betaFeedback

WorksPhysics and Astronomy

Scattering of charged particles within Efros method utilizing oscillator series expansion of wave functions

Abstract

We apply the version of the Efros method utilizing oscillator expansion of wave functions to the Coulomb scattering problem using our recent developments of the HORSE formalism. The approach yields accurate phase shifts and cross sections with significantly reduced computational cost compared to the full HORSE method, while maintaining agreement with exact solutions. These results demonstrate the efficiency of the Efros method and its prospect for applications in ab initio nuclear reaction calculations.

Introduction

Static properties of light atomic nuclei are currently calculated within various ab initio approaches which do not use any model assumptions about nuclear structure. It is especially worth noting No-Core Shell Model (NCSM){sup}1` as one of the most advanced and promising ab initio methods which utilizes the harmonic oscillator basis.

The nuclear continuum spectrum has been successfully studied so far within different approaches, e. g., RR-matrix method{sup}2 and $J$-matrix method.{sup}3,4 The formalism of Harmonic Oscillator Representation of Scattering Equations (HORSE){sup}5–8` is a particular version of JJ-matrix method utilizing harmonic oscillator basis which is preferable for many-body nuclear applications. The HORSE is the discrete analogue of PP-matrix which is the inverse of RR-matrix.[8]

Various versions of the HORSE method have been successfully applied to the studies of nuclear resonant states,[9, 10] photodisintegration,[11] three-body continuum within phenomenological cluster models[12, 13] and various problems within the Resonating Group Model (RGM).[14, 15, 16] However, accounting for the Coulomb interaction between charged particles within HORSE encounters difficulties due to the long-range nature of the Coulomb potential. Several methods have been suggested to account for Coulomb asymptotics within the oscillator expansion of wave functions,[4, 8, 17] however, all of them have significant difficulties in combining with many-body ab initio approaches like the NCSM for studying the nuclear continuum. Recently we have suggested[18] the method for studying scattering of charged particles based on analysis of Coulomb matrix elements in oscillator basis performed in Ref. 17, which is simple and convenient to use within the HORSE formalism and is prospective for many-body applications in combination with NCSM.

Unfortunately, one cannot use the HORSE in combination with NCSM for the studies of nuclear reactions due to a prohibitive computational cost as it requires calculating extremely large number of eigenstates, even for light nuclei reactions. The method proposed recently by V. D. Efros[19] is based on variational Hulthén–Kohn method[20] and can be used for ab initio calculations of nuclear reactions in combination with NCSM. Recently a modified version of the Efros method has been suggested[21] which utilizes the oscillator expansion of non-Coulomb scattering wave functions developed within the HOSRE formalism and is well-adapted for the use in combination with NCSM in many-body applications. If one uses the full basis of eigenstates of the truncated Hamiltonian matrix in the oscillator basis as short-range functions (SRFs) within the modified Efros method, it appears to be equivalent to the HORSE, however, it allows accounting for a small number of SRFs to obtain adequate results. Using a small number of SRFs not only simplifies calculations significantly but just opens possibilities for studying nuclear reactions within the framework of ab initio approaches.

In this contribution, we combine our method for accounting for the Coulomb distortion of wave function asymptotics proposed in Ref. 18 with the modified Efros method[19] for the studies of charged particles scattering. Using model problems, we demonstrate that this approach provides quite accurate results even for small number of SRFs and basis truncations accessible by modern NCSM codes.

Coulomb scattering within HORSE formalism

In this section, we sketch the key expressions obtained within HORSE formalism for Coulomb scattering following the approach described in Ref. 18. Consider a single-channel scattering of two particles with charges eZ1eZ_1 and eZ2eZ_2. We use the conventional partial wave expansion of the wave function and describe the system in terms of partial amplitudes ul(k,r)u_l(k,r) each of which obeys the radial Schrödinger equation with the Hamiltonian HlH^l,

