39.

Kinematics 1

39. KINEMATICS
Revised January 2000 by J.D. Jackson (LBNL) and June 2008 by
D.R. Tovey (Sheeld).
Throughout this section units are used in which  = c = 1. The
following conversions are useful: c = 197.3 MeV fm, (c)2 = 0.3894
(GeV)2 mb.

39.1. Lorentz transformations

The energy E and 3-momentum p of a particle of mass m form a 4-vector p = (E, p)
whose square p2 E 2 |p|2 = m2 . The velocity of the particle is = p/E. The energy
and momentum (E , p ) viewed from a frame moving with velocity f are given by
  
 
E
E
f
f f
(39.1)
=
, pT = pT ,

p
p
f f
f
where f = (1 f2 )1/2 and pT (p ) are the components of p perpendicular (parallel) to
f . Other 4-vectors, such as the space-time coordinates of events, of course transform in
the same way. The scalar product of two 4-momenta p1 p2 = E1 E2 p1 p2 is invariant
(frame independent).

39.2. Center-of-mass energy and momentum

In the collision of two particles of masses m1 and m2 the total center-of-mass energy
can be expressed in the Lorentz-invariant form

1/2
2
2
,
Ecm = (E1 + E2 ) (p1 + p2 )
1/2

,
(39.2)
= m21 + m22 + 2E1 E2 (1 1 2 cos )
where is the angle between the particles. In the frame where one particle (of mass m2 )
is at rest (lab frame),
(39.3)
Ecm = (m21 + m22 + 2E1 lab m2 )1/2 .
The velocity of the center-of-mass in the lab frame is
where plab p1 lab and

cm = plab /(E1 lab + m2 ) ,

(39.4)

cm = (E1 lab + m2 )/Ecm .

(39.5)

The c.m. momenta of particles 1 and 2 are of magnitude

m2
pcm = plab
.
(39.6)
Ecm
For example, if a 0.80 GeV/c kaon beam is incident on a proton target, the center of mass
energy is 1.699 GeV and the center of mass momentum of either particle is 0.442 GeV/c.
It is also useful to note that
Ecm dEcm = m2 dE1 lab = m2 1 lab dplab .

K. Nakamura et al., JPG 37, 075021 (2010) (http://pdg.lbl.gov)

July 30, 2010

14:36

(39.7)

39. Kinematics

39.3. Lorentz-invariant amplitudes

The matrix elements for a scattering or decay process are written in terms of an
invariant amplitude iM . As an example, the S-matrix for 2 2 scattering is related
to M by
p1 p2 |S| p1 p2  = I i(2)4 4 (p1 + p2 p1 p2 )
M (p1 , p2 ; p1 , p2 )
.

(2E1 )1/2 (2E2 )1/2 (2E1 )1/2 (2E2 )1/2

(39.8)

The state normalization is such that

p |p = (2)3 3 (p p ) .

(39.9)

39.4. Particle decays

The partial decay rate of a particle of mass M into n bodies in its rest frame is given
in terms of the Lorentz-invariant matrix element M by
d =

(2)4
|M |2 dn (P ; p1 , . . . , pn ),
2M

(39.10)

where dn is an element of n-body phase space given by

4

dn (P ; p1 , . . . , pn ) = (P

n


pi )

i=1

n

i=1

d3 pi
.
(2)3 2Ei

(39.11)

This phase space can be generated recursively, viz.

dn (P ; p1 , . . . , pn ) = dj (q; p1 , . . . , pj )
(39.12)
dnj+1 (P ; q, pj+1 , . . . , pn )(2)3 dq 2 ,
2

j

j
where q 2 = ( i=1 Ei )2 i=1 pi . This form is particularly useful in the case where a
particle decays into another particle that subsequently decays.
39.4.1. Survival probability : If a particle of mass M has mean proper lifetime
(= 1/) and has momentum (E, p), then the probability that it lives for a time t0 or
greater before decaying is given by
P (t0 ) = et0 / = eMt0 /E ,

(39.13)

and the probability that it travels a distance x0 or greater is

P (x0 ) = eMx0 /|p| .

