Introduction to polarization physics

Chapter 2
Spin in Strong Interactions
In this chapter, the basic theoretical relations underlying the design and analysis of
polarization experiments involving strong interactions are reported on classical ex-
amples of pion–nucleon and nucleon–nucleon scattering. After the introduction of
the density matrix and reaction matrix, we discuss such notions as the complete set
of experiments and the equality of the polarization P in the direct reaction and the
asymmetry A in the inverse reaction on the example of the simple pion–nucleon
system. On the example of the nucleon–nucleon system, we present the method for
explicitly constructing the reaction matrix, formulate the unitarity condition, and
point to the possibilities of seeking the effects of parity and time reversal violation.
The presentation is primarily based on the technique of nonrelativistic quantum me-
chanics. Relativistic pion–nucleon and nucleon–nucleon elastic scattering matrices
are discussed in the concluding sections. The inclusion of the relativistic effects
insignificantly changes the nonrelativistic results.
2.1 Density Matrix
As known, a particle with spin s is described by the wave function Ψ having 2s + 1
components Ψα, where α = s,s − 1,...,0,...,−(s − 1),−s. If only one of these
components with a certain α value is nonzero, the particle is in a pure spin state. In
this case, the mean value of any operator Ô is given by the expression
Ô = Ψ ∗
α
Ô Ψα .
However, components for several α values are most often nonzero in reality. In
this case, the state is mixed. A simple example is a polarized proton beam. If its
polarization is 100 %, this is a pure spin state: the spins of all protons are oriented
identically. When the beam is partially polarized, the protons are in a mixed spin
state. This means that the spins of some protons are directed upwards and the spins
S.B. Nurushev et al., Introduction to Polarization Physics,
Lecture Notes in Physics 859, DOI 10.1007/978-3-642-32163-4_2,
© Moskovski Inzhenerno-Fisitscheski Institute, Moscow, Russia 2013
59

60 2 Spin in Strong Interactions
of other protons are directed downwards. In this case, the mean value of an arbitrary
spin operator Ô in the mixed spin state Ψ is given by the expression
Ô =
α
Wα Ψ ∗
α
Ô Ψα , (2.1)
where Wα is the weight of the pure state α. The Dirac brackets mean integration
(summation) with respect to continuous (discrete) variables.
Let us consider the set of other (2s + 1) components {χm}, which are the orthog-
onal eigenfunctions of a certain spin operator as the basis functions. If this set is the
complete orthogonal set, it can be used for the expansion
Ψα =
s
m=−s
Cα
mχm. (2.2)
The substitution of this expansion into Eq. (2.1) yields
Ô =
αmn
WαCα∗
m Cα
n χ∗
m
Ôχn =
mn α
WαCα
n Cα∗
m · χ∗
m
Ôχn
=
mn
ρnmOmn, (2.3)
where Omn is the matrix element of the operator Ô and
ρnm =
α
Wα Cα∗
m Cα
n (2.4)
is the matrix element of a certain operator ˆρ, which is called the density matrix. The
operator ˆρ can be represented in the matrix form
ˆρ =
α
WαΨαΨ +
α . (2.5)
Indeed, let us represent the matrix element of the operator ˆρ in the form
ρmn = χ+
m ˆρχn =
α
Wα χ+
m ΨαΨ +
α χn . (2.6)
Since the eigenfunctions χ are orthogonal, expansion (2.2) provides
χ+
m Ψα =
n
Cα
n χ+
m χn = Cα
m, Ψ +
α χ+
m =
n
Cα∗
n χ+
n χm = Cα∗
m . (2.7)
The substitution into Eq. (2.6) finally yields the expression
ρmn =
α
WαCα
mCα∗
n , (2.8)
which coincides with Eq. (2.4).
Using Eqs. (2.4) or (2.5), one can show that the matrix ρ is Hermitian:
ˆρ+
= ρ; (2.9)

2.1 Density Matrix 61
i.e., the mean value of ρ is a real number. Let us define the sum of the diagonal
elements of an arbitrary matrix C (its trace) as
Tr C =
i
Cii. (2.10)
If the operator ˆC is the product of two operators Â and ˆB, its matrix element is
given by the expression
Cij =
m
AimBmj . (2.11)
In this case, the sum of the diagonal elements is
Tr ˆC =
mn
AnmBmn = Tr Â ˆB. (2.12)
Comparison with Eq. (2.3) provides
Ô = Tr Ô ˆρ = Tr ˆρ Ô. (2.13)
The latter equality shows that two operators can be transposed in the trace even
if they do not commute. However, in the general case, only a clockwise or counter-
clockwise cyclic permutation without the transposition of the operators is possible
(if the operators do not commute).
Thus, to determine the mean value of the operator Ô, it is necessary to multiply
this operator by the density matrix and to calculate the sum of the diagonal elements
of the resulting matrix.
According to the theory of matrices, any matrix of rank m = n = (2s + 1) can be
expanded in a complete set of (2s + 1)2 matrices {sν} of the same rank satisfying
the orthogonality condition
Tr sνsμ = δνμ(2s + 1). (2.14)
Therefore, we can write the expansion
ˆρ =
(2s+1)2
μ=1
Cμsμ. (2.15)
Multiplying this relation by sν from the right and calculating the trace, we obtain
Tr ˆρsν =
μ
CμTr sμsν = (2s + 1)
μ
Cμδμν = (2s + 1)Cν. (2.16)
The substitution of the coefficients Cμ determined from Eq. (2.16) into Eq. (2.15)
finally gives
ˆρ =
1
2s + 1
(2s+1)2
μ=1
Tr ( ˆρˆsμ)ˆsμ. (2.17)

Substituting Ô = ˆsμ (spin operator) into Eq. (2.13), we obtain the polarization
vector P :
1
2
P = ˆsμ = Tr ( ˆρˆsμ). (2.18)
Hence, the final expression for the density matrix in terms of observable P = ˆs
has the form
ˆρ =
1
2s + 1
(2s+1)2
μ=1
ˆsμ ˆsμ. (2.19)
Thus, the density matrix is completely determined by the mean value of the op-
erators ˆsμ. With the normalization of the mean value of the density matrix to unity,
we obtain
Ô = Tr ˆρ Ô/Tr ˆρ. (2.20)
With this normalization condition, the final expression for the density matrix has
the form
ˆρ =
1
2s + 1
(2s+1)2
μ=1
Tr ˆρ ˆsμ ˆsμ. (2.21)
The above presentation was based on works Martin and Spearman (1970) and
Nurushev (1983).
2.2 Reaction Matrix
When considering reactions involving particles with spin, we use the wave functions
Ψ and Φ of the initial and final states, respectively. The transition matrix from the
initial to final state, M, is defined as follows (Nurushev 1983):
Φ = MΨ. (2.22)
Then, we have two density matrices:
ρi =
α
WαΨαΨ +
α (2.23)
for the initial state and
ρf =
α
WαΦαΦ+
α (2.24)
for the final state. Here, α stands for averaging over the initial spin states and
summation over the final spin states. In view of relation (2.22), the relation between
these two matrices is obtained from Eq. (2.24) in the form
ρf =
α
WαMΨαΨ +
α M+
= M
α
WαΨαΨ +
α M+
(2.24a)

2.2 Reaction Matrix 63
or
ρf = MρiM+
. (2.25)
Hence, the solution of the problem of the interaction between particles is re-
duced to the determination of ρi in terms of the mean values of the complete set of
the spin matrices sν (the mean values sν are called the observables) and the oper-
ator ˆM. Then, the density matrix of the final state is unambiguously determined by
Eq. (2.25), and a necessary observable in the final state can be calculated.
Let us find the operator ˆρf as a function of sν and M. To this end, we multiply
Eq. (2.25) by sμ from the right and calculate the trace of the product of the matrices:
Tr ˆρf sμ = Tr M ˆρiM+
sμ
= Tr M
1
2s + 1
(2s+1)2
ν=1
Tr ˆρi sν isν M+
sμ
=
1
2s + 1
Tr ˆρi
(2s+1)2
ν=1
sν i · Tr MsνM+
sμ . (2.26)
In view of the relation
Tr ( ˆρf sμ) = sμ f Tr ˆρf , (2.27)
we obtain
sμ f · Tr ˆρf =
1
2s + 1
Tr ˆρi
(2s+1)2
ν=1
sν i · Tr MsνM+
sμ . (2.28)
Denoting the differential cross section as
I =
Tr ˆρf
Tr ˆρi
, (2.29)
we arrive at the following expression for the mean value sμ f of the spin operator
sμ final state:
sμ f · I =
1
2s + 1
·
(2s+1)2
ν=1
sν i · Tr MsνM+
sμ . (2.30)
This expression allows one to calculate the mean value of any spin operator sμ in
the final state using the known parameters of the initial state sν i and the scattering
matrix M.
Let us consider reactions of the type
π + N = π + N, (2.31)
or, in the spin notation, 0 + 1/2 → 0 + 1/2 (spins of the pion and nucleon are 0
and 1/2, respectively). In the corresponding two-dimensional spin space, the Pauli
matrices σ (σx,σy,σz) together with the identity matrix 1 can be used as a complete

set of spin operators. In this case, the density matrix of the initial state can be written
in the form
ˆρi = C0 · 1 + C1 · σ. (2.32)
Let us determine the coefficients C0 and C1. From the normalization condition
of the trace of the density matrix ˆρi to unity, we obtain
Tr ˆρi = Co · Tr 1 + C1 · Tr σ = 2C0 = 1, (2.33)
because
Tr σ = 0. (2.34)
Relation (2.34) is easily verified by writing the Pauli matrices in the explicit form
σx =
0 1
1 0
, σy =
0 −i
i 0
, σz =
1 0
0 −1
. (2.35)
Let the nucleon in the initial state be polarized and have polarization vector Pt
(the subscript t means the target). For the initial state,
Pt = σt =
Tr ˆρσt
Tr ˆρ
= Tr σt ·
1
2
+ C1 · σ = Tr σt (C1 · σ)
= C1k · ek · Tr σkσi = C1k · ek · 2δik = 2C1t . (2.36)
Here, ek are the unit coordinate vectors in the Cartesian coordinate system and
we use the relation
(σ · A)(σ · B) = A · B + iσ · (A × B), (2.37)
which is easily proved using relations (2.35). Thus, we obtain
C0 =
1
2
, C1 =
1
2
Pt . (2.38)
Then, the density matrix of the initial state ˆρi, is represented in the form
ˆρi =
1
2
(1 + Pt · σ). (2.39)
Hence, the density matrix ˆρ is completely determined by the target polarization
vector Pt (or the beam polarization vector PB in the case of the 1/2 + 0 → 1/2 + 0
reaction).
Note that, in view of the properties of the Pauli matrices, expression (2.39) cannot
include operators above the first order, because all such operators are reduced to an
operator maximally of the first order.
2.3 R, P , and T Transformations
In the next section, we consider nucleon–nucleon elastic scattering and, following
Wolfenstein and Ashkin (1952), construct the elastic scattering matrix. Here, as a

2.3 R, P , and T Transformations 65
preparation to this consideration, we discuss the constraints on this matrix that fol-
low from the physical requirements of the isotropy of space (R operation), space
inversion (P operation), and time reversal (T operation). Below in this section, we
follow Bilenky et al. (1964). In the interaction (or Heisenberg) representation, the
S-matrix is defined in terms of the Dirac brackets as
ψ(+∞) = S ψ(−∞) . (2.40)
Here, |ψ(−∞) is the wave function of the system in the initial state at t → −∞
and |ψ(+∞) is the wave function of the system in the final state. According to this
definition, the S-matrix transforms the initial state of two free nucleons to the final
state with allowance for their interactions. Thus, the S-matrix contains all informa-
tion on the interaction between the nucleons. If the nucleons do not interact, it is
reasonable to set S = 1. Then, we formulate the necessary physical requirements.
1. R operation. Let |ψ(t) be the wave function of the system at time t in an
arbitrary reference frame called base. We introduce the second reference frame R
rotated by a certain angle and denote the wave function of the system in this ref-
erence frame as |ψ(R,t) . The wave functions in two reference frames should be
related by a unitary transformation (according to the requirement that the numbers
of particles in both reference frames should be the same). The unitarity of the oper-
ator implies the equality U+(R) = U−1(R).
Hence,
ψ(R,t) = U(R) ψ(t) . (2.41)
The matrix U(R) is obviously a function of the rotation angles of the R frame with
respect to the base frame. The multiplication of Eq. (2.40) by U(R) from the left
gives
ψ(R,+∞) = U(R)SU−1
(R) ψ(R,−∞) . (2.42)
The wave functions |ψ(R,−∞) and |ψ(R,+∞) describe the initial and final
states of the nucleons, respectively, in the rotated reference frame R. Hence, in this
reference frame, by the definition of the S-matrix, which is determined only by the
interaction dynamics but is independent of the choice of the reference frame, the
wave functions should be related by the same S-matrix as in Eq. (2.40):
ψ(R,+∞) = S ψ(R,−∞) . (2.43)
Comparison of Eqs. (2.42) and (2.43) shows that
U(R)SU−1
(R) = S. (2.44)
Since the matrix U(R) is unitary, this equality can be represented in the other
form
U−1
(R)SU(R) = S. (2.45)
Relation (2.44) (or (2.45)) expresses the invariance of strong interactions under
rotations of the reference frame in physical space. An application of this relation
will be considered elsewhere.

2. P operation. It is postulated that strong interactions are invariant under space
inversion. This property is also called the parity conservation law. Let us consider
the constraints imposed by this postulate on the S-matrix. As in the above consid-
eration, we take the initial reference frame as base. As above, the wave function
in this frame is denoted as |ψ(t) . We introduce the reference frame I, in which
all coordinate axes are inverted, i.e., x → −x, y → −y, and z → −z. If the base
frame is left-handed, the frame I is right-handed. Let U(I) be a unitary operator
transforming the function |ψ(t) in the base frame to the wave function in the I
frame:
ψ(I,t) = U(I) ψ(t) . (2.46)
In this case, both functions describe the same physical state, but in different coor-
dinates. Hence, the S-matrices in two frames should be the same. Let us determine
the matrix S(I). By the deﬁnition,
ψ(I,+∞) = S(I) ψ(I,−∞) . (2.47)
At the same time, from Eq. (2.46) we obtain
ψ(I,+∞) = U(I) ψ(+∞) = U(I)S ψ(−∞) = U(I)SU−1
ψ(I,−∞) .
(2.48)
Comparison of Eqs. (2.47) and (2.48) shows that
S(I) = U(I)SU−1
(I). (2.49)
Thus, the postulate of the invariance of strong interactions under space inversion
leads to the relation
S = U−1
(I)SU(I). (2.50)
These two cases imply that the invariance under the R and P transformations
reduces to the commutativity of the S-matrix and the corresponding transformation
U matrices.
3. T operation. The principle of the invariance of the strong interaction under
time reversal is formulated as follows. Let us consider the Schrödinger equation in
the interaction representation
i
∂|ψ(t)
∂t
= H(t) ψ(t) . (2.51)
Changing t → −t and taking complex conjugation in this equation, we arrive at
the equation
i
∂|ψ(−t) ∗
∂t
= H∗
(−t) ψ(−t)
∗
. (2.52)
Since H∗(−t) = H(t) in the general case, this new equation is not a Schrödinger
equation. However, let us assume that there is a unitary operator U(T ) providing
the transformation
ψ(T,t) = U(T ) ψ(−t)
∗
. (2.53)

The substitution of this relation into Eq. (2.52) yields
i
∂|ψ(T,t)
∂t
= U(T )H∗
(−t)U−1
(T ) ψ(T,t) . (2.54)
Setting
H(t) = U(T )H∗
(−t)U−1
(T ), (2.55)
we arrive at the Schrödinger equation
i
∂|ψ(T,t)
∂t
= H(t) ψ(T,t) . (2.56)
Thus, relation (2.55) presents a mathematical formulation of the physical postu-
late of the invariance of the interaction under time reversal.
According to Eq. (2.55), for each wave function |ψ(t) that is a solution
of Schrödinger equation (2.51), there is another function |ψ(T,t) that satisﬁes
Eq. (2.56) and describes the motion of the system in the reverse time direction.
Let us obtain the requirement imposed on the S-matrix by the time reversibility
condition.
We have two Schrödinger equations with two wave functions, but with the same
S-matrix, because the S-matrix is independent of the initial state of the system, but
is completely determined by the dynamics of interaction.
According to the above consideration and by analogy with relation (2.40), we
write
ψ(T,+∞) = S ψ(T,−∞) .
Using relation (2.53), we represent this equality in the form
U(T ) ψ(−∞)
∗
= SU(T ) ψ(+∞)
∗
.
The multiplication of this relation by U−1(T ) from the left yields
ψ(−∞)
∗
= U−1
(T )SU(T ) ψ(+∞)
∗
. (2.57)
Let us take into account that the S-matrix is unitary, i.e., S+S = 1 and that
S+ = ˜S∗ by the deﬁnition, where the asterisk and tilde mean the complex conju-
gation and transposition of the matrix, respectively. Multiplying Eq. (2.40) by S+
from the left, taking complex conjugation, and taking into account the properties of
the S-matrix, we obtain
ψ(−∞)
∗
= ˜S ψ(+∞)
∗
. (2.58)
The comparison of Eqs. (2.57) and (2.58) shows that
U−1
(T )SU(T ) = ˜S. (2.59)
This is the requirement imposed on the S-matrix by the time-reversal invariance
of the interaction.
The results are summarized as follows:
• the invariance of the interaction under rotation of the reference frame, the corre-
sponding R operation is given by relation (2.45): U−1(R)SU(R) = S;

