Solving the Lippmann-Schwinger equation in momentum space

Renato Higa
July 8, 2017

Abstract
Draft, improve the text later. Previous versions:
• May/31/2010. Changes: δ± misprints in Eqs. (120,121).
• Jun/24/2017. Changes: incorporates Tomio/Adhikari’s trick, Sec. 1.5.

1 Lippmann-Schwinger equation

(Ĥ0 + V̂ )|Ψi = E|Ψi ⇒ (E − Ĥ0 )|Ψi = V̂ |Ψi

1 1
|Ψi = |Φi + V̂ |Ψi = |Φi + V̂ |Ψi (1)
E − Ĥ0 ± i E± − Ĥ0
Wave function equation:
d3 q
Z
0 0 1
hp |Ψp i = hp |Φp i + hp0 |V̂ |qihq|Ψp i
E± − p02 /2µ (2π)3
d3 q
Z
0 1
0
Ψ(p , p) = Φ(p , p) + V (p0 , q) Ψ(q, p) . (2)
E± − p02 /2µ (2π)3
Scattering amplitude:
d3 q
Z
0 0 1
hp |V̂ |Ψp i = hp |V̂ |Φp i + 3
hp0 |V̂ |qi hq|V̂ |Ψp i
(2π) E± − q 2 /2µ
d3 q
Z
0 0 1
T (p , p) = V (p , p) + 3
V (p0 , q) T (q, p) . (3)
(2π) E± − q 2 /2µ

1.1 Partial wave decomposition

Since the spherical harmonics Ylm (θ, ϕ) form a complete basis it is legitimate to partial-
wave expand the initial and final states of the physical quantity X(p0 , p) as
XX
X(p0 , p) = (4π)Yl∗0 m0 (Ωp0 )Ylm (Ωp )xl0 m0 ,lm (p0 , p) , (4)
l0 ,m0 l,m

where X stands for T , V , Ψ or Φ. The inverse relation reads

Z Z
0 1 ∗
xl0 m0 ,lm (p , p) = dΩp0 dΩp Yl0 m0 (Ωp0 )Ylm (Ωp )X(p0 , p) . (5)
4π

1
The same operation on Eq. (3) yields
Z Z
0 0 1 ∗
tl0 m0 ,lm (p , p) = vl0 m0 ,lm (p , p) + dΩp0 dΩp Yl0 m0 (Ωp0 )Ylm (Ωp )
4π
Z ∞ 2 Z
q dq 1
× 3 E − q 2 /2µ
dΩq V (p0 , q)T (q, p)
0 (2π) ±

1 X X X X ∞ q 2 dq
Z
0 1
= vl0 m0 ,lm (p , p) +
4π a0 ,b0 a,b c0 ,d0 c,d 0 (2π)3 E± − q 2 /2µ
Z Z h i
× dΩq dΩp0 Yl0 m0 (Ωp0 ) (4π)Ya∗0 b0 (Ωp0 )Yab (Ωq )va0 b0 ,ab (p0 , q)
Z h i
∗ ∗
× dΩp Ylm (Ωp ) (4π)Yc0 d0 (Ωq )Ycd (Ωp )tc0 d0 ,cd (q, p)

∞
0
XXZ q 2 dq 1
= vl0 m0 ,lm (p , p) + 4π
a,b c0 ,d0 0 (2π) E± − q 2 /2µ
3

Z
× dΩq Yab (Ωq )vl0 m0 ,ab (p0 , q) Yc∗0 d0 (Ωq )tc0 d0 ,lm (q, p)

∞
0
XZ q 2 dq vl0 m0 ,c0 d0 (p0 , q) tc0 d0 ,lm (q, p)
= vl0 m0 ,lm (p , p) + . (6)
c0 ,d0 0 2π 2 E± − q 2 /2µ

1.2 Spinless particles

In the scattering of two spinless particles (or cases where spin does not play a role) the
orbital angular momentum is conserved. That implies that both the interaction and the
amplitude are diagonal in l and m, i.e.,
xl0 m0 ,lm (p0 , p) = δl0 ,l δm0 ,m xlm (p0 , p) . (7)
From the azimuthal isotropy of the reference frame, and consequently, independence of
xlm (p0 , p) on the quantum number m, one has
l
X X
∗
0
X(p , p) = (4π)Ylm (Ωp0 )Ylm (Ωp )xl (p0 , p)
l m=−l
X
(2l + 1)Pl (cos θp0 p )xl (p0 , p) . (8)
l

Eq. (6) then simplifies to

∞
q 2 dq vl (p0 , q) tl (q, p)
Z
0 0
tl (p , p) = vl (p , p) + . (9)
0 2π 2 E± − q 2 /2µ

2
1.3 Spin 1/2 particles with tensor-like forces
TODO

1.4 Numerical solution I: basics

The numerical solution outlined here is independent of the type of the interaction, as-
suming only the absence of couplings among different partial waves. The generalization
to the case of coupled waves is straightforward, and if I feel like I’ll write about it later.
The separable interaction case can be used to check the correct implementation of the
numerical code.
The equation one wishes to solve is Eq. (9). The standard idea is to discretize the
external p0 and integration q variables using some quadrature (e.g., Gauss-Legendre), √ for
a given value of p. It is common to choose p at the on-shell point, p = 2µE, for the
scattering case (E > 0). Eq. (9) then becomes a matrix equation,
tl (p01 , p) vl (p01 , p) Kl (p01 , q1 ) Kl (p01 , q2 ) · · · Kl (p01 , qN ) tl (q1 , p)
      
 tl (p02 , p)   vl (p02 , p)   Kl (p02 , q1 ) Kl (p02 , q2 ) · · · Kl (p02 , qN )   tl (q2 , p) 
 .. = .. + .. .. ... ..  .. 
 .   .   . . .  . 
tl (p0N , p) vl (p0N , p) Kl (p0N , q1 ) Kl (p0N , q2 ) · · · Kl (p0N , qN ) tl (qN , p)
(10)
where
qj2 vl (p0i , qj )
Kl (p0i , qj ) = wj , (11)
2π 2 E± − qj2 /2µ
p0i = qi = xi , with xi and wi the usual quadrature points and weights, respectively. The
matrix equation to solve is conveniently rearranged into
1 − Kl (p01 , q1 ) −Kl (p01 , q2 ) · · · −Kl (p01 , qN ) tl (q1 , p) vl (p01 , p)
    
 −Kl (p02 , q1 ) 1 − Kl (p02 , q2 ) · · · −Kl (p02 , qN )   tl (q2 , p)   vl (p02 , p) 
 .. .. .. ..  .. = .. 
 . . . .  .   . 
−Kl (p0N , q1 ) −Kl (p0N , q2 ) · · · 1 − Kl (p0N , qN ) tl (qN , p) vl (p0N , p)
(12)
and can be solved, for instance, via LU-decomposition and backward substitution [1].
One can use the homogeneous version of Eq. (9) to find the bound state(s) for a given
potential vl . We denote τl (p0 ) as tl (p0 , p) for the bound state situation [it does not depend
on p, see similar arguments for the wave function after Eq. (49)]. A non-trivial solution of
1 − Kl (p01 , q1 ) −Kl (p01 , q2 ) · · · −Kl (p01 , qN ) τl (q1 )
  
 −Kl (p02 , q1 ) 1 − Kl (p02 , q2 ) · · · −Kl (p02 , qN )   τl (q2 ) 
 .. .. .. ..  .  = 0 (13)
 . . . .   .. 
−Kl (p0N , q1 ) −Kl (p0N , q2 ) · · · 1 − Kl (p0N , qN ) τl (qN )
requires det(I − Kl ) = 0. One therefore combines a code that calculates det(I − Kl ) for
any value of Bl = −E with another that performs a search for zeros of a given function.

