This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

11institutetext: Departamento de Física, Universidade Federal Rural de Pernambuco, 52171-900, Recife, PE, Brazil
Departamento de Física, Universidade Federal de Pernambuco, 50670-901, Recife, PE, Brazil
Departamento de Estatística e Informática, Universidade Federal Rural de Pernambuco, 52171-900, Recife, PE, Brazil

Wave optics Geometrical optics Multilayers; superlattices; photonic structures; metamaterials

A simple light-trapping device from a hyperbolic metamaterial on a catenoid

Frankbelson dos Santos Azevedo 0000-0002-4009-0720 11    José Diêgo M. de Lima 0000-0002-4296-6001 1122    Antônio de Pádua Santos 0000-0003-1262-0875 11    Tiago A. E. Ferreira 0000-0002-2131-9825 33    Fernando Moraes 0000-0001-7045-054X 11112233
Abstract

By using both ray and wave optics, we show that a simple device which consists on a film of hyperbolic metamaterial on the surface of a catenoid can be used to trap light. From the study of the trajectories, we observe a tendency for the light rays to wrap, and eventually be trapped, around the neck of the device. The wave equation appears to have an effective attractive potential, and their solutions confirm the bound states suggested by the trajectories. The relevant equations are solved numerically using neural networks.

pacs:
42.25.-p
pacs:
42.15.-i
pacs:
78.67.Pt

Metamaterials are a class of optical materials that may present a negative ratio between the refractive index components [1]. Hence, they may provide many applications for technological devices and new discoveries in the science of transformation optics [2]. From metamaterial devices, we can get analogies with cosmological systems as a way to verify and discuss ideas of cosmology in the laboratory. For example, with metamaterials, it is possible to mimic curved spacetime [3], to observe gravitational lens by trapping light [4], as well as to simulate black holes [5] and spinning cosmic string spacetime [6]. Also, they may be used to mimic a discontinuous change of metric signature [7].

An important class of metamaterials, called “hyperbolic” [8], can be found in nature [9] but are mostly artificial. These materials can be realized through layered metal–dielectric structures [12] or, alternatively, from nematic liquid crystals with metallic nanorods mixed in [13]. Among the applications of such materials, we mention the design of hyperlenses [10] and analogies with cosmological systems [11]. In a study made by two of us and coworkers, a hyperbolic liquid crystal metamaterial with molecules circularly arranged in a cylinder, shows optical concentrator behavior, for light is focused on its axis while propagating along the device [14]. In addition, wormhole representations based on metamaterials are well-known in the literature; Maslovski et al. [15] showed that superabsorbent metamaterial wormholes are conceivable with meshes of loaded transmission lines. Moreover, electromagnetic wormholes from metamaterials objects behave as virtual magnetic monopoles [16]: electromagnetic waves propagate between two points in space through an invisible tunnel. This was later experimentally demonstrated by using magnetic metamaterials and metasurfaces [17].

In a recent work [18], we showed that an oriented thin nematic liquid crystal film on a catenoid has an optical metric that corresponds to the geometry of a two-dimensional section of a conical wormhole spacetime. In that article, it was suggested the existence of zero angular momentum bound states, which would be soon verified by Atanasov et al. [19], that showed the possibility of having wormholes as waveguides for quantum particles with zero angular momentum, having an optical wormhole as one of their feasible realizations. From this standpoint, it is reasonable to expect to find bound states for nonzero angular momentum for a catenoid covered with a hyperbolic metamaterial thin film.

In this work, by using the optical metric, we study the trajectories and wave behavior of light through a thin, nematic-based, hyperbolic metamaterial film on a catenoid. We assume the director field lines oriented as shown in Fig. 1. Away from the throat of the catenoid, since the surface is asymptotically flat, the director field will look like that of a disclination on a plane with a hole (similar to a zz constant section of the cylindrical device described in Refs. [11, 14]). Here, we use a geometrical method to obtain the optical metric that describes the propagation of light along the nematic film on the catenoid, as in Ref. [18]. By specifying the director field n\vec{n} of the molecular arrangement, we can study optics in the device. From the optical point of view, this device is (locally) a uniaxial medium with ordinary and extraordinary indices given in terms of the permittivities no2=ϵn_{o}^{2}=\epsilon_{\perp} and ne2=ϵn_{e}^{2}=\epsilon_{\parallel} [20]. We remind that \perp and \parallel refer to the direction of the electric field of the light with respect to the director field. The material is made of an ordinary nematic liquid crystal with an admixture of metallic nanorods aligned along the director field, resulting in a negative component of the permittivity (ϵ<0\epsilon_{\parallel}<0) [13].