• the invariance of the interaction under space inversion, the corresponding P op-
eration is given by relation (2.50): U−1(I)SU(I) = S;
• the invariance of the interaction under time reversal, the corresponding T opera-
tion is given by expression (2.59): U−1(T )SU(T ) = ˜S.
In applications, another matrix M determined only by the interaction between
the nucleons is used. The initial state of the particles at t → −∞ is denoted as |i ,
and the final state at t → +∞ is denoted as |f . The particles in the initial and final
states do not interact and have the relative momenta p and p , total momenta Q and
Q , and total energies E and E , respectively. The matrix M is expressed in terms
of the S matrix as
S − 1 = M, (2.60)
or in terms of the matrix elements, taking into account the conservation of the energy
and momentum:
f |S|i = f |i − 2πiδ Q − Q δ E − E f |M|i . (2.61)
Here, the δ functions ensure the conservation of the momentum and energy. The
matrix M transforms the initial state |i to the final state |f and acts only in the
spin space. By the definition of the matrix elements,
f |M|i = χ +
M p ,p χ . (2.62)
Here, χ and χ are the spin wave functions of the initial and final states of the
nucleons, respectively. The matrix M, as well as the S matrix, is determined by the
dynamics of interaction and satisfies the requirements of the R, P , and T invari-
ances. Let us consider their in more detail.
From the postulate of the invariance of the interaction under rotation of the ref-
erence frame, the following constraint on the S-matrix was obtained (see (2.45)):
U−1
(R)SU(R) = S.
Let us represent this relation in terms of the matrix elements in the base and
rotated R frames
f U−1
(R)SU(R) i = f,R|S|R,i = f |S|i . (2.63)
Here, |R,i and |R,f are the wave functions in the rotated R frame. The total,
Q, and relative, p, momenta in the base reference frame are related to the respective
momenta QR and pR in the rotated reference frame as
(QR)i = ailQi, (pR)i = ailpi. (2.64)
Here, a is the rotation matrix from the base frame to the R frame and ali are its
matrix elements (in this case, the cosines and sines of the rotation angle from the
old to new reference frame).
Relation (2.63) indicates that the respective elements of the S matrix in different
frames are the same. According to M-matrix definition (2.61), this is valid for its
matrix elements:
χ +
(R)M pR,pR χ(R) = χ +
(p)M p ,p χ(p) , (2.65)

where χ(R) and χ (R) are the spin wave functions in the rotated frame and χ
and χ , in the base frame. Since these sets of functions describe the same spin state,
but in different frames, they should be related by a unitary transformation:
χ(R) = U(R)χ, χ (R) = U (R)χ . (2.66)
The nucleon–nucleon scattering under consideration involves two spin particles
in the initial and final states of the reaction. This means that the spin functions in the
initial and final states of the reaction are the products of the functions of individual
nucleons. As a result, the unitary operators U(R) and U (R) are the direct products
of the matrices acting on the spin functions of the individual particles.
The mean value of the spin operator in quantum mechanics is an observable,
namely, the polarization vector (more precisely, the polarization vector is P = σ =
2s, where σ is the Pauli operator and s is the spin vector). The mean value of the
spin operator should be transformed as a vector:
χ+
(R)slχ(R) = aliχ+
siχ.
Here, sl is the spin operator of one of the initial nucleons. Therefore, the matrix
U(R) should satisfy the condition
U−1
(R)slU(R) = alisi. (2.67)
The same condition is obviously imposed on the matrix U (R). For a given rota-
tion angle of the frame R, the matrix U can be reconstructed from these conditions.
From Eqs. (2.65) and (2.66), it follows that
U −1
(R)M pR,pR U (R) = M p ,p . (2.68)
This is the mathematical expression of the postulate of the invariance of the in-
teraction under space rotation.
Then, we consider the P operation, i.e., space inversion. Under this operation,
the momenta, being polar vectors, change signs, whereas the spin vector, being an
axial vector, does not change sign. Therefore,
sl = U−1
(I)slU(I), sl = U −1
(I)slU (I), (2.69)
where the unitary matrix U(I)(U (I)) ensures the transformation of the wave func-
tion of the initial (final) state from the base frame to the inverted frame I. This
transformation is written as follows:
χ(I) = U(I)χ, χ (I) = U (I)χ .
The condition on the S-matrix provides
χ +
M p ,p χ = IiI∗
f χ +
(I)M −
←
p ,−p χ(I) . (2.70)
Here, Ii and If are the internal parities of two initial and two final nucleons,
respectively. Using the relation between the wave functions χ and χ , we obtain
M p ,p = IiI∗
f U−1
(I)M −p ,−p U(I). (2.71)

2.4 Unitarity Condition
The Schrödinger equation for the wave function Ψk has the form
∇2
Ψk +
2μ
2
(E − û)Ψk = 0, (2.72)
where μ is the reduced mass of colliding particles and û is the potential energy of
their interaction in the operator form (can include the spin and isospin operators).
For the case of elastic scattering, Ψk can be represented in the form of the sum of
two terms:
Ψk = eik·r
χ +
1
r
eikr
M k ,k χ. (2.73)
The first term is an incident plane wave; the second term is the divergent scat-
tering waves; k and k are the wave vectors of the initial (before the interaction)
and final states, respectively; χ is the spinor function of the initial state; and M is
the reaction matrix. Since the elastic scattering is considered in the center-of-mass
frame,
k = |k|, û+
= û. (2.74)
The second condition is the Hermitian condition imposed on the interaction po-
tential, because its eigenvalues should be real.
Then, the interaction in the intermediate state of the colliding particles leads
to the transition of the wave vector k to the wave vector k , determining a given
direction (e.g., the direction to a recording detector).
At large distances from the collision center, the following expansion can be used
(to define the absolute phases, it is necessary to include a known, for example, elec-
tromagnetic or weak interaction in addition to strong interactions):
Ψ 0
k = eik·r
= eikzcosθ
=
∞
l=0
il
(2l + 1)Pl(cosθ)
sin(kr − 1
2 lπ)
kr
=
1
ikr
∞
l=0
(2l + 1)
2
Pl(cosθ)eikr
−
1
ikr
∞
l=0
(−1)l (2l + 1)
2
Pl(cosθ)e−ikr
. (2.75)
It can be shown that
∞
l=0
(2l + 1)
2
Pl(cosθ) = δ(1 − cosθ),
∞
l=0
(−1)l (2l + 1)
2
Pl(cosθ) = δ(1 + cosθ).
(2.76)

2.4 Unitarity Condition 71
These formulas are verified by multiplying the both sides by Pl(cosθ) and inte-
grating with respect to cosθ. The validity of relations (2.76) is obvious in view of
the orthogonality condition of the Legendre polynomials,
1
−1
Pl (cosθ)Pl(cosθ)d cosθ =
2
2l + 1
δll (2.77)
and the definition of the Dirac δ functions
f (x)δ(x − a)dx = f (a). (2.78)
Taking into account Eq. (2.76), from Eq. (2.75) we obtain
eik·r
=
1
ikr
eikr
δ(1 − cosθ) −
1
ikr
e−ikr
δ(1 − cosθ). (2.79)
Therefore,
Ψk(r) =
1
r
eikr
· χ
1
ik
δ(1 − cosθ) + M k ,k −
1
ikr
e−ikr
δ(1 + cosθ) · χ.
(2.80)
Similarly,
Ψk (r) =
1
r
e−ikr
· χ+
M k ,k −
1
ik
δ(1 − cosθ) +
1
ikr
eikr
δ(1 + cosθ) · χ+
.
(2.81)
From the Schrödinger equation
∇2
Ψk +
2μ
2
(E − û)Ψk = 0, ∇2
Ψ +
k +
2μ
2
· Ψ +
k · (E − û) = 0, (2.82)
we can obtain the relation
Ψ +
k ∇2
Ψk − ∇2
Ψ +
k Ψk = 0, (2.83)
which can be modified to the form
∇ Ψ +
k ∇Ψk − ∇Ψ +
k Ψk = 0. (2.84)
The integration over the volume V provides
∇ Ψ +
k ∇Ψk − ∇Ψ +
k Ψk dv = ∇ Ψ +
k ∇Ψk − ∇Ψ +
k Ψk ds = 0. (2.85)
Using Eqs. (2.80) and (2.81), as well as the operator ∇ = ∂
∂r , we obtain
1
kr2
δ(1 + cosθ)δ 1 + cosθ −
1
r2
M+
k ,k +
i
k
δ 1 − cosθ
· M k ,k −
1
k
δ(1 − cosθ) ds = 0, (2.86)
where ds = r2dωk . After the integration, we arrive at the formula
1
2i
M k ,k − M+
k,k =
k
4π
M+
k ,k · M k ,k dωk . (2.87)

This condition can be generalized to the case of inelastic reactions:
1
2i
1
ka
Mab k ,k −
1
kb
M+
ba k,k =
C
1
4π
M+
bc k ,k · Mac k ,k dωk ,
(2.88)
where the summation over all possible reaction channels is implied.
Relation (2.87) provides a number of important consequences. At k = k (elastic
scattering at zero angle), this relation gives the so-called optical theorem
Ima(0) =
k
4π
σTOT, (2.89)
which is the relation between the imaginary part of the forward elastic scattering
amplitude a(0) and the total cross section σTOT.
The application of relation (2.88) to the pion–nucleon scattering matrix gives two
relations
Ima k ,k =
k
4π
a∗
k ,k · a k ,k
+ ib∗
k ,k × b k ,k · n × n dωk , (2.90)
Reb k ,k =
k
4π
a∗
k ,k · b k ,k + b∗
k ,k · a k ,k
+ Re b∗
k ,k × b k ,k · n × n · n dωk . (2.91)
Similar relations are also valid for nucleon–nucleon scattering. These relations
are particularly suitable at low energies, when only the elastic channel is opened.
As a result, the number of necessary experiments is halved (by two and ﬁve exper-
iments for the πN and nucleon–nucleon scatterings, respectively). However, these
statements should be considered carefully, because they are valid only under ideal
conditions, which are almost inaccessible in actual experiments.
The presentation in this section follows Nurushev (1983).
2.5 Pion–Nucleon Scattering
Let us consider reactions of the type
π + N = π + N, (2.92)
or, in the spin notation, 0 + 1/2 → 0 + 1/2. In the corresponding two-dimensional
spin space, the Pauli matrices σ(σx,σy,σz) together with the identity matrix 1 can
be used as a complete set of spin operators. In this case, the density matrix of the
initial state can be written in the form (see Eq. (2.32) in Sect. 2.2)
ˆρi = C0 · 1 + C1 · σ. (2.93)
Hence, the density matrix ˆρ is completely determined by the target, Pt , or beam,
PB, polarization vector.

2.5 Pion–Nucleon Scattering 73
Then, it is necessary to determine the reaction matrix M. In the general case, it
should be, first, a function of two variables, for example, momenta ki and kf before
and after reactions, respectively, and, second, a two-row matrix in spin space. As
the matrix, it can be expanded in the complete set consisting of the Pauli matrices
and identity matrix:
M ki,kf = a ki,kf · I + b ki,kf · σ. (2.94)
According to the experimental data, we require that strong interactions satisfy
the following conditions (Nurushev 1983):
1. Parity conservation law. Since the parities of the initial and final systems are
identical, the M-matrix should be a scalar function of the initial energy and scat-
tering angle of the particles. This means that the vector b, as well as σ, should
also be axial. The only axial vector that can be composed of two vectors ki and
kf has the form
b (ki,kf ) = bn, (2.95)
where n = ki ×kf /|ki ×kf | is the unit vector perpendicular to the reaction plane.
The quantity a(ki,kf ) should obviously be a scalar function.
2. Time reversibility. Under the time reversal operation, ki → −kf and kf → −ki,
so that n changes sign. Under this operation, σ also changes sign, so that the
quantity σ · n is a scalar. For the (0 + 1/2) system, this requirement is satisfied
simultaneously with requirement 1. However, for more complex systems (e.g.,
(1/2 + 1/2)), time reversibility gives rise to additional constraints.
Thus, the πN elastic scattering matrix M(ki,kf ) has the form
M(ki,kf ) = a(ki,kf ) + b(ki,kf ) · σ · n. (2.96)
Here, b and a are called the amplitudes with and without spin flop, respectively,
are complex quantities; therefore, four real functions (for a given initial energy and
a given scattering angle) should be determined from experiments. A set of inde-
pendent experiments necessary for the unambiguous determination of all reaction
amplitudes is called the complete set of experiments. Hence, the complete set of ex-
periments for the πN system should include at least four independent experiments.
However, reality is much more complicated that the above description. First,
since the experimentally measured quantities are quadratic combinations of the am-
plitudes a and b, only the difference of the phases of the amplitudes a and b for
strong interactions, rather than their absolute values, can be determined from an ex-
periment. To determine the absolute values of the phases, it is necessary to include
a known (for example, electromagnetic or weak) interaction in addition to strong
interactions. Thus, the complete set of experiments should include more than four
experiments. Second, if elastic scattering is the single allowed reaction channel, the
unitarity condition gives rise to two additional relations between the amplitudes a
and b; for this reason, to determine these amplitudes, it is sufficient to perform two
independent experiments.

However, at energies above the pion production threshold, the number of experi-
ments in the complete set is larger than four.
In reality, there are three reactions induced by charged pions:
π+ + p → π+ + p(a), π− + p → π− + p(b),
π− + p → π0 + n(c).
(2.97)
These three reactions are related by the requirement of the isotopic invariance of
strong interactions. As a result, the matrices of these reactions are related as
M(a) = M1, M(b) =
1
2
(M1 + M0), M(c) =
1
2
(M1 − M0). (2.98)
Here, the matrices M1 and M0 correspond to the isotopic states of the pion–
nucleon system with T = 3/2 and 1/2, respectively. These matrices are recon-
structed similarly to the matrix M. Disregarding isotopic invariance, three reactions
(2.97) are described by the set of 12 experiments. Allowance for isotopic invariance
reduces this number to 8.
For the system of two particles with spin 1/2 (example is pion + nucleon), we
present the proof of the equality P = A, where P and A are the polarization of the
particle and its asymmetry in the binary reactions, respectively; this relation is very
important in applications (Bilenky et al. 1964).
Let us consider the wave function of the system with the inverted direction of the
wave vector that satisﬁes the Schrödinger equation with reversed time (t → −t).
We represent it in the form
Ψ−k = e−ik r
+
1
r
eikr
M k ,−k
=
i
kr
e−ikr
· δ 1 − cosθ
+
1
r
eikr
M k ,−k −
i
k
· δ 1 − cosθ . (2.99)
The substitution of this wave function into the following Eq. (2.85) from
Sect. 2.4:
∇ Ψ +
k ∇Ψk − ∇Ψ +
k Ψk dv = ∇ Ψ +
k ∇Ψk − ∇Ψ +
k Ψk ds = 0,
gives
M k ,k · δ 1 − cosθ +
1
ik
· δ 1 − cosθ · δ(1 − cosθ)
− M k ,−k · δ(1 + cosθ) dωk = 0. (2.100)
After the integration, we obtain
M k ,k = M −k,−k . (2.101)
This relation is the condition of the time reversibility of the process.

2.5 Pion–Nucleon Scattering 75
Scattering matrix (2.96) constructed above satisfies this condition. Using the ex-
plicit form of the scattering matrix, we prove the following statement widely used
in applications.
Let the polarization of the particle d be measured in the reaction
a(0) + b(1/2) → c(0) + d(1/2) (2.102)
(the spins of the particles are given in the parentheses). This is usually achieved by
the rescattering of the particle d on a certain nucleus with a known analyzing power.
To explain new terms, we make a brief digression.
Let a beam with a given energy and unit polarization be scattered on a nuclear
target at a given angle. The number of the scattered particles per unit flux of the
incident beam whose polarization is directed upward from the scattering plane is
denoted as N1. Under the same conditions, this number for the beam polarization
downwards to the scattering plane is denoted as N2. In this notation, the analyzing
power of the target is defined by the formula
AN =
N1 − N2
N1 + N2
.
If the beam is partially polarized, i.e., P = 1, raw asymmetry (directly measured
in experiment), can be defined as
ε = P · A.
The quantity ε is also called left–right asymmetry or, sometimes, “raw” asym-
metry. According to the definition, the asymmetry ε coincides with the analyzing
power AN at the 100 % polarization of the beam.
We denote the polarization of the particle d as P . Let us consider the reverse
reaction
c(0) + d↑(1/2) → a(0) + b(1/2), (2.103)
where the particle d is polarized. Let the left–right asymmetry AN is measured in
the formation of the particle a (or b). Theorem: the polarization P of the particle d
in reaction (2.102) is equal to the asymmetry A of the particle b (or a) in reaction
(2.103). This statement is expressed by the equality
P = AN . (2.104)
Let us prove this statement. Indeed, by the definition of polarization (Bilenky et
al. 1964).
P =
Tr (MρiM+σ)
Tr (MρiM+)
(2.105)
since the particle b in initial system (2.102) is unpolarized, ρi = 1/2 and, calculat-
ing P , we obtain
P · I0 = 2Rea∗
b = Tr MM+
σ , (2.106)
where
I0 = |a|2
+ |b|2
(2.107)

is the differential cross section for reaction (2.102). Since the particle d in reaction
(2.103) is polarized, ρi = 1
2 (1 + P0· · σ) and the scattering cross section is given by
the expression
If = Tr M ˆρiM+
=
1
2
Tr MM+
+
1
2
P0Tr MσM+
. (2.108)
The direct calculation shows that MM+ = M+M and
Tr MσM+
= Tr MM+
σ = 2I0P, (2.109)
where P is determined by expression (2.106). Then,
If = I0(1 + P0 · P). (2.110)
By the definition of left–right asymmetry
AN =
1
P0
If (+) − If (−)
If (+) + If (−)
= P, (2.111)
quod erat demonstrandum.
When deriving relation (2.111), we explicitly take into account the detailed bal-
ance principle, i.e., the equality of the cross sections for the direct and inverse reac-
tions.
Note that the positive sign in relation (2.111) appears for a reaction in which the
initial and final states have the same parity. For the case of different parities, the sign
in relation (2.111) is negative.
Theorem (2.104) is also proved for the case of the binary reaction when both
initial (and final) particles have spin 1/2. This theorem is invalid for the inclusive
reactions.
A test of the relation P = A is simultaneously a test of the T invariance in strong
interactions. At present, this problem is of very great interest in view of the creation
of an absolute polarimeter for the RHIC collider (see Chap. 8 in the second part of
the book, which is devoted to the polarimetry of beams).
It is known that a reaction with the production of hyperons is a convenient reac-
tion for verifying the relation P = A. If a target is unpolarized, the polarization of a
hyperon is determined from its decay. If the target is polarized, left–right asymme-
try can be measured with the same instruments. The same is true for the beam. An
example of such interactions is the reaction
π−
+ p → K0
+ Λ(↑)(a), π−
+ p(↑) → K0
+ Λ(b). (2.112)
This reaction is very convenient because channels (a) and (b) can be measured
simultaneously if a polarized target is used. In this case, it is very important to
detect both K mesons and Λ hyperons. Averaging the experimental results over
the target polarizations (as if the target polarization vanishes), one can determine
the polarization of the Λ hyperons (channel (a)). Averaging over the polarization
of the Λ hyperons for the polarized target (channel (b)), we determine asymmetry.
Comparison of these two observables ensures the direct test of the equality P = AN .
Such an experiment has not yet been carried out.