3
 For the scattering case there is a technical detail due to the fact that 2µE± can get very
close to one of qj ’s. This can be overcome via a contour deformation (Hetherington-Schick
method) or more sophisticated Splines interpolations.

1.4.1 bound state wave function

The calculation of the bound state wave function ψl requires another numerical approach.
Calling K the matrix formed by the elements
qj2 vl (p0i , qj )
K̄l (p0i , qj ) = 2 wj (14)
2π E± − p0 2i /2µ
and Ψ the vector formed by ψl (qi ) one wishes to solve the eigenvalue problem
K · xn = λn xn (15)
for the particular case where λn = 1 and xn = Ψ. The technique outlined in Ref. [1] relies
on the following ideas. One start with
(K − τ I) · y = b , (16)
where b is a random vector and τ is an arbitrary number close to λn . The procedure
consists of
1. solving Eq. (16) for y,
2. normalize it,
3. replace b by the normalized solution,
4. solve for a new y,
5. and iterate the procedure until satisfactory convergence is reached.
It is expected that y will converge to xn for τ sufficiently close to λn . To understand why,
expand y and b in the base formed by the eigenvectors of K:
X X
y= αj xj , b= β j xj . (17)
j j

Replacing into (16) gives

X X βj
αj (λj − τ )xj = βj x j ⇒ αj = (18)
j j
λj − τ
X βj
∴y= xj . (19)
j
λj − τ
The larger component in the expansion of y is therefore expected to be in the xn -direction,
were the coefficient is enhanced by a factor (λn − τ ). After each iteration one increases
the alignment of y in the xn direction by a factor of 1/(λ − τ ), thus converging to the
desired result.

4
 In case one wishes to obtain also λn with an initial guess τ0 one proceeds as follows.
We know that
K · xn = λn xn ⇒ (K − τ I) · xn = (λn − τ )xn . (20)
In the k-th iteration one gets (K − τk I) · yk = bk . Replacing xn by yk in Eq. (20) we have,
apart from a normalization factor,
yk
(K − τk I) · yk = (λkn − τk )yk , where λkn → λn as → xn . (21)
|yk |
This provides the next value for τk in the iteration, τk+1 = λkn , as follows:
(K − τk I) · yk = bk = (τk+1 − τk )yk ⇒ (τk+1 − τk )yk · bk = bk · bk ≡ 1
1
⇒ τk+1 = τk + . (22)
y k · bk
• Summary of the algorithm:
y0 1
(K − τ0 I) · y0 = b0 ⇒ b1 = , τ 1 = τ0 + ⇒ (K − τ1 I) · y1 = b1 ⇒ ···
|y0 | y 0 · b0
(23)
• Alternative when λn is already known (which is our bound-state case): do not update
τ , then the matrix (K − τ I) needs to be LU-decomposed only once.

1.5 Numerical solution II: advanced

When dealing the numerics of scattering (E > 0), Eq. (9) poses a challenge, since the
denominator of the integrand is likely to develop a numerical singularity at the on-shell
point q 2 = 2µE. This is a physical singularity —at this point the denominator becomes
purely imaginary, E± − q 2 /2µ = E − q 2 /2µ ± i =→ ±i, and contributes to generate the
imaginary part of the scattering amplitude. This imaginary term guarantees the unitarity
of the S-matrix, and for this reason is often referred as the unitarity term of the amplitude.
One way to deal with the unitarity term avoiding numerical instabilities is to use a
contour deformation in the complex-momentum plane, proposed by Hetherington-Schick
(names/references later!!!).
Another technique was developed by Tomio and Adhikari [2]. The dynamical equation
for coupled-channel scattering can be written as
XZ ∞
0 0
tβα (p , p) = vβα (p , p) + f (q)q 2 dq vβσ (p0 , q)G(q, p)tσα (q, p) , (24)
σ 0

where, for two-body scattering, G(q, p) = 2µ/(p2 − q 2 + i) and f (q) = 1/(2π 2 ). The
above equation can be rearranged as
tβα (p0 , p) = vβα (p0 , p)
XZ ∞ nh i o
+ f (q)q 2 dq vβσ (p0 , q)−vβσ (p0 , k)γσ (k, q) +vβσ (p0 , k)γσ (k, q) G(q, p)tσα (q, p)
σ 0

5
 XZ ∞ h i
0
= vβα (p , p) + f (q)q 2 dq vβσ (p0 , q) − vβσ (p0 , k)γσ (k, q) G(q, p) tσα (q, p)
σ 0
XZ ∞
+ f (q)q 2 dq vβσ (p0 , k)γσ (k, q)G(q, p) tσα (q, p)
σ 0

XZ ∞
= f (q)q 2 dq Aβσ (p0 , q, p; k) tσα (q, p) + vβα (p0 , p)
σ 0
X Z ∞
0
+ vβσ (p , k) f (q)q 2 dq γσ (k, q)G(q, p) tσα (q, p) . (25)
σ 0

where the function γσ (k, q) is arbitrary and satisfies γσ (k, k) = 1. Rearranging Eq. (25),
XZ ∞ h δ(q − p0 ) i
2 0
f (q)q dq δβσ − Aβσ (p , q, p; k) tσα (q, p)
σ 0 q 2 f (q)
X Z ∞
0 0
= vβα (p , p) + vβσ0 (p , k) f (q 0 )q 0 2 dq 0 γσ0 (k, q 0 )G(q 0 , p) tσ0 α (q 0 , p) . (26)
σ0 0

One now defines an auxiliary function Γβα (p0 , p) via

XZ ∞ h δ(q − p0 ) i
0
f (q)q 2 dq 2 f (q)
δ βσ − A βσ (p , q, p; k) Γσα (q, p) = vβα (p0 , p) . (27)
σ 0 q

The expression for vβα (p0 , p) in the above equation can be used into Eq. (26), leading to
XZ ∞ h δ(q − p0 ) i
0
f (q)q 2 dq 2 f (q)
δ βσ − A βσ (p , q, p; p) tσα (q, p)
σ 0 q
XZ ∞ h δ(q − p0 ) i
0
= f (q)q 2 dq 2 f (q)
δ βσ − A βσ (p , q, p; p) Γσα (q, p)
σ 0 q
XXZ ∞ h δ(q − p0 ) i
0
+ f (q)q 2 dq 2 f (q)
δβσ − A βσ (p , q, p; p) Γσσ0 (q, p)
0 0 q
Zσ ∞ σ
× f (q 0 )q 0 2 dq 0 γσ0 (p, q 0 )G(q 0 , p) tσ0 α (q 0 , p)
0

XZ ∞ h δ(q − p0 ) i
= f (q)q 2 dq δβσ − Aβσ (p0 , q, p; p)
σ 0 q 2 f (q)
n X Z ∞ o
× Γσα (q, p) + Γσσ0 (q, p) f (q 0 )q 0 2 dq 0 γσ0 (p, q 0 )G(q 0 , p) tσ0 α (q 0 , p) . (28)
σ0 0

where, since
√ k is an arbitrary momentum variable, we have set it to the on-shell point,
k = p = 2µE.