(a)Refer to caption (b)Refer to caption

Figure 1: (a) Hyperbolic metamaterial on the catenoid, with nematic liquid crystal molecules and metallic nanorods circularly aligned as indicated by the dashed circles. (b) Same device as as in (a) viewed from the top.

A coordinate system on the catenoid can be conveniently obtained from its definition in terms of the parametric equations

x=b0cosh(z/b0)cosϕ,y=b0cosh(z/b0)sinϕ,z=z,\begin{split}x&=b_{0}\cosh(z/b_{0})\cos\phi,\\ y&=b_{0}\cosh(z/b_{0})\sin\phi,\\ z&=z,\end{split} (1)

where b0b_{0} is the throat radius and ϕ[0,2π]\phi\in\left[0,2\pi\right]. Intersections of ϕ=const.\phi=const. planes with the catenoid are catenaries, whose arc lengths measured from the throat (z=0z=0) are given by τ=b0sinh(z/b0)\tau=b_{0}\sinh(z/b_{0}). This defines the coordinate system on the catenoid, composed of τ\tau and ϕ\phi. Following Ref. [18], we obtain for the device shown in Fig. 1 the optical metric

ds2=dτ2+α2(τ2+b02)dϕ2,ds^{2}=-d\tau^{2}+\alpha^{2}(\tau^{2}+b_{0}^{2})d\phi^{2}, (2)

where

α=ϵ|ϵ|.\alpha=\sqrt{\frac{\epsilon_{\perp}}{|\epsilon_{\parallel}|}}. (3)

The metric (2) is very similar to the metric of the device using an ordinary nematic liquid crystal of Ref. [18]. But, due to the introduction of the metallic nanorods and consequently, an imaginary extraordinary refractive index, we obtain a (,+)(-,+) signature of the metric instead (+,+)(+,+) as before.

Now, we investigate the trajectories of light traveling on the metamaterial film on the catenoid. For this, we start taking the geodesic equation

d2xαdλ2+Γαμνdxμdλdxνdλ=0,\frac{d^{2}x^{\alpha}}{d\lambda^{2}}+{\Gamma^{\alpha}}_{\mu\nu}\,\frac{dx^{\mu}}{d\lambda}\,\frac{dx^{\nu}}{d\lambda}=0, (4)

where λ\lambda is a continuous parameter, and

Γαμν=12gαβ(gβμxν+gβνxμgμνxβ).{\Gamma^{\alpha}}_{\mu\nu}=\frac{1}{2}g^{\alpha\beta}\left(\frac{\partial g_{\beta\mu}}{\partial x^{\nu}}+\frac{\partial g_{\beta\nu}}{\partial x^{\mu}}-\frac{\partial g_{\mu\nu}}{\partial x^{\beta}}\right). (5)

From the line element given in Eq. (2), we obtain the metric tensor as

gμν=(100α2(τ2+b02)),g_{\mu\nu}=\begin{pmatrix}-1&0\\ 0&\alpha^{2}(\tau^{2}+b_{0}^{2})\end{pmatrix}, (6)

such that the Christoffel symbols are

Γτϕϕ=α2τ{\Gamma^{\tau}}_{\phi\phi}=\alpha^{2}\tau (7)

and

Γϕτϕ=Γϕϕτ=ττ2+b02.{\Gamma^{\phi}}_{\tau\phi}={\Gamma^{\phi}}_{\phi\tau}=\frac{\tau}{\tau^{2}+b_{0}^{2}}. (8)

Substituting Eqs. (7) and (8) into Eq. (4), we obtain the geodesic equations

d2τdλ2+α2τ(dϕdλ)2=0\frac{d^{2}\tau}{d\lambda^{2}}+\alpha^{2}\tau\,\bigg{(}\dfrac{d\phi}{d\lambda}\bigg{)}^{2}=0 (9)

and

d2ϕdλ2+(2ττ2+b02)dτdλdϕdλ=0,\dfrac{d^{2}\phi}{d\lambda^{2}}+\bigg{(}\frac{2\tau}{\tau^{2}+b_{0}^{2}}\bigg{)}\,\dfrac{d\tau}{d\lambda}\,\dfrac{d\phi}{d\lambda}=0, (10)

where ττ(λ)\tau\equiv\tau(\lambda) and ϕϕ(λ)\phi\equiv\phi(\lambda).