Hlul(k,r)=Eul(k,r).(1)H^l u_l(k,r) = E u_l(k,r). \tag*{(1)}

Here, ll is the orbital quantum number, rr is the distance between the particles, k=2μE/ℏk = \sqrt{2\mu E}/\hbar is the momentum, μ\mu is the reduced mass, and EE is the relative motion energy. The interaction potential between the particles is represented as Vl=VNucl,l+VCoulV^l = V^{Nucl,l} + V^{Coul}, where VNucl,lV^{Nucl,l} and VCoulV^{Coul} are nuclear and Coulomb potentials, respectively. The asymptotic behavior of partial amplitudes ul(k,r)u_l(k,r) is described by the superposition of regular Fl(η,kr)F_l(\eta,kr) and irregular Gl(η,kr)G_l(\eta,kr) Coulomb wave functions,[22]

ul(k,r)⟶r→∞Fl(η,kr)+tan⁡δl(k)Gl(η,kr),(2)u_l(k,r) \underset{r\to\infty}{\longrightarrow} F_l(\eta,kr) + \tan\delta_l(k)G_l(\eta,kr), \tag*{(2)}

where η=μZ1Z2e2/ℏ2k\eta= \mu Z_1 Z_2 e^2/\hbar^2 k is the Sommerfeld parameter and δl(k)\delta_l(k) is the phase shift. Within the HORSE formalism, we expand the partial amplitudes ul(k,r)u_l(k,r) into a series of harmonic oscillator functions φnl(r)\varphi_{nl}(r),

ul(k,r)=∑n=0∞anl(k) φnl(r),(3)u_l(k,r) = \sum_{n=0}^{\infty} a_{nl}(k)\,\varphi_{nl}(r), \tag*{(3)}

where

φnl(r)=(−1)n2n!bΓ(n+l+32)(rb)l+1e−r22b2Lnl+12(r2b2).(4)\varphi_{nl}(r) = (-1)^n \sqrt{\frac{2n!}{b\Gamma(n+l+\frac{3}{2})}}\left(\frac{r}{b}\right)^{l+1} e^{-\frac{r^2}{2b^2}} L_n^{l+\frac{1}{2}}\left(\frac{r^2}{b^2}\right). \tag*{(4)}

Here, Γ(x)\Gamma(x) is the gamma function,[22] Lnα(x)L_n^\alpha(x) is the associated Laguerre polynomial,[22] b=ℏ/μωb = \sqrt{\hbar/\mu\omega} is the oscillator radius, and ω\omega is the oscillator frequency. The expansion coefficients anl(k)a_{nl}(k) satisfy an infinite system of linear equations,

∑n′=0∞(Hnn′l−δnn′E)an′l(k)=0,(5)\sum_{n'=0}^{\infty}\left(H^l_{nn'}-\delta_{nn'}E\right)a_{n'l}(k)=0, \tag*{(5)}

where Hnn′l=Tnn′l+Vnn′lH^l_{nn'} = T^l_{nn'} + V^l_{nn'} are the Hamiltonian matrix elements in the harmonic oscillator basis with Tnn′lT^l_{nn'} and Vnn′lV^l_{nn'} being the matrix elements of the kinetic TlT^l and potential VlV^l energies, respectively.

In the harmonic oscillator basis, the kinetic energy matrix Tnn′lT^l_{nn'} is tridiagonal,

Tnn′l=0,∣n−n′∣>1,Tnnl=ℏω2(2n+l+32),Tn+1,nl=Tn,n+1l=−ℏω2(n+1)(n+l+32).(6)\begin{aligned} T^l_{nn'} &= 0,\quad|n-n'|>1,\\ T^l_{nn} &= \frac{\hbar\omega}{2}\left(2n+l+\frac{3}{2}\right),\\ T^l_{n+1,n} &= T^l_{n,n+1} = -\frac{\hbar\omega}{2}\sqrt{(n+1)\left(n+l+\frac{3}{2}\right)}. \tag*{(6)} \end{aligned}

with its non-zero elements TnnlT^l_{nn} and Tn,n±1lT^l_{n,n\pm1} increasing linearly with nn for large values of nn. At the same time, the matrix elements of nuclear potential Vnn′Nucl,l→0V^{Nucl,l}_{nn'} \to0 as $n and/or n′→∞n' \to\infty. This allows us to replace VNucl,lV^{Nucl,l} by the potential V~Nucl,l\widetilde{V}^{Nucl,l} defined by its matrix in harmonic oscillator basis truncated at some N,

V~nn′Nucl,l={Vnn′Nucl,lfor n and n′≤N,0for n or n′>N.(7)\widetilde{V}^{Nucl,l}_{nn'} = \begin{cases} V^{Nucl,l}_{nn'} & \text{for } n \text{ and } n' \le N,\\ 0 & \text{for } n \text{ or } n' > N. \end{cases} \tag*{(7)}