6
 From Eqs. (27) and (28) one gets
XZ ∞
0 0
Γβα (p , p) = vβα (p , p) + f (q)q 2 dq Aβσ (p0 , q, p; k) Γσα (q, p) , (29)
σ 0

X Z ∞
0 0 0
tβα (p , p) = Γβα (p , p) + Γβσ (p , p) f (q)q 2 dq γσ (p, q)G(q, p) tσα (q, p)
σ 0
X X h i
= Γβα (p0 , p) + Γβσ (p0 , p) Bσα (p) = Γβσ (p0 , p) δσα + Bσα (p) . (30)
σ σ

Since for elastic two-body scattering the potential vβα is real, the auxiliary function Γβα is
real —the integral equation that defines it (29) has a kernel that is real and non-singular:
h i
2 0 0
f (q)q vβσ (p , q) − vβσ (p , p)γσ (p, q)
lim 2µ = finite number.
q→p p2 − q 2 + i
The expression for Bβα (p) is obtained as follows,
Z ∞
Bβα (p) = f (q)q 2 dqγβ (p, q)G(q, p)tβα (q, p)
0
Z ∞ h X i
2
= f (q)q dqγβ (p, q)G(q, p) Γβα (q, p) + Γβσ (q, p) Bσα (p)
0 σ

Xh Z ∞ i X
2
⇒ δβσ − f (q)q dqγβ (p, q)G(q, p)Γβσ (q, p) Bσα (p) = Dβσ (p)Bσα (p)
σ 0 σ
Z ∞
= f (q)q 2 dqγβ (p, q)G(q, p)Γβα (q, p) = Cβα (p) , (31)
0

and can be recast into a matrix equation, D · B = C. Note that D = 1 − C. Note also
that such matrix equation is in the number of channels coupled instead of the number
of gaussian quadrature points (B, C, and D are functions of the on-shell momentum p
only). For instance, in a two coupled-channel situation like the deuteron, B, C, and D
are 2 × 2 matrices with indices β, α, σ, γ, running from 1 to 2. From Eq. (31) defining
Dβσ (p), one finds formally the solution for Bσα (p),
Bσα (p) = [D−1 (p)]σγ Cγα (p) or B = D−1 · C .
One can now obtain the half off-shell scattering amplitude as follows,
δσα = [D−1 (p)]σγ Dγα (p) , Dγα (p) + Cγα (p) = δγα ,
X h i X
tβα (p0 , p) = Γβσ (p0 , p)[D−1 (p)]σγ Dγα (p) + Cγα (p) = Γβσ (p0 , p)[D−1 (p)]σα .
σ,γ σ
(32)

7
In order to obtain obtain the inverse of the matrix D one needs to evaluate the elements
of the matrix C via Eq. (31). However,
√ due to the two-particle propagator G(q, p),
the integrand finds a pole at q = 2µE + iδ that generates the imaginary term of the
amplitude tβα . To deal with that numerically, the most appropriate procedure is to
calculate separately the real and the imaginary terms, where the latter can have its integral
evaluated exactly. An easier way to visualize this is to separate the integration region as
follows,
Z ∞ Z p−δ Z Z ∞
J(q, p) J(q, p)
dq 2 2
= + + dq
0 q − p − i 0 lower p+δ (q − p − iδ)(q + p + iδ)
Z ∞ Z
J(q, p) J(z, p)
= P dq 2 2
+ dz (33)
0 q −p lower (z − p)(z + p)
where G(q, p) = −2µ/(q 2 −p2 −i) and we have defined J(q, p) = −2µ f (q)q 2 γβ (p, q)Γβα (q, p).
The first term in the above equation is the principal value of the integral and contributes
R
to the real part, since J(q, p) is also real in the real q-axis. The integration path lower is
√ of infinitesimal radius δ iθ≈  in the complex q-plane just below
along the lower semi-circle
the pole q = p + iδ = 2µE + iδ. With z = p + δe and in the limit δ → 0 one has
Z 0
J(p + δeiθ , p) J(p, p) 0
Z Z
J(z, p) iθ J(p, p)
dz = i δe dθ iθ iθ
≈ i dθ = iπ .
lower (z − p)(z + p) −π δe (2p + δe ) 2p −π 2p
(34)
As for the principal value (real term), its computation is quite unstable numerically. The
trick is to use
Z p−δ Z ∞  Z p−δ   Z ∞  
1 dq 1 1 dq 1 1
+ dq 2 = − − + −
0 p+δ q − p2 0 2p p−q p+q p+δ 2p q − p q+p

1h ip−δ 1h i∞
= ln(p − q) − ln(p + q) + ln(q − p) − ln(q + p)
2p 0 2p p+δ

  p−δ   ∞    
1 p−q 1 q−p 1 δ 1 δ
= ln + ln = ln − ln
2p p+q 0 2p q + p p+δ 2p 2p − δ 2p 2p + δ
 
1 2p + δ
= ln → 0, (35)
2p 2p − δ
multiply the above 0 by J(p, p), and subtract it from the principal value,
Z ∞ Z ∞ Z ∞
J(q, p) J(q, p) − J(p, p) J(q, p) − J(p, p)
P dq 2 2
=P dq 2 2
= dq . (36)
0 q −p 0 q −p 0 q 2 − p2
In the last step, the principal-value prescription can be eliminated, since the zero in the
denominator of the integrand is exactly canceled by the zero in the numerator.