2.6 Nucleon–Nucleon Scattering 77
An important application of the relation P = AN is the measurement of asym-
metry in the reaction
π−
+ p(↑) → π0
+ n. (2.113)
For the direct measurement of the neutron polarization on an unpolarized target,
it is necessary to scatter the neutron by another target and to detect the scattered
neutron. This is a difficult experimental problem owing to loss in the yields for the
second scattering process and low neutron detection efficiency. Therefore, the use
of the polarized proton target made it possible to measure the neutron polarization
in reaction (2.113). These measurements provided first doubts in the Regge pole
model very popular in the 1960s.
2.6 Nucleon–Nucleon Scattering
In this section, we present the method for determining the reaction matrix in the
nonrelativistic case proposed in Wolfenstein and Ashkin (1952). We will see later
that the relativistic approach does not change the results, but leads to the kinematic
rotations of the observables lying in the reaction plane. The observables perpendic-
ular to the reaction plane remain unchanged in this case.
2.6.1 Construction of the Reaction Matrix
The system of two nucleons is described by two spin operators σ1 and σ2 and two
identity operators I1 and I2 acting in the spin spaces of the first and second par-
ticles, respectively. As a result, the scattering matrix is a four-dimensional matrix
depending on the physical vectors σ1 and σ2, ki and kf (relative momenta of two
nucleons in the initial and final states, respectively). When constructing the NN-
scattering matrix, we follow Wolfenstein and Ashkin (1952). Since the initial (two-
nucleon) and final (also two-nucleon) systems in our case have the same internal
parity, the scattering matrix M(σ1,σ2;ki,kf ) should be a scalar function composed
of the combination of spin operators and momenta. Using the spin operators, we can
generally compose 16 combinations (complete set):
1 (scalar)
(σ1 · σ2 − 1) (scalar)
(σ1 + σ2) (axial vector)
(σ1 − σ2) (axial vector)
(σ1 × σ2) (axial vector)
lαβ = (σ1ασ2β + σ1βσ2α) (symmetric tensor).
(A)
In view of the properties of the sigma operators, these combinations cannot in-
clude terms of the orders higher than the first order.

The combinations that can be composed of the momenta ki and kf are as follows:
1 (scalar)
kf − ki = K (polar vector)
kf × ki = n (axial vector)
n × K = P (polar vector)
KαKβ,nαnβ (symmetric tensors)
PαPβ,KαPβ + KβPα (symmetric tensors).
(B)
Multiplying the quantities from sets (A) and (B), we take into account the re-
quirement of the invariance of the matrix M(σ1,σ2;ki,kf ) under rotation and inver-
sion of space. Thus, the following combinations can be included in the amplitudes
of the scattering matrix:
1, (σ 1 · σ 2 −1), (σ1 + σ2) · n, (σ1 − σ2) · n, (2.114)
(σ1 × σ2) · n, (2.115)
αβ
lαβKαKβ,
αβ
lαβnαnβ,
αβ
lαβPαPβ,
αβ
lαβ(KαPα + KβPα).
(2.116)
Combinations (2.116) can be represented in the form
σ1 · Kσ2 · K, σ1 · nσ2 · n, σ1 · Pσ2 · P; (2.117)
σ1 · Kσ2 · P + σ1 · Pσ2 · K. (2.118)
Three vectors K, n, and P are mutually orthogonal. As a result, the sum of three
terms in (2.117) is equal to the dot product σ1 · σ2, as can be veriﬁed by direct
calculations. Hence, only two of three terms in (2.117) are independent.
Now, we require the time-reversal invariance of these terms. Under time reversal
t → −t, the spin operator and momentum are transformed as follows (prime marks
the quantities with reversed time):
σ = −σ, ki = −kf , kf = −ki. (2.119)
Using Eq. (2.119) and the deﬁnition of vectors K, n, and P (see (B)), we can
show that
K = K, n = −n, and P = −P. (2.120)
Under this transformation, terms (2.115) and (2.118) change sign and, corre-
spondingly, are rejected. Finally, the nucleon–nucleon elastic scattering matrix is
expressed in the form
M(σ1,σ2;ki,kf ) = A + B(σ 1 · σ 2 −1) + C(σ1 + σ2) · n
+ D(σ1 − σ2) · n + E(σ1 · K)(σ2 · K)
+ F(σ1 · P)(σ2 · P). (2.121)

Let us introduce the triple of orthogonal unit vectors in the center-of-mass frame:
n =
k × k
|k × k |
, m =
k − k
|k − k|
, l =
k + k
|k + k |
, (2.122)
where k = ki
|ki|
and k =
kf
|kf |
are the unit vectors.
It is convenient to introduce these unit vectors, because the vectors l and m in the
nonrelativistic approximation coincide with the directions of the momenta of the
scattered and recoil particles in the laboratory frame, respectively. The scattering
matrix is rewritten in the new notation as
M(σ1,σ2;ki,kf ) = a + b(σ 1 ·n)(σ 2 ·n) + c(σ1 + σ2) · n
+ d(σ1 − σ2) · n + e(σ1 · m)(σ2 · m)
+ f (σ1 · l)(σ2 · l). (2.123)
The amplitudes a, b, c, d, e, and f are complex functions of the energy and
scattering angle (kk ) = cosθ.
The term with the amplitude d should be absent for nucleon–nucleon scattering.
This is proved as follows (Bilenky et al. 1964). Two nucleons in the initial state have
the internal parity (−1)l, total spin S, and total isospin T . According to the Pauli
exclusion principle, the wave function of two nucleons should be antisymmetric
under permutation, i.e., should change sign:
Pi = (−1)l
(−1)S+1
(−1)T +1
= −1. (2.124)
A similar relation can also be obtained for the ﬁnal nucleons with the orbital
angular momentum l , spin S , and isospin T :
Pf = (−1)l
(−1)S +1
(−1)T +1
= −1. (2.125)
To satisfy the condition Pi = Pf , we take into account that the parities of the nu-
cleons in the interaction remain unchanged and the terms containing orbital angular
momenta can be canceled. We also accept the hypothesis of the isotopic invariance
of the strong interaction; for this reason, the terms with isotopic spin T can also be
canceled. As a result, we arrive at the condition
(−1)S
= (−1)S
. (2.126)
Since the possible values of S and S are 0 (singlet state) and 1 (triplet state),
then S = S . This means that transitions only within triplets and singlets separately
are allowed in nucleon–nucleon scattering, whereas mixed singlet–triplet and in-
verse transitions are forbidden. This leads to the exclusion of the term with d in the
scattering matrix. The nucleon–nucleon scattering matrix has the ﬁnal form
M(σ1,σ2;ki,kf ) = a + b(σ 1 ·n)(σ 2 ·n) + c(σ1 + σ2) · n
+ e(σ1 · m)(σ2 · m) + f (σ1 · l)(σ2 · l). (2.127)

The singlet and triplet projection operators are introduced as
ˆS =
1
4
1 − (σ1 · σ2) , ˆT =
1
4
3 + (σ1 · σ2) . (2.128)
Then, expression (2.127) can be represented in the form
M(σ1,σ2;ki,kf ) = B ˆS + C(σ1 + σ2) · n +
1
2
G(σ1 · m)(σ2 · m)
+ (σ1 · l)(σ2 · l) +
1
2
H(σ1 · m)(σ2 · m) − (σ1 · l)(σ2 · l)
+ N(σ1 · n)(σ 2 ·n) ˆT. (2.129)
The amplitude B corresponds to singlet scattering, whereas the remaining four
amplitudes describe triplet scattering.
The amplitudes in expressions (2.127) and (2.129) are related as:
B = a − b − e − f, C = c, G = 2a + e + f,
H = e − f, N = a + b.
(2.130)
For joint description of all possible types of nucleon–nucleon scattering (pp, nn,
and np), the general matrix can be written taking into account isotopic invariance:
M(σ1,σ2;ki,kf ) = M0 ˆT0 + M1 ˆT1. (2.131)
Here,
ˆT0 =
1
4
(1 − τ1 · τ2), ˆT1 =
1
4
(3 + τ1 · τ2) (2.132)
are the isosinglet and isotriplet projection operators, respectively; and τ1 and τ2
are the isospin operators of the first and second nucleons, respectively. Each of the
matrices M0 and M1 is a scattering matrix of form (2.129).
The final wave function of the system of two nucleons can be written in the form
χf = M σ1,σ2;k,k χiSχiT , (2.133)
where χiS and χiT are the spin and isospin wave functions of the initial system of
two nucleons. In view of Eqs. (2.129) and (2.132), the requirement of the antisym-
metry of this function provides the following conditions on the amplitude under the
change θ → π − θ:
(a) the isotriplet amplitudes B, C, and H do not change their signs, whereas G and
N change their signs;
(b) on the contrary, the isosinglet amplitudes B, C, and H change their signs,
whereas G and N do not change their signs.
These relations allow one to investigate pp and nn scatterings only in the an-
gular range 0 ≤ θ ≤ 90° . Moreover, the amplitude analysis for angles 0°, 90°, and
180° can be performed only with three rather than five amplitudes; this significantly
reduces the number of necessary experimental observables.
In the case of np scattering, where both isotopic matrices M0 and M1 are used,
the measurements should be performed in a wider angular range, namely, 0 ≤
θ ≤ 180°.

2.6.2 Some Ways for Experimentally Seeking P - and
T -noninvariant Terms in the Matrices of the Strong
Interaction
The nucleon–nucleon scattering matrix given by Eq. (2.129) can be written in the
form
M(0)
= (u + v) + (u − v)(σ1 · n)(σ2 · n) + C (σ1 · n) + (σ2 · n)
+ (g − h)(σ1 · m)(σ2 · m) + (g + h)(σ1 · l)(σ2 · l), (2.134)
where l, m, and n were deﬁned in Eq. (2.122).
The aim of the complete set of experiments on nucleon–nucleon scattering is to
reconstruct the amplitudes u, v, c, g, and h from the experimental data. In the case of
parity violation or time reversal violation, the scattering matrix contains additional
terms, which are considered below.
2.6.2.1 Parity Violation
The forward NN-scattering matrix in the case of parity violation can be written in
the form
M = M0
+ (i/4)(Mos − Mso)(σ1 × σ2) · k
+ (i/4)(Mos + Mso)(σ1 − σ2) · k. (2.135)
Here, M0 is given by expression (2.134) and the amplitudes Mos and Mso de-
termine P -violating triplet–singlet and singlet–triplet transitions, respectively. The
total cross section for the interaction between polarized particles corresponding to
the matrix M is written in the form (Bilenky and Ryndin 1963; Philips 1963):
σP1P2 = σ(0)
P1P2
+ (1/4)(σos − σso)(P1 × P2) · k
+ (1/4)(σos + σso)(P1 − P2) · k, (2.136)
where P1 and P2 are the beam and target polarizations, respectively; and σos(σso)
is the total cross section for the P -odd interaction with the triplet–singlet (singlet–
triplet) transition. The cross section for the P -invariant interaction, σ
(0)
P1P2
, can be
written in the form
σ
(0)
P1P2
= σ0 + σ1(P1 · P2) + σ2(P1 · k)(P2 · k)
= σ0 + σT P1T · P2T + σLP1L · P2L, (2.137)
where the subscripts L and T mean the longitudinal and transverse polarization
components of the beam. Formula (2.136) shows that, to detect a P -odd effect, it is
necessary to measure the total cross section for the interaction of the longitudinally
polarized beam with the unpolarized target or the unpolarized beam with the trans-

versely polarized target (the third term). Such experiments have been performed,
and they will be discussed in the next sections of this book.
The second term in the formula (2.136) corresponds to simultaneous parity and
time-reversal violation. To measure it, the polarized beam and polarized target,
whose polarization vectors are perpendicular to each other and to the beam mo-
mentum, should be used. Such experiments have not yet been carried out.
2.6.2.2 T -odd Terms
If the interaction is invariant under space inversion, the term corresponding to the
T -odd effect has the form
M(1)
= MT (σ1 · l)(σ2 · m) + (σ1 · m)(σ2 · l). (2.138)
In the case of the matrix M(0), it can be shown that the polarization P of the final
nucleon for the case of the unpolarized initial states is equal to left–right asymmetry
AN in the scattering of the polarized nucleon on the unpolarized nucleon: P = AN .
Note that this equality holds for the more general case of the direct
a + b → c + d (2.139)
and inverse
c + d → a + b (2.140)
reactions. In this case, the polarization P refers to the particle c in the direct reac-
tion with the unpolarized particles a and b, whereas the asymmetry A refers to the
particle a in the inverse reaction with polarized particles (Baz 1957).
If the interaction contains the term M(1), the equality P = AN is violated and is
replaced by the relation (for NN elastic scattering)
σ0(P − A) = −8Im M∗
T · h . (2.141)
This relation should be tested at the angles at which h is noticeably nonzero.
Such a test is simplified if the unambiguous phase or amplitude analysis has been
performed.
2.7 Complete Set of Experiments
The idea of the complete set of experiments for the set of observables that com-
pletely and unambiguously determine the reaction matrix elements was first pro-
posed in Puzikov et al. (1957) and Smorodinskii (1960). In application to nucleon–
nucleon elastic scattering, the possible ways for reconstructing the matrix elements
were first proposed in Schumacher and Bethe (1961). We use these works to recon-
struct the nucleon–nucleon scattering amplitudes.

2.7 Complete Set of Experiments 83
The nucleon–nucleon elastic scattering matrix was constructed in Sect. 2.6,
where it was also shown that the total number of the independent complex am-
plitudes necessary for describing the p + p → p + p reaction at a fixed angle and a
fixed energy is five. This means that ten real quantities, namely, five absolute values
of the amplitudes and five their phases, should be measured at a fixed angle and a
fixed initial energy. Hence, these ten observables constitute a minimum set for the
complete experiment. Pion–nucleon scattering is described by two amplitudes, and
the complete set should consist of no less than four observables. The complete set
for pion–pion scattering consists of only two observables. The number of the com-
ponents of the complete experiments generally depends on the spin of the interacting
particles. The number of observables in the complete set increases with the spin. It
is seen that most observables in the complete experiment are associated with spin,
i.e., spin carries rich information on interaction.
Let us consider the examples of complete experiments at fixed angles and fixed
initial energies.
2.7.1 Complete Experiment on pp Elastic Scattering at 0° in the
Center-of-Mass Frame. The Total Cross Sections for
Nucleon–Nucleon Interactions
First, the total cross section σT for the interaction between two particles with spin
1/2 should be a scalar. Second, it should be a linear function of the polarizations
of the initial particles, P1 and P2. Third, it should be composed of the kinematic
quantities determining the reaction (Bilenky and Ryndin 1963; Philips 1963). Thus,
σ = σ0 + σ1(P1 · P2) + σ2(P1 · k)(P2 · k). (2.142)
Here, k is the unit vector in the incident beam direction and σ0, σ1, and σ2 are
the experimentally measured parameters depending only on the initial beam energy.
Their meaning is as follows. By the definition, the polarization is the mean value of
the Pauli operator; hence,
(P1 · P2) = (σ1 · σ2) = 2S2 − 3 ,
(P1 · k)(P2 · k) = (σ1 · k)(σ2 · k) = 2(S · k)2 − 1 .
(2.143)
Here, S = 1
2 (σ1 +σ2) is the total spin of two initial interacting nucleons. Accord-
ing to Eqs. (2.142) and (2.143),
(P1 · P2) =
m
wt
m − 3ws
, (P1 · k)(P2 · k) =
m
(−1)1+m
wt
m − ws
, (2.144)

where ws and wt are the probabilities of finding the system of two nucleons in
the singlet and triplet states, respectively. Taking into account the normalization
condition ws + m wt
m = 1, it follows from Eq. (2.145) that
ws
=
1
4
1 − (P1 · P2) ,
wt
0 =
1
4
1 + (P1 · P2) − 2(P1 · k)(P2 · k) ,
wt
+ + wt
− =
1
2
1 + (P1 · k)(P2 · k) .
(2.145)
Using expression (2.142) and the definition of the triplet projection operator, we
can show that wt
+ = wt
−.
Representing the total cross section in the form of the sum of the weighted cross
sections for the singlet and triplet states
σ = ws
σs
+ wt
mσt
m (2.146)
and substituting the expressions for ws and wt
m, we obtain
σ = σ0 +
1
4
σt
0 − σs
(P1 · P2) +
1
2
σt
+ − σt
0 (P1 · k)(P2 · k). (2.147)
Comparison of Eqs. (2.142) and (2.147) provides the following relation between
the coefficients:
σ1 =
1
4
σt
0 − σs
and σ2 =
1
2
σt
+ − σt
0 . (2.148)
Thus, in an experiment with polarized nucleons, three total cross sections can
be measured for the cases: (a) both nucleons are unpolarized, (b) both nucleons are
polarized transversely to the beam, and (c) both nucleons are polarized along the
beam. These three experiments constitute the complete set for determining the total
cross sections for nucleon–nucleon interactions. As a result, σs, σt
0, and σt
+ can be
reconstructed and their individual contributions to the usual (unpolarized) total cross
section σ0 can be determined:
σ0 =
1
4
σs
+
1
4
σt
0 +
1
2
σt
+. (2.149)
The applications of the above relations are discussed in the section “Polarization
experiments and results.”
2.7.2 Forward NN-Scattering Amplitudes
In view of the symmetry condition, the forward scattering amplitudes satisfy the
conditions c(0) = d(0) = 0, b(0) = e(0) and the forward scattering matrix has the
form
M(σ1,σ2;ki,kf ) = a(0) + e(0)(σ1 · σ2) + f (0) − e(0) (σ1 · k)(σ2 · k). (2.150)

