dNdpT
0
( dpT ) dpT
R∞
0
pT dpT pT ( pTdNdpT
)
= R∞ dN
0
pT dpT ( pT dpT )
R∞
pT dpT pT f (pT )
= 0R∞ (167)
0
pT dpT f (pT )
Relativistic Kinematics Raghunath Sahoo 39

where 2π pT dpT is the phase space factor and the pT -distribution function is given by
dN dN
f (pT ) = = . (168)
dpT pT dpT
Example Experimental data on pT -spectra are sometimes fitted to the exponential Boltz-
mann type function given by
1 dN
f (pT ) = ' C e−mT /Tef f . (169)
pT dpT
The hmT i could be obtained by
R∞
pT dpT mT exp.(−mT /Tef f )
hmT i = 0 R ∞
0
pT dpT exp.(−mT /Tef f )
2Tef f + 2m0 Tef f + m20
2
= (170)
m0 + Tef f
(Tef f )2
⇒ hmT i = Tef f + m0 + (171)
m0 + Tef f
where m0 is the rest mass of the particle. It can be seen from the above expression that for
a massless particle
hmT i = hpT i = 2Tef f . (172)
This also satisfies the principle of equipartition of energy which is expected for a massless
Boltzmann gas in equilibrium. However, in experiments the lower (higher) limit of pT is a
finite quantity. In that case the integration will involve an incomplete gamma function.

Q. How is the radial flow measured from pT -spectra?

In central heavy-ion collisions, radial flow is supposed to play a vital role in the thermo-
dynamic expansion of the produced fireball. Radial flow is related to the initial pressure
produced just after the collision. This could be extracted from the analysis of the trans-
verse momentum spectra. Assuming a thermalized non-relativistic plasma (for simplicity),
particle velocity [14]

~v = ~vf low + ~vth , (173)

where ~vf low ≡ transverse velocity of the expanding fluid, which is independent of the
particle species and is the collective component of ~v .

~vf low ≡ thermal component of ~v , which is generated due to random thermal motion of
the quanta of the system.

Hence for a particle of mass m0

   
1 1 1
m0 v 2 = 2
m0 vf low + 2
m0 vth
2 2 2
1 3
= m0 vf2low + kT, (174)
2 2
Relativistic Kinematics Raghunath Sahoo 40

where T is the temperature of the fluid. Hence the average kinetic energy (K.E) is given by

1 3
hK.E.i = m0 vf2low + kT
2 2
3 1 3
⇒ kTef f = m0 vf2low + kTth
2 2 2
1
⇒ Tef f = Tth + m0 vf2low (taking k = 1) (175)
3
Because the final state particle are measured at freeze out (after they stream out to
reach the detectors), the extracted values of ~vf low and T correspond to the instant of freeze
out.

Taking k = 1, in 2-dimension:

1
Tef f = Tth + m0 vf2low , (176)
2
and in 1-dimension:

Tef f = Tth + m0 vf2low (177)

Note that these formulae are used essentially when the spectra is well described by
an exponential function (low-pT regime). However, when one goes to high-pT regime,
the following formula could be used for extracting the radial flow from the pT or mT -spectra.

s
1 + vf low
Tef f = Tth (178)
1 − vf low

[1] Introduction to High Energy Physics, D. H. Perkins, Cambridge University Press, 4th edition
(2000).
[2] Relativistic Kinematics, R. Hagedorn, W. A. Benjamin, Inc (1964).
[3] Hadrons and Quark-Gluon Plasma, J. Letessier and J. Rafelski, Cambridge (2005).
[4] Introduction to High-Energy Heavy-Ion Collisions, C-Y. Wong, World Scientific (1994).
[5] M. Kliemant, R. Sahoo, T. Schuster and R. Stock, Lect. Notes Phys. 785, 23 (2010),
[arXiv:0809.2482 [nucl-ex]].
[6] B.B. Back et al., (PHOBOS Collaboration), Nucl. Phys. A 757 28 (2005).
[7] C. Alt et al. (NA49 Collaboraion), Phys. Rev. C 77, 024903 (2008).
[8] S.V. Afanasiev et al. (NA49 Collaboration), Phys. Rev. C 66, 054902 (2002).
[9] P. Carruthers and Minh Duong-van, Phys. Rev. D 8, 859 (1973).
Relativistic Kinematics Raghunath Sahoo 41

[10] C.Y. Wong, Phys. Rev. C 78, 054902 (2008).

[11] P. Carruthers, Annals N.Y. Acad.Sci 229, 91 (1974).
[12] E.V. Shuryak, Yad. Fiz 16, 395 (1972).
[13] H. Petersen and M. Bleicher, PoS CPOD2006 (2006) 025.
[14] R.S. Bhalerao, Lecture Notes in SERC School on Nuclear Physics, IoP, Bhubaneswar (2006)
[15] B.I. Abelev et al., (STAR Collaboration), Phys. Rev. C 75, 064901 (2007).
[16] L. van Hove, Phys. Lett. B 118, 138 (1982).
[17] J. Gosset et al., Phys. Rev. C 16, 629 (1977).
[18] J. Adams et al., (STAR Collaboration), Phys. Rev. C 70, 054907 (2004).
[19] F. Karsch, Prog. Part. Nucl. Phys. 62, 503 (2009).