The matrix elements of Coulomb potential Vnn′Coul,lV^{Coul,l}_{nn'}, however, decrease slowly along the main diagonal[17] and should be accounted for at large values of radial quantum numbers nn. In the Kiev group study,[17] it was stated that for large values of nn, the asymptotic coefficients anlas(k)≡anl(k)a^{as}_{nl}(k) \equiv a_{nl}(k) (n≥Nn \ge N) in the expansion (3) fit the three-term recurrence relation (TRR)

Tn,n−1lan−1,las(k)+(Tnnl+Vnnad,l−E)anlas(k)+Tn,n+1lan+1,las(k)=0,(8)T^{l}_{n,n-1}a^{as}_{n-1,l}(k) + \left(T^{l}_{nn} + V^{ad,l}_{nn} - E\right)a^{as}_{nl}(k) + T^{l}_{n,n+1}a^{as}_{n+1,l}(k) = 0, \tag*{(8)}

where the additional Coulomb term

Vnnad,l=ℏωηkb4n+2l+3.(9)V^{ad,l}_{nn} = \hbar\omega\frac{\eta kb}{\sqrt{4n+2l+3}}. \tag*{(9)}

The calculations, however, demonstrate that TRR (8) remains highly accurate even for small n∼0n \sim0.[18] As it was shown in Ref. 18, this fact allows one to replace VCoulV^{Coul} and TlT^l by the operators V~Coul,l\widetilde{V}^{Coul,l} and T~l\widetilde{T}^l, respectively, defined by their matrices in oscillator basis,

V~nn′Coul,l={Vnn′Coul,lfor n and n′≤N,0for n or n′>N.(10)\widetilde{V}^{Coul,l}_{nn'} = \begin{cases} V^{Coul,l}_{nn'} & \text{for } n \text{ and } n' \le N,\\ 0 & \text{for } n \text{ or } n' > N. \end{cases} \tag*{(10)}
T~nn′l={Tnn′lfor n and n′≤N,Tnn′l+Vnnad,lδnn′for n or n′>N.(11)\widetilde{T}^{l}_{nn'} = \begin{cases} T^{l}_{nn'} & \text{for } n \text{ and } n' \le N,\\ T^{l}_{nn'} + V^{ad,l}_{nn}\delta_{nn'} & \text{for } n \text{ or } n' > N. \end{cases} \tag*{(11)}

The conventional HORSE formulas are then applied to the problem with Hamiltonian H~l=V~Nucl,l+V~Coul,l+T~l\widetilde{H}^{l} = \widetilde{V}^{Nucl,l} + \widetilde{V}^{Coul,l} + \widetilde{T}^{l} to obtain the phase shift δl(k)\delta_l(k) (see Ref. 18 for details).

The asymptotic coefficients anlas(k)a^{as}_{nl}(k) can be expressed as a superposition of two linearly independent solutions of TRR (8), Snl(k)S_{nl}(k) and Cnl(k)C_{nl}(k),

anlas(k)=Snl(k)+tan⁡δl(k)Cnl(k).(12)a^{as}_{nl}(k) = S_{nl}(k) + \tan\delta_l(k) C_{nl}(k). \tag*{(12)}

To satisfy the asymptotics (2) of functions ul(k,r)u_l(k,r), we should define the solutions Snl(k)S_{nl}(k) and Cnl(k)C_{nl}(k) so that

∑n=0∞Snl(k) φnl(r)=Fl(η,kr),(13)\sum_{n=0}^{\infty} S_{nl}(k)\,\varphi_{nl}(r) = F_l(\eta,kr), \tag*{(13)}
∑n=0∞Cnl(k) φnl(r)=G~l(η,kr)→r→∞Gl(η,kr),(14)\sum_{n=0}^{\infty} C_{nl}(k)\,\varphi_{nl}(r) = \widetilde{G}_l(\eta,kr) \xrightarrow{r\to\infty} G_l(\eta,kr), \tag*{(14)}

where G~l(η,kr)\widetilde{G}_l(\eta,kr) is some regular function which fits an inhomogeneous Schrödinger equation.[8, 23]

At large nn, harmonic oscillator functions φnl(r)\varphi_{nl}(r) have a delta-like behaviour in the vicinity of classical turning point rturn=νnlbr_{turn} = \nu_{nl}b, where νnl=4n+2l+3\nu_{nl} = \sqrt{4n+2l+3}.[23] Using this property, we obtain the asymptotic expressions for the coefficients Snl(k)S_{nl}(k) and Cnl(k)C_{nl}(k),[18]