8
 In summary, the Tomio-Adhikari trick involves the computation of the following steps:

1. solve the integral equation for the (real) auxiliary function Γβα (p0 , p):
h i
0 0
X Z ∞ vβσ (p , q) − vβσ (p , k)γσ (k, q)
Γβα (p0 , p) = vβα (p0 , p) + f (q)q 2 dq 2 /2µ
Γσα (q, p) ,
σ 0 E − q

2. iterates one more time the above equation to get the on-shell function Γβα (p, p),
3. build the elements of the nch × nch matrix C, where nch is the number of coupled-
channels,
h i
Z ∞ f (q)q 2 γβ (p, q)Γβα (q, p) − f (p)p2 γβ (p, p)Γβα (p, p)
Cβα (p) = dq
0 E − q 2 /2µ

−iπ µpf (p)Γβα (p, p) ,

4. build (perhaps simultaneously) the elements of the matrix D = 1 − C, and invert D,

5. finally, get the half off-shell scattering amplitude
X h i
tβα (p0 , p) = Γβσ (p0 , p) D−1 (p) .
σα
σ

2 Separable interactions
2.1 uncoupled waves
We write the interaction in the partial wave l as
vl (p0 , p) = αl fl (p0 )gl (p) , (37)
where gl and fl play the role of form factors. This separable interaction allow us to obtain
a solution to Eq. (9):
Z
0 0 0
tl (p , p) = αl fl (p )gl (p) + αl fl (p ) gl (q)tl (q, p) = αl fl (p0 )gl (p) + αl fl (p0 )Fl (E, p) , (38)
qE

where we used the abreviation

∞
q 2 dq
Z Z
1
= . (39)
qE 0 2π E± − q 2 /2µ
2

Iterating the definition of Fl (E, p) once one gets

Z h i
Fl (E, p) = gl (q) αl fl (q)gl (p) + αl fl (q)Fl (E, p)
qE
h iZ h i
= αl gl (p) + Fl (E, p) fl (q)gl (q) = αl gl (p) + Fl (E, p) Gl (E) . (40)
qE

9
This allows one to obtain an explicit solution not only for Fl , but also for tl ,
αl gl (p)Gl (E)
Fl (E, p) = , (41)
1 − αl Gl (E)
αl fl (p0 )gl (p)
 
0 0 αl gl (p)Gl (E)
tl (p , p) = αl fl (p ) gl (p) + = , (42)
1 − αl Gl (E) 1 − αl Gl (E)
Z ∞ 2 Z ∞ 2
q dq fl (q)gl (q) q dq fl (q)gl (q)
with Gl (E) = 2 2
= −2µ . (43)
0 2π E± − q /2µ 0 2π 2 q 2 − 2µE±
An expression for the wave function can be obtained following analogous steps. The
partial wave equation for the wave function reads
Z ∞ 2
0 0 1 q dq
ψl (p , p) = φl (p , p) + 02 2
vl (p0 , q)ψl (q, p) (44)
E± −p /2µ 0 2π
Z ∞ 2
0 αl fl (p0 ) q dq 0 αl fl (p0 )
= φl (p , p) + gl (q)ψ l (q, p) = φl (p , p) + Hl (p)
E± −p02 /2µ 0 2π 2 E± −p02 /2µ
Z ∞ 2  
q dq αl fl (q)
Hl (p) = gl (q) φl (q, p) + Hl (p) = Il (p) + αl Hl (p)Gl (E) , (45)
0 2π 2 E± −q 2 /2µ
∞
q 2 dq
Z
Il (p)
⇒ Hl (p) = , with Il (p) = gl (q)φl (q, p) . (46)
1 − αl Gl (E) 0 2π 2

1 αl fl (p0 )
⇒ ψl (p0 , p) = φl (p0 , p) + Il (p)
E± −p02 /2µ 1 − αl Gl (E)
1 Il (p)
= φl (p0 , p) + 02
tl (p0 , p) . (47)
E± −p /2µ gl (p)
Note that, if φl (p0 , p) = (2π 2 /p02 )δ(p0 − p), one has
2π 2 0 1
ψl (p0 , p) = 02
δ(p − p) + tl (p0 , p) . (48)
p E± −p02 /2µ
From Eq. (42) it is clear that the condition for the existence of a bound state is αl =
1/G(−Bl ), where Bl = γl2 /2µ is the (positive) binding energy. The bound state wave
function ψ̄l is the solution of the homogeneous version of Eq. (44). From Eq. (47),

0 αl Il (p) 2µfl (p0 )

ψ̄l (p ) = −N , (49)
1 − αl Gl (−Bl ) γl2 +p02
where N is a normalization constant. As shown in the following with an explicit example,
this normalization constant removes the dependence on p in the above expression, as it
should be—the bound state wave function is expected to depend only on p0 . The argument
goes as follows: the p dependence on the wave function can never be determined by the
homogeneous version of Eq. (44), since the kernel depends only on p0 and on the integration

10
variable q. Any solution multiplied by an arbitrary function of p will also be a solution.
In order to have an unique solution one imposes the integral of ψl∗ ψl over the p0 -space to
be a constant (in the case where l is the only partial wave in the process, such constant
is the unity). This restriction forces the solution to be p-independent. This is a general
property of bound states, and not just specific to separable potentials.
An alternative (and more correct way) to obtain the bound state wave function is
starting directly with the homogeneous equation,
αl fl (p0 ) αl fl (p0 )
Z 2
0 q dq
ψ̄l (p ) = − g l (q) ψ̄l (q) = − H̄l . (50)
Bl + p02 /2µ 2π 2 Bl + p02 /2µ
In this case the p0 -dependence of the wave function is already determined. If one tries to
go further to determine H̄l as before,
Z 2  
q dq αl fl (q)
H̄l = gl (q) − H̄l (51)
2π 2 Bl + q 2 /2µ
one rather ends up with the condition for the presence of a bound state,
Z 2
q dq gl (q)fl (q)
1 = αl = αl Gl (−Bl ) . (52)
2π 2 −Bl − q 2 /2µ
H̄l can only be restricted via the normalization condition of the wave function, condition
that makes it independent of p.

2.1.1 Yamaguchi form factors

A Hulthén-type of interaction, which employs Yamaguchi form factors,

1 1
fl (q) = , gl (q) = , (53)
q 2 + Λ2 q 2 + Σ2
allows for an explicit solution of the integral Gl (E). With the help of the relation
q2 A
2 2 2
=
(q + A)(q + B)(q + C) (B − A)(A − C)
 
1 B(A − C) 1 C(A − B) 1
× 2 − + (54)
q + A A(B − C) q 2 + B A(B − C) q 2 + C
and defining r2 = −2µE± one gets
µ Λ2
G(E) = −
π 2 (Σ2 − Λ2 )(Λ2 − r2 )
Z ∞ 
Σ2 (Λ2 − r2 ) r2 (Λ2 − Σ2 )

1 1 1
× dq 2 − +
0 q + Λ2 Λ2 (Σ2 − r2 ) q 2 + Σ2 Λ2 (Σ2 − r2 ) q 2 + r2

Λ2 Σ2 (Λ2 − r2 ) 1 r2 (Λ2 − Σ2 ) 1
 
µ 1
= − − +
2π (Σ2 − Λ2 )(Λ2 − r2 ) Λ Λ2 (Σ2 − r2 ) Σ Λ2 (Σ2 − r2 ) r

11
 Σ(Λ2 − r2 ) r(Λ2 − Σ2 )
 
µ 1
= − Λ− +
2π (Σ2 − Λ2 )(Λ2 − r2 ) (Σ2 − r2 ) (Σ2 − r2 )
µ 1
Λ(Σ2 − r2 ) − Σ(Λ2 − r2 ) + r(Λ2 − Σ2 )
 
= − 2 2 2 2 2 2
2π (Σ − Λ )(Λ − r )(Σ − r )