The above unitarity condition (see matrix relation (2.87) in Sect. 2.4) applied to
matrix (2.150) leads to the following relations between the imaginary parts of the
amplitudes and total cross sections (optical theorem):
Ima(0) =
k
4π
σT , Ime(0) =
k
4π
σ1, Im f (0) − e(0) =
k
4π
σ2. (2.151)
Here, k is the wave number in the center-of-mass frame. Thus, the measurements
of three observables σ0, σ1, and σ2 allow one to reconstruct the imaginary parts of
three amplitudes a, e, and f of the forward pp elastic scattering.
The determination of three real parts of these amplitudes is necessary to com-
plete the reconstruction of the scattering matrix. This requires the measurements of
additional three parameters at very small angles (in the so-called Coulomb–nuclear
interference region). One of them is the differential cross section. Such a measure-
ment makes it possible to reconstruct the real part of the spin-independent ampli-
tude a(0). The measurements of two other parameters, for example, σ1 and σ2,
allow one to reconstruct the real parts of the amplitudes e and f through the dis-
persion relations. Another way is to measure the spin–spin correlation parameters
Aik (i,k = N,S,L) in the Coulomb–nuclear interference region. Here, N, S, and
L denote the initial-proton polarizations that are (N) perpendicular to the reaction
plane, (S) transverse, and longitudinal (L) with respect to the initial momentum in
the reaction plane. Thus, the complete experiment on forward pp elastic scattering
is ﬁnished. This approach is obviously universal and can be applied at any initial
energy.
Unfortunately, such a complete experiment has not yet been performed at any
energy (except for the low-energy region, where the phase analysis has been per-
formed).
Relations (2.151) are useful in a number of cases. They impose additional con-
ditions on the phases in the phase analysis, which can be substantial when choosing
between several sets of phase solutions. The use of the dispersion relation allows the
reconstruction of the real parts Re a(0), Ree(0), and Ref (0), if the imaginary parts
of these amplitudes are known from the experimental data (these imaginary parts
appear in the integrands in the dispersion relations). The differential cross section
for forward proton–proton scattering is of great interest for theoretical analysis. At
the same time, a method for its direct measurement is absent and it is necessary to
use an extrapolation method; i.e., the differential cross sections for elastic scatter-
ing are measured down to extremely small angles for which reliable data are yet
obtained and, then, they are extrapolated to zero angle using a certain function, for
example, an exponential. To test the resulting cross section, relations (2.151) are
used as follows. The differential cross section for forward scattering can be written
in the form
dσ
dΩ
=
k
4π
2
|a|2
+ 2|e|2
+ |f |2
. (2.152)

This expression contains the squares of the real and imaginary parts of each of
three amplitudes. The substitution of only imaginary parts from relations (2.151)
gives the inequality
dσ
dΩ
≥
k
4π
2
(σT )2
+ 2(σ1)2
+ (σ1 + σ2)2
. (2.153)
This is the test relation that provides a lower limit and is widely used to measure
the differential cross sections for forward scattering.
2.7.3 Complete Set of Experiments on pp Elastic Scattering at 90°
in the Center-of-Mass Frame
One of the first attempts to solve this problem was made as early as in 1959 in
Nurushev (1959). We briefly repeat the way proposed in that work.
The following five parameters of pp elastic scattering were measured at an en-
ergy of 660 MeV and an angle of 90° (Azhgirey et al. 1963): differential cross
section I = (2.07 ± 0.03) mb/sr, spin correlation parameter Cnn = (0.93 ± 0.21),
depolarization parameter D = (0.93 ± 0.17), transverse polarization rotation pa-
rameter R = (0.26 ± 0.07), and longitudinal polarization rotation parameter A =
(0.20 ± 0.06). Using these five observables, one can reconstruct three absolute val-
ues and two relative phases of the nonzero amplitudes B, C and H (see Sect. 2.6
“Nucleon–nucleon scattering”). The phase of the amplitude B is taken to be zero.
Thus, the amplitudes (dimensionless) normalized to the cross section are given by
the formulas
|b|2
=
|B|2
4I
=
1
2
(1 − Cnn), |c|2
=
2|C|2
I
=
1
4
(1 + Cnn + 2D),
|h|2
=
|H|2
2I
=
1
4
(1 + Cnn − 2D), sin(δC − δB) = −
R + A
2dc
,
cos(δH − δB) =
A − R
2bh
.
(2.154)
The substitution of the numerical values of the observables gives (Kumekin et al.
1954)
|b|2
= 0.35 ± 0.11, |c|2
= 1.00 ± 0.10, |h|2
= 0.02 ± 0.10,
sinδc = 0.39 ± 0.21, cosδh = −0.36 ± 0.18.
Comparison with similar data for lower energies shows (Nurushev 1959) that
the contributions from the triplet amplitudes c and h prevail in the energy range
under consideration, whereas the contribution from the singlet term b is smaller. The
terms h (tensor interaction) and c (spin–orbit interaction) dominate in the upper and
lower energy ranges, respectively.

2.7.4 Complete Set of Experiments on pp Elastic Scattering at an
Arbitrary Angle in the Center-of-Mass Frame
In this section, we try to answer the question: How many and what particular ob-
servables should be measured at a given initial energy in order to reconstruct the
amplitude of nucleon–nucleon elastic scattering at an arbitrary angle θ in the center-
of-mass frame? More briefly, how many measurements constitute the complete set
of experiments? Unfortunately, the complete set of experiments has not yet been
performed for any energy above 3 GeV. The direct reconstruction of the elements of
the nucleon–nucleon scattering matrix from experimental data is the single method
of analysis at energies above the meson production threshold. In such an approach,
one common phase in the scattering matrix remains undetermined. It can be deter-
mined at energies below the meson production threshold by means of the unitarity
relation. The possibility of directly reconstructing the scattering amplitudes was dis-
cussed in Puzikov et al. (1957), Smorodinskii (1960), and Schumacher and Bethe
(1961). In the last work, 11 experimental observables (the differential cross section,
polarization, and components of the depolarization, polarization transfer, and polar-
ization correlation tensors) were used and the absolute values of five amplitudes, as
well as four relative phases, were unambiguously reconstructed. In agreement with
the expectations, one common phase was undetermined. According to Schumacher
and Bethe (1961), the complete set includes tensors up to the second order. The pos-
sibilities of simplifying the procedure for reconstructing the amplitudes with the use
of the polarization tensors of the third and fourth orders were considered in Bilenky
et al. (1965) and Vinternitts et al. (1965).
It is not excluded that theoretical ideas can noticeably reduce the number of nec-
essary experiments of the complete set at asymptotic energies.
Below, we discuss the method for reconstructing the scalar amplitudes of
nucleon–nucleon scattering in the relativistic case and present the particular sets
of the complete set of experiments. In this presentation, we follow Bilenky et al.
(1966).
We write the nucleon–nucleon scattering matrix in the form (Wolfenstein and
Ashkin 1952; Dalitz 1952)
M p ,p = (u + v) + (u − v)(σ1 · n)(σ2 · n) + c(σ1 + σ2) · n
+ (g − h)(σ1 · m)(σ2 · m) + (g + h)(σ1 · l)(σ2 · l). (2.155)
Here, the complex scalar scattering amplitudes u,v,c,g, and h are functions of
the energy and scattering angle θ. Our main aim is to express these amplitudes in
terms of experimental quantities. The unit vectors n, l, m are given by the expres-
sions
l =
p + p
|p + p|
, m =
p − p
|p − p|
, n = l × m =
p × p
|p × p |
. (2.156)
These unit vectors defined in the center-of-mass frame are mutually orthogo-
nal. In the nonrelativistic approximation, the vectors l and m are directed along the

scattered and recoil particle momenta in the laboratory frame, respectively. In the
relativistic case, this relation is invalid and an additional rotation angle appears.
The proton–proton scattering matrix should satisfy the Pauli exclusion principle:
M p ,p = −Π(1,2)M −p ,p = −M p ,−p Π(1,2). (2.157)
Here,
Π(1,2) =
1
2
(1 + σ1 · σ2) (2.158)
is the operator of the permutation of the spin variables. The substitution of pp-
scattering matrix (2.155) into relation (2.157) indicates that the scalar scattering
amplitudes satisfy the following symmetry conditions:
u(π − θ) = −u(θ), h(π − θ) = h(θ),
c(π − θ) = c(θ), v(π − θ) = −g(θ).
(2.159)
It follows from these relations that, for example, three amplitudes, c, h and, e.g.,
g, are nonzero for an angle of 90°. Their reconstruction was considered above.
Neutron–proton elastic scattering was not discussed above. Such a discussion is
most appropriate with the isotopic invariance hypothesis. In this case, two isotopic
scattering matrices M1(p ,p) and M0(p ,p) with isospins 1 and 0, respectively, can
be introduced. Both matrices are written in the form, where the scalar amplitudes
have subscripts 1 and 0, and the matrix M1(p ,p) coincides with the pp-scattering
matrix. In this case, the np-scattering matrix is determined by the expression
Mnp p ,p =
1
2
M1 p ,p + M0 p ,p . (2.160)
The generalized Pauli exclusion principle can be applied to the isotopic scattering
matrices. Introducing the subscript i = 0,1, we write the Pauli condition as follows:
Mi p ,p = (−1)i
Π(1,2)Mi −p ,p = (−1)i
Mi p ,−p Π(1,2). (2.161)
This condition provides the following constraints on the scalar scattering ampli-
tude in different isotopic states:
ui(π − θ) = (−1)iui(θ), hi(π − θ) = (−1)i+1hi(θ),
ci(π − θ) = (−1)i+1ci(θ), vi(π − θ) = (−1)igi(θ).
(2.162)
In the nonrelativistic approximation, the problem of the joint analysis of the pp-
and np-scattering data for reconstructing the scattering matrix was considered in
Kazarinov (1956) and Golovin et al. (1959).
A theoretical analysis of collisions between particles is usually performed in the
center-of-mass frame. However, observables such as cross sections and polarizations
are measured in the laboratory frame. Therefore, it is necessary to determine the
rules for the transition from one frame to the other taking into account the relativistic
kinematics and the speciﬁcity of the spin transformations. Let us consider particular
examples.

According to the general rules, the mean value of any spin operator σ1L is
expressed as
σ1L aL = Tr σ1 aL R
ρf /Tr ρf . (2.163)
Here, ρf is the density matrix of the final state and aL is an arbitrary unit vector
in the laboratory frame (L frame). It is intended to measure the projection of the po-
larization vector of the first particle on this direction. The vector (aL)R = Rn(Ω )aL
includes the relativistic transformation of the spin vector and is obtained from the
vector aL by means of its rotation by the angle Ω = θ − 2θL about a vector perpen-
dicular to the scattering plane. In the nonrelativistic limit, the center-of-mass scat-
tering angle θ is equal to 2θL, where θL is the laboratory scattering angle. Hence,
Ω = 0 in this case.
For the measurement of the projection of the polarization vector of the recoil
particle (particle 2) on the direction of the unit vector bL in the L frame, we have
σ2L bL = Tr σ2 bL R
ρf /Tr ρf . (2.164)
It is intended to measure the projection of the polarization vector of the second
particle on this direction. The vector (bL)R = Rn(Ω )bL includes the relativistic
transformation of the spin vector and is obtained from the vector bL by means of its
rotation by the angle Ω = 2ϕL − ϕ about a vector perpendicular to the scattering
plane, where ϕ = π − θ and ϕL are the emission angles of the second particle in the
center-of-mass frame and L frame, respectively.
The correlation of polarization projections on the same unit vectors (aL,bL) can
be measured in the experiment:
σ1 · aL σ2 · bL = Tr σ1 · aL R
σ2 · bL R
ρf /Tr ρf . (2.165)
Let us introduce the following three sets of the orthonormalized unit vectors in
the laboratory frame:
nL, kL, sL = nL × kL, (2.166)
nL, kL, sL = nL × kL, (2.167)
nL, kL, sL = nL × kL. (2.168)
Here, kL, kL, kL are the unit vectors in the directions of the momenta of the inci-
dent, scattered, and recoil nucleons, respectively, and nL = kL ×kL/|
←
kL ×kL| is the
unit vector perpendicular to the scattering plane; nL = n, where n is the unit vector
perpendicular to the scattering plane in the center-of-mass frame. In the subsequent
presentation, the initial, scattered, and recoil particles are described in coordinate
systems (2.166), (2.167), and (2.168), respectively. When calculating observables,
we use tensors up to the second rank.
1. The zero-rank tensor is the differential cross section. It is given by the expression
σ0 = Tr ρf =
1
4
Tr MM+
= 2 |u|2
+ |v|2
+ |c|2
+ |g|2
+ |h|2
. (2.169)
2. The first-rank tensors:

(a) The initial particles are unpolarized. The i-th component of the polarization
of the scattered particle is measured. It is given by the expression
P1iσ0 =
1
4
Tr σ1iMM+
. (2.170)
(b) The initial particles are unpolarized. The i-th component of the polarization
of the recoil particle is measured. It is given by the expression
P2iσ0 =
1
4
Tr σ2iMM+
= P1iσ0. (2.171)
(c) The incident (first) particle is polarized in the i-th direction. The target (sec-
ond) particle is unpolarized. The i-th component of the asymmetry is mea-
sured. It is given by the expression
A1iσ0 =
1
4
Tr Mσ1iM+
. (2.172)
(d) The incident (first) particle is unpolarized. The target (second) particle is
polarized in the i-th direction. The i-th component of the asymmetry of the
recoil particle is measured. It is given by the expression
A2iσ0 =
1
4
Tr Mσ2iM+
= A1iσ0. (2.173)
3. The second-rank tensors:
(a) The depolarization tensor Dik. The first particle is polarized (component i)
in the initial state, and its polarization (component k) after scattering is mea-
sured. The expression for the tensor Dik has the form
Dikσ0 =
1
4
Tr σ1iMσ1kM+
. (2.174)
(b) The depolarization tensor of the second particle D
(2)
ik . The second particle is
polarized (component i, the first particle is unpolarized) in the initial state,
and its polarization (component k) after scattering is measured. The expres-
sion for the tensor D
(2)
ik has the form
D
(2)
ik σ0 =
1
4
Tr σ2iMσ2kM+
= Dikσ0. (2.175)
(c) The polarization transfer tensor from the first particle to the second Kik. The
first particle is polarized (the second particle is unpolarized) in the initial
state, and the polarization of the second particle after scattering is measured.
The expression for the tensor Kik has the form
Kikσ0 =
1
4
Tr σ1iMσ2kM+
. (2.176)
(d) The polarization transfer tensor from the second particle to the first one K
(2)
ik .
The second particle is polarized (the first particle is unpolarized) in the initial

state, and the polarization of the first particle after scattering is measured.
The expression for the tensor K
(2)
ik has the form
K
(2)
ik σ0 =
1
4
Tr σ2iMσ1kM+
= Kikσ0. (2.177)
(e) The polarization correlation tensor Cik. Both particles in the initial state are
unpolarized, and the correlation of their polarizations after scattering is mea-
sured. The expression for the tensor Cik has the form
Cikσ0 =
1
4
Tr σ1iσ2kMM+
. (2.178)
(f) The polarization correlation tensor C
(2)
ik . Both particles in the initial state
are unpolarized, and the correlation of their polarizations after scattering is
measured. The difference from (e) is that the Pauli operators are enclosed by
M matrices. The expression for the tensor C
(2)
ik has the form
C
(2)
ik σ0 =
1
4
Tr σ1iσ2kMM+
. (2.179)
(g) The two-spin asymmetry tensor or asymmetry correlation tensor Aik. Both
particles in the initial state are polarized: the first and second particles are
polarized in the directions i and k, respectively. Asymmetry after scattering
is measured. The expression for the tensor Aik has the form
Aikσ0 =
1
4
Tr Mσ1iσ2kM+
. (2.180)
(h) The two-spin asymmetry tensor A
(2)
ik . Both particles in the initial state are
polarized: the first and second particles are polarized in the directions k and i,
respectively. Asymmetry after scattering is measured. The expression for the
tensor A
(2)
ik has the form
A(2)
ik σ0 =
1
4
Tr Mσ1kσ2iM+
= Aikσ0. (2.181)
The requirements of the invariance of strong interactions under a number of con-
tinuous (isotropy and uniformity of space) and discrete (space inversion and time
reversal) transformations lead to relations between measured quantities. For exam-
ple,
Pi = Ai = Pni Cik p ,p = Aik −p,−p . (2.182)
For the case of the beam with polarization P1 and the polarized target with polariza-
tion P2 (or two colliding polarized proton beams with the indicated polarizations),
the general requirements of the invariance of the scattering matrix under the trans-
formations listed above in the parentheses provide the following expression for the
differential cross section
σ(P1,P2) = σ0 1 + P(P1 + P2) · nL + Ann(P1 · nL)(P2 · nL)
+ Ass(P1 · sL)(P2 · sL) + Akk(P1 · kL)(P2 · kL)
+ Ask (P1 · sL)(P2 · kL) + (P1 · kL)(P2 · sL) . (2.183)

Here, P is the polarization appearing after the scattering of the unpolarized par-
ticles and
Aab = (aL)iAik(bL)k. (2.184)
For the measurements of the depolarization parameters Dik or the polarization
transfer tensor Kik, additional scattering is necessary in order to determine the po-
larization of the scattered particles. Such experiments are very difficult because,
first, luminosity in the second scattering process is low and, second, it is difficult to
find analyzers with a high analyzing power at high energies. This is one of the main
reasons why, for example, such experiments are not planned at the RHIC collider.
In contrast to these parameters, the asymmetry correlation tensor Aik is directly
measured at RHIC. For the measurements of, for example, the parameter All, the
polarization of both initial particles should be oriented along their momentum; i.e.,
both beams should be longitudinally polarized. To measure, for example, Asl, one
beam of particles should be polarized along the vector s, and the other, along the
vector l. All these possibilities can be implemented at the RHIC polarized particle
collider.
As mentioned above, the measurements of observables are carried out in the
laboratory frame, whereas all theoretical jobs with these quantities are performed in
the center-of-mass frame. Let us determine the transformations between these two
frames. First, we recall the relation between the triples of unit orthogonal vectors in
the laboratory and center-of-mass frames:
nL = n, kL = k, sL = s. (2.185)
Here, k = p/|p| is the unit vector in the direction of the incident particle momen-
tum in the center-of-mass frame, n is the unit vector perpendicular to the scattering
plane in this frame, and s = n × k is the vector that is perpendicular to the mo-
mentum of the initial particles and lies in the scattering plane. Thus, we obtain the
relations
Ass = C+ − Clm sinθ − C− cosθ,
Akk = C+ + Clm sinθ + C− cosθ,
Ask = −Clm cosθ + C− sinθ.
(2.186)
Here,
C+ =
1
2
(Cll + Cmm), C− =
1
2
(Cll − Cmm). (2.187)
The components Ann and Cnn perpendicular to the scattering plane are the same
in both frames. The other components of the tensors satisfy the relations
C+ =
1
2
(Ass + Akk), (2.188)
Clm = −Ask cosθ +
1
2
(Akk − Ass)sinθ, (2.189)
C− = Ask sinθ +
1
2
(Akk − Ass)cosθ. (2.190)

