betaFeedback

WorksPhysics and Astronomy

Barrier penetration in a discrete basis formalism

Abstract

A standard way to solve a Schrödinger equation is to discreteize the radial coordinates and apply a numerical method for a differential equation, such as the Runge-Kutta method or the Numerov method. Here I employ a discrete basis formalism based on a finite mesh method as a simpler alternative, with which the numerical computation can be easily implemented by ordinary linear algebra operations. I compare the numerical convergence of the Numerov integration method to the finite mesh method for calculating penetrabilities of a one-dimensional potential barrier, and show that the latter approach has better convergence properties.

Introduction

In most of physics problems, a Schrödinger equation cannot be solved analytically but has to be solved numerically. For a bound state problem, one may expand wave functions on some finite basis and diagonalize the resultant Hamiltonian matrix. Alternatively, one may discretize the radial coordinates and successively obtain a wave function at the mesh points with e.g., the Runge-Kutta method or the Numerov method [1].

A yet different method, referred to as a discrete-basis formalism1, has been proposed in Ref. [2]. In this method, one first forms a Hamiltonian matrix based on discretized radial meshes and solve it with a linear algebra with appropriate boundary conditions. An advantage of this method is that the method is well compatible with a many-body Hamiltonian, in particular in a configuration-interaction formulation [6, 7, 8]. Notice that the discrete-basis formalism is referred to as a 3D mesh method in the context of nuclear density functional theory[9, 10, 11, 12, 13, 14].

Even though the discrete-basis formalism has been applied to an induced fission problem [2, 3, 4, 5, 7], its applicability has not yet been clarified, at least for a scattering problem. In this paper, I therefore apply the discrete-basis formalism to a simple one-dimensional barrier penetration problem, and carry out a comparative study of the numerical accuracy. To this end, I shall consider a Gaussian barrier and compare the penetrabilities obtained with the discrete-basis formalism to those with the standard Numerov method.

The paper is organized as follows. In Sec. II, I will detail the discrete-basis formalism for a one-dimensional problem. In § 3, I will apply it to a barrier penetration of a one-dimensional Gaussian barrier and discuss the applicability of the discrete-basis formalism. I will then summarize the paper in Sec. IV.

Discrete-basis formalism for barrier penetration

Consider a one-dimensional system for a particle with mass mm under a potential V(x)V(x). The Hamiltonian for this system reads,

H=−ℏ22md2dx2+V(x).(1)H=-\frac{\hbar^2}{2m}\frac{d^2}{dx^2}+V(x). \tag*{(1)}

I discretize the radial coordinate as xi=xmin⁡+(i−1)Δxx_i=x_{\min}+(i-1)\Delta x, and consider the model space from x1=xmin⁡x_1=x_{\min} to xN≡xmax⁡x_N\equiv x_{\max}. Using the 3-point formula for the kinetic energy in HH, the Hamiltonian (1) is transformed to a matrix form of

Hij=−tδi,j+1+(2t+Vi)δi,j−tδi,j−1,(2)H_{ij}=-t\delta_{i,j+1}+(2t+V_i)\delta_{i,j}-t\delta_{i,j-1}, \tag*{(2)}

where tt is defined as t=ℏ22m(Δx)2t=\frac{\hbar^2}{2m(\Delta x)^2} and Vi=V(xi)V_i=V(x_i). The wave function ϕi=ϕ(xi)\phi_i=\phi(x_i) then obeys

−tϕ0δi,1+∑j=1NHijϕj−tδN+1,iϕN=Eϕi.(3)-t\phi_0\delta_{i,1}+\sum_{j=1}^{N}H_{ij}\phi_j-t\delta_{N+1,i}\phi_N=E\phi_i. \tag*{(3)}

In the absence of the potential VV, the wave function ϕn(0)\phi_n^{(0)} obeys the equation

−t(ϕn+1(0)−2ϕn(0)+ϕn−1(0))=Eϕn(0).(4)-t\left(\phi_{n+1}^{(0)}-2\phi_n^{(0)}+\phi_{n-1}^{(0)}\right)=E\phi_n^{(0)}. \tag*{(4)}

I consider a free-particle solution given by

ϕn(0)∝e−iknΔx−eiknΔx.(5)\phi_n^{(0)}\propto e^{-ikn\Delta x}-e^{ikn\Delta x}. \tag*{(5)}

Substituting this into Eq. (4), one finds

cos⁡(kΔx)=1−E2t.(6)\cos(k\Delta x)=1-\frac{E}{2t}. \tag*{(6)}

In the presence of the potential VV, I consider the case where the particle is incident from the left hand side of the potential. Assuming that the potential VV almost vanishes at xmax⁡x_{\max}, the wave function ϕN+1\phi_{N+1} is given by ϕN+1=eikΔxϕN\phi_{N+1}=e^{ik\Delta x}\phi_N. Substituting this into Eq. (3), one finds

ϕi=[(H~−E)−1]i1tϕ0≡Gi1tϕ0.(7)\phi_i=[(\widetilde{H}-E)^{-1}]_{i1}t\phi_0\equiv G_{i1}t\phi_0. \tag*{(7)}

where H~\widetilde{H} is defined as H~ij=Hij−teikΔxδi,Nδj,N\widetilde{H}_{ij}=H_{ij}-te^{ik\Delta x}\delta_{i,N}\delta_{j,N} and the Green’s function GG is defined as G=(H~−E)−1G=(\widetilde{H}-E)^{-1}.