Snl(k)⟶n→∞2bνnlFl(η,νkb),(15)S_{nl}(k) \underset{n\to\infty}{\longrightarrow} \sqrt{\frac{2b}{\nu_{nl}}}F_l(\eta,\nu kb), \tag*{(15)}
Cnl(k)⟶n→∞2bνnlGl(η,νkb).(16)C_{nl}(k) \underset{n\to\infty}{\longrightarrow} \sqrt{\frac{2b}{\nu_{nl}}}G_l(\eta,\nu kb). \tag*{(16)}

The high accuracy of TRR (8) allows us to reproduce the coefficients Snl(k)S_{nl}(k) and Cnl(k)C_{nl}(k).[18] For some starting value nsn_s, the coefficients Sns+2,l(k)S_{n_s+2,l}(k) and Sns+1,l(k)S_{n_s+1,l}(k) are obtained from the asymptotic formulas (13a). The coefficients Snl(k)S_{nl}(k) (n=0,…,nsn=0,\ldots,n_s) are then calculated according to TRR (8). The same procedure is used to obtain the coefficients Cnl(k)C_{nl}(k) (n=0,…,nsn=0,\ldots,n_s).

Efros method

The approach to finding solutions Snl(k)S_{nl}(k) and Cnl(k)C_{nl}(k) of TRR (8) described in § 2 allows us to apply Efros method[19, 21] to Coulomb scattering problem. Suppose that solution of the Schrödinger equation (1) has the following form,

u~l(k,r)=Fl(η,kr)+tan⁡δl(k)G~l(η,kr)+∑n=0v−1bnl(k)βnl(r),(17)\widetilde{u}_l(k,r)=F_l(\eta,kr)+\tan\delta_l(k)\widetilde{G}_l(\eta,kr)+\sum_{n=0}^{v-1}b_{nl}(k)\beta_{nl}(r), \tag*{(17)}

where βnl(r)\beta_{nl}(r) (n=0,…,v−1n=0,\ldots,v-1) is the set of vv linearly independent SRFs. The equation (17) includes (v+1)(v+1) unknowns: vv coefficients bnl(k)b_{nl}(k) and tan⁡δl(k)\tan\delta_l(k). Introduce the vector cc of unknowns with its components

cn={bnl(k)for n=0,1,…,v−1,tan⁡δl(k)for n=v.(18)c_n=\begin{cases} b_{nl}(k) & \text{for } n=0,1,\ldots,v-1,\\ \tan\delta_l(k) & \text{for } n=v. \end{cases} \tag*{(18)}

To find the unknowns cnc_n, consider (v+1)(v+1) equations

⟨βˉnl∣H~l−E∣u~l⟩=0,n=0,…,v.(19)\langle\bar{\beta}_{nl}|\widetilde{H}_l-E|\widetilde{u}_l\rangle=0,\quad n=0,\ldots,v. \tag*{(19)}

Here βˉnl(r)\bar{\beta}_{nl}(r) comprise a set of (v+1)(v+1) linearly independent SRFs. Following Ref. 21, we choose the first vv functions as βˉnl(r)=βnl(r)\bar{\beta}_{nl}(r)=\beta_{nl}(r) (n=0,…,v−1n=0,\ldots,v-1). The last function βˉvl(r)\bar{\beta}_{vl}(r) is arbitrarily chosen so that it satisfies the requirement of linear independence with the rest βˉnl(r)\bar{\beta}_{nl}(r) with n≤v−1n\leq v-1. Equation (19) can be written in matrix form,

Ac=B,(20)Ac=B, \tag*{(20)}

where matrix AA and vector BB have the following matrix elements and components,

Ann′=⟨βˉnl∣H~l−E∣βn′l⟩,{n=0,…,v,n′=0,…,v−1,(21)A_{nn'}=\langle\bar{\beta}_{nl}|\widetilde{H}_l-E|\beta_{n'l}\rangle,\qquad \begin{cases} n=0,\ldots,v,\\ n'=0,\ldots,v-1, \end{cases} \tag*{(21)}
Anv=⟨βˉnl∣H~l−E∣G~l⟩,n=0,…,v;(22)A_{nv}=\langle\bar{\beta}_{nl}|\widetilde{H}_l-E|\widetilde{G}_l\rangle,\quad n=0,\ldots,v; \tag*{(22)}
Bn=−⟨βˉnl∣H~l−E∣Fl⟩,n=0,…,v.(23)B_n=-\langle\bar{\beta}_{nl}|\widetilde{H}^{l}-E|F_l\rangle,\quad n=0,\ldots,v. \tag*{(23)}

Solving the matrix equation (20), we obtain tan⁡δl(k)\tan\delta_l(k) and unknown coefficients bnl(k)b_{nl}(k).