July 30, 2010

14:36

(39.14)

39. Kinematics 3
p1, m1
P, M
p2, m2
Figure 39.1: Denitions of variables for two-body decays.
39.4.2. Two-body decays :
In the rest frame of a particle of mass M , decaying into 2 particles labeled 1 and 2,
M 2 m22 + m21
,
E1 =
2M
|p1 | = |p2 |

 2

 1/2
M (m1 + m2 )2 M 2 (m1 m2 )2
,
=
2M

(39.15)

(39.16)

and

|p |
1
|M |2 12 d ,
(39.17)
2
32
M
where d = d1 d(cos 1 ) is the solid angle of particle 1. The invariant mass M can be
determined from the energies and momenta using Eq. (39.2) with M = Ecm .
d =

39.4.3.

Three-body decays :

p1, m1
p2, m2

P, M

p3, m3
Figure 39.2: Denitions of variables for three-body decays.
Dening pij = pi + pj and m2ij = p2ij , then m212 + m223 + m213 = M 2 + m21 + m22 + m23
and m212 = (P p3 )2 = M 2 + m23 2M E3 , where E3 is the energy of particle 3 in the
rest frame of M . In that frame, the momenta of the three decay particles lie in a plane.
The relative orientation of these three momenta is xed if their energies are known. The
momenta can therefore be specied in space by giving three Euler angles (, , ) that
specify the orientation of the nal system relative to the initial particle [1]. Then
d =

1
1
|M |2 dE1 dE2 d d(cos ) d .
5
(2) 16M
July 30, 2010

14:36

(39.18)

39. Kinematics

Alternatively
1
1
|M |2 |p1 | |p3 | dm12 d1 d3 ,
5
(2) 16M 2

d =

(39.19)

where (|p1 |, 1 ) is the momentum of particle 1 in the rest frame of 1 and 2, and 3 is the
angle of particle 3 in the rest frame of the decaying particle. |p1 | and |p3 | are given by


|p1 |

and


|p3 | =


 1/2
m212 (m1 + m2 )2 m212 (m1 m2 )2
,
2m12

M 2 (m12 + m3 )2

M 2 (m12 m3 )2
2M

(39.20a)

 1/2
.

(39.20b)

[Compare with Eq. (39.16).]

If the decaying particle is a scalar or we average over its spin states, then integration
over the angles in Eq. (39.18) gives
d =

1
1
|M |2 dE1 dE2
3
(2) 8M

1
1
|M |2 dm212 dm223 .
3
(2) 32M 3

(39.21)

This is the standard form for the Dalitz plot.

39.4.3.1. Dalitz plot: For a given value of m212 , the range of m223 is determined by its
values when p2 is parallel or antiparallel to p3 :
(m223 )max =
(E2

+ E3 )2



E22

m22

E32

m23

2
,

(39.22a)

(m223 )min =

2


2
2
2
2
2
(E2 + E3 )
E2 m2 + E3 m3
.

(39.22b)

Here E2 = (m212 m21 + m22 )/2m12 and E3 = (M 2 m212 m23 )/2m12 are the energies
of particles 2 and 3 in the m12 rest frame. The scatter plot in m212 and m223 is called a
Dalitz plot. If |M |2 is constant, the allowed region of the plot will be uniformly populated
with events [see Eq. (39.21)]. A nonuniformity in the plot gives immediate information
on |M |2 . For example, in the case of D K, bands appear when m(K) = mK (892) ,
reecting the appearance of the decay chain D K (892) K.

July 30, 2010

14:36

39. Kinematics 5
10

(m1+m2) 2
(Mm1) 2

8
2
m23
(GeV 2)

(m223)max

(Mm3) 2

4
2 )
(m23
min

(m2+m3) 2

2
3
2 (GeV 2)
m12

Figure 39.3: Dalitz plot for a three-body nal state. In this example, the state
is + K 0 p at 3 GeV. Four-momentum conservation restricts events to the shaded
region.
39.4.4.

Kinematic limits :

39.4.4.1. Three-body decays: In a three-body decay (Fig. 39.2) the maximum of |p3 |,
[given by Eq. (39.20)], is achieved when m12 = m1 + m2 , i.e., particles 1 and 2 have
the same vector velocity in the rest frame of the decaying particle. If, in addition,
m3 > m1 , m2 , then |p3 |max > |p1 |max , |p2 |max . The distribution of m12 values possesses
an end-point or maximum value at m12 = M m3 . This can be used to constrain the
mass dierence of a parent particle and one invisible decay product.
39.4.4.2.