Assuming that the potential V(x)V(x) is negligible at x=x1x=x_1 and x=x2x=x_2, the wave functions at these points are given as linear superpositions of e±ikΔxne^{\pm ik\Delta x n} with n=1n=1 and 2, respectively. I parameterize the coefficients of the linear superpositions in terms of tt and the wave function ϕ0\phi_0 as,

ϕ1=(AeikΔx+Be−ikΔx)tϕ0,(8)\phi_1=(Ae^{ik\Delta x}+Be^{-ik\Delta x})t\phi_0, \tag*{(8)}
ϕ2=(Ae2ikΔx+Be−2ikΔx)tϕ0.(9)\phi_2=(Ae^{2ik\Delta x}+Be^{-2ik\Delta x})t\phi_0. \tag*{(9)}

This is equivalent to assume

G11=AeikΔx+Be−ikΔx.(10)G_{11}=Ae^{ik\Delta x}+Be^{-ik\Delta x}. \tag*{(10)}

Substituting Eqs. (8) and (9) into Eq. (3) and using Eq. (6), one finds

Ae2ikΔx+Be−2ikΔx=2cos⁡(kΔx)G11−1t.(11)Ae^{2ik\Delta x}+Be^{-2ik\Delta x}=2\cos(k\Delta x)G_{11}-\frac{1}{t}. \tag*{(11)}

Combining this with Eq. (10), one finds

A=e−ikΔxeikΔx−e−ikΔx(eikΔxG11−1t),(12)A=\frac{e^{-ik\Delta x}}{e^{ik\Delta x}-e^{-ik\Delta x}}\left(e^{ik\Delta x}G_{11}-\frac{1}{t}\right), \tag*{(12)}
B=−eikΔxeikΔx−e−ikΔx(e−ikΔxG11−1t).(13)B=-\frac{e^{ik\Delta x}}{e^{ik\Delta x}-e^{-ik\Delta x}}\left(e^{-ik\Delta x}G_{11}-\frac{1}{t}\right). \tag*{(13)}

Writing the wave function ϕN\phi_N as ϕN=GN1tϕ0≡TeikΔxtϕ0\phi_N=G_{N1}t\phi_0\equiv Te^{ik\Delta x}t\phi_0, the penetrability P(E)P(E) reads,

P(E)=∣TA∣2=∣2sin⁡(kΔx)GN1eikΔxG11−1/t∣2.(14)P(E)=\left|\frac{T}{A}\right|^2=\left|\frac{2\sin(k\Delta x)G_{N1}}{e^{ik\Delta x}G_{11}-1/t}\right|^2. \tag*{(14)}

Figure 1. The penetrabilities of a Gaussian barrier given by Eq. (15) with V0=100V_0=100 MeV and s=2s=2 fm. The mass is set to be m=29mNm=29m_N, where mNm_N is the nucleon mass. The upper panel is obtained with the Numerov method (the dashed line) and the discrete-basis formalism (the filled circles) with the mesh size of Δx=0.05\Delta x=0.05 fm. On the other hand, the lower panel shows the results of the Numerov method (the dashed line), the modified Numerov method (the dotted line), and the discrete-basis formalism (the solid line) with the mesh size of Δx=0.15\Delta x=0.15 fm.

Penetrability of a Gaussian barrier

Let us now numerically evaluate the penetrability for a given potential. For this purpose, I consider a Gaussian potential,

V(x)=V0e−x2/2s2.(15)V(x)=V_0e^{-x^2/2s^2}. \tag*{(15)}

Following Refs. [15, 16, 17], the parameters are chosen to be V0=100V_0=100 MeV and s=2s=2 fm together with m=29mNm=29m_N, where mNm_N is the nucleon mass, to mimic the fusion reaction of 58Ni+58Ni^{58}\mathrm{Ni}+^{58}\mathrm{Ni}. We set xmin⁡=−10x_{\min}=-10 fm and xmax⁡=10x_{\max}=10 fm.