As for the choice of SRFs βnl(r)\beta_{nl}(r), it is natural to use either the set of vv oscillator functions φn′l(r)\varphi_{n'l}(r) (n′≤Nn'\leq N), or the set of vv eigenfunctions γλ(r)\gamma_\lambda(r) of truncated Hamiltonian H~tr,l\widetilde{H}^{tr,l}.[21] Here, the Hamiltonian H~tr,l\widetilde{H}^{tr,l} is defined by its (N+1)×(N+1)(N+1)\times(N+1) matrix H~nn′tr,l=H~nn′l\widetilde{H}^{tr,l}_{nn'}=\widetilde{H}^{l}_{nn'} (n,n′=0…,Nn,n'=0\ldots,N) and eigenfunctions γλ(r)\gamma_\lambda(r) satisfy the following equations,

H~tr,l∣γλ⟩=Eλ∣γλ⟩,λ=0,…,N,(24)\widetilde{H}^{tr,l}|\gamma_\lambda\rangle=E_\lambda|\gamma_\lambda\rangle,\quad\lambda=0,\ldots,N, \tag*{(24)}
⟨γλ∣γλ′⟩=δλλ′,λ,λ′=0,…,N,(25)\langle\gamma_\lambda|\gamma_{\lambda'}\rangle=\delta_{\lambda\lambda'},\quad\lambda,\lambda'=0,\ldots,N, \tag*{(25)}

where EλE_\lambda are the eigenenergies corresponding to eigenfunctions γλ(r)\gamma_\lambda(r). The eigenfunctions are convenient to be ordered so that lower indices λ\lambda correspond to lower eigenenergies EλE_\lambda, i. e., Eλ≤Eλ′E_\lambda\leq E_{\lambda'} if λ<λ′\lambda<\lambda'. The set of eigenfunctions is especially convenient for many-body applications in combination of the Efros method with NCSM since otherwise one needs to construct the SRFs βnl(r)\beta_{nl}(r) as non-excited center-of-mass superpositions with the definite value of the total angular momentum JJ of the NCSM basis functions which are Slater determinants that are not characterized by a definite value of the total angular momentum JJ and may have contributions of the center-of-mass excitations. However, generally, one can define the set of SRFs βnl(r)\beta_{nl}(r) as any set of vv linearly independent superpositions of short-range functions. As for the function βˉvl(r)\bar{\beta}_{vl}(r), it was shown in Ref. 21 that it is reasonable to define it as the first oscillator function outside the truncation boundary NN, i. e., βˉvl(r)=φN+1,l(r)\bar{\beta}_{vl}(r)=\varphi_{N+1,l}(r). Such a choice is made in order to provide a necessary coupling of the interaction (nn and n′≤Nn'\leq N) and the continuum (nn or n′>Nn'>N) regions of the Hamiltonian matrix H~nn′l\widetilde{H}^{l}_{nn'}.

It is noteworthy that the modified Efros method of Ref. 21 is equivalent to the HORSE approach if v=N+1v=N+1. However, the advantage of the Efros method is that one can obtain adequate results which converge to numerical solution if v<N+1v<N+1 and in many cases even if v≪N+1v\ll N+1. Besides, matrix elements Ann′A_{nn'} and vector components BnB_n are readily obtained if one uses expressions (13) for the Coulomb functions Fl(η,kr)F_l(\eta,kr) and Gl(η,kr)G_l(\eta,kr) with coefficients Snl(k)S_{nl}(k) and Cnl(k)C_{nl}(k) that are easily numerically calculated as discussed in § 2. Using φn′l(r)\varphi_{n'l}(r) or γλ(r)\gamma_\lambda(r) as SRFs βˉnl(r)\bar{\beta}_{nl}(r), one also obtains finite sums in the right-hand sides of expressions (19) and (23). Thus, Efros method provides a simple and numerically efficient approach to finding phase shifts δl(k)\delta_l(k) and wave functions in scattering of charged particles.

