PHYSICAL REVIEW C 105, 054003 (2022)

Accuracy of ground-state solutions of boson systems in the revised few-body

integrodifferential equations approach

G. J. Rampho
Department of Physics, University of South Africa, Private Bag X6, Florida 1710, South Africa

(Received 26 December 2021; accepted 4 May 2022; published 16 May 2022)

In this work, the traditional two-variable few-body integrodifferential equations approach is reviewed. Bound-
ary conditions are implemented in both the hyperradial and hyperangular domains. A novel procedure of
including effects of higher partial waves of the interaction potential is introduced. The new approach reproduces
results obtained by an exact method for boson systems. These results suggest that many-body correlations are
negligibly small in systems with short-range interactions and that effect of higher partial waves are adequately
accounted for by the revised hypercentral potential.

DOI: 10.1103/PhysRevC.105.054003

I. INTRODUCTION results in regularized integrodifferential equations for s-wave

potentials.
Many of the methods in few-body physics are capable
In the IDEA, effects of higher partial waves are accounted
of treating known features of the systems and degrees of
for by introducing a zeroth order hypercentral potential
freedom of constituents exactly. However, the implementation
multipole to the s-projected integrodifferential equations. Tra-
of their full capacity is prohibited by resulting complexities
ditionally, the potential multipoles are determined using a
as the number of constituents increases. Examples of such
parameter that depends on the size of the system. However,
methods are the widely used Faddeev formalism [1] and
such a procedure faces challenges when the size of the system
the hyperspherical harmonics expansion (HHE) method [2,3].
is large [12]. In this work, the procedure for determining
The exact implementation of the Faddeev formalism is limited
the hypercentral potential is revised to be independent of the
to, at most, four-particle systems, while the HHE method is
number of particles, and the procedure for the inclusion of
limited to less than seven-particle systems.
the hypercentral potential in the s-projected integrodifferential
The few-body integrodifferential equations approach
equations is revised to be physically meaningful.
(IDEA) [4] circumvent shortcomings associated with few-
In Sec. II, the many-body Faddeev integrodifferential equa-
body methods by considering two approximations. The
tions are presented. The derivation entails the elimination of
IDEA [4,5] uses the Faddeev formalism to treat explicitly only
first-order derivatives from the kinetic energy operator from
two-particle correlations, and approximates effects of higher
the standard IDEA equations, introducing explicit boundary
partial waves. These approximations reduce the many-body
conditions in the angular domain. Test calculations with the
Schrödinger equation in hyperspherical coordinates to a sys-
new integrodifferential equations are conducted on A-boson
tem of coupled two-variable Faddeev-type integrodifferential
systems with A = 10 at most. Results are presented and dis-
equations, the form of which does not change as the number
cussed in Sec. III. Conclusions are reported in Sec. IV.
of constituents of the system changes. The system of coupled
equations reduces to one two-variable equation in the case of
systems of identical particles. This single integrodifferential II. FORMALISM
equation can treat systems of up to a million particles [6]. The The derivation of the integrodifferential equations for few-
description of quantum few-body systems using this approach body systems is now commonplace in the literature. In this
has shown considerable success in studies of various nuclear section, only features of the derivation that are relevant to the
and atomic few-body as well as many-body quantum mechan- goal of revising the equations are recalled.
ical systems [7–9].
Although the IDEA is extensively discussed in the lit-
erature, an explicit implementation of boundary conditions A. Regularization
in the angular domain [10] is still lacking. The direct solu- Consider a system of A identical bosons, each with mass
tion of the integrodifferential equation explored in Ref. [11] m and position vector ri (i = 1, 2, . . . , A). Internal properties
for bound states reveals that the convergence of the nu- of this system are extracted by eliminating the center-of-mass
merical solution is sensitive to the boundary behavior of kinetic energy. That is achieved in the Jacobi coordinates
√ 
the amplitudes. This paper proposes a modification to the defined by the vectors ηi = 2 i/(i + 1) ( ri+1 − ij=1 rj /i ),
IDEA that accounts for boundary conditions of the ampli- where ηN = r1 − r2 and N = A − 1. However, the dynami-
tudes in both the hyperradial and hyperangular domains. This cal equation for the system is more conveniently solved in

2469-9985/2022/105(5)/054003(5) 054003-1 ©2022 American Physical Society

G. J. RAMPHO PHYSICAL REVIEW C 105, 054003 (2022)

hyperspherical coordinates (r, N ) where r ∈ [0, ∞) is the where L̂2 (N ) is the grand orbital angular momentum op-
hyperradius and N ≡ {ω1 , . . . , ωN ; ϕ2 , . . . , ϕN } a set of an- erator. This operator is constructed through the recurrence
gles, constituted by the spherical polar angles ωi (θi , φi ) of the relation
Jacobi vectors and hyperangles ϕi ∈ [0, 21 π ]. The definitions
of these coordinates are readily found in the literature [3,10]. ∂2 ∂
In hyperspherical coordinates, ηN = r cos ϕN . L̂2 (N ) = + [(3N − 4) cot ϕN − 2 tan ϕN ]
∂ϕN2 ∂ϕN
When the bosons interact only through two-body potentials
V (ri j ), where ri j = ri − rj , the total interactions in the sys- ˆ2 (ωN ) L̂2 (N−1 )
 + + , (4)
tem are given by Aij V (ri j ). In the Faddeev formalism, the cos ϕN
2
sin2 ϕN
Schrödinger wave function for the system is expressed in the
form [13] with L̂2 (1 ) = ˆ2 (ω1 ) and N = {ωN , ϕN ; N−1 }. The opera-
tors ˆ2 (ωi ) are orbital angular momentum operators associated

A
(r, N ) = ψi j (r, N ), (1) with the Jacobi vectors ηi .
ij It is noted that in (3) and (4), the first-order derivatives
have singularities associated with them. To eliminate these
where ψi j (r, N ) are Faddeev-type two-body amplitudes. first-order derivatives, one introduces the reduced two-body
These 21 A ( A − 1 ) amplitudes satisfy the coupled amplitudes
equations [4]

A ψ̄i j (r, N )
ψi j (r, N ) = (5)
(T̂0 − E ) ψi j (r, N ) = −V (ri j ) ψkl (r, N ), (2) r (3A−4)/2 (sin ϕN )(3A−7)/2 cos ϕN
kl
and the corresponding reduced total wave function
obtained from the Schrödinger equation for the system, where 
¯ (r, N ) = Aij ψ̄i j (r, N ). The reduced amplitudes
T̂0 is the internal kinetic energy operator and E is the energy
of the system. The kinetic energy woperator has the general satisfy the boundary conditions ψ̄i j (r, N ) = 0 for r = 0,
form ϕN = 0, and ϕN = 21 π . The condition ψ̄i j (∞, ) = 0 is also
  considered when determining bound states. The reduced
h̄2 ∂ 2 3A − 4 ∂ L̂2 (N )
T̂0 = − + + , (3) amplitudes solve the coupled equations
m ∂r 2 r ∂r r2
  
h̄2 1 ∂2 ˆ2 (ωN ) L̂2 (N−1 ) − ρA 
A
− T̂r + 2 + + ψ̄i j (r, N ) − E ψ̄i j (r, N ) = −V ( 
r i j ) ψ̄kl (r, N ), (6)
m r ∂ϕN2 cos2 ϕN sin2 ϕN kl

where ρA = (3N − 4)(3N − 6)/4 = (3A − 7)(3A − 9)/4 and harmonics and [L] a set of quantum numbers. However,
Y[0] (N ) = constant for states that satisfy (8). F (r, ri j ) can
∂ 2
1
T̂r = + 2. (7) be expanded in spherical harmonics Y m (ωN ) as
∂r 2 4r 
Note that Eq. (6) do not involve any approximation. Ap- F (r, ri j ) = F (r, ri j )Y m (ωN ), (9)
proximations accounting for only two-body correlations are
introduced, in the next subsection. were is the orbital angular momentum quantum number with
azimuthal projection m . Note that the same spherical harmon-
B. Two-body correlation ics arise in the case of the two-body interaction potentials.
Two-body correlation are dominant in systems in which The equations defining the amplitudes F (r, ri j ) are ob-
constituents are weakly interacting or interacting through tained by applying condition (8) and the expansion (9) in (6),
short-range forces. The summation over {kl} in (6) consti- and then project the equations onto the ri j space through the
tutes permutations of particles in, or kinematic rotations of, overlap integral ωN N−1 )| Eq. (6) to obtain
the system. When three-body and higher order correlations  
h̄2 1 ∂2 ( + 1) ρA
are neglected, the resulting wave functions of the system are − T̂r + 2 − − F (r, ϕN )
invariant under rotations in the D − 3 subspace spanned by the m r ∂ϕN2 cos2 ϕN sin2 ϕN
vectors η1 , . . . , ηN−1 [14]. Such invariance is expressed by the − E F (r, ϕN )
property 

A
L̂ (N−1 )ψ̄i j (r, N ) = 0,
2
(8) = −V (r, ϕN ) N ψ̄kl (r, N ) , (10)
kl
which also implies that the amplitudes depend mainly on
the global coordinate r and the local vector ri j . That is, where N = ωN N−1 and [ ˆ2 (ωN ) − ( + 1) ] Y m (ωN ) =
the amplitudes can be expressed in the form ψ̄i j (r, N ) = 0 was used. The evaluation of the integral on the right-hand
Y[L] (N ) F (r, ri j ), where Y[L] (N ) are hyperspherical side of (10) is discussed, for example, in Refs. [14,15]. In

054003-2
ACCURACY OF GROUND-STATE SOLUTIONS OF BOSON … PHYSICAL REVIEW C 105, 054003 (2022)

terms of the new variable z = cos 2ϕN , the integration leads account by introducing the hypercentral potential multipole
to   +1
 V (ri j )dN V (r, z)(1 − z)α (1 + z)β dz
A V0 (r) =  = −1  +1
α β
−1 (1 − z) (1 + z) dz
G (r, z) = N ψ̄kl (r, N ) (11) dN
kl (17)
 +1 in (14). This definition of V0 (r) utilizes the parameter α which
= F (r, z) + u (z) f (z, z ) F (r, z ) w(z ) dz , depends on the number of particles in the system. However, in
−1 weakly interacting systems, effects of two-body higher partial
(12) waves should be more significant in the interaction region
than anywhere else in the system. That is, two-body higher
where u (z) f (z, z ) is a projection kernel, u (z) = (1 + z) /2 , partial waves are felt more by the spectator particles nearest
and w(z) = (1 − z)α (1 + z)β is the weight function with α = to the interacting pair, as explained in Ref. [16]. This is to be
(3A − 8)/2 and β = + 1/2. The nonlocal kernel has the expected when only two-body correlations are dominant.
form The above argument suggests that the evaluation of (17)
∞
   with α = 21 ≡ (A = 3) and β = 21 , for any A, should ade-
f (z, z ) = Nc 0 |0 cK + Nd 0 |0 dK quately account for the effects of higher partial waves in the
K system. For this reason, the hypercentral potential is evaluated
by
Pnα,β (z)Pnα,β (z )
× , (13) 
hnα,β (−1)δA3 +1
V0 (r) = V (r, z)(1 − z2 )1/2 dz, (18)
π −1
where K = 2n + ,  0 | 0 νKL are Raynal-Revai coefficients,
Nc = 2(A − 2) with c = π /3, and Nd = 21 (A − 2)(A − 3) where δA 3 is the Kronecker delta function, for A  3, which
with d = π /2. Here Nc is the number of pairs of bosons when is independent of variation of α and free of the challenges
either k or l labels the same boson as either i or j, and Nd is highlighted in Ref. [12]. This form of hypercentral potential is
the number of pairs when {kl} and {i j} label different pairs, the zeroth multipole of the expansion of √V (r, z) in Chebyshev
i.e., {kl} = {i j}. polynomials of the first kind, Tn (z) ( = 1 − z2 Un−1 (z) ) re-
Equations (10) are now transformed by introducing the lating to (15), where Um (z) are Chebyshev polynomials of the
variable z ∈ [−1, +1] and considering (12). This leads to the second kind.
regularized two-variable integrodifferential equations Since the purpose of the hypercentral potential is to
  include effects of higher partial waves of the interaction
h̄2 4 potential, V0 (r) should be added to the two-body poten-
− T̂r + 2 T̂z − E F (r, z) = −V (r, z)G (r, z), (14)
m r tial and V (r, z) should be replaced with V (r, z) + V0 (r)
 subtract the total 2 A(A − 1) V0 (r) =
1
in (14). One then can
where the functional form of V (r, z) is known and  
i j V (ri j ) dN / dN from the left-hand side so that the
∂2 ∂ α 2 − 41 β 2 − 41 resulting equations still sum up to the Schrödinger equation.
T̂z = (1 − z2 ) − z − − (15) Note that the opposite approach is implemented in the tradi-
∂z2 ∂z 2(1 − z) 2(1 + z)
tional IDEA [17]. The procedure outlined above leads to the
is the parameterized angular kinetic energy operator. These regularized integrodifferential equations
equations are to be solved with the constraints  
h̄2 4 1
− T̂r + 2 T̂z − A(A − 1)V0 (r) − E F (r, z)
F (0, z) = F (∞, z) = 0, m r 2
F (r, −1) = F (r, +1) = 0, (16) = −[V (r, z) + V0 (r)]G (r, z). (19)
as indicated in (5). The solution to Eq. (14) is spatially Equations (19) revise the traditional few-body IDEA by in-
symmetric [13] with respect to particle permutations, which corporating angular boundary conditions explicitly as well as
is consistent with the symmetry requirement for systems of modify the incorporation of the hypercentral potential in the
bosons. Equations (14) with the boundary conditions (16), equations. Equations (19) are here called regularized IDEA
which are here called regularized s-projected IDE (rSIDE), (rIDEA).
form part of the main results of this work. These results These equations can be solved with any of the numerical
improve the traditional few-body s-projected IDE by incor- methods that are applicable the Faddeev integrodifferential
porating angular boundary conditions explicitly. In the next equations. In this paper, the Lagrange-mesh method [11,18–
subsection, the procedure of including effects of higher partial 20] is used. Illustrative applications of (14) and (19) are dis-
waves in (14) is discussed. cussed in the next section.

C. Effects of higher partial waves III. ILLUSTRATIONS

Equations (14) hold for s-projected potentials where all the The integrodifferential equations (14) were solved directly
bosons, except the interacting pair, are in the s state [4]. Ef- using the Lagrange-mesh method [11]. Ground-state energies
fects of higher partial waves of the potential are then taken into E0 for A-boson systems interacting through the central Volkov

054003-3
G. J. RAMPHO PHYSICAL REVIEW C 105, 054003 (2022)

TABLE I. Calculated ground-state energy E0 (MeV) for few- the s-wave results are identical for A = 3, but differ by about
boson systems with the central Volkov potential. 5% for A = 4. However, the rSIDE results are 1% closer to the
Kmax = 0 results. In the absence of the s-wave results for the
−E0 (MeV) A = 5 and A = 6 systems, it is tempting to compare the rSIDE
A rSIDE rIDEA Ref. results with the intermediate all waves results of Ref. [3]. The
rSIDE results differ by <1% with those of the Kmax = 6 and
3 8.430 9 8.460 2 Kmax = 8 all waves, respectively, for the A = 5 and A = 6
8.430 9 8.464 6 [3] systems.
4 28.845 29.844 In general, it observed that there are small differences
30.252 30.418 [3] between the rSIDE and the rIDEA results. This could be an
5 66.831 68.340
indication that effects of higher partial waves of short-range
66.893a 68.280 [3]
interaction potentials in few-body systems are small but sig-
6 121.662 123.180
nificant. This agrees with the findings of Ref. [5]. Also, small
121.738b 122.776 [3]
7 193.129 194.453
differences are observed between the rIDEA results and the
200.136 [12] HHE method results. These small differences in the results of
8 281.176 282.253 the two approaches can be confidently interpreted as an indi-
285.808 [12] cation that the IDEA and the HHE method are equivalent and,
9 385.809 386.649 therefore, generate similar solutions [13], and that A-body
386.640 [12] correlations (A > 3) are not significant in systems interacting
10 507.064 507.698 via short-range potentials. These claims could not be made
512.193 [12] with confidence in the past because the comparisons were
made with interpolated results from the adiabatic solutions of
a
Kmax = 6 (all waves). IDEA [5]. The results presented above are direct solutions of
b
Kmax = 8 (all waves). the rIDEA.
In Ref. [12], ground-state energies for A = 7, 8, 9, 10
nucleon-nucleon potential [21] with A ranging from 3 to 10 bosonic systems are presented for the Volkov potential. The
were calculated. The description of nuclear interactions with results are obtained by solving the integrodifferential equa-
central potentials is not realistic [3,22] because it does not take tions using the perturbation method. As explained in the
into account significant degrees of freedom of the nucleons. reference, the perturbation solution is not always reliable,
However, this potential is widely used in the literature as a but generates satisfactory solutions. On the other hand, the
test case in many methods and provides benchmark results to projection method, although exact, has convergence chal-
compare with. The Volkov potential has the form lenges. It should be noted that in both solution methods,
the angular domain is not regularized, higher potential mul-
V (ri j ) = VR e−ri j /aR + VA e−ri j /aA ,
2 2 2 2
(20) tipoles are employed, and the A-dependent weight function
w(z) is used. The results of this work are compared with
where VR = 144.86 MeV, VA = −83.34 MeV, aR = 0.82 fm, the perturbation results of Ref. [12]. It is declared in the
and aA = 1.6 fm. The nucleon mass m such that h̄2 /m = reference that the results from the projection method had
41.47 MeV fm2 was used and a maximum hyperradial range not converged. Since the present results are converged and
of rm = 30 fm was adopted. Convergence was reached quite have been shown to consistently compare very well with
rapidly with a very low number of basis size (Nr = Nz = 30) exact results of Ref. [3], they provide a test of accuracy
for all the systems considered. to the results of Ref. [12]. Whereas the energies for A = 8
The calculated ground-state energies for boson systems and A = 9 are reproduced to within 1% and 0.01%, respec-
with A = 3 to A = 10 are presented in Table I. In Table, it can tively, the energies for the A = 7 and A = 10 systems are
be notice that the rIDEA results do not differ much (≈1 MeV) overestimated by 3% and 1%, respectively. The perturbation
from the rSIDE results, in all the systems. This observation, solutions of the integrodifferential equation can be accepted as
also reported in Refs. [5,16], suggests that contributions of satisfactory.
higher partial waves to ground-state energies do not appear
to depend on the number of particles in the system when
IV. CONCLUSION
two-body correlations are dominant.
In Ref. [3], ground-state energies of boson systems with The traditional few-body integrodifferential equations ap-
A = 3, 4, 5, 6 are presented for the Volkov potential. The proach has been reviewed. Dirichlet boundary conditions
results were obtained by solving the Schrödinger equation ex- were implemented in the hyperradial and hyperangular do-
actly for the systems using the HHE method. The “s-wave” mains. Procedures for the evaluation of the hypercentral
results of this reference correspond to the rSIDE results while potential and for the inclusion of effects of higher partial
the “all waves” results correspond to the rIDEA results of this waves of the interaction potential have been revised. Results
work. The results of this work are compared with those pre- obtained with the modified integrodifferential equations ap-
sented in the reference. As can be seen, the rIDEA reproduces proach converge rapidly and reproduce those obtained with
the all waves exact results almost exactly (<1% differences) the hyperspherical harmonics approach, an exact method,
for all the A  6 systems. On the other hand, the rSIDE and to within 5% for boson systems interacting through the

054003-4
ACCURACY OF GROUND-STATE SOLUTIONS OF BOSON … PHYSICAL REVIEW C 105, 054003 (2022)

Volkov potential. These results suggest that contributions potentials, and that the IDEA is equivalent to the HHE
from effects of higher partial waves to the ground state of method. Work is under way to clarify the impact of these
few-body systems are small, that many-body correlations changes on other features of the integrodifferential equations
are not significant in systems interacting through short-range approach.

[1] A. K. Motovilov, Progress in methods to solve the Faddeev [11] G. J. Rampho, L. C. Mabunda, and M. Ramantswana, Few-body
and Yakubovsky differential equations, Few-Body Syst. 43, 121 integrodifferential equation on Lagrange mesh, J. Phys.: Conf.
(2008). Ser. 915, 012005 (2017).
[2] N. Barnea, Hyperspherical functions with arbitrary permuta- [12] T. K. Das and R. Chattopadhyay, Perturbative solution of
tional symmetry: Reverse construction, Phys. Rev. A 59, 1135 integro-differential equation for a system of identical bosons,
(1999). Fizika B 2, 73 (1993).
[3] M. Gattobigio, A. Kievsky, and M. Viviani, Nonsymmetrized [13] M. Fabre de la Ripelle, A bridge between hyperspherical and
hyperspherical harmonic basis for an A-body system, Phys. Rev. integro-differential approaches to the many-body bound states,
C 83, 024001 (2011). Few-Body Syst. 1, 181 (1986).
[4] M. Fabre de la Ripelle, H. Fiedeldey, and S. A. Sofianos, Inte- [14] M. Fabre de la Ripelle, The potential harmonic expansion
grodifferential equation for few- and many-body systems, Phys. method, Ann. Phys. 147, 281 (1983).
Rev. C 38, 449 (1988). [15] R. M. Adam and H. Fiedildey, 5λ He and 6 He calculations by
[5] M. Fabre de la Ripelle, H. Fiedeldey, and S. A. Sofianos, means of the integrodifferential equation approach, J. Phys. G:
Several versions of the integro-differential equation approach Nucl. Part. Phys. 19, 703 (1993).
to bound systems, Few-Body Syst. 6, 157 (1989). [16] M. Fabre de la Ripelle, Contributions of l = 0 two-body orbitals
[6] S. A. Sofianos, T. K. Das, B. Chakrabarti, M. L. Lekala, R. M. and three-body correlations in few-body systems, Phys. Rev. C
Adam, and G. J. Rampho, Beyond mean-field ground-state 42, R823 (1990).
energies and correlation properties of a trapped Bose-Einstein [17] W. Oehm, S. A. Sofianos, H. Fiedeldey, and M. Fabre de la
condensate, Phys. Rev. A 87, 013608 (2013). Ripelle, Integrodifferential equation approach. I. Triton and α-
[7] S. A. Sofianos, G. J. Rampho, and R. M. Adam, Two- particle binding energies, Phys. Rev. C 42, 2322 (1990).
dimensional integrodifferential equations for unequal mass [18] D. Baye, The Lagrange-mesh method, Phys. Rep. 565, 1 (2015).
particle systems, Phys. Part. Nucl. 40, 757 (2009). [19] G. J. Rampho, Perspective on the Lagrange-Jacobi mesh,
[8] M. L. Lekala, G. J. Rampho, S. A. Sofianos, and R. M. Adam, J. Phys. A: Math. Theor. 49, 295202 (2016).
An integrodifferential equation for Bose-Einstein condensates, [20] G. J. Rampho, The Schrödinger equation on a Lagrange mesh,
Few-Body Syst. 50, 427 (2011). J. Phys.: Conf. Ser. 905, 012037 (2017).
[9] M. L. Lekala, G. J. Rampho, R. M. Adam, S. A. Sofianos, and [21] A. B. Volkov, Equilibrium deformation calculations of the
V. B. Belyaev, The double-λ hypernucleus 11  Be, Phys. Atom. ground state energies of 1p shell nuclei, Nucl. Phys. 74, 33
Nucl. 77, 472 (2014). (1965).
[10] O. Sørensen, D. V. Fedorov, and A. S. Jensen, Two-body cor- [22] R. Brizzi, M. Fabre de la Ripelle, and M. Lassaut, Model cal-
relations in n-body boson systems, Phys. Rev. A 66, 032507 culations of doubly closed-shell nuclei in the integrodifferential
(2002). equation approach, Nucl. Phys. A 596, 199 (1996).

054003-5