The upper panel of Fig. 1 shows the penetrabilities of the Gaussian barrier obtained with Δx=0.05\Delta x=0.05 fm. The dashed line and the solid circles denote the results with the standard Numerov method and the discrete-basis formalism, respectively. The value of Δx\Delta x is small enough in this case, and both the method lead to accurate results. The lower panel shows the results with a larger value of Δx\Delta x, that is, Δx=0.15\Delta x=0.15 fm. In this case, the numerical error is significantly large with the Numerov method: the penetrabilities do not reach unity even at energies well above the barrier (see the dashed line). This is the case also with the modified Numerov method [18], with which the penetrability even exceeds unity at high energies with a non-monotonic behavior (see the dotted line). In marked contrast, the results with the discrete-basis formalism is rather robust and the penetrabilities are almost the same as the one with Δx=0.05\Delta x=0.05 fm shown in the upper panel. Notice that the discrete-basis formalism employs the simple 3-point formula for the kinetic energy, while a more sophisticated formula is used in the Numerov and the modified Numerov methods. Yet, it is interesting to notice that the discrete-basis method is numerically more stable than the Numerov and the modified Numerov methods. We point out that Δx\Delta x cannot be taken larger than 2ℏ2/Em\sqrt{2\hbar^2/Em}, though. If Δx\Delta x exceeds this value, the right hand side of Eq. (6) exceeds unity and the wave number kk cannot be defined unless it is extended to a complex number.

Summary

I examined the applicability of the discrete-basis method for a reaction theory. To this end, I considered barrier penetration of a one-dimensional Gaussian barrier. It was demonstrated that the discrete-basis method provides a more accurate and stable method than the standard Numerov method. This property may be helpful in obtaining numerically stable solutions of coupled-channels equations [19, 20].

The discrete-basis formalism has a good connection to a many-body Hamiltonian. As a matter of fact, there have been several applications of this method to microscopic descriptions of induced fission. In such applications, absorbing potentials, or imaginary energies, are introduced to a model Hamiltonian, and the absorbing probability is computed with the so called Datta formula [2]. Even though the model setup is somewhat different from a barrier problem in one-dimension, in which there is no absorbing part in the Hamiltonian, the conclusion in this paper would remain the same in the fission problem as well.

Acknowledgments

The author thanks G.F. Bertsch for helpful discussions and for a careful reading of the manuscript. This work was supported in part by JSPS KAKENHI Grant Numbers JP19K03861 and JP23K03414.

References

References

  1. [1]S.E. Koonin and D.C. Meredith, Computational Physics (Addison-Wesley, Reading, 1990).
  2. [2]P. Fanto, G.F. Bertsch, and Y. Alhassid, Phys. Rev. C98, 014604 (2018).
  3. [3]G.F. Bertsch and W. Younes, Ann. Phys. 403, 68 (2019).
  4. [4]Y. Alhassid, G.F. Bertsch, and P. Fanto, Ann. of Phys.419, 168233 (2020).
  5. [5]Y. Alhassid, G.F. Bertsch, and P. Fanto, Ann. of Phys.424, 168381 (2021).
  6. [6]G.F. Bertsch and K. Hagino, Phys. Rev. C105, 034618 (2022).
  7. [7]G.F. Bertsch and K. Hagino, Phys. Rev. C107, 044615 (2023).
  8. [8]K. Uzawa and K. Hagino, Phys. Rev. C108, 024319 (2023).
  9. [9]K.T.R. Davies, H. Flocard, S. Krieger, and M.S. Weiss, Nucl. Phys. A 342, 111 (1980).
  10. [10]P. Bonche, H. Flocard, and P.-H. Heenen, Comput. Phys. Commun. 171, 49 (2005).
  11. [11]W. Ryssens, V. Hellemans, M. Bender, and P.-H. Heenen, Comput. Phys. Commun. 187, 175 (2015).
  12. [12]Y. Tanimura, K. Hagino, and H.Z. Liang, Prog. Theo. Exp. Phys. 2015, 073D01 (2015).
  13. [13]Z.X. Ren, S.Q. Zhang, and J. Meng, Phys. Rev. C95, 024313 (2017).
  14. [14]B. Li, Z.X. Ren, and P.W. Zhao, Phys. Rev. C102, 044307 (2020).
  15. [15]C.H. Dasso, S. Landowne, and A. Winther, Nucl. Phys. A405, 381 (1983).
  16. [16]C.H. Dasso, S. Landowne, and A. Winther, Nucl. Phys. A407, 221 (1983).
  17. [17]K. Hagino and A.B. Balantekin, Phys. Rev. A70, 032106 (2004).
  18. [18]K. Hagino, N. Rowley, and A.T. Kruppa, Comput. Phys. Comm. 123, 143 (1999).
  19. [19]K. Hagino, K. Ogata, and A.M. Moro, Prog. in Part. and Nucl. Phys. 125, 103951 (2022).
  20. [20]K. Hagino and N. Takigawa, Prog. Theo. Phys. 128, 1061 (2012).

Paper details

vlnc:W_emvpj4ah
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