Sequential two-body decays:

Figure 39.4: Particles participating in sequential two-body decay chain. Particles

labeled 1 and 2 are visible while the particle terminating the chain (a) is invisible.

July 30, 2010

14:36

39. Kinematics

When a heavy particle initiates a sequential chain of two-body decays terminating

in an invisible particle, constraints on the masses of the states participating in the
chain can be obtained from end-points and thresholds in invariant mass distributions of
the aggregated decay products. For the two-step decay chain depicted in Fig. 39.4 the
invariant mass distribution of the two visible particles possesses an end-point given by:
2
(mmax
12 )

(m2c m2b )(m2b m2a )

=
,
m2b

(39.23)

provided particles 1 and 2 are massless. If visible particle 1 has non-zero mass m1 then
Eq. (39.23) is replaced by
2
2
(mmax
12 ) = m1 +

m21

+ m2b

m2a


+

(m21

(m2c m2b )

2m2b

+ m2b

m2a )2

4m21 m2a


.

(39.24)

See Refs. 2 and 3 for other cases.

39.4.5. Multibody decays : The above results may be generalized to nal states
containing any number of particles by combining some of the particles into eective
particles and treating the nal states as 2 or 3 eective particle states. Thus, if
pijk... = pi + pj + pk + . . ., then
mijk...


= p2 ijk... ,

(39.25)

and mijk... may be used in place of e.g., m12 in the relations in Sec. 39.4.3 or Sec. 39.4.4
above.

39.5. Cross sections

p3, m3

p1, m1
.
.
.
p2, m2

pn+2, mn+2

Figure 39.5: Denitions of variables for production of an n-body nal state.

July 30, 2010

14:36

39. Kinematics 7
The dierential cross section is given by
(2)4 |M |2
d = 
4 (p1 p2 )2 m21 m22
dn (p1 + p2 ; p3 , . . . , pn+2 ) .

(39.26)

[See Eq. (39.11).] In the rest frame of m2 (lab),


(p1 p2 )2 m21 m22 = m2 p1 lab ;

(39.27a)

while in the center-of-mass frame



(p1 p2 )2 m21 m22 = p1cm s .

(39.27b)

39.5.1.

Two-body reactions :

p1, m1

p3, m3

p2, m2

p4, m4

Figure 39.6: Denitions of variables for a two-body nal state.

Two particles of momenta p1 and p2 and masses m1 and m2 scatter to particles of
momenta p3 and p4 and masses m3 and m4 ; the Lorentz-invariant Mandelstam variables
are dened by
s = (p1 + p2 )2 = (p3 + p4 )2
= m21 + 2E1 E2 2p1 p2 + m22 ,
2

t = (p1 p3 ) = (p2 p4 )
= m21 2E1 E3 + 2p1 p3 + m23 ,
u = (p1 p4 )2 = (p2 p3 )2

and they satisfy

(39.28)

(39.29)

= m21 2E1 E4 + 2p1 p4 + m24 ,

(39.30)

s + t + u = m21 + m22 + m23 + m24 .

(39.31)

The two-body cross section may be written as

1
1
d
=
|M |2 .
2
dt
64s |p1cm |
July 30, 2010

14:36

(39.32)

39. Kinematics
In the center-of-mass frame
t = (E1cm E3cm )2 (p1cm p3cm )2

4p1cm p3cm sin2 (cm /2)

= t0 4p1cm p3cm sin2 (cm /2) ,

(39.33)

where cm is the angle between particle 1 and 3. The limiting values t0 (cm = 0) and
t1 (cm = ) for 2 2 scattering are


m21 m23 m22 + m24

t0 (t1 ) =
2 s

2

(p1 cm p3 cm )2 .

(39.34)

In the literature the notation tmin (tmax ) for t0 (t1 ) is sometimes used, which should
be discouraged since t0 > t1 . The center-of-mass energies and momenta of the incoming
particles are
s + m21 m22
s + m22 m21

E1cm =
,
E2cm =
,
(39.35)
2 s
2 s
For E3cm and E4cm , change m1 to m3 and m2 to m4 . Then

m
p
2 .
pi cm = Ei2cm m2i and p1cm = 1 lab
s

(39.36)

Here the subscript lab refers to the frame where particle 2 is at rest. [For other relations
see Eqs. (39.2)(39.4).]
39.5.2. Inclusive reactions : Choose some direction (usually the beam direction) for
the z-axis; then the energy and momentum of a particle can be written as
E = mT cosh y , px , py , pz = mT sinh y ,

(39.37)

where mT , conventionally called the transverse mass, is given by

m2T = m2 + p2x + p2y .