We numerically solve Eqs. (9) and (10) to find geodesics for some chosen initial conditions. The geodesics are shown in Fig. 2. The numeric computational procedure employed here to solve those equations is based on Neural Networks [41, 42]. In particular, the python library NeuroDiffEq [43] was used with the boundary conditions described in Fig. 2. Figure 2(a) shows the case of light being wrapped around the throat of the catenoid. In Fig. 2(b), light is reflected while in Fig. 2(c), light winds around the throat, but it is still transmitted. The values of b0b_{0} and α\alpha used were, respectively, 1.0 and 0.85.

(a)Refer to caption (b) Refer to caption (c)Refer to caption

Figure 2: Trajectories of light, obtained numerically from Eqs. (9) and (10), with ϕ(0)=0\phi(0)=0, ϕ(0)=1\phi^{\prime}(0)=1, τ(0)=5\tau(0)=5, and the following values of τ(0)\tau^{\prime}(0) (which sets the shooting angle) for incident light rays: (a) τ(0)=0.1\tau^{\prime}(0)=0.1, (b) τ(0)=2\tau^{\prime}(0)=2, and (c) τ(0)=0.5\tau^{\prime}(0)=-0.5.

Now, to examine the wave behavior of the extraordinary rays, we begin with the scalar Helmholtz equation for monochromatic light propagating on a curved surface [23]

(Δg+k2)Ψ=(H2K)Ψ,\left(\Delta_{g}+k^{2}\right)\Psi=-(H^{2}-K)\Psi, (11)

where Δg\Delta_{g} is the Laplace-Beltrami operator, HH is the mean curvature, KK is the Gaussian curvature, Ψ\Psi is the amplitude of the electric field, and kk is the wavenumber. For the catenoid, H=0H=0, and |K|1014|K|\sim 10^{-14} nm-2 for surfaces of the size of centimeters (107 nm). Since for visible light k2104k^{2}\sim 10^{-4} nm-2, we can neglect the right-hand side of Eq. (11). Using the metric tensor (6), we get

Δg=2τ2ττ2+b02τ+1α2(τ2+b02)2ϕ2,\Delta_{g}=-\dfrac{\partial^{2}}{\partial\tau^{2}}-\dfrac{\tau}{\tau^{2}+b_{0}^{2}}\dfrac{\partial}{\partial\tau}+\dfrac{1}{\alpha^{2}(\tau^{2}+b_{0}^{2})}\dfrac{\partial^{2}}{\partial\phi^{2}}, (12)

and from Eq. (11), we find

2Ψτ2ττ2+b02Ψτ+1α2(τ2+b02)2Ψϕ2+k2Ψ=0,-\dfrac{\partial^{2}\Psi}{\partial\tau^{2}}-\dfrac{\tau}{\tau^{2}+b_{0}^{2}}\dfrac{\partial\Psi}{\partial\tau}+\dfrac{1}{\alpha^{2}(\tau^{2}+b_{0}^{2})}\dfrac{\partial^{2}\Psi}{\partial\phi^{2}}+k^{2}\Psi=0, (13)

which, with the ansatz Ψ(τ,ϕ)=eimϕZ(τ)\Psi(\tau,\phi)=e^{im\phi}Z(\tau), where m=0,±1,±2,m=0,\pm 1,\pm 2,\dots, becomes

d2Zdτ2+ττ2+b02dZdτ+[m2α2(τ2+b02)k2]Z=0.\dfrac{d^{2}Z}{d\tau^{2}}+\dfrac{\tau}{\tau^{2}+b_{0}^{2}}\dfrac{dZ}{d\tau}+\left[\dfrac{m^{2}}{\alpha^{2}(\tau^{2}+b_{0}^{2})}-k^{2}\right]Z=0. (14)

The solution to this equation can be obtained in terms of the modified Mathieu function since, by substitution of τ=b0sinh(z/b0)\tau=b_{0}\sinh\left({z}/{b_{0}}\right) in Eq. (14), one obtains the modified Mathieu equation [21, 18].