µ (Σ − Λ)
ΛΣ + r2 − r(Λ + Σ)
 
= − 2 2 2 2 2 2
2π (Σ − Λ )(Λ − r )(Σ − r )

µ (Σ − Λ)(Λ − r)(Σ − r) µ 1
= − = − . (55)
2π (Σ2 − Λ2 )(Λ2 − r2 )(Σ2 − r2 ) 2π (Σ + Λ)(Λ + r)(Σ + r)
In the presence of a bound state (Bl = γl2 /2µ) interaction strength αl is given by
2π
αl = − (Σ + Λ)(Λ + γl )(Σ + γl ) . (56)
µ
With these results the scattering amplitude (42) is completely determined.
In order to obtain the bound state wave function one determines H̄l in Eq. (50) via
Z ∞ 2 Z ∞ 2
q dq ∗ 2 q dq 4µ2 1
2
ψ̄l (q)ψ̄l (q) = αl H̄l 2
= 1. (57)
0 2π 0 2π (γl +q ) (q + Λ2 )2
2 2 2 2

(Ist das immer gültig? Ergebnis von Partialwelle unitarität?) Using

q2 2γl2 2Λ2 (Λ2 −γl2 )q 2 (Λ2 −γl2 )q 2
 
1
= − + + 2 , (58)
(q 2 +γl2 )2 (q 2 +Λ2 )2 (Λ2 −γl2 )3 q 2 +γl2 q 2 +Λ2 (q 2 +γl2 )2 (q +Λ2 )2
Z ∞  q ∞
dq 1 π
2 2
= arctan = , (59)
0 q +a a a 0 2a
Z ∞ ∞ Z ∞
q 2 dq
 
1 q dq π
2 2 2
= − 2 2
+ 2 2
= , (60)
0 (q + a ) 2 q +a 0 0 q +a 4a
one gets
∞
q 2 dq
Z  
π 2 2 (Λ+γl )
= 4(γl −Λ) + (Λ −γl )
0 (q 2 + γl2 )2 (q 2 + Λ2 )2 4(Λ2 −γl2 )3 Λγl
 
π(Λ−γl ) 1 π
= 2 2 3
− 4Λγl + (Λ+γl )2 = . (61)
4(Λ −γl ) Λγl 4Λγl (Λ+γl )3
Therefore,
p
2 4µ2 π 8πΛγl (Λ+γl )3
αl H̄l = 1 ⇒ αl H̄l = . (62)
2π 2 4Λγl (Λ+γl )3 2µ
Eq. (50) then becomes
p p
8πΛγ (Λ+γ ) 3 8πΛγl (Λ+γl )3 1
l l
ψ̄l (p0 ) = 2 02
f l (p 0
) = 2 02 02
. (63)
γl + p γl + p p + Λ2

12
 It is interesting to check that the result above is the correct expression for the wave
function. Its Fourier transform has to satisfy the Schrödinger equation in coordinate
space, Z
−∇ ψ̃(r) + 2µ d3 r0 Ṽ (r, r 0 )ψ̃(r 0 ) = 2µE ψ̃(r) ,
2
(64)
with the Laplacian in spherical coordinates given by
∂2
     
2 1 ∂ 2 ∂ 1 ∂ ∂
∇ = 2 r + 2 2 sin θ sin θ + 2 . (65)
r ∂r ∂r r sin θ ∂θ ∂θ ∂φ
To keep it simple we assume the potential to be in S-wave and ignore the orbital angular
momentum subscript. The wave function and potential in coordinate space read
p
d3 q i q · r 8πΛγ(Λ + γ)3
Z
ψ̃(r) = e
(2π)3 (q 2 + γ 2 )(q 2 + Λ2 )
p
d3 q iq·r
 
8πΛγ(Λ + γ)3
Z
1 1
= e −
(Λ2 − γ 2 ) (2π)3 (q 2 + γ 2 ) (q 2 + Λ2 )
p
8πΛγ(Λ + γ)3 e−γr e−Λr
 −γr
e−Λr
  
e
= − = Ñ − , (66)
4π(Λ2 − γ 2 ) r r r r
0
d3 q i q · r d3 q 0 −iq0 ·r0 e−Λr e−Σr
Z Z
0 α
Ṽ (r, r ) = e e = α . (67)
(2π)3 (2π)3 (q 2 + Λ2 )(q 02 + Σ2 ) 4πr 4πr0
Substitution into the l.h.s. of the Schrödinger equation yields
Z 3 0 −Σr0  −γr0 0
1 ∂  −γr −Λr
 e−Λr dr e e e−Λr
−Ñ 2 −e (1+γr) + e (1+Λr) + 2µα Ñ − 0
r ∂r 4πr 4π r0 r0 r
e−γr e−Λr µα e−Λr ∞ 0 −Σr0 h −γr0
  Z i
0
= −Ñ γ 2 − Λ2 + Ñ dr e e − e−Λr
r r 2π r 0
−γr −Λr −Λr
   
2e 2e µα e 1 1
= −Ñ γ −Λ + Ñ −
r r 2π r Σ+γ Σ+Λ
−γr
e−Λr
  
2e 2 µ (Λ − γ)
= −Ñ γ − Λ +α . (68)
r r 2π (Σ + γ)(Σ + Λ)
Using the condition (56) on α one gets
−γr
e−Λr
  
2e 2 µ (Λ − γ)
−Ñ γ − Λ +α
r r 2π (Σ + γ)(Σ + Λ)
−γr
e−Λr  2
 −γr
e−Λr
  
2e 2 2 2 e

= −Ñ γ − Λ − (Λ − γ ) = −γ Ñ −
r r r r
= −γ 2 ψ̃(r) , (69)
which is the correct result (64) as E = −B = −γ 2 /2µ.

13
2.2 two coupled waves
For a more realistic situation of nucleon-nucleon scattering one deals with the coupling of
waves generated by tensor forces. For each conserved value of the total angular momentum
J rises three possibilities of orbital angular momenta L that couple to total spin S = 1:
L = J and L = J ±1. The first case has the opposite parity of the other two, and therefore
do not couple to them or any other wave. The other two form the two-dimensional
representation Z
0 0
tJ (p , p) = vJ (p , p) + vJ (p0 , q) tJ (q, p) , (70)
qE
where
   
0 xJ−1,J−1 (p0 , p) xJ−1,J+1 (p0 , p) x11 (p0 , p) x12 (p0 , p)
xJ (p , p) = = , (71)
xJ+1,J−1 (p0 , p) xJ+1,J+1 (p0 , p) x21 (p0 , p) x22 (p0 , p)
with x = t or v. Each element in the above matrices has a specific summation over
the z-direction projected orbital (ml ) and spin (ms ) quantum numbers. With separable
interactions of the form
vij (p0 , p) = αij fij (p0 )gij (p) (72)
the equation for tij reads
2
X Z
0 0 0
tij (p , p) = αij fij (p )gij (p) + αin fin (p ) gin (q)tnj (q, p)
n=1 qE

2
X
0
= αij fij (p )gij (p) + αin fin (p0 )Finj (p) . (73)
n=1

For Finj , in analogy to the uncoupled waves case, one has

Z " 2
#
X
Finj (p) = gin (q) αnj fnj (q)gnj (p) + αns fns (q)Fnsj (p)
qE s=1
2
X
= αnj gnj (p)Ginj + αns Gins Fnsj (p) , (74)
s=1

where Z
Ginj = gin (q)fnj (q) (75)
qE
is a known function once the form factors are chosen. The Finj ’s satisfy
F11j = α1j g1j G11j + α11 G111 F11j + α12 G112 F12j ,
F21j = α1j g1j G21j + α11 G211 F11j + α12 G212 F12j ,
F12j = α2j g2j G12j + α21 G121 F21j + α22 G122 F22j ,
F22j = α2j g2j G22j + α21 G221 F21j + α22 G222 F22j .