Results

We use the Woods–Saxon potential as the nuclear potential,

VNuc,l=V01+exp⁡(r−R0α0)+(1⋅s)1rddrVls1+exp⁡(r−R1α1),(26)V^{Nuc,l}=\frac{V_0}{1+\exp\left(\frac{r-R_0}{\alpha_0}\right)}+(1\cdot s)\frac{1}{r}\frac{d}{dr}\frac{V_{ls}}{1+\exp\left(\frac{r-R_1}{\alpha_1}\right)}, \tag*{(26)}

where l\mathbf{l} and s\mathbf{s} are the orbital momentum and the spin, respectively. To improve the convergence, as it was shown in Ref. 18, it is important to use smoothing[24] of the truncated matrix of the nuclear interaction (7), that is V~nn′Nucl,l\widetilde{V}^{\mathrm{Nucl},l}_{nn'} is replaced by

V~nn′Nucl,l={σnVnn′Nucl,lσn′,for n and n′≤N,0,for n or n′>N,(27)\widetilde{V}^{\mathrm{Nucl},l}_{nn'} = \begin{cases} \sigma_n V^{\mathrm{Nucl},l}_{nn'}\sigma_{n'}, & \text{for } n \text{ and } n' \le N,\\ 0, & \text{for } n \text{ or } n' > N, \end{cases} \tag*{(27)}

where

σn=1−exp⁡(−[α(n−N−1)N+1]2)1−exp⁡(−α2).(28)\sigma_n = \frac{1-\exp\left(-\left[\frac{\alpha(n-N-1)}{N+1}\right]^2\right)}{1-\exp(-\alpha^2)}. \tag*{(28)}

In what follows, the smoothing parameter α=5\alpha= 5 is used. Note, the Coulomb interaction V~nn′Coul,l\widetilde{V}^{\mathrm{Coul},l}_{nn'} [see Eq. (10)] should not be smoothed. Following the Efros formalism, we calculate the phase shifts δl(k)\delta_l(k) and corresponding partial elastic scattering cross sections

σl(k)=4πk2(2l+1)sin⁡2δl(k),(29)\sigma_l(k) = \frac{4\pi}{k^2}(2l+1)\sin^2\delta_l(k), \tag*{(29)}

and compare the results with those obtained within HORSE method for Coulomb scattering discussed in Ref. 18 as well as with the results found by numerical integration of Schrödinger equation using Numerov method which will be referred to as exact.

Let us consider the pp–α\alpha scattering phase shifts in the p3/2p_{3/2} partial wave. We use the Woods–Saxon potential parameterization suggested in Ref. 25 with V0=−43.0V_0=-43.0 MeV, R0=2.0R_0=2.0 fm, α0=0.70\alpha_0=0.70 fm, Vls=40.0V_{ls}=40.0 MeV ⋅\cdot fm2^2, R1=1.50R_1=1.50 fm, and α1=0.35\alpha_1=0.35 fm. The calculations of phase shifts were performed for different numbers of SRFs vv and various sets of SRFs βnl(r)\beta_{nl}(r) and β‾nl(r)\overline{\beta}_{nl}(r) (note, we always choose the first vv functions β‾nl(r)\overline{\beta}_{nl}(r) the same as βnl(r)\beta_{nl}(r): β‾nl(r)=βnl(r)\overline{\beta}_{nl}(r)=\beta_{nl}(r) for n=0,…,v−1n=0,\ldots,v-1) at oscillator energy ℏω=27\hbar\omega=27 MeV which yields the best results. We started with choosing the set of SRFs β‾nl(r)\overline{\beta}_{nl}(r) as

β‾nlEig(r)={γn(r),n=0,…,v−1,φN+1,l(r),n=v.(30)\overline{\beta}^{\mathrm{Eig}}_{nl}(r) = \begin{cases} \gamma_n(r), & n=0,\ldots,v-1,\\ \varphi_{N+1,l}(r), & n=v. \end{cases} \tag*{(30)}