(39.38)

and the rapidity y is dened by

1
y = ln
2

= ln

E + pz
mT

E + pz
E pz

= tanh1

p 
z

(39.39)

Note that the denition of the transverse mass in Eq. (39.38) diers from that used
by experimentalists at hadron colliders (see Sec. 39.6.1 below). Under a boost in the
z-direction to a frame with velocity , y y tanh1 . Hence the shape of the rapidity
July 30, 2010

14:36

39. Kinematics 9
distribution dN/dy is invariant, as are dierences in rapidity. The invariant cross section
may also be rewritten
E

d3
d2
d3
=
.
=
d dy pT dpT
d3 p
dy d(p2T )

(39.40)

The second form is obtained using the identity dy/dpz = 1/E, and the third form
represents the average over .
Feynmans x variable is given by
x=

pz
pz max

E + pz
(E + pz )max

(pT
|pz |) .

(39.41)

In the c.m. frame,

x
and

2mT sinh ycm

2pz cm

=
s
s

= (ycm )max = ln( s/m) .

(39.42)

(39.43)

The invariant mass M of the two-particle system described in Sec. 39.4.2 can be
written in terms of these variables as
M 2 = m21 + m22 + 2[ET (1)ET (2) cosh y pT (1) pT (2)] ,


where
ET (i) =

|pT (i)|2 + m2i ,

(39.44)

(39.45)

and pT (i) denotes the transverse momentum vector of particle i.

For p  m, the rapidity [Eq. (39.39)] may be expanded to obtain
y=

cos2 (/2) + m2 /4p2 + . . .

1
ln
2
sin2 (/2) + m2 /4p2 + . . .
ln tan(/2)

(39.46)

where cos = pz /p. The pseudorapidity dened by the second line is approximately
equal to the rapidity y for p  m and  1/, and in any case can be measured when
the mass and momentum of the particle are unknown. From the denition one can obtain
the identities
sinh = cot , cosh = 1/ sin , tanh = cos .
(39.47)

July 30, 2010

14:36

10

39. Kinematics

39.5.3. Partial waves : The amplitude in the center of mass for elastic scattering of
spinless particles may be expanded in Legendre polynomials
1
(2 + 1)a P (cos ) ,
(39.48)
f (k, ) =
k


where k is the c.m. momentum, is the c.m. scattering angle, a = ( e2i 1)/2i,
0  1, and  is the phase shift of the th partial wave. For purely elastic scattering,
 = 1. The dierential cross section is
d
= |f (k, )|2 .
d

(39.49)

The optical theorem states that

tot =

4
Im f (k, 0) ,
k

(39.50)

and the cross section in the th partial wave is therefore bounded:

4
4(2 + 1)
(2 + 1)|a |2
.
(39.51)
2
k
k2
The evolution with energy of a partial-wave amplitude a can be displayed as a trajectory
in an Argand plot, as shown in Fig. 39.7.
 =

Im A

/2
1/2 2

al

Re A
1/2
0
1/2
Figure 39.7: Argand plot showing a partial-wave amplitude a as a function of
energy. The amplitude leaves the unitary circle where inelasticity sets in ( < 1).
The usual Lorentz-invariant matrix element M (see Sec. 39.3 above) for the elastic
process is related to f (k, ) by

(39.52)
M = 8 s f (k, ) ,
so
tot =

Im M (t = 0) ,
(39.53)
2plab m2
where s and t are the center-of-mass energy squared and momentum transfer squared,
respectively (see Sec. 39.4.1).
July 30, 2010

14:36

39. Kinematics 11
39.5.3.1. Resonances: The Breit-Wigner (nonrelativistic) form for an elastic amplitude
a with a resonance at c.m. energy ER , elastic width el , and total width tot is
a =

el /2
,
ER E itot /2

(39.54)

where E is the c.m. energy. As shown in Fig. 39.8, in the absence of background the
elastic amplitude traces a counterclockwise circle with center ixel /2 and radius xel /2,
where the elasticity xel = el /tot . The amplitude has a pole at E = ER itot /2.
The spin-averaged Breit-Wigner cross section for a spin-J resonance produced in the
collision of particles of spin S1 and S2 is
BW (E) =