We can write Eq. (14) as a Schrödinger-like equation [24] by making Z(τ)=(τ2+b02)1/4χ(τ)Z(\tau)=(\tau^{2}+b_{0}^{2})^{-1/4}\chi(\tau), such that

d2χdτ2[k2+V(τ)]χ=0,\dfrac{d^{2}\chi}{d\tau^{2}}-\left[k^{2}+V(\tau)\right]\chi=0, (15)

with the “effective potential” (see Fig. 3)

V(τ)=2b02τ24(τ2+b02)2m2α2(τ2+b02).V(\tau)=\dfrac{2b_{0}^{2}-\tau^{2}}{4(\tau^{2}+b_{0}^{2})^{2}}-\dfrac{m^{2}}{\alpha^{2}(\tau^{2}+b_{0}^{2})}. (16)

The second term in Eq. (16) works as a “centripetal” potential for the angular momentum mm. The function (16) goes to zero when τ±\tau\rightarrow\pm\infty, in agreement with the flatness of the surface in this region. For m=0m=0, it has a maximum at τ=0\tau=0, and two shallow minima at τ=±5b0\tau=\pm\sqrt{5}b_{0}, respectively. However, the potential has always a minimum at τ=0\tau=0 for m0m\neq 0. This allows the existence of bound states or trapping of the light in the throat region. We proceed now to find some of those states by applying a variational method to (15) and then refining the obtained estimates with the neural network method described below Eq. (10).

(a) Refer to caption
(b) Refer to caption

Figure 3: Effective potential, as given by Eq. (16), for m=±1,±2,±3m=\pm 1,\pm 2,\pm 3 with α=0.85\alpha=0.85 and b0=1b_{0}=1. In graph (a), for m=0m=0, there are two minima at τ=±5\tau=\pm\sqrt{5} and a maximum at τ=0\tau=0. In graph (b), for m=±1m=\pm 1 and m=±2m=\pm 2, the potential has only a minimum at τ=0\tau=0. For the three curves, the potential goes to zero asymptotically.

Following the quantum mechanics analogy provided by Eq. (15), we consider normalized “wavefunctions” obeying the boundary condition limτ±χ=0\displaystyle{\lim_{\tau\rightarrow\pm\infty}}\chi=0. We also define a bound state “energy” En,m=k2E_{n,m}=-k^{2} corresponding to the wavefunction χn,m\chi_{n,m}, where n=1,2,3,n=1,2,3,\dots is the quantum number that describes the order of the energy eigenvalues. The eigenvalues En,mE_{n,m} are then the expectation values of the operator d2dτ2+V-\dfrac{d^{2}}{d\tau^{2}}+V, i.e.,

En,m=+χn,md2χn,mdτ2𝑑τ++χn,mVχn,m𝑑τ.\begin{split}E_{n,m}=&-\int_{-\infty}^{+\infty}\chi_{n,m}^{*}\dfrac{d^{2}\chi_{n,m}}{d\tau^{2}}d\tau\\ &+\int_{-\infty}^{+\infty}\chi_{n,m}^{*}V\chi_{n,m}d\tau.\end{split} (17)

The shape of the effective potential suggests a Gaussian as trial function for the ground state, or n=1n=1 case. Including the variational parameter βn,m>0\beta_{n,m}>0, we have for n=1n=1,

χ1,m(τ)=(β1,mπb02)1/4exp(β1,mτ22b02).\chi_{1,m}(\tau)=\left(\dfrac{\beta_{1,m}}{\pi b_{0}^{2}}\right)^{1/4}\exp\left(-\dfrac{\beta_{1,m}\tau^{2}}{2b_{0}^{2}}\right). (18)

For the first excited state, which must be orthogonal to the ground state, we propose

χ2,m(τ)=(4β2,m3πb06)1/4τexp(β2,mτ22b02),\chi_{2,m}(\tau)=\left(\dfrac{4\beta_{2,m}^{3}}{\pi b_{0}^{6}}\right)^{1/4}\tau\exp\left(-\dfrac{\beta_{2,m}\tau^{2}}{2b_{0}^{2}}\right), (19)

obtained from the derivative of (18) with respect to τ\tau.