14
One then sees that they are solutions of
(I − A) · uj = yj (76)
where I is a 4x4 identity matrix,
F11j α1j g1j G11j α11 G111 0 α12 G112 0
     
F  α g G  α G 0 α12 G212 0
uj =  21j  , yj =  1j 1j 21j  , A =  11 211 .

F12j α2j g2j G12j 0 α21 G121 0 α22 G122 
F22j α2j g2j G22j 0 α21 G221 0 α22 G222
(77)
Renaming
a = α11 G111 , b = α12 G112 , c = α11 G211 , d = α12 G212 ,
e = α21 G121 , f = α22 G122 , g = α21 G221 , h = α22 G222 ,
κ = (1−h) , % = (1−a) , ς = bc+d−ad , ϑ = e+f g−eh ,
∆ = κ%−ςϑ , (78)
one gets
a g11 b g12 a 0 b 0
     
cg  dg  c 0 d 0
y1 =  11  , y2 =  12  , A = 
e g21 f g22 0 e 0 f
g g21 h g22 0 g 0 h
κ−dϑ bϑ κb fb
 
1  cκ %κ κς fς 
⇒ (I − A)−1 =  . (79)
∆ cϑ %ϑ κ% f% 
cg %g gς %−eς
With the above inverted matrix one determines the functions Finj (p) and, consequently,
the elements of the scattering amplitude. The former read
F111 = {[aκ − (ad − bc)ϑ] g11 + bϑ g21 }/∆ ,
F211 = {cκ g11 + ςϑ g21 }/∆ ,
F121 = {cϑ g11 + %ϑ g21 }/∆ ,
F221 = {cg g11 + g% g21 }/∆ ,

F112 = {bκ g12 + f b g22 }/∆ ,

F212 = {ςκ g12 + f ς g22 }/∆ ,
F122 = {ςϑ g12 + f % g22 }/∆ ,
F222 = {ςg g12 + [h% − (eh − f g)ς] g22 }/∆ .
Writing Eq. (73) explicitly,
t11 (p0 , p) = α11 f11 (p0 )g11 (p) + α11 f11 (p0 )F111 (p) + α12 f12 (p0 )F121 (p)

15
 1n
= α11 f11 (p0 ){[aκ − (ad − bc)ϑ + (κ% − ςϑ)] g11 (p) + bϑ g21 (p)}
∆ o
+α12 f12 (p0 )ϑ[c g11 (p) + % g21 (p)]
1n 0 0
o
= α11 f11 (p )[(κ−dϑ) g11 (p) + bϑ g21 (p)] + α12 f12 (p )ϑ[c g11 (p) + % g21 (p)] ,
∆
(80)

t21 (p0 , p) = α21 f21 (p0 )g21 (p) + α21 f21 (p0 )F211 (p) + α22 f22 (p0 )F221 (p)
1n
= α21 f21 (p0 ){cκ g11 (p) + [ςϑ + (κ%−ςϑ)] g21 (p)}
∆ o
+α22 f22 (p0 )g[c g11 (p) + % g21 (p)]
1n o
= α21 f21 (p0 )κ[c g11 (p) + % g21 (p)] + α22 f22 (p0 )g[c g11 (p) + % g21 (p)]
∆
1h 0 0
ih i
= κ α21 f21 (p ) + g α22 f22 (p ) c g11 (p) + % g21 (p) , (81)
∆
t12 (p0 , p) = α12 f12 (p0 )g12 (p) + α11 f11 (p0 )F112 (p) + α12 f12 (p0 )F122 (p)
1n
= α11 f11 (p0 )[bκ g12 (p) + f b g22 (p)]
∆ o
+α12 f12 (p0 ){[ςϑ + (κ%−ςϑ)] g12 (p) + f % g22 (p)}
1n 0 0
o
= α11 f11 (p )b[κ g12 (p) + f g22 (p)] + α12 f12 (p )%[κ g12 (p) + f g22 (p)]
∆
1h ih i
= b α11 f11 (p0 ) + % α12 f12 (p0 ) κ g12 (p) + f g22 (p) , (82)
∆
t22 (p0 , p) = α22 f22 (p0 )g22 (p) + α21 f21 (p0 )F212 (p) + α22 f22 (p0 )F222 (p)
1n
= α21 f21 (p0 )ς[κ g12 (p) + f g22 (p)]
∆ o
0
+α22 f22 (p ){ςg g12 (p) + [h% − (eh−f g)ς + (κ%−ςϑ)] g22 (p)}
1n o
= α21 f21 (p0 )ς[κ g12 (p) + f g22 (p)] + α22 f22 (p0 )[ςg g12 (p) + (% − eς) g22 (p)] .
∆
(83)
This is the most general solution for interactions of the form (72). However, we know that
strong interactions are invariant under the time-reversal operation. That constraints the
interaction matrix to be symmetric, namely, vij (p0 , p) = vji (p, p0 ). That means
α21 = α12 , g21 (p) = f12 (p) , f21 (p) = g12 (p) , gii (p) = fii (p) . (84)

16
The functions Ginj then become
R R 2
R R
G111 = qE g11 f11 = qE f11 , G112 = g f =
qE 11 12 qE
f11 f12 ,
R R 2
R
G121 = qE g12 f21 = qE g12 , G122 = qE
g12 f22 ,
R R R R (85)
G211 = qE g21 f11 = qE f12 f11 = G112 , G212 = g f =
qE 21 12
f2 ,
qE 12
R R R R 2
G221 = qE g22 f21 = qE f22 g12 = G122 , G222 = qE
g22 f22 = qE
f22 .
The relations G211 = G112 and G221 = G122 lead to
α11 α22
c= b, and f= g. (86)
α12 α12
With the above constraints the elements of the amplitude matrix read
1n
t11 (p0 , p) = α11 (κ−dϑ)f11 (p0 )f11 (p)
∆ o
+α12 cϑ [f11 (p0 )f12 (p) + f12 (p0 )f11 (p)] + α12 ϑ% f12 (p0 )f12 (p) , (87)

1 h ih i
t21 (p0 , p) = α12 κ g12 (p0 ) + f f22 (p0 ) c f11 (p) + % f12 (p) , (88)
∆
0 1 h
0 0
ih i
t12 (p , p) = α12 c f11 (p ) + % f12 (p ) κ g12 (p) + f f22 (p) , (89)
∆
0 1n
t22 (p , p) = α12 ςκ g12 (p0 )g12 (p)
∆ o
+α12 f ς[g12 (p0 )f22 (p) + f22 (p0 )g12 (p)] + α22 (% − eς) f22 (p0 )f22 (p) . (90)

Note that tij (p0 , p) = tji (p, p0 ), as required by time-reversal invariance.

2.2.1 wave function

The scattering wave function equation for two coupled waves reads
Z ∞ 2
0 0 1 q dq
ψ J (p , p) = φJ (p , p) + 02
vJ (p0 , q)ψ J (q, p) , (91)
E± − p /2µ 0 2π 2
using the notation (71) for ψ J and φJ . With the separable interaction of Eq. (72) one has
2 Z ∞ 2
0 0 1 X
0 q dq
ψij (p , p) = φij (p , p) + 02
αin fin (p ) gin (q)ψnj (q, p)
E± − p /2µ n=1 0 2π 2
2
0 1 X
= φij (p , p) + αin fin (p0 ) Hinj (p) ,
E± − p02 /2µ n=1
Z ∞ 2 " 2
#
q dq 1 X
Hinj (p) = gin (q) φnj (q, p) + αns fns (q) Hnsj (p)
0 2π 2 E± − q 2 /2µ s=1

17
 2
X
= Iinj (p) + αns Gins Hnsj (p) , (92)
s=1

with the same Ginj integrals defined by Eq. (75). In analogy with the calculations for the
scattering amplitude, the Hinj solutions are given by
vj = (I − A)−1 · zj (93)
where
H11j I11j
   
H  I 
vj =  21j  and zj =  21j  . (94)
H12j I12j
H22j I22j
The explicit results read
H11j = {(κ − dϑ)I11j + bϑI21j + b(κI12j + f I22j )}/∆ ,
H21j = {κ(cI11j + %I21j ) + ς(κI12j + f I22j )}/∆ ,
H12j = {ϑ(cI11j + %I21j ) + %(κI12j + f I22j )}/∆ ,
H22j = {g(cI11j + %I21j ) + gςI12j + (% − eς)I22j }/∆ ,

0 0 1 1n
ψ1j (p , p) = φ1j (p , p) +
(E± − p02 /2µ) ∆
h i
α11 f11 (p0 ) (κ−dϑ) I11j (p) + bϑI21j + b[κ I12j (p)+f I22j (p)]
h io
+α12 f12 (p0 ) ϑ[c I11j (p)+% I21j (p)] + %[κ I12j (p)+f I22j (p)] , (95)

1 1n
ψ2j (p0 , p) = φ2j (p0 , p) +
(E± − p02 /2µ) ∆
h i
0
α21 f21 (p ) κ[c I11j (p)+% I21j (p)] + ς[κ I12j (p)+f I22j (p)]
h io
+α22 f22 (p0 ) g[c I11j (p)+% I21j (p)] + [ςg I12j (p)+(%−eς) I22j (p)] . (96)

It is straightforward to check that if φij (p0 , p) = δij (2π 2 /p02 )δ(p0 − p) one has
Z ∞ 2
q dq
Iinj (p) = gin (q)φnj (q, p) = δnj gin (p)
0 2π 2
2π 2 0 1
⇒ ψij (p0 , p) = δij 02
δ(p − p) + tij (p0 , p) . (97)
p E± − p02 /2µ
For the bound state case one starts with the homogeneous version of Eq. (91),
2 Z ∞ 2
0 1 X
0 q dq
ψ̄ij (p ) = 02
αin fin (p ) gin (q)ψ̄nj (q)
−BJ − p /2µ n=1 0 2π 2

18
 2
1 X
= αin fin (p0 ) H̄inj ,
−BJ − p02 /2µ n=1
Z ∞ 2 " 2
# 2
q dq 1 X X
H̄inj = gin (q) αns fns (q) H̄nsj = αns Ḡins H̄nsj .(98)
0 2π 2 −BJ −q 2 /2µ s=1 s=1

The integrals Ḡinj are the same defined by Eq. (75), but evaluated at the bound state
energy E = −BJ . The constants H̄inj satisfy
H̄11j
 
 H̄ 
Ā · h = h where h =  21j  (99)
H̄12j
H̄22j
and Ā is the matrix A in Eq. (77) evaluated at E = −BJ . At this point it is clear that
H̄inj , and consequently ψ̄ij , do not depend on the index j. This is somewhat expected:
while for scattering it makes sense to talk about transition probabilities between initial
and final states with well-defined orbitals J − 1 and J + 1, for bound states one talks
about the “amount” of J − 1 and J + 1 waves that are present in the bound state. Or,
holding on the scattering picture, one talks about probabilities of J − 1 and J + 1 virtual
states (with imaginary momenta) to end up in the bound state.
The solution of Eq. (99) can be easily obtained in terms of the variables (78), with a
bar on top to indicate that they are evaluated at E = −BJ . The equations read
H̄11 = āH̄11 + b̄H̄12 , (100)
H̄21 = c̄H̄11 + d¯H̄12 , (101)
H̄12 = ēH̄21 + f¯H̄22 , (102)
H̄22 = ḡ H̄21 + h̄H̄22 . (103)
Solving
ḡ ḡ
from Eq. (103): H̄22 = H̄21 = H̄21 , (104)
(1 − h̄) κ̄
(ēκ̄ + f¯ḡ) ϑ̄
from Eqs. (102,104): H̄12 = H̄21 = H̄21 , (105)
κ̄ κ̄
b̄ b̄ϑ̄
from Eqs. (100,105): H̄11 = H̄12 = H̄21 , (106)
(1 − ā) κ̄ %̄
b̄c̄ϑ̄ d¯ϑ̄
from Eqs. (101,105,106): H̄21 = H̄21 + H̄21 ⇒ κ̄ %̄ = ς¯ϑ̄ . (107)
κ̄ %̄ κ̄
The equations above determine h apart from a normalization constant. The last equation,
instead of determining H̄21 , provides the condition for a bound state, Det(I − Ā) = 0.
One then writes h as
N
hT = − ( b̄ϑ̄ κ̄ %̄ %̄ϑ̄ %̄ḡ ) . (108)
2µ

19
The sign and 2µ factor are for later convenience. The bound state wave functions become
N ϑ̄ 
ψ̄J−1 (p0 ) = 0 0

α11 f 11 (p ) b̄ + α12 f 12 (p ) %̄ , (109)
γJ2 + p02
N %̄
ψ̄J+1 (p0 ) = [α21 f21 (p0 ) κ̄ + α22 f22 (p0 ) ḡ] . (110)
γJ2 + p02
Note that these expressions resemble Eqs. (95,96) if all Iinj = 0 but I21j = 1.
Now we consider form factors like Eq. (53). To get N one needs to evaluate
Z ∞
q2
L(γ, Λ, Σ) = dq 2 . (111)
0 (q + γ 2 )2 (q 2 + Λ2 )(q 2 + Σ2 )
Using
q2
=
(q 2 +γ 2 )2 (q 2 +Λ2 )(q 2 +Σ2 )
q2 γ 2 (Λ2 +Σ2 −2γ 2 )
 
1 1
+
(Λ2 −γ 2 )(Σ2 −γ 2 ) (q 2 +γ 2 )2 (Λ2 −γ 2 )(Σ2 −γ 2 ) (q 2 +γ 2 )
Λ2 Σ2
 
1 1 1
+ 2 − (112)
(Λ −Σ2 ) (Λ2 −γ 2 )2 (q 2 +Λ2 ) (Σ2 −γ 2 )2 (q 2 +Σ2 )
and Eqs. (59,60) one finds
h i
π (Λ2 −γ 2 )(Σ2 −γ 2 ) + 2γ 2 (Λ2 +Σ2 −2γ 2 )
L(γ, Λ, Σ) =
4γ[(Λ2 −γ 2 )(Σ2 −γ 2 )]2
 
π Λ Σ πD
+ 2 2 2 2 2
− 2 2 2 = , (113)
2(Λ −Σ ) (Λ −γ ) (Σ −γ ) 4γ(Λ −Σ )[(Λ2 −γ 2 )(Σ2 −γ 2 )]2
2 2

D = (Λ2 −Σ2 )(Λ2 −γ 2 )(Σ2 −γ 2 ) + 2γ 2 (Λ2 −Σ2 )(Λ2 +Σ2 −2γ 2 )

+2γ[Λ(Σ2 −γ 2 )2 − Σ(Λ2 −γ 2 )2 ]
= (Λ2 −Σ2 )(Λ2 −γ 2 )(Σ2 −γ 2 ) + 2γ 2 (Λ2 −Σ2 )[(Λ2 −γ 2 ) + (Σ2 −γ 2 )]
+2γ{Λ(Σ2 −γ 2 )[(Σ2 −Λ2 ) + (Λ2 −γ 2 )] − Σ(Λ2 −γ 2 )[(Λ2 −Σ2 ) + (Σ2 −γ 2 )]}
= (Λ2 −γ 2 )(Σ2 −γ 2 )[(Λ2 −Σ2 ) + 2γ(Λ−Σ)]
+2γ(Λ2 −Σ2 )[γ(Λ2 −γ 2 ) + γ(Σ2 −γ 2 ) − Λ(Σ2 −γ 2 ) − Σ(Λ2 −γ 2 )]
= (Λ2 −γ 2 )(Σ2 −γ 2 )(Λ−Σ)(Λ+Σ+2γ)
−2γ(Λ2 −Σ2 )[(Λ2 −γ 2 )(Σ−γ) + (Σ2 −γ 2 )(Λ−γ)]
= (Λ+γ)(Σ+γ)(Λ−γ)(Σ−γ)(Λ−Σ)(Λ+Σ+2γ)
−2γ(Λ+Σ)(Λ−Σ)(Λ−γ)(Σ−γ)(Λ+Σ+2γ)
= (Λ−γ)(Σ−γ)(Λ−Σ)(Λ+Σ+2γ)[ΛΣ + γ(Λ+Σ) + γ 2 − 2γ(Λ+Σ)]
(Λ−γ)2 (Σ−γ)2 (Λ−Σ)(Λ+Σ+2γ) . (114)

20
Therefore,
 
π(Λ + Σ + 2γ) π 1 1
L(γ, Λ, Σ) = = + .
4γ(Λ+Σ)(Λ+γ)2 (Σ+γ)2 4γ(Λ+Σ)(Λ+γ)(Σ+γ) (Λ+γ) (Σ+γ)
(115)
The norm of ψ̄J−1 reads
d3 q ∗ (N ϑ̄)2 h
Z
0 0
ψ̄ (p )ψ̄J−1 (p ) = (α11 b̄)2 L(γJ , Λ11 , Λ11 )
(2π)3 J−1 2π 2
i
+2α11 α12 b̄%̄ L(γJ , Λ11 , Λ12 ) + (α12 %̄)2 L(γJ , Λ12 , Λ12 )

N 2 ϑ̄2 (α11 b̄)2 (α12 %̄)2

 
2α11 α12 b̄%̄(Λ11 +Λ12 +2γJ )
= + +
8πγJ Λ11 (Λ11 +γJ )3 (Λ11 +Λ12 )(Λ11 +γJ )2 (Λ12 +γJ )2 Λ12 (Λ12 +γJ )3
= N 2 NJ−12
, (116)

d3 q ∗ (N %̄)2 h
Z
0 0
3
ψ̄J+1 (p )ψ̄J+1 (p ) = 2
(α21 κ̄)2 L(γJ , Λ21 , Λ21 )
(2π) 2π
i
+2α21 α22 κ̄ḡ L(γJ , Λ21 , Λ22 ) + (α22 ḡ)2 L(γJ , Λ22 , Λ22 )

N 2 %̄2 (α21 κ̄)2 (α22 ḡ)2

 
2α21 α22 κ̄ḡ(Λ21 +Λ22 +2γJ )
= + +
8πγJ Λ21 (Λ21 +γJ )3 (Λ21 +Λ22 )(Λ21 +γJ )2 (Λ22 +γJ )2 Λ22 (Λ22 +γJ )3
= N 2 NJ+1
2
. (117)
The normalization constant then becomes
1
N=q . (118)
2 2
NJ−1 + NJ+1

2.2.2 phase shifts

With the definition (3) for the scattering amplitude the S-matrix is given by
µ
S=I− 2ik T , (119)
2π
√
where k = 2µE+ is the on-shell momentum for a given positive energy E+ added with an
infinitesimal and positive imaginary part. The elements of T are the amplitudes (87-90)
evaluated at the on-shell point, tij (k, k). For this two-dimensional coupled wave case it is
convenient to write the elements of S in the SYM (or nuclear-bar) parametrization,
 
e2iδ̄− cos 2¯ iei(δ̄− +δ̄+ ) sin 2¯
S= , (120)
iei(δ̄− +δ̄+ ) sin 2¯ e2iδ̄+ cos 2¯
where δ̄± is the phase shift associated to the state with orbital angular momentum L =
J ± 1 and ¯ the mixing parameter. It requires just a simple algebra to obtain these

21
quantities. Calling sij , ∆S , and RS the elements, the determinant, and the ratio between
the first and second diagonal elements of S one has
   
ln(∆S RS ) ln(∆S /RS ) 1 s12 1 s12
δ̄− = , δ̄+ = , ¯ = arctan √ = arcsin √ .
4i 4i 2 is22 RS 2 i ∆S
(121)

References
[1] Numerical Recipes.
[2] L. Tomio and S. K. Adhikari, Phys. Rev. C 22, 28 (1980); Phys. Rev. C 22, 2359
(1980); Phys. Rev. C 24, 43 (1981).

22