Bin Bout 2tot

(2J + 1)
,
(2S1 + 1)(2S2 + 1) k 2 (E ER )2 + 2tot /4

(39.55)

where k is the c.m. momentum, E is the c.m. energy, and B in and B out are the
branching fractions of the resonance into the entrance and exit channels. The 2S + 1
factors are the multiplicities of the incident spin states, and are replaced by 2 for photons.
This expression is valid only for an isolated state. If the width is not small, tot cannot
be treated as a constant independent of E. There are many other forms for BW , all of
which are equivalent to the one given here in the narrow-width case. Some of these forms
may be more appropriate if the resonance is broad.

Im A

xel /2
ixel /2

1/2

1/2

Re A

Figure 39.8: Argand plot for a resonance.

The relativistic Breit-Wigner form corresponding to Eq. (39.54) is:
a =

mel
.
s m2 + imtot

(39.56)

A better form incorporates the known kinematic dependences, replacing mtot by

s tot (s), where tot (s) is the width the resonance particle would have if its mass
July 30, 2010

14:36

12

39. Kinematics

were s, and correspondingly

mel by s el (s) where el (s) is the partial width in the
incident channel for a mass s:

s el (s)

.
a =
s m2 + i s tot (s)

(39.57)

For the Z boson, all the decays are to particles

whose masses are small enough to be
ignored, so on dimensional grounds tot (s) = s 0 /mZ , where 0 denes the width
of the Z, and el (s)/tot (s) is constant. A full treatment of the line shape requires
consideration of dynamics, not just kinematics. For the Z this is done by calculating the
radiative corrections in the Standard Model.

39.6. Transverse variables

At hadron colliders, a signicant and unknown proportion of the energy of the incoming
hadrons in each event escapes down the beam-pipe. Consequently if invisible particles
are created in the nal state, their net momentum can only be constrained in the plane
transverse to the beam direction. Dening the z-axis as the beam direction, this net
momentum is equal to the missing transverse energy vector
ETmiss =


i

pT (i) ,

(39.58)

where the sum runs over the transverse momenta of all visible nal state particles.
39.6.1.

Single production with semi-invisible final state :

Consider a single heavy particle of mass M produced in association with visible

particles which decays as in Fig. 39.1 to two particles, of which one (labeled particle 1)
is invisible. The mass of the parent particle can be constrained with the quantity MT
dened by
MT2 [ET (1) + ET (2)]2 [pT (1) + pT (2)]2

= m21 + m22 + 2[ET (1)ET (2) pT (1) pT (2)] ,

where

pT (1) = ETmiss .

(39.59)

(39.60)

This quantity is called the transverse mass by hadron collider experimentalists but it
should be noted that it is quite dierent from that used in the description of inclusive
reactions [Eq. (39.38)]. The distribution of event MT values possesses an end-point at
MTmax = M . If m1 = m2 = 0 then
MT2 = 2|pT (1)||pT (2)|(1 cos 12 ) ,

(39.61)

where ij is dened as the angle between particles i and j in the transverse plane.
July 30, 2010

14:36

39. Kinematics 13
39.6.2.

Pair production with semi-invisible final states :

p1 , m1

p3 , m1
M

p2 , m2

p4 , m4

Figure 39.9: Denitions of variables for pair production of semi-invisible nal

states. Particles 1 and 3 are invisible while particles 2 and 4 are visible.
Consider two identical heavy particles of mass M produced such that their combined
center-of-mass is at rest in the transverse plane (Fig. 39.9). Each particle decays to a
nal state consisting of an invisible particle of xed mass m1 together with an additional
visible particle. M and m1 can be constrained with the variables MT 2 and MCT which
are dened in Refs. [4] and [5].
References:
1. See, for example, J.J. Sakurai, Modern Quantum Mechnaics, Addison-Wesley (1985),
p. 172, or D.M. Brink and G.R. Satchler, Angular Momentum, 2nd ed., Oxford
University Press (1968), p. 20.
2. I. Hinchlie et al., Phys. Rev. D55, 5520 (1997).
3. B.C. Allanach et al., JHEP 0009, 004 (2000).
4. C.G. Lester and D.J. Summers, Phys. Lett. B463, 99 (1999).
5. D.R. Tovey, JHEP 0804, 034 (2008).

July 30, 2010

14:36