By replacing (18) or (19) in Eq. (17), we obtain the function En,m(βn,m)E_{n,m}(\beta_{n,m}) whose minimum provides the energy eigenvalue in each state. The aim is then to find the value of βn,m\beta_{n,m} that minimizes En,mE_{n,m}, for each choice of n,mn,m. The wavenumbers are finally obtained from kn,m=En,mk_{n,m}=\sqrt{-E_{n,m}}. Table 1 shows kn,mk_{n,m} obtained numerically for some values of nn and mm, using α=0.85\alpha=0.85 and b0=1b_{0}=1. Using the wavenumbers supplied by Table 1 as input, we utilized neural networks again. This time to solve the wave equation (15), improving then the precision of the wavefunction. Figure 4 shows the radial wave functions thus obtained for n=1n=1 and n=2n=2 and m=±1,±2,±3m=\pm 1,\pm 2,\pm 3. As the wavefunctions indicate, the light rays are bound to the throat region of the catenoid. That could already be visualized in Fig. 2(a) obtained from ray optics.

Table 1: Variational results for some wavenumbers kn,mk_{n,m} associated with bound states of Eq. (15).
nn mm kn,mk_{n,m}
11 ±1\pm 1 0.691500.69150
11 ±2\pm 2 1.853891.85389
11 ±3\pm 3 3.028963.02896
22 ±1\pm 1 0.169270.16927
22 ±2\pm 2 1.080961.08096
22 ±3\pm 3 2.175092.17509

(a) Refer to caption
(b) Refer to caption

Figure 4: (a) Radial wave functions for n=1n=1 and m=±1,±2,±3m=\pm 1,\pm 2,\pm 3. (b) The same for n=2n=2.

To conclude, we want to emphasize that we have studied here a device made from a hyperbolic metamaterial based on nematic liquid crystals under a particular arrangement. This may lead to bound states, or trapping of light around the throat of the device, as shown both by ray and wave optics (respectively, classical trajectories and the wavefunctions in the quantum mechanics analogy). This could be the starting point for the design of light storage devices. Furthermore, the trajectories that are not trapped, like the ones that are reflected and transmitted, corresponding to scattering states of the wave equation, may also have a technological application in the control of the transmission of light. We also mention that, in Ref. [18], we proposed a similar device made from a nematic liquid crystal without the metallic nanorods which are included here. From the effective geometry given by the optical metric, we observed there that we had an analogue wormhole in an asymptotically flat Lorentzian (i.e. with metric signature (-1,1,1,1)) spacetime. Here, with the nanorods giving the hyperbolic metamaterial character to the system, the optical metric suggests again a wormhole but, this time in an asymptotically flat Kleinian (i.e. with metric signature (-1,-1,1,1)) spacetime [7]. Kleinian geometry leads to an alternate Special Relativity [44] and the possibility of using optics to simulate exotic objects in such context is an exciting research line which we are presently pursuing.

Acknowledgements.
This work was partially supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico (F.M. and T.A.E.F.) and Fundação de Amparo à Ciência e Tecnologia do Estado de Pernambuco (J.D.M.L.).