Here, the polarization correlation tensors Cik are expressed in terms of the asym-
metry correlation tensors Ann. This representation is reasonable. Indeed, the tensors
Cik can also be measured with the use of unpolarized initial particles. However,
this measurement requires the analysis of the final polarization (double scattering)
and, therefore, analyzing scattering. As mentioned above, such experiments are very
difficult. At the same time, it is easier to measure the parameters Ann (in single scat-
tering) and to determine Cik in terms of these parameters by the above formulas.
In the first experiments at low energies (beam kinetic energy 100–600 MeV), the
spin correlation tensor Cnn was measured and other parameters were measured later.
We consider these tensors:
Cs s = sL Ri
Cik sL Rk
, Cs k = sL Ri
Cik kL Rk
,
Ck s = kL Ri
Cik sL Rk
, Ck k = kL Ri
Cik kL Rk
.
(2.191)
Let us apply the rotation operator about the vector n to an arbitrary vector a and
represent the result in terms of three vectors:
Rn(Ω)a = (a · n)n(1 − cosθ) + a cosΩ + (n × a)sinΩ. (2.192)
Using this relation and formulas (2.183), we can find that
kL R
= Rn Ω kL = l cosα + msinα, (2.193)
sL R
= Rn Ω sL = −l sinα + mcosα. (2.194)
Here, the relativistic spin rotation angle is α = θ/2 − θL, where θ and θL are the
particle scattering angles in the center-of-mass and laboratory frames, respectively.
Similar transformation formulas can be obtained for the recoil particle:
kL R
= Rn Ω kL = −l sinα − mcosα , (2.195)
sL R
= Rn Ω sL = l cosα − msinα . (2.196)
Here, the relativistic spin rotation angle is α = ϕ/2−ϕL, where ϕ and ϕL are the
scattering angles of the recoil particle in the center-of-mass and laboratory frames,
respectively.
We emphasize two features. First, relativistic spin precession (Thomas preces-
sion) concerns only the polarization components lying in the scattering plane and
does not involve its normal component. Second, in the nonrelativistic limit, angles
are α = α = 0 and Thomas precession at low energies does not affect observables.
In the nonrelativistic case, we have the equalities
kL R
= l, sL R
= m, kL R
= −
←
m, sL R
= l. (2.197)
Let us express experimentally measured quantities (2.189) in terms of the tensors
in the center-of-mass frame using relations (2.191)–(2.194). The desired formulas
have the form
Cs s = −C+ sin α + α + Clm cos α − α − C− sin α − α . (2.198)
Cs k = −C+ cos α + α + Clm sin α − α + C− cos α − α . (2.199)

Ck s = C+ cos α + α + Clm sin α − α + C− cos α − α . (2.200)
Ck k = −C+ sin α + α − Clm cos α − α + C− sin α − α . (2.201)
It is easy to verify that the observables are related as
(Cs s + Ck k )/(Cs k − Ck s ) = tan α + α . (2.202)
Hence, the number of the independent observables is three rather than four.
Let us solve the inverse problem, i.e., express three parameters C+,, C−, Clm in
terms of the observables Cs s , Cs k , and Ck s . Using relations (2.196)–(2.199), we
obtain
C+ = (Ck s − Cs k )/2cos α + α , (2.203)
C− =
1
2
(Cs k + Ck s )cos α − α
− Cs s +
1
2
tan α + α (Ck s − Cs k ) sin α − α , (2.204)
Clm =
1
2
(Cs k + Ck s )sin α − α
+ Cs s +
1
2
tan α + α (Ck s − Cs k ) cos α − α . (2.205)
Now, we consider the case with the polarized beam and unpolarized target. The
polarization components of the scattered particles are measured after collisions. In
view of invariance, these components can be written in the form
σ(P1) σ1 L · nL = σ0 P + Dnn(P1 · nL) , (2.206)
σ(P1) σ1 L · kL = σ0 Dk k(P1 · kL) + Dk s(P1 · sL) , (2.207)
σ(P1) σ1 L · sL = σ0 Ds k(P1 · kL) + Ds s(P1 · sL) . (2.208)
Here, σ(P1) is the differential cross section for the scattering of particles with
polarization P1 arbitrarily oriented in space on the unpolarized target; it is given by
the expression
σ(P1) = σ0 1 + P(P1 · n) . (2.209)
Taking certain components of the initial and ﬁnal polarizations, we arrive at the
known Wolfenstein parameters:
Dnn = D = (nL)iDik(nL)k, Ds s = R = sL Ri
Dik(sL)k; (2.210)
Ds k = A = sL Ri
Dik(kL)k, Dk s = R = kL Ri
Dik(sL)k; (2.211)
Dk k = A = kL Ri
Dik(kL)k. (2.212)
The spin rotation parameters (A,R,A ,R ) contain the ﬁnal unit vectors with
the subscript R, which means the necessity of the inclusion of the relativistic spin
rotation in the reaction plane.

Referring readers interested in the details of the derivation of the following for-
mulas to Bilenky et al. (1965), we present the expressions of physical observables
in terms of the scattering amplitude in the relativistic case:
σ0 = 2 |u|2
+ |v|2
+ |c|2
+ |g|2
+ |h|2
, (2.213)
σ0Dnn = 2 |u|2
+ |v|2
+ |c|2
− |g|2
− |h|2
, (2.214)
σ0Knn = 2 |u2
| − |v|2
+ |c|2
+ |g|2
− |h|2
, (2.215)
σ0Cnn = 2 |u2
| − |v|2
+ |c|2
− |g|2
+ |h|2
, (2.216)
σ0P = 4Recu∗
, (2.217)
σ0D+ = 4Reuv∗
, (2.218)
σ0D− = 4Regh∗
, (2.219)
σ0Dlm = 4Imcv∗
, (2.220)
σ0K+ = 4Reug∗
, (2.221)
σ0K− = 4Revh∗
, (2.222)
σ0Klm = 4Imcg∗
, (2.223)
σ0C+ = 4Revg∗
, (2.224)
σ0C− = 4Reuh∗
, (2.225)
σ0Clm = −4Imch∗
. (2.226)
These 14 experimental observables are used to reconstruct five amplitudes and
their phases. In this case, one common phase remains undetermined. Fundamentally,
it also can be reconstructed. However, this problem is not discussed here.
One of the variants of reconstruction of the amplitudes is as follows (for details,
see Bilenky et al. 1966):
|g|2
=
1
8
σ0(1 + Knn − Dnn − Cnn); (2.227)
|h|2
=
1
8
σ0(1 − Knn − Dnn + Cnn); (2.228)
|v|2
=
1
8
σ0(1 − Knn + Dnn − Cnn); (2.229)
|u|2
+ |c|2
=
1
8
σ0(1 + Knn + Dnn + Cnn). (2.230)
For further analysis, it is necessary to fix the phase of any of five amplitudes. For
example, let the amplitude c be real positive. This means that the scattering matrix
is determined up to the phase of the amplitude c. Under this condition, we obtain
Reu =
1
4c
σ0P, Imh =
1
4c
σ0Clm, (2.231)
Imv = −
1
4c
σ0Dlm, Img = −
1
4c
σ0Klm. (2.232)

For further calculations, we write the following identity for any two complex
quantities:
|x|2
|y|2
− Rexy∗ 2
= |x|2
(Imy)2
+ |y|2
(Imx)2
− 2Rexy∗
Imx Imy. (2.233)
Taking x = g and y = h and using Eqs. (2.231) and (2.232), we obtain
c2
=
|g|2M2 − |h|2N2 − 2Regh∗MN
|g|2|h|2 − (Regh∗)2
. (2.234)
Here, the quantities |g|2, |h|2, and Regh∗ were determined above and
M =
1
4
σ0Clm, N = −
1
4
σ0Klm. (2.235)
It remains to determine the signs of the quantities
Imu, Reh, Rev, Reg. (2.236)
One relation can be written immediately in the form
Regh∗
= Reg Reh + Img Imh =
1
4
σ0D−. (2.237)
Using Eq. (2.218), we can determine the signs of Imu and Rev. Any unused
equation from the set of Eqs. (2.213)–(2.226) can be used to eliminate the remaining
ambiguity.
Thus, the problem of the relativistic reconstruction of the nucleon–nucleon scat-
tering matrix has been solved in the general form. Since the volume of the book is
limited, we omit such interesting problems as the reconstruction of the amplitudes
by means of the measurements of the polarization parameters of the recoil particles
and the joint analysis of pp and np scatterings using isotopic invariance.
2.8 Partial Wave Analysis
The results of polarization experiments on the elastic scattering of nucleons in the
low-energy range (0.1–10 GeV) were considered using the partial wave analysis
(Hoshizaki 1968; Matsuda 1993). In this method, the scattering amplitude is ex-
panded in terms of the eigenfunctions of the complete set of conserving operators
and the expansion coefﬁcients are the elements of the scattering matrix S.
These elements are expressed in terms of the phase shifts, which contain com-
plete information on the interaction process. There are several reasons to apply this
method. First, the number of the phases directly depends on the maximum orbital
angular momentum Lmax in the interaction. According to nonrelativistic quantum
mechanics, Lmax is related to the impact parameter b as (Lmax + 1
2 ) ≈ bpi, where
pi is the incident particle momentum in the center-of-mass frame. According to this
relation, Lmax increases with energy; thus, the phase analysis becomes impossible
when the number of free parameters is equal to or larger than the number of ex-
perimental points. The situation is further complicated at energies above the meson

2.8 Partial Wave Analysis 97
production threshold, when the phases become complex. This is the main reason
why the phase analysis is not applied for high energies.
Nevertheless, the phase analysis was widely used for low energies in 1950–1960
for several reasons. First, the number of the phases and, correspondingly, the num-
ber of free parameters are small for low energies. For example, taking b = 1.5mπ c
for the impact parameter, we can estimate Lmax = 1 and 3 at the kinetic energy
T = 50 and 300 MeV, respectively. Hence, it is easy to perform the phase analysis.
The second important reason to perform this analysis is that the phases, being de-
pendent on the impact parameter, make it possible to scan the internal structure of
the nucleon. If the particles pass at a large distance from each other, the interaction
between them is weak and the phases are small. On the contrary, for the central col-
lision (L = 0), the phases are expected to be large. Deviations from this picture are
obviously possible, for example, in the presence of repulsion forces or when reso-
nances are formed. The third reason to perform the phase analysis at low energies
is that the angular dependence of any observable can be calculated (predicting its
behavior) using a few phases and, then, the calculation can be compared with exper-
imental data. Fourth, any theoretical model should be tested on the phase analysis
data if the phase analysis is unambiguous.
Let us express the scattering matrix elements in terms of the phase.
In view of the unitarity of the S-matrix, it can be written in terms of the phase
operator δ as follows:
S = eiδ
. (2.238)
The nucleon–nucleon interaction matrix should conserve the total angular mo-
mentum J , total spin S, and parity Π = (−1)l. The requirement of the antisymmetry
under the permutation of two nucleons leads to the relation
(−1)S+1+T +1
Π = (−1)S+T +L
= −1. (2.239)
Here, T is the isospin of the system of two nucleons. Relation (2.239) should be
applied separately for the system of two nucleons in the initial and final states. Tak-
ing into account this relation, the elements of the S-matrix can be characterized by
three quantum numbers: the total angular momentum J , the total spin of the system
of two nucleons S, and the orbital angular momentum L, because T is unambigu-
ously determined from the above relation; i.e., T can be omitted when specifying
the elements of the S-matrix.
Let us consider the matrix M defined by the expression
M(pf ,pi) =
2π
ik
θf ϕf |S − 1|θiϕi . (2.240)
Here, |θiϕi and θf ϕf | are the wave functions of the initial and final system of
two nucleons, respectively; and θn, ϕn, where n = i, f , are the angles of the mo-
menta pi, pf . The elements of this matrix in spin space are given by the expression
Sms M(pf ,pi) S ms =
2π
ik
θf ϕf ,Sms e2iδ
− 1 S ms,θiϕi . (2.241)

This expression can be rewritten in terms of the spherical functions of the angles
using the properties of the completeness and orthogonality of the wave functions
2π
ik
θf ϕf |LmL LmLSmS|LSJmJ LSJmJ |e2iδ
− 1 L S J mJ
× L S J mJ L mLS mS L mL θiϕi . (2.242)
Here,
θϕ|LmL = YmL
L (θ,ϕ). (2.243)
Below, we use the following notation for the Clebsch–Gordan coefficients ap-
pearing in expression (2.242):
CLS(JmJ mLmS) = LmLSmS|LSJmJ . (2.244)
The quantization axis is usually taken in the direction of the incident particle
momentum; in this case, the angles θi and ϕi are zero and the angles θf and ϕf are
the scattering angles of the final particles. After these simplifications, since S, J ,
and mJ are conserving quantum numbers, expression (2.242) can be represented in
the form
Sms M(pf ,pi) S ms
= δSS
4π
ik
L
L+S
J=|L−S|
L+S
L =|J−S|
2L + 1
4π
Y
ms−ms
L (θ,ϕ)
× CLS J,ms,ms − ms,ms CLS J,ms,0,ms LSJms e2iδ
− 1 L SJms .
(2.245)
The summation should be performed taking into account antisymmetry condition
(2.239) and that the scattering matrix elements are independent of the projection of
the total angular momentum, mJ , due to the isotropy of space. The nonzero elements
of the matrix S − 1 are denoted as
RL = L0LmJ e2iδ
− 1 L0LmJ ;
RLJ = L1JmJ e2iδ
− 1 L1JmJ ; (2.246)
−RJ
± = RJ
= J ± 1,1,J,mJ e2iδ
− 1 J ∓ 1,1,J,mJ .
Time reversal invariance leads to the equality RJ
+ = RJ
− = RJ . For further sim-
plification, we introduce the following notation for the triplet and singlet elements:
Mmsms
= 1ms M 1ms , Mss = 00|M|00 . (2.247)
As a result, the nonzero matrix elements for the singlet and triplet states can be
written in the form
Mss =
2π
ik
2L + 1
4π
RLY0
L(θ,ϕ) (2.248)
for the singlet transitions and

Mmsms
=
2π
ik
L
L+1
J=L−1
2L + 1
4π
CL1 J,ms,ms − ms,ms CL1 J,ms,0,ms RLJ
−
1
J=L±1
2L + 1
4π
CL1 J,ms,ms − ms,ms CL1 J,ms,0,ms RJ
× Y
ms−ms
L (θ,ϕ) (2.249)
for the triplet transitions. Here, L = 2J − L in the second term. When applying
these formulas to pp elastic scattering, it is necessary to take into account two cir-
cumstances. First, two protons are identical; therefore, measuring instruments can-
not determine which proton, from the beam or from the target, is detected. As a
result, the number of counts is twice as large as that following from the above con-
sideration. Thus, to compare with the theoretical cross section, the measured cross
sections should be halved. The second circumstance is associated with the antisym-
metry condition (Pauli exclusion principle). Owing to this circumstance, the partial
amplitudes with spins S = 0 and 1 include only even and odd orbital angular mo-
menta, respectively. Since the orbital angular momentum is not conserved, for a
given total angular momentum J in the triplet state, there are two states differing
in the orbital angular momentum. As an example, we point to the states 3P2–3F2,
3F4–3H4, etc. of the pp system. For each pair of such states, the mixed transitions
3P2 ↔3 F2, 3F4 ↔ 3H4 are possible in addition to the direct transitions 3P2 → 3P2,
3F2 → 3F2. Correspondingly, the phases describing the direct transitions should
be supplemented by additional parameters for describing the mixed transitions.
Such parameters are called mixing parameters and are usually denoted as εJ . To
describe the mixed transitions in the absence of inelastic channels, the following
two-dimensional symmetric unitary submatrix is introduced:
SJ − 1 =
RJ−1,J −RJ
−RJ RJ+1,J
. (2.250)
An unambiguous method for parameterizing this matrix in terms of the phase is
absent. One of the methods was proposed by Blatt and Weisskopf (1952) and Blatt
and Biedenharn (1952). This method is the diagonalization of the matrix by means
of a unitary transformation
SJ = GSJ G−1
, (2.251)
where
SJ =
e2iδJ−1,j 0
0 e2iδJ+1,j
, G =
cosεJ −sinεJ
sinεJ cosεJ
. (2.252)
This set of phase shifts is called the proper phase shift and is convenient if the
Coulomb interaction can be neglected.
Another parameterization variant was proposed in Stapp et al. (1957) in the form
SJ = ˜SJ
˜G ˜SJ , (2.253)

where
˜SJ =
ei ¯δJ−1,j 0
0 ei ¯δJ+1,j
, ˜G =
cos2¯εJ i sin2¯εJ
i sin2¯εJ cos2¯εJ
. (2.254)
This parameterization is favorable over parameterization (2.252), because it pro-
vides the estimation of the mixing parameters for low orbital angular momenta
(where the nuclear interaction prevails over the Coulomb interaction) in the pure
form (without the Coulomb contribution). An additional advantage of this param-
eterization is that it allows one to more clearly separate the nuclear and Coulomb
contributions. For this reason, this parameterization is more often used in the phase
analysis.
The relation between representations (2.251) and (2.253) can be found by equat-
ing them to each other, because they are the elements of the same matrix. This
relation is expressed as follows:
δJ−1,J + δJ+1,J = ¯δJ−1,J + ¯δJ+1,J ,
sin(δJ−1,J − δJ+1,J ) = sin2¯εJ /sin2εJ ,
sin(¯δJ−1,J − ¯δJ+1,J ) = tan2¯εJ /tan2εJ .
(2.255)
Table 2.1 presents the elements of the matrix M in terms of the partial waves h.
These matrix elements can also be used in the case of neutron–proton elastic scat-
tering with three changes: (a) the Coulomb amplitudes are neglected, (b) all sums
over the even or odd L values are extended over all L values (even and odd), and
(c) the resulting sums are multiplied by a factor of 1/2.
The partial nuclear amplitudes h are expressed in terms of the scattering phases
by the formulas
2ikhl = e2i ¯δN
l − 1 e2iΦl (2.256)
for the singlet states and
2ikhlj = e2i ¯δN
lj − 1 e2iΦl (2.257)
for the triplet states.
For the mixed singlet–triplet states,
2ikhj±1,j = cos2εN
j e
2i ¯δN
j±1,j − 1 e2iΦl (2.258)
2khj
= sin2εN
j e
i(¯δN
j−1,j +¯δN
j+1,j )
, (2.259)
where εN
j is the mixing parameter for the total angular momentum j and the super-
script N means that this parameter refers to pure nuclear scattering.
The Coulomb amplitudes are deﬁned as follows:
fc(θ) =
−n
k(1 − cosθ)
e−inlog[(1−cosθ)/2]
, (2.260)
where n = e2
v and v is the relative velocity in the center-of-mass frame. The sym-
metrized and antisymmetrized Coulomb amplitudes used in the partial wave analy-
sis are presented in Table 2.1.

Table 2.1 Singlet–triplet matrix elements for pp elastic scattering in terms of the partial ampli-
tudes h. The Coulomb interaction contributions are presented in the explicit form with allowance
for the identity of protons
1 Mss = fc,s + 2 even l(2l + 1)hlPl
2 M11 = fc,a + odd l[(l + 2)hl,l+1 + (2l + 1)hl,l + (l − 1)hl,l−1
−
√
(l + 1)(l + 2)hl+1 −
√
(l − 1)lhl−1]Pl
3 M00 = fc,a + 2 odd l[(l + 1)hl,l+1 + lhl,l−1 + (l − 1)hl,l−1
+
√
(l + 1)(l + 2)hl+1 +
√
(l − 1)lhl−1]Pl
4 M01 =
√
2 odd l[−l+2
l+1 hl,l+1 + 2l+1
l(l+1) hl,l + l−1
l hl,l−1
+ l+2
l+1 hl+1 − l−1
l hl−1]P 1
l
5 M10 =
√
2 odd l[hl,l+1 − hl,l−1 + l+2
l+1 hl+1 − l−1
l hl−1]P 1
l
6 M1−1 = odd l[ 1
l+1 hl,l+1 − 2l+1
l(l+1) hl,l + 1
l hl,l−1
− 1√
(l+1)(l+2)
hl+1 − 1√
(l−1)l
hl−1]P 2
l
7 M11 − M00 − M1−1 −
√
2ctgθ(M10 + M01) = 0
8 fc,s = fc(θ) + fc(π − θ),fc,a = fc(θ) − fc(π − θ), where fc is
the Coulomb amplitude
Pl, P 1
l , and P 2
l are the associated Legendre polynomials of the zeroth, ﬁrst, and second orders,
respectively
Let us consider the separation of the Coulomb and nuclear contributions in the
phase analysis. Since Coulomb forces are long-range and nuclear forces are short-
range, i.e., they weakly overlap, it is usually accepted that the pure nuclear, ¯δN , and
Coulomb, φ, phases are added. In this case,
¯δN
L = ¯δL − φL, ¯δN
JL = ¯δJL − φL, ¯εN
J = ¯εJ . (2.261)
These phases denoted by the overlined symbols with the superscript N are called
pure nuclear phases. The Coulomb phases are calculated by the formula (Stapp et
al. 1957):
φL ≡ ηL − η0 =
L
x=1
atan
n
x
. (2.262)
Here, n = e2
v and v is the relative velocity. Let us introduce the Coulomb scatter-
ing matrix by the formula Rc = Sc − 1. Then, the general reaction matrix is written
in the form
R = S − 1 = ε + Rc, α = S − Rc, (2.263)
where ε is the mixing parameter for a given j value.
The matrix α corresponding to pure nuclear scattering can be expanded in partial
waves, whereas Rc is calculated exactly and is given by the expression

Table 2.2 Expressions for
experimentally measured
quantities in terms of the
amplitude of the pp elastic
scattering matrix in the
nonrelativistic case
1 σ0 = |a|2 + |b|2 + 2|c|2 + |e|2 + |f |2
2 σ0Dnn = |a|2 + |b|2 + 2|c|2 − |e|2 − |f |2
3 σ0Dll = |a|2 − |b|2 − |e|2 + |f |2
4 σ0Dmm = |a|2 − |b|2 + |e|2 − |f |2
5 σ0Dml = 2Imc∗(a − b)
6 σ0P0 = 2Rec∗(a + b)
7 σ0Cml = 2Imc∗(e − f )
8 σ0Kml = 2Imc∗(e + f )
9 σ0Cnn2 = Reab∗ + |c|2 − Reef ∗
10 σ0Knn2 = Reab∗ + |c|2 + Reef ∗
11 σ0Cll2 = Reaf ∗ − Rebe∗
12 σ0Kll2 = Reaf ∗ + Rebe∗
13 σ0Cmm2 = Reae∗ − Rebf ∗
14 σ0Kmm2 = Reae∗ + Rebf ∗
f |Rc|i =
ik
2π
fc(θ), fc(θ) = −
n
k(1 − cosθ)
exp −inlog
1 − cosθ
2
.
(2.264)
The partial amplitudes h are calculated using formulas (2.246), (2.253), and
(2.261), as well as the relation α = 2ikh. These expressions have the form
hL =
1
2ik
exp 2iδN
L − 1 exp(2iφL) (2.265)
for the singlet state and
hLJ =
1
2ik
exp 2i ¯δN
LJ − 1 exp(2iφL),
hJ±1,J =
1
2ik
cos2εN
J exp 2i ¯δN
J±1,J − 1 exp(2iφJ±1),
hJ
=
1
2ik
sin2εN
J exp i ¯δN
J−1,J + i ¯δN
J+1,J
(2.266)
for the triplet state.
These relations make it possible to express the elements of the matrix M in terms
of the phase shifts; hence, experimental observables are determined in terms of the
phases. Thus, the phase analysis is possible.
Table 2.2 presents the expressions for measured quantities in terms of the ampli-
tudes a,b,c,e, and f expressed through linear relations with the matrix elements
presented in Table 2.1. Hence, experimental data can be used either for the phase
analysis or for direct reconstruction of the amplitudes.

2.9 Relativistic Pion–Nucleon Scattering Matrix 103
2.9 Relativistic Pion–Nucleon Scattering Matrix
In the preceding sections, we consider pion–nucleon scattering in the nonrelativistic
case, where the kinetic energy of a particle is much lower than its rest energy. This
approach was appropriate in the early 1950s, when synchrocyclotrons accelerated
protons up to energies 200–300 MeV. However, the kinetic energy of accelerated
protons reached the rest energy in the mid-1950s and, then, became much higher
than the rest energy. Theoreticians foresaw this situation and developed the covari-
ant formulation for the density and scattering matrices, which made it possible to
analyze processes at relativistic energies. The main results of such an analysis con-
ﬁrmed the applicability of the nonrelativistic approach in the center-of-mass frame,
and the relativistic corrections were reduced to an additional rotation angle. Such
corrections refer to the observables that involve the polarization components in the
scattering plane (parameters A and R), whereas the parameters that involved only
polarization components perpendicular to the scattering plane (parameters P , DNN,
and CNN) remain unchanged.
Below, we apply the relativistic description to the reaction
a(0) + b(1/2) = a(0) + b(1/2) (2.267)
(the spins of the particles are given in the parentheses) proposed in Stapp (1956).
Since the particle b is a Dirac particle with spin 1/2, it is described by a four-
component wave function ψ. The wave function of a free incident (or initial) particle
(with positive energy) can be written in the form
ψ = exp(if · x)
2
i=1
AiUi, (2.268)
whereas the wave function of an antiparticle (with negative energy) has the form
ψ = exp(−if · x)
4
i=3
AiUi. (2.269)
Here, f (f,f0) is the four-momentum of the particle in the base frame, where it
is measured (for example, in the laboratory frame), such that f0 > 0; x are four-
dimensional space coordinates. Each spinor Ui has four components Usi given by
the expressions
Usi(f ) = (∓if · γsi + m)/ 2m(f0 + m) . (2.270)
Hereinafter, the upper sign (−) refers to the subscripts i = 1,2 (positive energy),
whereas the lower sign (+), to the subscripts i = 3,4 (negative energy). The sub-
scripts in the four-vector γ (−iβα,β) denote its matrix elements and m is the mass
of the Dirac particle. The spinors are normalized in the covariant form as follows:
U+
i (f )Uj (f ) = U∗
i (f )βUj (f ) = ±δij . (2.271)
The sign (+) in the spinor means complex conjugation and transposition (in-
terchange of the columns and rows of the matrix), i.e., Hermitian conjugation

U+ = U∗β. Using formula (2.270), it is easy to verify that the spinors Ui satisfy
the Dirac equation
(±if · γ + m)Ui(f ) = 0. (2.272)
To make the below expressions shorter, we introduce the notation
γ (ν) = (γ · ν)/ (v · v). (2.273)
The denominator on the right-hand side can be either a positive real number or a
positive imaginary number. With this symbol, the Dirac equation is represented in
the shorter form
γ (f )Ui = ±Ui. (2.274)
The wave function of the initial state of the pion–nucleon system was introduced
above (see Eqs. (2.268) and (2.269)). Let Φ be the wave function of the ﬁnal state
of the same system. The relation between these two functions is determined by the
reaction matrix S(f ,t,f ):
Φ f = S f ,t,f Ψ (f ). (2.275)
The theory of “holes” requires that the Dirac particle described by a plane wave
with the momentum f at time T = −∞ and by a plane wave with the momentum
f at time T = +∞ has the same sign of energy. This means that the transition
of the particle to the antiparticle and vise versa is forbidden. This does not mean
that the scattering matrix cannot describe the production of a particle. However, in
this particular case, we analyze elastic processes and imply this exclusion, which is
mathematically written in the form
S f ,t,f = γ f S f ,t,f γ (f ). (2.276)
Introducing the new symbol
γ (u,w) = γ u/ |u · u| + w/ |w · w| = γ (u) + γ (w) (2.277)
and taking into account the relations
γ (u)γ (u) = 1 = γ (w)γ (w), (2.278)
we obtain
γ (u)γ (u,w) = γ (u,w)γ (w). (2.279)
The new scattering matrix Sq(k ,t,k) is introduced through the relation
S f ,t,f = γ f ,t Sq k ,t,k γ (t,f ). (2.280)
This matrix Sq(k ,t,k) is the scattering matrix in the center-of-mass frame with
the relative momenta k and k before and after scattering, respectively, and the total
energy t in the center-of-mass frame.
Then, the substitution of Eq. (2.280) into Eq. (2.276) provides the condition of
the exclusion of the particle–antiparticle transition in the form
Sq k ,t,k = γ (t)Sq k ,t,k γ (t). (2.281)

For a deeper insight into the meaning of this matrix, we substitute relation (2.281)
into relation (2.280) and obtain
S f ,t,f = γ f ,t γ (t) Sq k ,t,k γ (t)γ (t,f ) . (2.282)
The transformation (γ (t)γ (t,f )) relates the rest frame of the initial particle (Ri
frame) with the center-of-mass frame (C frame for brevity). The transformation
(γ (f ,t)γ (t)) relates the Rf frame to the C frame. To verify this, we write the
Lorentz transformation
L(f ) = exp −
1
2
θ(α · f )/|f | . (2.283)
This expression is modified as
L(f ) = β(−iγ · f + mβ)/ 2m(f0 + m) . (2.284)
In the C frame, t(0,t0) and γ (t) = β and it can be shown that the following
equality takes place:
γ (t1)γ (t1,f1) = L(f1), γ f1,t1 γ (t1) = L−1
f1 . (2.285)
Subscript 1 means that the quantities are taken in the C frame. Thus, the scatter-
ing matrix can be written in the form
S f1,t1,f1 = L−1
f1 Sq k1,t1,k1 L(f1). (2.286)
Now, this expression can be interpreted as follows. The S-matrix in the C frame
is the product of two Lorentz transformations and the scattering matrix Sq. The
Lorentz transformation L(f1) transforms the spinor of the initial particle from the
C frame to the Rf1 frame, i.e., to the rest frame of the initial particle, where spin is
physically defined. Then, the unitary operator Sq describes the effect of scattering
on this spinor. Finally, the second Lorentz transformation transforms the spinor of
the final particles to the C frame.
Since the S-matrix specified by expression (2.275), as well as the γ matrix, has
the covariant form, the new matrix Sq(k ,t,k) should also be covariant.
It is a 4 × 4 matrix and can be expanded in 16 Dirac matrices constituting a
complete set:
Sq k ,t,k = A + Bμγμ +
1
2
Cμνσμν + Dμ(iγ5γμ) + Eγ5. (2.287)
As expected in the general case, we have 16 coefficients, which are functions of
the relative momenta k and k and the total energy in the center-of-mass frame, t.
These 16 coefficients are the scalar parameter A, pseudoscalar E, 4 components
of the vector B,4 components of the pseudovector D, and 6 components of the
antisymmetric tensor C. Condition (2.281) implying the conservation of the sign of

the total energy before and after reaction (a particle cannot become the antiparticle
and vise versa) leads to the following constraints on the coefficients:
Bμ = −iNB(Btμ),
Cμν = NCC kμkν − kνkμ
(m2 − μ2)
|t · t|
· tμ kν − kν − tν kμ − kμ ,
Dμ = NDD(−i)kλkρtσ εμλρσ ≡ Dnμ, E = 0.
(2.288)
Here, εμλρσ is the antisymmetric tensor and n is the four-dimensional unit pseu-
dovector, which is the generalization of the nonrelativistic three-dimensional pseu-
dovector n perpendicular to the scattering plane. This relativistic pseudovector n has
the properties
(k · n) = k · n = (t · n) = (1 − n · n) = 0. (2.289)
The coefficients B,C, and D, as well as A, are scalar functions of the energy and
scattering angle and μ is the pion mass. The normalization coefficients NB, NC, and
ND are chosen so that the following equalities are satisfied:
BμBμ
= B2
, CμCμ
= 2C2
, DμDμ
= D2
. (2.290)
In the center-of-mass frame, expressions (22) are strongly simplified:
Bμγμ = Bβ,
1
2
Cηνσμν = C(σ · N), Dμiγ5γμ = Dβ(σ · N), E = 0.
(2.291)
Here, (σ · N) is the dot product of the three-dimensional vectors σ and N, where
N is the unit vector perpendicular to the scattering plane in the center-of-mass
frame. The 4 × 4 Dirac matrix σi (i = 1,2,3) has the form
σi =
σi 0
0 σi
. (2.292)
Here, the two-dimensional Pauli matrices appear in the parentheses. Combining
the terms in expressions (2.291) with the term A (see Eq. (2.287)), we obtain the
following expression for the scattering matrix:
Sq k ,t,k =
(F+ + G+(σ · N)) 0
0 (F− + G−(σ · N))
, (2.293)
where
F±
= A ± B, G±
= D ± C. (2.294)
Thus, pion–nucleon scattering in the relativistic case is also described by two
complex amplitudes F and G; the upper (+) and lower (−) superscripts refer to the
scattering of particles and antiparticles, respectively. With the use of the projection
operator
Λ±
(t) =
1
2
1 ± γ (t) , (2.295)

the expression for pion–nucleon scattering matrix can be easily transformed to the
covariant form
Sq(k ,t,k) =
±
Λ±
(t) F±
+ G±
iγ5(γ · n) . (2.296)
After the substitution of this relation into Eq. (2.287), the resulting S-matrix is
covariant, and the states with positive and negative energies are clearly separated.
This form of the scattering matrix is often used when discussing the problem of spin
interactions.
2.9.1 Covariant Density Matrix
As mentioned above when considering nonrelativistic scattering theory, the density
matrix is convenient when a partially polarized beam is used. The same is true for
relativistic theory. For the physical system state ψα, the density matrix is defined by
the expression
ρ =
α
|ψα Wα ψα|, (2.297)
where Wα is the probability of finding the system in this state; therefore,
α Wα = 1.
The probability of finding the system in the region R is determined by the for-
mula
w(R) = Tr (ρΠ), (2.298)
where Π is the projection operator separating the region R. If the region is an ele-
ment of the three-dimensional momentum space (df = df1df2df3), then
w(df ) = df Tr ρs(f ). (2.299)
Here, Tr stands for the trace of the matrix ρs(f ) in spin space and
ρs(f ) = aα(f )
2
Uα(f ) Wα U∗
α (f ) . (2.300)
The amplitude aα(f ) is related to the wave function in momentum space as
ψα(f ) = aα(f ) Uα(f ) . (2.301)
Here, the spinor Uα(f ) of the particle moving with the momentum f can be
expressed in terms of the spinor of the particle in its R frame by means of the
Lorentz transformation:
Uα(f ) = L−1(f )Uα(0),
w(df ) = (df ) a(f )
2
(γ )f = (df ) Wα aα(f )
2
(γ )f .
(2.302)
In this expression, the ratio (df )/(γ )f is Lorentz invariant (this is the invariant
dp/dE). Hence, the quantity |aα(f )γ f |2 should be Lorentz invariant.

Note that both the density matrix and volume element are not invariant sepa-
rately. However, if the particle is in a definite energy state, it is possible to make the
transformation (sum signs are omitted)
ρs(f ) = aα(f )
2
Uα(f ) Wα U∗
α (f )
= aα(f )
2
Uα(f ) Wα U∗
α (f ) Uα(f ) (±) U+
α (f )
= γ f
aα(f )
2
Uα(f ) (±Wα) U+
α (f ) /(γ )f
≡ ρf /(γ )f
. (2.303)
The product of the factor 1/(γ )f and the volume element in the momentum space
df is invariant. The matrix ρ defined by the above expression is covariant, and its
matrix elements are calculated by the formula
ρij (f ) = U+
i (f ) ρ(f ) Uj (f ) . (2.304)
The square of the absolute value of the amplitude specifies the probability of
finding the system in the spin state α and with the momentum f . The mean value of
the operator Ô in the initial state with the momentum f is given by the expression
Ô f Tr ρ(f ) = Tr ρ(f ) Ô . (2.305)
The mean value of the operator in the final state with the momentum f is written
similarly
Ô f Tr ρ f = Tr ρ f Ô . (2.306)
Two density matrices are related by the scattering matrix as
ρ f = S f ,t,f ρ(f )S+
f ,t,f . (2.307)
The differential cross section is determined by the expression
I = Tr ρ f /Tr ρ(f ). (2.308)
A Hermitian conjugate operator Â+ can be defined as: (AU)+ = U+A+, where
U is the spinor and U+ = U∗β. In application to the S-matrix, we obtain the relation
S+ = βS∗β, where the asterisk in S means complex conjugation.
The application of relativistic formulas (2.307) and (2.308) leads to the differ-
ential cross section in the center-of-mass frame, as in the nonrelativistic case. The
same conclusion is also valid for polarization. In the case of the spin rotation pa-
rameters in the horizontal plane, only an additional kinematic factor appears.
2.10 Relativistic Nucleon–Nucleon Scattering
In the preceding section, we consider the case where both the initial and final states
include only one Dirac particle. Now, we discuss the case where both the initial
and final states of the reaction include two particles with spin 1/2. Specifically, we
analyze nucleon–nucleon elastic scattering.

2.10 Relativistic Nucleon–Nucleon Scattering 109
Each nucleon has its complete set of the spin operators acting in two independent
spin spaces. We present these operators (Bjorken and Drell 1964).
In the spin space of the first particle, these are
I(1)
, γ (1)
μ ,
1
2
σ(1)
μν , iγ
(1)
5 γ (1)
μ , γ (1)
μ , γ
(1)
5 , (2.309)
in the spin space of the second particle,
I(2)
, γ (2)
μ ,
1
2
σ(2)
μν , iγ
(2)
5 γ (2)
μ , γ (2)
μ , γ
(2)
5 . (2.310)
Each of sets (2.309) and (2.310) contains 16 terms. The scattering matrix is com-
posed of the direct product of these terms, i.e., contains 16 × 16 = 256 terms, as in
the nonrelativistic case. Correspondingly, the number of amplitudes is the same. It
is necessary to impose allowable physical conditions in order to reduce this number
to the minimum possible number.
By analogy with the case of pion–nucleon scattering (see the preceding sec-
tion), but taking into account the presence of four nucleons, we introduce the matrix
Sq(k ,t,k) through the expression (Stapp 1956)
S f ,k ,t,f,k = γ (1)
f ,t γ (1)
(t) γ (2)
k ,t γ (2)
(t) Sq k ,t,k
× γ (2)
(t)γ (2)
(k,t) γ (1)
(t)γ (1)
(f,t) . (2.311)
Here, k and k are the initial and final relative momenta, respectively. We can
introduce the condition of the theory of holes and obtain
γ (1)(t)Sq k ,t,k γ (1)(t) = Sq k ,t,k ,
γ (2)(t)Sq k ,t,k γ (2)(t) = Sq k ,t,k .
(2.312)
The matrix Sq(k ,t,k) can be expanded in the products of 16 spin matrices for
particle 1 by 16 spin matrices for particle 2 (see Eq. (2.310)). The main aim is to
significantly reduce the number of the resulting 256 terms. To this end, we consider
a term appearing from the product of tensor operators:
Cμνσρ
1
2
σ(1)
μν
1
2
σ(2)
σρ . (2.313)
From the first of expressions (2.312) applied to this term, we obtain
tμCμνσρ = −tμCνμσρ = 0. (2.314)
Therefore,
Cμνσρ
1
2
σ(1)
μν = γ (1)
(t)iγ
(1)
5 γ
(1)
λ Cλ;σρ, (2.315)
where we take into account that tλCλνσρ = 0. The dependence on σ
(2)
σρ can be mod-
ified to the form
Cμνσρ
1
2
σ(1)
μν
1
2
σ(2)
σρ = Cλσ γ (1)
(t) iγ
(1)
5 γ
(1)
λ γ (2)
(t) iγ
(2)
5 γ (2)
σ . (2.316)

Here, the equality Cληtη = −tλCλη is taken into account. Excluding similarly all
terms containing σμν, we obtain
Sq k ,t,k =
±±
Λ(1)±
(t)Λ(2)±
(t)
· F±±
+ G(1)±±
iγ
(1)
5 γ (1)
· n + G(2)±±
iγ
(2)
5
+ C±±
nλnρ + D±±
sλsρ + E±±
dλdρ
· iγ
(1)
5 γ
(1)
λ iγ
(2)
5 γ (2)
ρ . (2.317)
Here four vectors n,s, and d are defined as follows
nλ ∝ kρkσ tμερσμλ,
sλ = Ns kλ + kλ − tλ tρ kρ + kρ (t · t)−1 ,
dλ = Nd kλ − kλ .
(2.317a)
Here Ni (i = s,d) are normalization coefficients. The vectors n,s,d,t form an
orthogonal set. The vector d retains its sign under time inversion, whereas s changes
sign. Therefore all terms containing the products of these two vectors should reduce
to zero.
The scattering matrix is written as follows:
Sq k ,t,k = F + F(1)
γ (1)
(t) + F(2)
γ (2)
(t) + G
(1)
λ iγ (1)
5
γ
(1)
λ
+ G
(2)
λ iγ
(2)
5 γ
(2)
λ + C
(1)
λ γ (1)
(t) iγ (1)
5
γ
(1)
λ
+ C
(2)
λ γ (2)
(t) iγ (2)
5
γ (2)
λ
+ Gλρ iγ
(1)
5 γ
(1)
λ · iγ
(2)
5 γ (2)
ρ
+ Cλργ (1)
(t) iγ (1)
5
γ
(1)
λ γ (2)
(t) iγ
(2)
5 γ (2)
ρ
+ F(3)
γ (1)
(t) · γ (2)
(t) + E
(1)
λ γ (2)
(t) iγ
(1)
5 γ
(1)
λ
+ E
(2)
λ γ (1)
(t) iγ
(2)
5 γ
(2)
λ
+ D
(1)
λ γ (2)
(t)γ (1)
(t) iγ
(1)
5 γ
(1)
λ + D
(2)
λ γ (1)
(t)γ (2)
(t) iγ
(2)
5 γ
(2)
λ
+ H
(1)
λρ iγ
(1)
5 γ
(1)
λ γ (2)
(t) iγ
(2)
5 γ (2)
ρ
+ H
(2)
λρ (iγ
(2)
5 γ
(2)
λ )γ (1)
(t) iγ
(1)
5 γ (1)
ρ . (2.318)
All parameters appearing in the scattering matrix are functions of the momenta
k, k , and t. These parameters are orthogonal to t with respect to all subscripts. For
example tλHλρ = tρHλρ = 0 and similar is for the other terms.
It is easy to group the terms in expression (2.318). For example, the first two
terms can be rewritten in the form
F + F(1)
γ (1)
(t) =
±
1
2
1 ± γ (1)
(t) F±
, (2.318a)
where F+ = F + F(1), F− = F − F(1). The other terms can be pairwise grouped
similarly. As a result, we arrive at the expression

Sq k ,t,k =
±
1
2
1 ± γ (1)
(t) · F±
+ G
(1)±
λ iγ
(1)
5 γ
(1)
λ + F(2)±
γ (2)
(t)
+ G(2)±
λ iγ (2)
5 γ (2)
λ + C(2)±
λ γ (2)
(t) iγ (2)
5 γ (2)
λ
+ G±
λρ iγ
(1)
5 γ
(1)
λ iγ
(2)
5 γ (2)
ρ
+ H
(1)±
λρ iγ
(1)
5 γ
(1)
λ γ (2)
(t) iγ
(2)
5 γ (2)
ρ
+ E(1)±
λ γ (2)
(t) iγ (1)
5 γ (1)
λ . (2.319)
Grouping similarly the terms with respect to γ (2)(t), we obtain
Sq k ,t,k =
±±
1
2
1 ± γ (1)
(t)
1
2
1 ± γ (2)
(t)
· F±±
+ G
(1)±±
λ iγ
(1)
5 γ
(1)
λ + G
(2)±±
λ iγ
(2)
5 γ
(2)
λ
+ G±±
λρ iγ
(1)
5 γ
(1)
λ iγ
(2)
5 γ (2)
ρ . (2.320)
Here, the following orthogonality relations for Gλρ were used:
tλG
(1)±±
λ = tλG
(2)±±
λ = tλG±±
λρ = G±±
λρ tρ = 0. (2.320a)
Since the scattering matrix should be a scalar function, the parameters G
(1)±±
λ
and G
(2)±±
λ should be pseudovectors. The existing three momenta can provide the
single pseudovector
nλ ∝ kρkσ tμερσμλ. (2.321)
Hence, we can write
G
(1)±±
λ = G(1)
nλ and G
(2)±±
λ = G(2)
nλ. (2.322)
To transform the tensor terms G±±
λρ , two normalized vectors s and d are intro-
duced in addition to t and n as follows:
sλ = Ns kλ + kλ − tλ tρ kρ + kρ (t · t)−1
, dλ = Nd kλ − kλ . (2.323)
The four vectors t,n,s, and d constitute a set of orthonormalized vectors in
which the second-rank tensor G±±
λρ can be expanded. Relation (2.320a) imposes
the constraints on the number of the terms of this tensor. Additional conditions are
imposed by the requirement of the invariance of the scattering matrix under space
inversion. As a result, we obtain
G±±
λρ = C±±
nλnρ + D±±
sλsρ + E±±
dλρ
+ G ±±
(sλdρ + dλsρ) + G±±
(sλdρ − dλsρ). (2.324)
Since the matrix should be invariant under time reversal, two last terms should
be zero, because the vector d does not change sign under this operation, whereas
the vector s changes sign.
Finally, the relativistic nucleon–nucleon elastic scattering matrix is written in the
form

Sq k ,t,k =
±±
Λ(1)±
(t) Λ(2)±
(t)
· F±±
+ G(1)±±
iγ
(1)
5 γ (1)
· n + G(2)±±
iγ
(2)
5 γ (2)
· n
+ C±±
nλnρ + D±±
sλsρ + E±±
dλdρ iγ
(1)
5 γ
(1)
λ iγ
(2)
5 γ (2)
ρ .
(2.325)
This is the relativistic formula for nucleon–nucleon scattering in the center-of-
mass frame. Several conclusions follow from it. First, the number of free param-
eters is six, as in the nonrelativistic case. Second, the scattering of antiprotons is
described by the same formulas as the scattering of protons. Third, passing to the
nonrelativistic case and taking the states only with positive energy, we exactly arrive
at Wolfenstein nonrelativistic formulas. As will be shown below, the relativistic case
gives only kinematic corrections, which are easily taken into account.
In the center-of-mass frame, the relativistic matrix given by formula (2.325) re-
duces to the Wolfenstein–Ashkin nonrelativistic matrix (where I(1) and I(2) are the
identity matrices in the spaces of particles 1 and 2, respectively):
M = aI(1)
I(2)
+ c σ(1)
n + σ(2)
n + mσ(1)
n σ(2)
n
+ g σ
(1)
ˆp
σ
(2)
ˆp
+ σ
(1)
ˆk
σ
(2)
ˆk
+ h σ
(1)
ˆp
σ
(2)
ˆp
− σ
(1)
ˆk
σ
(2)
ˆk
. (2.326)
This matrix differs from the matrix used above in the notation: a = a, c = c,
b = m, l = g − h, and f = g + h.
The ordinary method provides the following expressions for the measured quan-
tities in terms of the elements of scattering matrix (2.326):
I0R =
1
2
Re M00 +
√
2cotθM10 (M11 + M1−1 − Mss)∗
× cos(θ − θL) −
√
2
sinθ
(M11 + M1−1)M∗
10 − MssM∗
01 cosθl ,
I0A = −
1
2
Re M00 +
√
2cotθM10 (M11 + M1−1 + Mss)∗
× sin(θ − θL) −
√
2
sinθ
(M11 + M1−1)M∗
01 − MssM∗
10 sinθl ,
I0R =
1
2
Re M00 +
√
2cotθM10 (M11 + M1−1 + Mss)∗
× sin(θ − θL) +
√
2
sinθ
(M11 + M1−1)M∗
10 − MssM∗
01 sinθl ,
I0A =
1
2
Re M00 +
√
2cotθM10 (M11 + M1−1 + Mss)∗
× cos(θ − θL) +
√
2
sinθ
(M11 + M1−1)M∗
01 − MssM∗
10 cosθl ,

I0Rt =
1
2
Re M00 +
√
2cotθM10 (M11 + M1−1 − Mss)∗
× cos θ − θL +
√
2
sinθ
(M11 + M1−1)M∗
10 + MssM∗
01 cosθl ,
I0At = −
1
2
Re M00 +
√
2cotθM10 (M11 + M1−1 − Mss)∗
× sin θ − θL −
√
2
sinθ
(M11 + M1−1)M∗
01 − MssM∗
10 sinθl ,
I0Rt =
1
2
Re M00 +
√
2cotθM10 (M11 + M1−1 − Mss)∗
× sin θ − θL −
√
2
sinθ
(M11 + M1−1)M∗
10 + MssM∗
01 sinθl ,
I0At =
1
2
Re M00 −
√
2cotθM10 (M11 + M1−1 − Mss)∗
× cos θ − θL −
√
2
sinθ
(M11 + M1−1)M∗
01 + MssM∗
10 cosθl ,
I0Ckp =
1
2sinθ
|M01|2
− |M10|2
cos α − α −
1
4
|M11 + M1−1|2
− |Mss|2
× cos α + α −
1
4cosθ
|M11 − M1−1|2
− |M00|2
sin α − α ,
I0Ckk = −
1
2sinθ
|M01|2
− |M10|2
sin α − α +
1
4
|M11 + M1−1|2
− |Mss|2
× cos α + α −
1
4cosθ
|M11 − M1−1|2
− |M00|2
cos α − α ,
I0Cpp = −
1
2sinθ
|M01|2
− |M10|2
sin α − α +
1
4
|M11 + M1−1|2
− |Mss|2
× cos α + α −
1
4cosθ
|M11 − M1−1|2
− |M00|2
cos α − α .
Here, θ and θl are the scattering angles in the center-of-mass and laboratory
frames, respectively, and θ and θL are the respective angles for the recoil particle.
Under relativistic transformations, the parameters that either are scalars or have
only components perpendicular to the reaction plane remain unchanged. These are
the following quantities: cross section I; polarization P ; depolarization tensors of
the scattered and recoil particles, D and Dt , respectively; and correlation parameters
Cnn and Ann.
In the above formalism of the relativistic reaction matrix, the wave functions are
represented in the space of angular momenta, where the quantization z axis is ﬁxed.
Jacob and Wick (1959) proposed a relativistic description of reactions using the
wave functions quantized along the momentum of the incident and scattered parti-
cles in the center-of-mass frame. In this case, the spin projection on the momentum
direction has two values, +1/2 and −1/2, and this projection is called helicity.

Denoting the helicities of the initial and final nucleons as λ1, λ2 and λ1, λ2, respec-
tively, we can write the scattering matrix in the form
λ1λ2 M λ1λ2 . (2.327)
These matrix elements are called helicity amplitudes. Let the scattered particle
move along the z axis inclined to the z axis at the angle θ (scattering angle). By
analogy with the spinning-top model, matrix element (2.327) can be expanded in
terms of the reduced wave functions of the symmetric top, dJ
μμ :
λ1λ2 M λ1λ2 =
1
2ik
J
(2J + 1) λ1λ2 S(J,E) − 1 λ1λ2 dJ
μμ (θ), (2.328)
where
μ = λ1 − λ2, μ = λ1 − λ2. (2.329)
As shown above, owing to the invariance of the scattering matrix under space
rotation and inversion and time reversal, as well as to the isotopic invariance, five
matrix elements are nonzero in the general case. The same requirements applied to
the scattering matrix in the helicity representation impose the following conditions:
parity conservation
λ1λ2 M λ1λ2 = −λ1 − λ2 M −λ1 − λ2 , (2.330)
time reversal
λ1λ2 M λ1λ2 = (−1)λ1−λ2−λ1+λ2 λ1λ2 M λ1λ2 , (2.331)
conservation of the total spin
λ1λ2 M λ1λ2 = λ2λ1 M λ2λ1 . (2.332)
Under these conditions, five matrix elements are also nonzero in the helicity rep-
resentation, and these elements are denoted as follows:
ϕ1 =
1
2
,
1
2
M
1
2
,
1
2
, ϕ2 =
1
2
,
1
2
M −
1
2
,−
1
2
,
ϕ3 =
1
2
,−
1
2
M
1
2
,−
1
2
,
ϕ4 =
1
2
,−
1
2
M −
1
2
,
1
2
, ϕ5 =
1
2
,
1
2
M
1
2
,−
1
2
.
(2.333)
The amplitudes ϕi, where i = 1,2,3,4,5, are classified according to the physics
of the process as the amplitudes without spin flip (ϕ1,ϕ3), with single spin flip (ϕ5),
and with double spin flip (ϕ2,ϕ4).
The relation between the matrix elements in the helicity representation and
angular-momentum representation can be obtained as follows (Jacob and Wick
1959). Let the direction of the incident-particle motion be the quantization axis for

the spin wave function. Then, the wave functions of the ﬁrst and second nucleons in
the helicity representation can be written in the form
χ(1)
1/2 =
1
0
, χ(1)
−1/2 =
0
1
(2.334)
in the initial state and
χ
(1)
1/2 = χ
(2)
−1/2 =
cos(θ/2)
sin(θ/2)
, χ
(1)
−1/2 = χ
(2)
1/2 =
−sin(θ/2)
cos(θ/2)
(2.335)
in the ﬁnal state, where θ is the particle scattering angle in the center-of-mass frame.
Let us rewrite matrix (2.326) as follows:
M = a + c σ(1)
+ σ2
· n + m σ(1)
· n σ(2)
· n
+ g σ(1)
· P σ(2)
· P + σ(1)
· K σ(2)
· K
+ h σ(1)
· P σ(2)
· P − σ(1)
· K σ(2)
· K . (2.336)
The unit vectors n, P , K have the following components:
n(0,1,0), K cosθ/2,0,−sinθ/2 , P sinθ/2,0,cos(θ/2) . (2.337)
Therefore, the vector n is perpendicular to the scattering plane, whereas the vec-
tors K and P lie in the scattering plane. Applying matrix (2.336) to wave functions
(2.334) and (2.335) and taking into account relations (2.337), we obtain the relation
between the amplitudes in the helicity and angular representations. They are given
below along with the relations with the amplitudes in the singlet–triplet representa-
tion.
2.10.1 Relation Between the Amplitudes in Different
Representations
A. In the helicity and angular representations:
ϕ1 − ϕ2 = a − m − 2g,
ϕ1 + ϕ2 = (a + m)cosθ + 2ic sinθ + 2h,
ϕ3 + ϕ4 = a − m + 2g,
ϕ3 − ϕ4 = (a + m)cosθ + 2ic sinθ − 2h,
ϕ5 = −
1
2
(a + m)sinθ + ic cosθ.

B. In the angular and helicity representations (inverse to case A):
a =
1
4
(ϕ1 − ϕ2 + ϕ3 + ϕ4) + (ϕ1 + ϕ2 + ϕ3 − ϕ4)cosθ − 4sinθϕ5 ,
ic =
1
4
(ϕ1 + ϕ2 + ϕ3 − ϕ4)sinθ + 4cosθϕ5 ,
m =
1
4
(−ϕ1 + ϕ2 − ϕ3 − ϕ4) + (ϕ1 + ϕ2 + ϕ3 − ϕ4)cosθ − 4sinθϕ5 ,
g =
1
4
(−ϕ1 + ϕ2 + ϕ3 + ϕ4),
h =
1
4
(−ϕ1 − ϕ2 + ϕ3 − ϕ4).
C. In the helicity and singlet–triplet representations:
ϕ1 − ϕ2 = Mss,
ϕ1 + ϕ2 = cosθM00 −
√
2sinθM10,
ϕ3 + ϕ4 = M11 + M1−1,
ϕ1 − ϕ2 = cosθM11 + sinθM01 − cosθM1−1,
ϕ5 = −
1
2
sinθM11 +
1
√
2
cosθM01 −
1
2
sinθM1−1
= −
1
2
sinθM00 −
1
√
2
cosθM10.
The helicity amplitudes can be expressed in terms of the phase shifts. To
this end, the matrix elements in helicity space should be transformed to the ma-
trix elements in singlet–triplet space that are already expressed in terms of the
phases.
D. The matrix elements in the helicity representation can be expressed in terms of
the elements in the singlet–triplet representation as follows:
λ1λ2 M|λ1λ2 = λ1λ2 |Sms Sms|M Sms Sms |λ1λ2 . (2.338)
Since the elements of the matrix M, as well as the Clebsch–Gordan coefﬁ-
cients, are known, the helicity amplitudes can be found from this relation (see
“Relation between the amplitudes in different representations”).
2.11 Isospin T , C and G Parities
2.11.1 Isotopic Invariance
There are many experimental evidences that the properties of the proton and neutron
in nuclear interaction are very similar. It can be assumed that they are the compo-
nents of the same object called the nucleon. First, their masses are very close: the

2.11 Isospin T , C and G Parities 117
proton and neutron masses are 938.27 and 939.57 MeV, respectively; hence, the
mass difference is as small as 1.3 MeV. This small difference (≈0.15 %) is assum-
ingly attributed to the electromagnetic interaction. Second, they strongly interact
with each other and are constituents of nuclei. Third, it is known that, excluding
the relatively weak electromagnetic interaction in the pp system, this system is al-
most equivalent to the system of two neutrons nn. This equality of the interactions
between two protons and two neutrons is called charge symmetry. In particular, ad-
ditional experimental evidence of charge symmetry follows from nucleon–nucleon
scattering experiments. It is shown that the pp and nn elastic scattering processes
are identical, excluding the Coulomb interaction. The same data are also obtained in
pion–nucleon and kaon–nucleon elastic scattering processes. In addition, the bind-
ing energies (B), energy levels, and other properties of the mirror nuclei in which the
protons and neutrons are interchanged are very close to each other. The closeness of
the binding energies of two mirror nuclei is exempliﬁed as follows:
H3
= (nnp) → B = 8.192 MeV, He3
= (ppn) → B = 7.728 MeV. (2.339)
Here, the nucleon compositions of the tritium and helium-3 nuclei are given in the
parentheses. The difference between the binding energies of these nuclei is as small
as B ≈ 0.5 MeV and can be attributed to the energy of the Coulomb repulsion
between two protons in the helium-3 nucleus:
VC(R) =
1
2
Z(Z − 1)
6e2
5R
, R ≈ 1.45 · 10−13
cm. (2.340)
The closeness of the energy levels of the mirror nuclei can be illustrated by the
example
B11 = (6n5p) → E = (1.98;2.14;4.46;5.03;6.76) MeV,
C11 = (6p5n) → E = (−;1.85;4.23;4.77;6.40) MeV.
(2.341)
Here, the numbers of protons and neutrons in a given nucleus are presented in the
ﬁrst parentheses and the level energies of this nucleus are indicated in the second
parentheses. The observed similarity of the energy levels of these mirror nuclei is
most simply explained by the hypothesis of the charge symmetry between the pro-
ton and neutron. The detailed discussion of these problems can be found in Schiff
(1968).
It is substantial that both the proton and neutron have a half-integer spin and are
governed by the same statistics (Dirac–Fermi statistics). This means that the system
of two protons or two neutrons is described by the wave function ψ(r1,s1;r2,s2)
(where r and s are the radius-vector and spin of the particles, respectively) anti-
symmetric under the simultaneous permutation of the coordinates and spins of the
particles. However, the experimentally observed charge symmetry is only one of the
manifestations of a deeper similarity of the proton and neutron. This new type of
symmetry was introduced by Heisenberg (1932) and was called isotopic invariance.
Its main meaning is that the forces between the pp, np, and nn pairs are identical
and the proton and neutron are two components of the nucleon. Let us illustrate
this hypothesis by an example from low-energy scattering. The interaction of the

proton with the proton or the neutron at low energies is characterized by two param-
eters: the scattering length a and effective scattering radius r0. They determine the
scattering phase δ in the S-state through the relation
k cotδ = −
1
a
+
1
2
r0k2
, (2.342)
where k is the wavenumber. The following parameters are experimentally deter-
mined (Nishijima 1964):
np : 3S1, r0t = (1.704 ± 0.028) · 10−13 cm,
at = (5.39 ± 0.03) · 10−13 cm;
1S1, r0S = (2.670 ± 0.023) · 10−13 cm,
aS = (−23.74 ± 0.09) · 10−13 cm;
pp : 1S1, r0S = (2.77) · 10−13 cm, aS = (−17.77) · 10−13 cm.
(2.343)
The parameters a and r0 of the np- and pp-scattering processes in the same 1S1
states are in qualitatively agreement. The quantitative difference can be ignored, be-
cause this difference can be eliminated by changing the potential-well depth by 3 %.
These experimental facts led to the notion of the nucleon isospin τ and to the
discovery of isotopic invariance in strong interactions. The isospin is a vector in
isotopic space and has the same properties as the Pauli matrices. Isospin space is
not a real space, but only a mathematical notion. Correspondingly, the nature of
isospin is unknown.
The isotopic invariance of strong interactions can be formulated as invariance
under “rotations in isospin space.” The isospin operator T (T1,T2,T3) and isospinor
ψ are written in the explicit form
T1 =
1
2
0 1
1 0
, T2 =
1
2
0 −i
i 0
, T3 =
1
2
1 0
0 −1
, ψ =
ψ1
ψ2
.
(2.344)
Let us consider the rotation by angle π about the x2 axis in this space. The
isospinor is transformed as:
ψ → eiπT2 ψ = ei π
2 τ2 ψ = cos
π
2
+ iτ2 sin
π
2
ψ = iτ2ψ. (2.345)
Here, Ti = 1/2τi, where i = 1,2,3; T is the nucleon isospin equal 1/2 with
the eigenvalues ±1/2; and τ is the analog of the Pauli operator in isotopic space.
It is usually accepted that the eigenvalue +1/2 corresponds to the proton and is
described by the function ψ1, whereas −1/2 corresponds to the neutron and is de-
scribed by the function ψ2 in Eq. (2.344). In particular, for the isospinors corre-
sponding to two components of the isodoublet (to the proton and neutron), we have
|p =
1
0
→
0
−1
= |n , |n =
0
1
→
1
0
= |p , (2.346)

which correspond to the transformations
τ+ψ =
1
2
(τ1 + iτ2)ψ =
0 0
−1 0
ψ, n → p,
τ−ψ =
1
2
(τ1 − iτ2)ψ =
0 0
−1 0
ψ, p → −n,
(2.347)
where p and n mean the charge states |p and |n of the nucleon. Similar transfor-
mations for the antinucleon have the form
¯p → −¯n, ¯n → ¯p. (2.348)
As seen, strong interactions are charge independent or, more widely, isotopically
invariant.
2.11.2 Charge Conjugation
The charge conjugation operation is deﬁned only in relativistic theory. Let us repre-
sent the ψ operator in the form of the expansion
ψ =
1
√
2ε
ape−i(ωt−p·r)
+ bpei(ωt−p·r)
, (2.349)
where ap and bp are the annihilation and creation operators for a particle with the
momentum p, respectively. The charge conjugation operation reduces to the change
of the particles to the antiparticles and vise versa; i.e.,
C : ap → bp, bp → ap. (2.350)
The action of operation (2.350) on operator (2.349) provides the charge conjugate
operator ψC; it is easy to see that
ψC
(t,r) = ψ+
(t,r). (2.351)
This equality expresses the property of the charge symmetry of the particles and
antiparticles. According to relation (2.349), the operator C changes the particle to
the antiparticle that is not identical to the particle. As a result, this operator has no
eigenfunctions and eigenvalues. For this reason, the charge conjugation operation
does not generally lead to new physical consequences. However, there is exclusion.
In the application to a system where the number of particles coincides with the
number of antiparticles, the operator C has eigenfunctions and eigenvalues. The
following transformations are illustrative:
C|Λ = | ¯Λ , C|n = |¯n , C|p = −| ¯p . (2.352)
The ﬁrst relation is obvious, because the Λ hyperon is the isotopic singlet. To
prove the last two relations, we write the wave function of a pair of nucleons and

antinucleons and the charge conjugation matrix in the explicit form (Lifshitz and
Pitaevskii 1973)
ψ =
ψ1
ψ2
=
p
n
¯n
¯p
(2.353)
and
C =
0 −iτ2
iτ2 0
=
0 0 0 −1
0 0 +1 0
0 +1 0 0
−1 0 0 0
. (2.354)
The action of operator (2.354) on wave function (2.353) gives
ψC
=
ψC
1
ψC
2
= Cψ =
0 −iτ2
iτ2 0
ψ1
ψ2
=
−iτ2ψ2
iτ2ψ1
. (2.355)
We obtain two matrix equations of the second rank, which are solved separately:
ψC
1 =
C(p)
C(n)
=
0 −1
1 0
·
¯n
¯p
=
− ¯p
¯n
. (2.356)
Therefore,
C(p) = − ¯p, C(n) = ¯n. (2.357)
We write and solve the second matrix equation:
ψC
2 =
C(¯n)
C( ¯p)
=
0 +1
−1 0
·
p
n
=
n
−p
. (2.358)
Hence (Pilkuhn 1979),
C(¯n) = n, C( ¯p) = −p. (2.359)
Since C2 = 1, it is easy to verify that results (2.358) and (2.359) coincide.
2.11.3 G Transformation
The simultaneous application of two conservation laws leads to new selection rules,
which do not follow from any individual law (Lee and Yang 1956).
The joint application of the isotopic transformation T and charge conjugation C
is described by the product of both operators and is denoted as G:
G = CeiπT3 . (2.360)
Since p ↔ − ¯p and n ↔ ¯n under charge conjugation, operator (2.360) provides
(Lifshitz and Pitaevskii 1973)
G : p → −¯n, n → ¯p, ¯p → −n, ¯n → p. (2.361)

The operator G commutes with the operators of all three isospin components
T1, T2, and T3. This is directly veriﬁed by writing the explicit expressions for the
operators in the form of four-row matrices transforming the nucleon and antinucleon
states. Let us represent these states in the form of the column
p
n
¯n
¯p
and generalize isospin to this case:
T1 =
1
2
τ1 0
0 τ1
, T2 =
1
2
τ2 0
0 τ2
, T3 =
1
2
τ3 0
0 τ3
,
C =
0 −iτ2
iτ2 0
, G =
0 I
−I 0
.
(2.362)
Here, 0 and I are two-row matrices.
If the operation G transforms a particle (or a system of particles) to itself, the
notion of the G parity appears: the state can remain unchanged or change sign. For
this, the baryon number and hypercharge Y (Y = B + S, where B is the baryon
number and S is strangeness) of the particles should be zero. Indeed, charge con-
jugation (transition from the particles to the antiparticles) changes the signs of the
electric charge Z and hypercharge. Rotation in isospace changes Z, but does not
change Y and B. Therefore, the joint application of both transformations changes
the numbers Y and B if they are nonzero.
An important property of the G parity is that it is the same for all components of
the same isomultiplet. This follows from the commutativity of the operator G with
all components of T and, therefore, with all rotations in isospace.
At Y = 0, we have Z = T3; therefore, T3 and thereby T are integers. The iso-
multiplet with an integer T value is described by a symmetric isospinor of the even
rank 2T , which is equivalent to the irreducible isotensor of the rank T . One of the
components of such isomultiplet is a neutral particle (T3 = 0). It corresponds to the
isotensor ψih with the nonzero component ψ33. The rotation by the angle π about
the x2 axis leads to the multiplication of this isotensor by (−1)T . The G parity of a
neutral particle with the charge parity C is given by the expression
G = C(−1)T
. (2.363)
According to the above consideration, the G parity of all components of the
isomultiplet is thus deﬁned.
For example, let us consider the pion isotriplet (T = 1). The charge parity of
the π0 meson is C = +1. This follows from the fact that the π0 meson decays
into an even number of particles, namely, into two charge-odd particles (photons).
Therefore, the G parity of the pions is G = −1. In particular, it follows from this
that strong interactions can transfer the system of pions to another system of pions
only without change of the parity of the number of particles.

Table 2.3 G and C parities of mesons and baryons
π+ π0 π− η K+ K0 ¯K0 K−
G −π+ −π0 −π− η ¯K0 K− −K+ −K0
C −π− π0 −π+ η −K− ¯K0 K0 −K+
Σ+ Σ0 Σ− Λ p n ¯n ¯p
G − ¯Σ+ − ¯Σ0 − ¯Σ− ¯Λ ¯n ¯p −p −n
C − ¯Σ− ¯Σ0 − ¯Σ+ ¯Λ − ¯p ¯n n −p
The η meson is an isosinglet (T = 0), and its charge parity is C = +1, because
the η meson, as well as the π0 meson, decays into two photons. Therefore, the η
meson has the positive G parity (G = +1). Hence, strong interactions cannot lead
to the η → 3π decay.
It is desirable to extend the notion of the G parity to other single-particle states.
The state of the particle m of the charge multiplet with the momentum p, helic-
ity λ, and third isospin component T3 is described by the function |m,T3,p,λ , and
the state of the corresponding antiparticle, by the function | ¯m,T3,p,λ . Since the
operation G changes the particle to the antiparticle and does not change the other
variables, we can write (Pilkuhn 1979):
G|m,T3,p,λ = ηG| ¯m,T3,p,λ . (2.364)
Here, ηG is independent of T3; the function of the state of the antiparticle is
transformed in isotopic space similarly to the function of the state of the particle. As
a result, for the multiplet of the particles, we have the relation
ηC = ηG(−1)T +T3 (2.365)
or in the other form
ηG = ηC(−1)T +T3 . (2.366)
For the π0 meson, C = +1, T = 1, and T3 = 0; therefore, G = −1. At the same
time, although the parity of the η meson is C = +1, but G = 1 because T = 0 and
T3 = 0. This agrees with the above results.
Formula (2.366) can be considered as the expansion of formula (2.363) to the
charged components of the multiplet, which should have the same G parity as the
truly neutral component of the multiplet. For the pion isotriplet, C|π± = G|π± =
−|π∓ . Let us consider the G and C parities for the hyperons and K mesons. It is
substantial that a baryon can emit (virtually) one pion and hyperon can emit one K
meson with the satisfaction of the selection rules in the isospin, hypercharge, and
baryon number. From the Σ → πΛ decay, it follows that ηG(Σ) = ηG(π)ηG(Λ) =
ηG(π), because the G parity of the Λ hyperon is positive. From the N → KΛ
decay, it follows that ηG(K) = ηG(N)ηG(Λ) = ηG(N). From the Ξ → KΛ decay,
it follows that ηG(Ξ) = ηG(K) = ηG(N). The results thus obtained are presented
in Table 2.3.

References 123
References
Azhgirey, L.S., et al.: Phys. Lett. 6, 196 (1963)
Baz, L.: Zh. Eksp. Teor. Fiz. 32, 628 (1957)
Bilenky, S.M., Ryndin, R.M.: Phys. Lett. 6, 217 (1963)
Bilenky, S.M., Lapidus, L.I., Ryndin, R.M.: Usp. Fiz. Nauk 84, 243 (1964)
Bilenky, S.M., Lapidus, L.I., Ryndin, R.M.: Zh. Eksp. Teor. Fiz. 49, 1653 (1965)
Bilenky, S.M., Lapidus, L.I., Ryndin, R.M.: Zh. Eksp. Teor. Fiz. 51, 891 (1966)
Bjorken, J.D., Drell, S.D.: Relativistic Quantum Mechanics. McGraw-Hill, New York (1964)
(Nauka, Moscow, 1978)
Blatt, J.M., Biedenharn, L.C.: Rev. Mod. Phys. 24, 258 (1952)
Blatt, J.M., Weisskopf, V.F.: Theoretical Nuclear Physics. Wiley, New York (1952)
Dalitz, R.H.: Proc. Phys. Soc. A 65, 175 (1952)
Golovin, B.M., Dzelepov, V.P., Nadezhdin, V.S., Satarov, V.I.: Zh. Eksp. Teor. Fiz. 36, 433 (1959)
Heisenberg, W.: Z. Phys. 77, 1 (1932)
Hoshizaki, N.: Prog. Theor. Phys. Suppl. 42, 107 (1968)
Jacob, M., Wick, G.C.: Ann. Phys. 7, 404 (1959)
Kazarinov, Yu.M.: Candidate’s Dissertation in Mathematical Physics, Dubna (1956)
Kumekin, Yu.P., et al.: Zh. Eksp. Teor. Fiz. 46, 51 (1954)
Lee, T.D., Yang, C.N.: Nuovo Cimento 3, 749 (1956)
Lifshitz, E.M., Pitaevskii, L.P.: Relativistic Quantum Theory, Part 2. Pergamon, Oxford (1973)
Martin, A.D., Spearman, T.D.: Elementary Particle Theory. North-Holland, Amsterdam (1970)
Matsuda, M.: In: Proceedings of Fifth Workshop on High Energy Spin Physics, Protvino, Russia,
p. 224 (1993)
Nishijima, K.: Fundamental Particles. Benjamin, New York (1964) (Mir, Moscow, 1965)
Nurushev, S.B.: Zh. Eksp. Teor. Fiz. 37(7), 301 (1959), vyp. 1
Nurushev, S.B.: Preprint No. 83-192, IFVE, Inst. for High Energy Physics, Serpukhov, Russia
(1983)
Philips, R.T.N.: Nucl. Phys. 43, 413 (1963)
Pilkuhn, H.: Relativistic Particle Physics. Springer, New York (1979) (Mir, Moscow, 1983)
Puzikov, L.D., Ryndin, R.M., Smorodinskii, Ya.A.: Zh. Eksp. Teor. Fiz. 32, 592 (1957)
Schiff, L.I.: Quantum Mechanics, 3rd edn. McGraw-Hill, New York (1968) (Russ. transl. of 2nd
edn., Inostrannaya Literatura, Moscow, 1957)
Schumacher, C.R., Bethe, H.A.: Phys. Rev. 191, 1534 (1961)
Smorodinskii, Ya.A.: In: Proceedings of Ninth International Conference on High Energy Physics,
Kiev, USSR (VINITI, Moscow, 1960)
Stapp, H.P.: Phys. Rev. 103, 425 (1956)
Stapp, H.P., et al.: Phys. Rev. 105, 302 (1957)
Vinternitts, P., Legar, F., Yanout, E.: Preprint No. R-2407, OIYaI, Joint Inst. for Nuclear Research,
Dubna (1965)
Wolfenstein, L., Ashkin, J.: Phys. Rev. 85, 947 (1952)

http://www.springer.com/978-3-642-32162-7

Introduction to polarization physics

More Related Content

What's hot (18)

Viewers also liked (20)

Similar to Introduction to polarization physics (20)

More from Springer (20)

Recently uploaded (20)

Introduction to polarization physics