We note that the same set of SRFs β‾nl(r)\overline{\beta}_{nl}(r) was used in Ref. 21 for the scattering of non-charged particles. The results are shown in Figure 1. One can see that the Efros method provides adequate phase shifts which converge to their exact values even for small v=4v=4. It seems, however, that choosing v=2v=2 leads to unsatisfying results for E>15E>15 MeV. This is not surprising since the second (first excited) eigenstate in this case has an energy of 10.25 MeV; hence one cannot expect an adequate description of the phase shifts at higher energies using only 2 SRFs and the phase shifts start deviating from the exact ones starting from approximately 11 MeV. Nevertheless, the elastic cross section σl(k)\sigma_l(k), which, contrary to the phase shifts, is an observable, does not differ much from exact values since sin⁡2δl(k)∼1\sin^2\delta_l(k)\sim1 in this region of energies [see Eq. (29)]. The difference from the exact solution can be actually expressed numerically by calculating the root-mean-square (RMS) deviation sσs_{\sigma} of the calculated cross section from the exact one which appears to be quite small for all choices for vv (see the inset in Figure 1).

Figure 1. Phase shift δl(k)\delta_l(k) and elastic p−αp-\alpha scattering cross section σl(k)\sigma_l(k) dependences on relative motion energy EE in the p3/2p_{3/2} partial wave obtained by different methods. Solid line: numerical integration of Schrödinger equation by Numerov method; crosses: HORSE method;[18] other lines: Efros method with different vv for SRFs βˉnl(r)=βˉnlEig(r)\bar{\beta}_{nl}(r) = \bar{\beta}^{Eig}_{nl}(r). For each method, sσs_{\sigma} is the cross-section RMS deviation from the exact values. The results were obtained for truncation boundary N=10N = 10, ℏω=27\hbar\omega= 27 MeV and starting number ns=200n_s = 200 for calculating Snl(k)S_{nl}(k) and Cnl(k)C_{nl}(k) using Eq. (8).

One can also use the first (v+1)(v + 1) oscillator functions φnl(r)\varphi_{nl}(r) as the set of SRFs βˉnl(r)\bar{\beta}_{nl}(r), i. e.,

βˉnlOsc(r)=φnl(r),n=0,…,v.(31)\bar{\beta}^{Osc}_{nl}(r) = \varphi_{nl}(r), \quad n = 0,\ldots,v. \tag*{(31)}

The results obtained in this case (see Figure 2) demonstrate also a good convergence. It is interesting that the use of the oscillator functions as SRFs provide closer results to the exact ones (smaller RMS deviations) in the case of small vv but the eigenfunction SRFs describe the exact cross sections better in the case of larger vv values.

Charged-particle scattering phase shifts and cross sections for different values of $v$

Figure 2. Same as Figure 1 but for SRFs βˉnl(r)=βˉnlOsc(r)\bar{\beta}_{nl}(r) = \bar{\beta}^{\mathrm{Osc}}_{nl}(r).

One can observe convergence of Efros method results with βˉnl(r)=βˉnlEig(r)\bar{\beta}_{nl}(r) = \bar{\beta}^{Eig}_{nl}(r) to exact solution for ℏω\hbar\omega in the wide range from ∼25\sim25 MeV to ∼50\sim50 MeV if v≳4v \gtrsim4 (see Figure 3). For small v∼2v \sim2, however, the method becomes more sensitive to the ℏω\hbar\omega variation.

Charged-particle scattering phase shifts for different values of $\hbar\omega$

Figure 3. Same as Figure 1 but the results of the Efros method are obtained for v=6v = 6 with different ℏω\hbar\omega values.

It is useful to study other types of scattering with broader or narrower resonances. For example, one can consider the p1/2p_{1/2} partial wave in pp–α\alpha scattering with a wider resonance (see Figure 4). In this case, the Efros method provides a satisfactory convergence to the exact values if v≥4v \ge4. Note, with N=10N = 10 and ℏω=22\hbar\omega= 22 MeV, the second state in this case has an eigenenergy of 10.99 MeV, and in the case v=2v = 2, the phase shifts and cross section start deviating at this energy from the exact results.

Figure 4. Same as Figure 1 but for p1/2p_{1/2} wave and ℏω=22\hbar\omega= 22 MeV.