References

  • [1] \NameSmith, D. R., Pendry, J. B. Wiltshire, M. C. K. \REVIEWScience3052004788–792.
  • [2] \NameChen, H., Chan, C. T. Sheng, P. \REVIEWNat. Mater.92010387–396.
  • [3] \NameGenov, D. A., Zhang, S. Zhang, X. \REVIEWNat. Phys.52009687–692.
  • [4] \NameSheng, C. L. et al. \REVIEWNat. Photonics72013902–906.
  • [5] \NameFernández-Núñez, I. Bulashenko, O. \REVIEWPhys. Lett. A38020161–8.
  • [6] \NameMackay, T. G. Lakhtakia, A. \REVIEWPhys. Lett. A37420102305–2308.
  • [7] \NameFigueiredo, D. et al. \REVIEWPhys. Rev. D942016044039.
  • [8] \NameGuo, Zhiwei, Haitao Jiang, and Hong Chen \REVIEWJournal of Applied Physics1272020071101
  • [9] \NameNarimanov, E. E. Kildishev, A. V. \REVIEWNat. Photonics92015214.
  • [10] \NameJacob, Z., Alekseyev, L. V. Narimanov, E. \REVIEWOpt. Express1420068247–8256.
  • [11] \NameFigueiredo, D. et al. \REVIEWPhys. Rev. D962017105012.
  • [12] \NamePoddubny, A. et al. \REVIEWNat. Photonics72013948–957.
  • [13] \NameXiang, J. Lavrentovich, O. D. \REVIEW Mol. Cryst. Liq.5592012106–114.
  • [14] \NameAzevedo, F. dos S. et al. \REVIEWEPL124201834006.
  • [15] \NameMaslovski, S. et al. \REVIEWPhys. Rev. B982018245143.
  • [16] \NameGreenleaf, A. et al. \REVIEWPhys. Rev. Lett.992007183901.
  • [17] \NamePrat-Camps, J., Navau, C. Sanchez, A. \REVIEWSci. Rep.5201512488.
  • [18] \NameAzevedo, F. dos S. et al. \REVIEWPhys. Rev. A1032021023516.
  • [19] \NameAtanasov, V., Dandoloff, R. Saxena, A. B. \REVIEWPhys. Scr.962021125001.
  • [20] \NameKleman, M. Laverntovich, O. D. \BookSoft matter physics: an introduction \Vol1 \PublSpringer Science & Business Media, New York \Year2003 .
  • [21] \NameAbramowitz, M. Stegun, I. A. \BookHandbook of mathematical functions with formulas, graphs, and mathematical tables \Vol55 \PublUS Government Printing Office, Washington \Year1964 .
  • [22] \NameGradshteyn, I. S. Ryzhik, I. M. \BookTables of integrals, series, and products \Vol7 \PublElsevier Science, New York \Year2007 .
  • [23] \NameSchultheiss, V. H. et al. \REVIEWPhys. Rev. Lett.1052010143901.
  • [24] \NameKar, S. Sahdev, D. Bhawal, B. \REVIEWPhys. Rev. D491994853
  • [25] \NameDunster, T. M. \REVIEWSIAM J. Math. Anal.211990995–1018.
  • [26] \NameOlver, F. W. J. et al. \BookNIST Handbook of Mathematical Functions \Vol1 \PublCambridge University Press, Cambridge \Year2010 .
  • [27] \NameSakurai, J. J. Napolitano, J. \BookModern Quantum Mechanics \Vol3 \PublCambridge University Press, Cambridge \Year2020 .
  • [28] \NameZettili, N. \BookQuantum Mechanics: Concepts and Applications \Vol2 \PublWiley, Chippenham \Year2009 .
  • [29] \NameAlty, L. J. \REVIEWClassical and Quant. Grav.1119942523.
  • [30] \NameBarrett, J. et al. \REVIEWInt J Mod Phys A919941457–1493.
  • [31] \NameFerraris, M. Francaviglia, M. \REVIEWGen. Relativ. Gravit.101979283–296.
  • [32] \NameSung, S.-T. \REVIEWgr-qc/9703039 preprint1997.
  • [33] \NameHellaby, C., Sumeruk, A. Ellis, G. F. R. \REVIEWInt. J. Mod. Phys. D61997211–238.
  • [34] \NameLemos, J. P. S., Lobo, F. S. N. de Oliveira, S. Q. \REVIEWPhys. Rev. D682003064004.
  • [35] \NameVisser, M. \REVIEWNucl. Phys. B3281989203–212.
  • [36] \NameVisser, M. \REVIEWPhys. Rev. D3919893182.
  • [37] \NameSharif, M Mumtaz, S. \REVIEWEur. Phys. J. Plus13220171–13.
  • [38] \NameLobo, F. S. N. Crawford, P. \REVIEWClassical and Quant. Grav.212003391.
  • [39] \NameLi, A.-C. and Xu, W.-L. Zeng, D.-F. \REVIEWJ. Cosmol. Astropart. Phys.20192019016.
  • [40] \NameHarris, E. G. \REVIEWAm. J. Phys.6119931140–1144.
  • [41] \NameLagaris I. E., Likas A. Fotiadis D. I. \REVIEWIEEE Transactions on Neural Networks 91998987.
  • [42] \NameMichoski C., Milosavljevic M., Oliver T. Hatch D. R. \REVIEWNeurocomputing 3992020193.
  • [43] \NameChen F., Sondak D., Protopapas P., Mattheakis M., Liu S., Agarwal D. Giovanni M. D. \REVIEWJournal of Open Source Software 520201931.
  • [44] \NameAlves-Júnior, F. A. P., A. B. Barreto, and F. Moraes \REVIEWPhysical Review D1032021044023