To analyze accuracy of Efros method in case of different charges and a narrower resonance, we examined the s1/2s_{1/2} wave in pp–15N^{15}\mathrm{N} scattering using the parameterization of the Woods–Saxon potential as in Ref. 8, i. e., V0=−55.91V_0 = -55.91 MeV, R0=3.083R_0 = 3.083 fm, α0=0.53\alpha_0 = 0.53 fm, Vls=0.9V_{ls} = 0.9 MeV·fm2^2, R1=3.083R_1 = 3.083 fm, and α1=0.53\alpha_1 = 0.53 fm (see Figure 5). Though the phase shift seems to converge well to exact values for v≥4v \ge4, in the region of small energies, the resonance position and its width value are sensible to the choice of vv. It is interesting that using a small number of SRFs v=2v = 2, the position of the resonance is closer to its numerical value than in the case of v=4v = 4, however, the phase shifts obtained with v=2v = 2 start deviating from the exact ones at E≥5E \ge5 MeV. To improve the description of this resonance, one needs to increase the truncation boundary of the potential energy matrix NN.

Figure 5. Same as Fig. 1 but obtained in the s1/2s_{1/2} wave for pp–15N^{15}\mathrm{N} scattering with ℏω=18\hbar\omega= 18 MeV.

Conclusion

In this paper, we developed the description of scattering of charged particles within the version of the Efros method utilizing the expansion of the scattering wave functions in oscillator functions. Using model problems which can be exactly solved numerically by other methods, we demonstrated that the proposed approach provides accurate results for scattering phase shifts and cross sections even when only a relatively small number of short-range functions is used. As a result, we obtained a promising tool for the ab initio description of reactions with charged particles based on the NCSM calculations.

Acknowledgements

This work is supported by the Russian Science Foundation under the grant No 24-22-00276.

ORCID

Ustin M. Yanikov - https://orcid.org/0009-0006-0532-7696

Vasily A. Kulikov - https://orcid.org/0000-0002-2770-3465

Andrey M. Shirokov - https://orcid.org/0000-0002-0331-6209

References

  1. [4]H. A. Yamani, A. D. Alhaidari and M. S. Abdelmonem, Phys. Rev. A 64 (Sep 2001) 042703.
  2. [8]J. M. Bang, A. I. Mazur, A. M. Shirokov, Y. F. Smirnov and S. A. Zaytsev, Ann. Phys. (NY) 280 (1999) 299.
  3. [9]A. M. Shirokov, A. I. Mazur, I. A. Mazur and J. P. Vary, Phys. Rev. C 94 (2016) 064320.
  4. [10]A. M. Shirokov, A. I. Mazur, I. A. Mazur, E. A. Mazur, I. J. Shin, Y. Kim, L. D. Blokhintsev and J. P. Vary, Phys. Rev. C 98 (2018) 044624.
  5. [11]W. Du, S. Pal, M. Sharaf, P. Yin, S. Sarker, A. M. Shirokov and J. P. Vary, Phys. Rev. C 106 (2022) 054608.
  6. [12]T. Y. Mikhelashvili, Y. F. Smirnov and A. M. Shirokov, J. Phys. G 16 (2018) 1241.
  7. [13]Y. A. Lurie and A. M. Shirokov, Ann. Phys. (NY) 312 (2004) 284.
  8. [14]V. Vasilevsky, A. V. Nesterov, F. Arickx and J. Broeckhove, Phys. Rev. C 63 (2001) 034607.
  9. [15]G. F. Filippov, Y. A. Lashko, S. V. Korennov and K. Katō, Few-Body Syst. 34 (2004) 209.
  10. [16]Y. A. Lashko, G. F. Filippov and V. S. Vasilevsky, Nucl. Phys. A 958 (2017) 78.
  11. [17]I. P. Okhrimenko, Nucl. Phys. A 424 (1984) 121.
  12. [18]U. M. Yanikov, V. A. Kulikov and A. M. Shirokov, Journal of Subatomic Particles and Cosmology 4 (2025) 100075.
  13. [19]V. D. Efros, Phys. Rev. C 99 (2019) 034620.
  14. [20]H. E. Moses, Phys. Rev. 92 (1953) 817.
  15. [21]M. A. Sharaf, A. M. Shirokov, W. Du and J. P. Vary, Phys. Rev. C 112 (Aug 2025) 024617.
  16. [22]M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables (U. S. Government Printing Office, 1964).
  17. [23]S. A. Zaitsev, Y. F. Smirnov and A. M. Shirokov, Theor. Math. Phys. 117 (1998) 1291.
  18. [24]J. J. Révai, M. Sotona and J. Žofka, J. Phys. G 11 (1985) 745.

Paper details

vlnc:W_96wd3cph
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents