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

Q-ball-like solitons on the M2-brane with worldvolume fluxes

P. Garcia pgarcial@ucab.edu.ve Departamento de Física, Facultad de Ingeniería, Universidad Católica Andrés Bello, Caracas 1020-A, Venezuela.    M.P. Garcia del Moral m-pilar.garciam@unirioja.es Área de Física, Departamento de Química, Universidad de la Rioja, La Rioja 26006, Spain.    J.M. Peña joselen@yahoo.com; joselen.pena@ce.ucn.cl Departamento de Física, Universidad Católica del Norte, Casa Central. Angamos 0610, Antofagasta, Chile.    R. Prado-Fuentes111The order of the authors is alphabetical reginaldo.prado@ua.cl; reginaldo.prado@umayor.cl Núcleo de Matemática, Física y Estadística, Facultad de Ciencias, Universidad Mayor, chile. Departamento de Física, Universidad de Antofagasta, Aptdo 02800, Chile.
Abstract

In this paper we obtain a family of analytic solutions to the nonlinear partial differential equations that describe the dynamics of the bosonic part of the mass operator of a M2-brane compactified on M9×T2M_{9}\times T^{2} in the LCG with worldvolume fluxes. Those fluxes can be induced by a constant and quantized supergravity 3-form. This sector of the theory, at supersymmetric level, has the interesting property of having a discrete spectrum. We have focused on the characterization of Q-ball-like (QBL) solitons on the M2-brane with worldvolume fluxes. Two scenarios are analysed: one in which the system is isotropic and the other anisotropic. In the isotropic case, we obtain analytic families of string-like solutions to the membrane equations of motion in the presence of a non-vanishing symplectic gauge field that satisfy all constraints. We explicitly show a localised family of QBL solutions. It is demonstrated that although the solutions generally exhibit dispersion, they also allow for dispersion-free solutions. In the non-isotropic case, we obtain full-fledged membrane QBL solutions by numerical methods. We characterize some other properties of the solutions found. The dynamics of the QBL solutions are also encountered. We analyze the Lorentz boosts and Galilean transformations. Since we work in the Light Cone Gauge, the Lorentz transformed solutions are not automatically solutions, rather some extra conditions must be imposed. Only a subset of the solutions remain. We discuss some examples. The QBL solitons of the M2-brane that have been discovered contain an interaction term between the Noether charge of the Q-ball and the topological monopole charge associated with the worldvolume flux. The monopole charge increases the stability of the analytic solutions against fission. The superposition law of the analytic solutions was obtained, which allows linear superposition, but only for very limited discrete frequencies. Further research is required to determine whether these solutions can be related to a new type of soliton, the Q-monopole ball.

preprint: APS/123-QED

I Introduction

Solitons are stable classical solutions with finite energy to nonlinear equations that are able to mimick the behaviour of particles as well as excitations. They are classified into topological and nontopological ones. Topological solitons like monopoles, vortices, instantons and kinks have been extensively studied in the literature. Non-topological solitons have also received a lot of interest for its potential to explain dark matter among others. One simple case corresponds to Q-balls. Q-balls [1] are the generic name given to solitons, originally with spherical topology, which are invariant under a U(1)U(1) global symmetry. This symmetry implies the existence of a Noether charge which provides stability against perturbations. Thus global symmetry has been associated to quantum invariants as lepton or barion number. There exists also gauged Q-balls [2, 3] in the case in which the complex scalar fields that parametrized the Q-ball are also charged under a gauge potential U(1)U(1). There are also other works in which the authors have considered more general global or gauged groups [4, 5]. Dynamics of Q-balls have been studied in [6, 7] and for the gauged case [2, 8]. Their stability has also been discussed in several papers, see for example [9]. Recently, it has been discovered that it is possible to obtain solitonic solutions that simultaneously share the two types of properties. These type of solitons have been denoted as Q-monopole-ball [10].

Solitons with other topologies different from the sphere have been described to less extent [11]. For example, Q-torus, which corresponds to Q-balls with toroidal topology have been also studied in [12]. Q-balls have also been used to model dark matter [13], in the context of string theory [14]. This paper describes Q-ball solitons from the open and closed string moduli as well as other types of solitons. However, there are very few results of solitons have been reported at the M-theory level.

Classical solutions of membrane theory can be useful to better understand the M-theory properties. First examples of classical solutions to the membrane theory firstly appeared in [15]. In the context of matrix models, it was discussed in [16]. The existence of membrane solitonic solutions has been discussed in several works. For example in terms of stringy solitary waves in [17], in the context of Q-ball matrix model [18], as instantonic solutions [19], or in terms of membranes formulated on hyperkähler backgrounds [20, 21]. There are bounds for the experimental detection of Q-balls [9], though they are model dependent.

This paper presents analytical and numerical methods to obtain solitonic solutions of the M2-brane with worldvolume fluxes, specifically Q-ball-like solutions. By Q-ball-like (QBL) solutions we refer to solitonic excitations of the membrane that satisfy the characteristic QBL ansatz. We consider a supermembrane on a M9×T2M_{9}\times T^{2} target space with a C±C_{\pm} flux background. This configuration possesses a magnetic U(1)U(1) monopole in the compact sector [22, 23] associated with the flux content that induces the presence of a magnetic monopole over the worldvolume [24]. We search solutions to the nonlinear equations of the M2-brane Hamiltonian on this particular background focusing on the existence of QBL solitons. These solutions are described by complex scalars that model the dynamics of the excitations on the membrane with fluxes, in the non-compact transverse space to the lightcone coordinates. The Hamiltonian that describes the dynamics of the M2-brane contains on top of nontrivial topological terms, non-linear cuartic scalar contributions and a symplectic field strength whose dynamics under certain assumptions becomes coupled. We analyze two different cases: Isotropic analytical solutions that generate string-like configurations with a non-vanishing cost in energy and numerical non-isotropical solutions for the membrane full-fledged configurations.
The string-like configurations, unlike with the string-like spikes associated with the case of a membrane without central charge, carry minimal energy associated with the M2-brane with worldvolume fluxes and the mass terms and thus have dynamics. These analytical solutions of the isotropic case are found by solving the equations of motion (EOM) and the constraints of the theory: the area preserving first class constraint and the flux constraint. The flux condition induces a topological central charge restriction. The different families of solutions found satisfy the Q-ball ansatz and include the case of localized solutions. With the aid of numerical simulation, we extend those results to the non-isotropic case in the absence of a symplectic gauge field. We then obtain, QBLnuev full-fledged membrane solutions.

Having established the existence of these solutions, we characterise some of their properties. We show that the analytic families of solutions considered exhibit dispersion generically, but they also admit dispersion-free solutions.

The dynamics of the Q-ball solitons are characterised by studying their Lorentz transformation and their nonrelativistic limit, the Galilean one. As we are in the Light Cone Gauge (LCG), the transformed solutions are not automatically valid and some extra restrictions must be imposed. We obtain it for some of the analytic families of solutions considered.

There are two types of charges for the solutions found: a non-topological charge related to the Noether charge of the Q-ball and a topological charge related to the existence of a monopole charge resulting from the pull-back of the target C±C_{\pm} fluxes. The monopole charge defined on the world volume is induced by the nontrivial embedding in the compact space, and the Q-ball solitons are related to the non-compact target space. There is a non-trivial interaction between them induced by the covariant derivative terms at the level of the mass operator. This interaction leads to an enhanced stability.

Finally, we describe their superposition law and demonstrate that, as expected, they cannot be linearly superposed. However, they can be linearly composed for very restricted values of the Fourier modes.

The paper is structured as follows: Section 2 reviews the formulation of the supermembrane toroidally wrapped on M9×T2M_{9}\times T^{2} subject to a C±C_{\pm} flux condition and the equations of motion to solve them. Sections 3 to 8 present our results. In Section 3, we obtain analitically a family of exact QBL string-like solutions for the isotropic case. The dispersion of solutions in the presence of a symplectic gauge field is characterized. Additionally, localized solutions for the QBL ansatz are demonstrated. Section 4 characterizes a membrane QBL soliton numerically. Firstly, we compare the numerical analysis of the isotropic case with the previously obtained analytical results by imposing the ansatz of the so-called ’family’. Then, we extend the analysis to the non-isotropic case, taking into account the complete membrane excitations, when the symplectic gauge vanishes. In the non-isotropic case, the nonlinear character of the complex scalar field becomes explicitly manifest. In section 5 the dynamics of the solution by Lorentz and Galilean transformations is discussed. Section 6 demonstrates a relation between topological and non-topological charges of the M2-brane, improving stability. In Sec. 7 we present a superposition law for the QBL family of solutions that was previously found for the isotropic case. Linear superposition of the solutions is only possible when the frequencies are strongly constrained, consistent with the nonlinear nature of the equations. The results obtained are discussed and concluded in section 8.

II The M2-brane with 2-form worldvolume C±C_{\pm} fluxes

In this work we analyze the existence of solutions to the dynamics of an M2-brane mass operator, focusing in those of Q-ball-type. The bosonic sector that we analyze, corresponds to the supermembrane with CC fluxes introduced in [22, 23] that induce a topological condition denoted central charges which is associated with the presence of a magnetic monopole over the worldvolume of the membrane [24]. The M2-brane is formulated on a target space M9×T2M_{9}\times T^{2} in the Light Cone Gauge (LCG) [25, 26]. We define the embedding maps on the non-compact space Xm(τ,σ,ρ):ΣM9X^{m}(\tau,\sigma,\rho):\Sigma\to M_{9} with m=2,,8m=2,...,8. The maps Xr(τ,σ,ρ):ΣT2X^{r}(\tau,\sigma,\rho):\Sigma\to T^{2}, where r=9,10r=9,10, describe the embedding on the compactified 2-torus. These maps act as scalars on the M2-brane worldvolume and as vectors on the target space. The coordinates of the supermembrane worldvolume are labeled by (τ,σ,ρ)(\tau,\sigma,\rho) parametrizing Σ×\Sigma\times\mathbb{R}, with Σ\Sigma representing a Riemann surface of genus one and \mathbb{R} parametrizing time. The maps XrX^{r} satisfy the winding condition

𝒞s𝑑Xr=2πRrmrs,\oint_{\mathcal{C}_{s}}dX^{r}=2\pi R^{r}m_{r}^{s}\ , (1)

where mrsm_{r}^{s} are the winding numbers and RrR^{r} are the torus radii.

The M2-brane is restricted by a worldvolume flux condition. It corresponds to a topological condition denoted by the authors as Central Charge condition(CC) [24], satisfying that n=det𝕎n=det\mathbb{W} where 𝕎\mathbb{W} denotes the winding matrix of an irreducible wrapping of the M2-brane around the compactified 2-torus. The central charge condition is associated to the existence of a monopole over the worldvolume which is given by the curvature F^=ϵrsdXhrdXhs\widehat{F}=\epsilon_{rs}dX^{r}_{h}\wedge dX^{s}_{h} associated to a nontrivial U(1)U(1) gauge field F^=dA^\widehat{F}=d\widehat{A} defined on the membrane worldvolume.

ΣF^=Σϵrs𝑑XrdXs=nAT2,n/{0},\int_{\Sigma}\widehat{F}=\int_{\Sigma}\epsilon_{rs}dX^{r}\wedge dX^{s}=nA_{T^{2}},\quad n\in\mathbb{Z}/\{0\}\ , (2)

where AT2A_{T^{2}} denotes the 2-torus target space area.

The worldvolume flux can be also induced by the pullback of a target-space 2-form flux condition generated by the quantization of the supergravity 3-form in the LCG once that the XX^{-} has been eliminated from the formulation, see, for example, in [22, 23].

T2F2=AT2n.with/{0}.\int_{T^{2}}F_{2}=A_{T^{2}}n.\quad\textrm{with}\quad\in\mathbb{Z}/\{0\}. (3)

The fluxes of the Hamiltonian can be induced by the presence of a constant quantized 3-form C±C_{\pm} background. The Hamiltonian of a supermembrane with central charges corresponds to the Hamiltonian of the M2-brane with worldvolume fluxes CC_{-}. Its formulation is equivalent to the Hamiltonian of the M2-brane with central charge, plus a constant shift given by the pullback of the C+C_{+} contribution [22, 23]

HC±fluxes=HC+ΣC+=HCC+ΣC+.H_{C_{\pm}}^{fluxes}=H_{C_{-}}+\int_{\Sigma}C_{+}=H_{CC}+\int_{\Sigma}C_{+}\ . (4)

Both cases are related, being duals for the CC_{-} case and the central charge one. They are defined in terms of an integer nn, so in the rest of the paper we will use the central charge formulation for simplicity in th3e resolution.

The closed one-forms dXrdX_{r} can be decomposed by a Hodge decomposition in terms of two harmonic one-forms dX^rd\widehat{X}_{r} and two exact one-forms dArdA_{r}, such that

dXr(σ,ρ,τ)=dXrh(σ,ρ)+dAr(σ,ρ,τ),dX_{r}(\sigma,\rho,\tau)=d{X}_{rh}(\sigma,\rho)+dA_{r}(\sigma,\rho,\tau)\,, (5)

The dXrhdX_{rh} is a closed one-form defined in terms of the basis of harmonic forms as dXrh=RrmrsdX^s(σ,ρ)d{X}_{rh}=R^{r}m_{r}^{s}d\widehat{X}_{s}(\sigma,\rho). They are not physical degrees of freedom in the theory but encode topological information of the compactified manifold. The central charge condition guarantees the existence of the two independent harmonic constant one-forms, dX^s(σ,ρ)d\widehat{X}_{s}(\sigma,\rho). The central charge condition also implies the existence a new dynamical degree of freedom ArA_{r} whose associated one-form 𝔸r=dAr=aArdσa\mathbb{A}_{r}=dA_{r}=\partial_{a}A_{r}d\sigma^{a}, with a=(σ,ρ)\partial_{a}=(\partial_{\sigma},\partial_{\rho}) describes a symplectic gauge field over the M2-brane worldvolume. It transforms as a symplectic connection under symplectomorphisms. dArdA_{r} are exact one-forms.

The LCG Hamiltonian of the M2-brane theory with central charge corresponds to [25, 26]

HCC\displaystyle H_{CC} =T2/3Σd2σW[12(PmW)2+12(PrW)2]\displaystyle=T^{-2/3}\int_{\Sigma}d^{2}\sigma\sqrt{W}\Big{[}\frac{1}{2}\Big{(}\frac{P_{m}}{\sqrt{W}}\Big{)}^{2}+\frac{1}{2}\Big{(}\frac{P_{r}}{\sqrt{W}}\Big{)}^{2}\Big{]} (6)
+T2/3Σd2σW[T24{Xm,Xn}2+\displaystyle+T^{-2/3}\int_{\Sigma}d^{2}\sigma\sqrt{W}\Big{[}\frac{T^{2}}{4}\left\{X^{m},X^{n}\right\}^{2}+
+T22(𝒟rXm)2+T24(rs)2],\displaystyle+\frac{T^{2}}{2}(\mathcal{D}_{r}X^{m})^{2}+\frac{T^{2}}{4}(\mathcal{F}_{rs})^{2}\Big{]},

where TT stands for the membrane tension and WW is the determinant of the induced spatial part component of the foliated metric on the membrane; {A,B}=1WϵabaAbB;\left\{A,B\right\}=\frac{1}{\sqrt{W}}\epsilon^{ab}\partial_{a}A\partial_{b}B;\ is the symplectic bracket defined in terms of the harmonic one-forms dX^d\widehat{X} of the Riemann surface with a,b=1,2a,b=1,2 and a=(σ,ρ)\partial_{a}=(\partial_{\sigma},\partial_{\rho}) and W=12ϵabϵrsaX^rbX^s\sqrt{W}=\frac{1}{2}\epsilon^{ab}\epsilon_{rs}\partial_{a}\widehat{X}^{r}\partial_{b}\widehat{X}^{s}. The scalar fields Xm,ArX^{m},A_{r} have respectively, canonical momenta PmP_{m} and PrP_{r}. The symplectic covariant derivative is defined as 𝒟r=Dr+{Ar,}\mathcal{D}_{r}\bullet=D_{r}\bullet+\{A_{r},\bullet\} with Dr=2πmruθuvRrϵabWaX^vb.D_{r}\bullet=2\pi m_{r}^{u}\theta_{uv}R_{r}\frac{\epsilon^{{a}{b}}}{\sqrt{W}}\partial_{{a}}\widehat{X}^{v}\partial_{{b}}\bullet.   The derivative DrD_{r} is defined in terms of the moduli of the 2-torus, the harmonic one-forms dX^rd\widehat{X}_{r}, and a matrix θuv\theta_{uv} with u,v=9,10u,v=9,10, related to the monodromy associated to its global description in terms of a torus bundle [27]. Therefore, the Hamiltonian contains a symplectic curvature \mathcal{F} defined by

rs=DrAsDsAr+{Ar,As}.\mathcal{F}_{rs}=D_{r}A_{s}-D_{s}A_{r}+\left\{A_{r},A_{s}\right\}.\, (7)

The constraints of the theory associated to the local Area Preserving Diffeomorphisms (APD) are:

𝒟rPr+{Xm,Pm}0,\mathcal{D}_{r}P_{r}+\left\{X^{m},P_{m}\right\}\approx 0\ , (8)

and to the two APD global constraints,

𝒞s((PrW)aXr+(PmW)aXm)𝑑σa0.\oint_{\mathcal{C}_{s}}\Big{(}\Big{(}\frac{P_{r}}{\sqrt{W}}\Big{)}\partial_{a}X^{r}+\Big{(}\frac{P_{m}}{\sqrt{W}}\Big{)}\partial_{a}X^{m}\Big{)}d\sigma^{a}\approx 0\ . (9)

In the dynamics of the M2-brane we will only be concerned with the local area preserving diffeomorphisms constraint and the flux constraint.

The dynamics of the M2-brane with worldvolume fluxes is determined by the equations of motion (EOM) and their constraints. The Lagrangian density of the theory can be obtained by performing a usual Legendre transformation of the Hamiltonian density defined as

T=Wc+Λϕ,\mathcal{H}_{T}=\sqrt{W}\mathcal{H}_{c}+\Lambda\phi, (10)

where Λ\Lambda represents a Lagrange multiplier and ϕ=ϕAPD\phi=-\phi_{APD} represents the APD constraint given by equation (8). It contains the Lagrangian of MIM2, previously obtained in [28], plus a constant term associated to the C+C_{+} flux contribution, denoted by KK.

S\displaystyle S =𝑑τ𝑑σ2T=\displaystyle=\int d\tau d\sigma^{2}\mathcal{L}_{T}= (11)
=dτdσ2W[12(DiXm)2+\displaystyle=-\int d\tau d\sigma^{2}\sqrt{W}\Big{[}\frac{1}{2}(D_{i}{X}_{m})^{2}+ 14{Xm,Xn}2\displaystyle\frac{1}{4}\left\{X_{m},X_{n}\right\}^{2}
+14ij2]+K.\displaystyle+\frac{1}{4}\mathcal{F}_{ij}^{2}\Big{]}+K.

The tension has been fixed to T=1T=1. Since the KK contribution gives a non dynamical constant term, added to the action, its equations of motion are equal to those of the M2-brane with central charges. The symplectic field strength, and the symplectic covariant derivative now run over the indices i,j=0,ri,j=0,r, with r=9,10r=9,10. A0=ΛA_{0}=\Lambda the lagrange multiplier and D0=τD_{0}=\partial_{\tau}. As explained in [29, Allen-Andersson-Restuccia-2010] the gauge freedom of the system allows to fix Λ=0\Lambda=0 and then the Lagrangian density function =TK\mathcal{L}=\mathcal{L}_{T}-K reduces to

W1\displaystyle\sqrt{W}^{-1}\mathcal{L} =12[(X˙m)2+(A˙r)2]14{Xm,Xn}2+\displaystyle=\frac{1}{2}[(\dot{X}_{m})^{2}+(\dot{A}_{r})^{2}]-\frac{1}{4}\left\{X_{m},X_{n}\right\}^{2}+ (12)
12(𝒟rXm)214rs2\displaystyle-\frac{1}{2}\left(\mathcal{D}_{r}X_{m}\right)^{2}-\frac{1}{4}\mathcal{F}_{rs}^{2}

The nonlinear system of equations that one has to solve is the following: From (12) we derive the equations of motion for the dynamical fields Xm(τ,σ,ρ)X^{m}(\tau,\sigma,\rho) and Ar(τ,σ,ρ)A_{r}(\tau,\sigma,\rho)

X¨m(τ,σ,ρ)={Xn,{Xn,Xm}}+{Xr,𝒟rXm},\ddot{X}^{m}(\tau,\sigma,\rho)=\left\{X_{n},\left\{X^{n},X^{m}\right\}\right\}+\left\{{X}_{r},\mathcal{D}^{r}{X}^{m}\right\}\ , (13)
A¨r(τ,σ,ρ)={𝒟rXm,Xm}+{rs,Xs},\displaystyle\ddot{A}_{r}(\tau,\sigma,\rho)=\left\{\mathcal{D}_{r}X^{m},X_{m}\right\}+\left\{\mathcal{F}_{rs},{X}^{s}\right\}\ , (14)

subject to satisfy the APD constraint of the theory

𝒟r(A˙r)+{Xm,X˙m}=0forr=9,10,\mathcal{D}_{r}\Big{(}\dot{A}_{r}\Big{)}+\left\{X^{m},\dot{X}^{m}\right\}=0\,\quad\text{for}\quad r=9,10, (15)

and the worldvolume flux condition (2), that restricts the winding numbers allowed for the M2-brane.

In order to find the admissible M2-brane solutions, the system of equations (13), (14), (15) and (2) must be solved. For simplicity we will also assume that one of the scalar fields is constant and we will fix the harmonic sector as in [BRR2005nonperturbative],

X2=constant,Xhr=Rr(nrσ+mrρ),r=9,10.X_{2}=constant,\quad X_{h_{r}}=R_{r}(n_{r}\sigma+m_{r}\rho),\quad r=9,10. (16)

For this ansatz W=1\sqrt{W}=1. There are some differences with the case analysed in [BRR2005nonperturbative]. The main one is associated with the fact that for the supersymmetric M2-brane, the worldvolume flux condition guarantees the discreteness of the spectrum. Other difference due to the topological restriction imposed is that appearance of new single-valued degrees of freedom, ArA_{r} that are components of a symplectic gauge connection 𝔸=dAr\mathbb{A}=dA_{r} in the theory. The EOM for the constant X2X_{2} can be expressed in terms of complex variables ZaZ_{a} with a=1,2,3a=1,2,3 where Za(τ,σ,ρ)=X2a+1+iX2a+2Z_{a}(\tau,\sigma,\rho)=X_{2a+1}+iX_{2a+2}. They are,

Z¨c\displaystyle\ddot{Z}_{c} =12a=13({{Zc,Za},Z¯a}+{{Zc,Z¯a},Za})+\displaystyle=\frac{1}{2}\sum_{a=1}^{3}\left(\left\{\left\{Z_{c},Z_{a}\right\},\overline{Z}_{a}\right\}+\left\{\left\{Z_{c},\overline{Z}_{a}\right\},{Z}_{a}\right\}\right)+ (17)
+r=910{Xr,𝒟rZc},\displaystyle+\sum_{r=9}^{10}\left\{X_{r},\mathcal{D}_{r}Z_{c}\right\},
A¨r\displaystyle\ddot{A}_{r} =12a=13({𝒟rZa,Z¯a}+{𝒟rZ¯a,Za})+\displaystyle=\frac{1}{2}\sum_{a=1}^{3}\left(\left\{\mathcal{D}_{r}Z_{a},\overline{Z}_{a}\right\}+\left\{\mathcal{D}_{r}\overline{Z}_{a},Z_{a}\right\}\right)+ (18)
+s=910{rs,Xs},\displaystyle+\sum_{s=9}^{10}\left\{\mathcal{F}_{rs},X^{s}\right\}\ ,

subject to the APD constraint of the theory:

12c=13({Z˙c,Z¯c}+{Z¯˙c,Zc})𝒟rA˙r=0.\frac{1}{2}\sum_{c=1}^{3}\left(\left\{\dot{Z}_{c},\bar{Z}_{c}\right\}+\left\{\dot{\bar{Z}}_{c},Z_{c}\right\}\right)-\mathcal{D}_{r}\dot{A}_{r}=0\ . (19)

In the following, we assume the Q-ball ansatz for the complex scalars ZaZ_{a}. We will denote this last ansatz as ’Q-ball like’ (QBL) irrespective of its toroidal topology associated to the membrane worldvolume

Za=fa(σ,ρ)eiωaτ,withf(σ,ρ).Z_{a}=f_{a}(\sigma,\rho)e^{i\omega_{a}\tau},\quad\text{with}\quad f(\sigma,\rho)\in\mathbb{R}\,. (20)

The EOM for ZaZ_{a} become

ωc2fc=\displaystyle\omega_{c}^{2}f_{c}= r=910[σ2fc[Rrmr+ρAr]2+ρ2fc(Rrnr+σAr)2]\displaystyle-\sum_{r=9}^{10}\left[\partial_{\sigma}^{2}f_{c}\left[R_{r}m_{r}+\partial_{\rho}A_{r}\right]^{2}+\partial_{\rho}^{2}f_{c}\left(R_{r}n_{r}+\partial_{\sigma}A_{r}\right)^{2}\right] (21)
+r=910σρ2fc[2(Rrnr+σAr)(Rrmr+ρAr)]\displaystyle+\sum_{r=9}^{10}\partial_{\sigma\rho}^{2}f_{c}\left[2\ \left(R_{r}n_{r}+\partial_{\sigma}A_{r}\right)\left(R_{r}m_{r}+\partial_{\rho}A_{r}\right)\right]
+r[σ2Arρfc(Rrmr+ρAr)+ρ2Arσfc(Rrnr+σAr)]\displaystyle+\sum_{r}\left[\partial_{\sigma}^{2}A^{r}\partial_{\rho}f_{c}\left(R_{r}m_{r}+\partial_{\rho}A_{r}\right)+\partial_{\rho}^{2}A^{r}\partial_{\sigma}f_{c}\left(R_{r}n_{r}+\partial_{\sigma}A_{r}\right)\right]
rσρ2Ar[σfc(Rrmr+ρAr)+ρfc(Rrnr+σAr)]\displaystyle-\sum_{r}\partial_{\sigma\rho}^{2}A_{r}\left[\partial_{\sigma}f_{c}\left(R_{r}m_{r}+\partial_{\rho}A_{r}\right)+\partial_{\rho}f_{c}\left(R_{r}n_{r}+\partial_{\sigma}A_{r}\right)\right]
a[σ2fc(ρfa)22σρ2fc[σfaρfa]+ρ2fc(σfa)2]\displaystyle-\sum_{a}\left[\partial_{\sigma}^{2}f_{c}(\partial_{\rho}f_{a})^{2}-2\partial_{\sigma\rho}^{2}f_{c}\left[\partial_{\sigma}f_{a}\partial_{\rho}{f}_{a}\right]+\partial_{\rho}^{2}f_{c}\left(\partial_{\sigma}f_{a}\right)^{2}\right]
+a[ρ2faσfcσfaσρ2fa(σfcρfa+ρfcσfa)].\displaystyle+\sum_{a}\left[\partial_{\rho}^{2}f_{a}\partial_{\sigma}f_{c}\partial_{\sigma}{f}_{a}-\partial_{\sigma\rho}^{2}f_{a}(\partial_{\sigma}f_{c}\partial_{\rho}{f}_{a}+\partial_{\rho}f_{c}\partial_{\sigma}{f}_{a})\right]\ .
+aσ2faρfcρfa.\displaystyle+\sum_{a}\partial_{\sigma}^{2}f_{a}\partial_{\rho}f_{c}\partial_{\rho}{f}_{a}\ .

The EOM for ArA_{r} are:

A¨r=\displaystyle\ddot{A}_{r}= \displaystyle- a=13[σ2faρfa(Rrmr+ρAr)+ρ2faσfa(Rrnr+σAr)]\displaystyle\sum_{a=1}^{3}\left[\partial_{\sigma}^{2}f_{a}\ \partial_{\rho}{f}_{a}\left(R_{r}m_{r}+\partial_{\rho}A_{r}\right)+\partial_{\rho}^{2}f_{a}\ \partial_{\sigma}{f}_{a}\left(R_{r}n_{r}+\partial_{\sigma}A_{r}\right)\right]
+\displaystyle+ a=13[σρ2fa[ρfa(Rrnr+σAr)+σfa(Rrmr+ρAr)]]\displaystyle\sum_{a=1}^{3}\left[\partial_{\sigma\rho}^{2}f_{a}\left[\partial_{\rho}{f}_{a}\left(R_{r}n_{r}+\partial_{\sigma}A_{r}\right)+\partial_{\sigma}{f}_{a}\left(R_{r}m_{r}+\partial_{\rho}A_{r}\right)\right]\right]
+\displaystyle+ a=13[σ2Ar(ρfa)22σρ2Arσfaρfa+ρ2Ar(σfa)2]\displaystyle\sum_{a=1}^{3}\left[\partial_{\sigma}^{2}A_{r}\left(\partial_{\rho}f_{a}\right)^{2}-2\ \partial^{2}_{\sigma\rho}A_{r}\ \partial_{\sigma}f_{a}\partial_{\rho}{f}_{a}+\partial_{\rho}^{2}A_{r}\left(\partial_{\sigma}f_{a}\right)^{2}\right]
+\displaystyle+ σ2Ar(Rsms+ρAs)2+ρ2Ar(Rsns+σAs)2\displaystyle\partial_{\sigma}^{2}A_{r}\left(R_{s}m_{s}+\partial_{\rho}A_{s}\right)^{2}+\partial_{\rho}^{2}A_{r}\left(R_{s}n_{s}+\partial_{\sigma}A_{s}\right)^{2}
\displaystyle- σρ2Ar[2(Rsns+σAs)(Rsms+ρAs)]\displaystyle\partial^{2}_{\sigma\rho}A_{r}\left[2\left(R_{s}n_{s}+\partial_{\sigma}A_{s}\right)\left(R_{s}m_{s}+\partial_{\rho}A_{s}\right)\right]
+\displaystyle+ σρ2As(Rrnr+σAr)(Rsms+ρAs)+\displaystyle\partial_{\sigma\rho}^{2}A_{s}\left(R_{r}n_{r}+\partial_{\sigma}A_{r}\right)\left(R_{s}m_{s}+\partial_{\rho}A_{s}\right)+
+\displaystyle+ σρ2As(Rrmr+ρAr)(Rsns+σAs)\displaystyle\partial_{\sigma\rho}^{2}A_{s}\left(R_{r}m_{r}+\partial_{\rho}A_{r}\right)\left(R_{s}n_{s}+\partial_{\sigma}A_{s}\right)
\displaystyle- σ2As(Rrmr+ρAr)(Rsms+ρAs)+\displaystyle\partial_{\sigma}^{2}A_{s}\left(R_{r}m_{r}+\partial_{\rho}A_{r}\right)\left(R_{s}m_{s}+\partial_{\rho}A_{s}\right)+
\displaystyle- ρ2As(Rrnr+σAr)(Rsns+σAs),\displaystyle\partial_{\rho}^{2}A_{s}\left(R_{r}n_{r}+\partial_{\sigma}A_{r}\right)\left(R_{s}n_{s}+\partial_{\sigma}A_{s}\right)\ ,

where the mixed partial second order derivatives are assumed to satisfy σρ2=ρσ2\partial^{2}_{\sigma\rho}=\partial^{2}_{\rho\sigma}.

The M2-brane is subject to the area preserving diffeomorphism (APD) constraint

0\displaystyle 0 =\displaystyle= r=910Rr(σA˙rmrρA˙rnr)+\displaystyle\sum_{r=9}^{10}R_{r}(\partial_{\sigma}\dot{A}_{r}m_{r}-\partial_{\rho}\dot{A}_{r}n_{r})+ (24)
+r=910(σA˙rρArρA˙rσAr).\displaystyle+\sum_{r=9}^{10}(\partial_{\sigma}\dot{A}_{r}\partial_{\rho}A_{r}-\partial_{\rho}\dot{A}_{r}\partial_{\sigma}A_{r}).

and subject to the flux condition that restricts the allowed winding numbers appearing in XrhX_{rh},

n=det(mrs)=n9m10n10m90withn.n=det(m_{r}^{s})=n_{9}m_{10}-n_{10}m_{9}\neq 0\quad\textrm{with}\quad n\in\mathbb{Z}. (25)

III Analytic String Q-ball-like (QBL) solutions to the M2-brane with fluxes.

In this section we obtain analytic solutions to the system of equations of the M2-brane with C±C_{\pm} fluxes (21)-(25) of the complex scalar field ZaZ_{a}. In this section, for simplicity, we assume the system to be isotropic Za=ZbZ_{a}=Z_{b} for aba\neq b, having fixed X2=0X_{2}=0. This assumption trivialises the symplectic bracket for the complex scalar embedding field and correspomd to strong -like configurations but with cost in energy. The solutions still contain a non-vanishing symplectic gauge field, 𝔸=dA\mathbb{A}=dA with a non-trivial curvature \mathcal{F}. To solve the equations of motion we assume that ArA_{r} and ZZ have the same spatial dependence associated with a single-valued function f(σ,ρ)f(\sigma,\rho). This assumption transforms the equations of motion into a system of three nonlinear coupled differential equations which must also satisfy the APD constraint and the flux constraint, i.e. a non-vanishing central charge.The configurations obtained depend non-trivially on a combination of the spatial variables. In the first two families we illustrate the simplest solutions, in which the dependence on the symplectic gauge cuvature becomes generalized and in the last family we show the existence of localized excitations with the characteristic breathing modes.

Interestingly, due to the central charge condition mass terms are generated, these dominate in the quartic potential along the flat directions and resemble a modified harmonic oscillator. They correspond to the minimum energy of the theory. These contributions are not present in compactifications without induced 2-form worldvolume fluxes. Furthermore, for the solutions obtained, the central charge condition is responsible for generating terms in the equations of motion that allow the presence of breathing modes different from zero.

In order to find analytic solutions to the M2-brane equations, we have inspected the symmetries of the theory. First of all, it is easy to see that there is a symmetry in the choice of the spatial coordinates. Furthermore, the equations are non-linear with triple and quadratic products in the partial derivatives of first and second order. By direct observation of the differential equations one can observe that a family of solutions for a given ff appears when first derivatives are proportionals among them and consequently the same type of relation holds for the second derivatives. Then, a natural choice is to convert the PDE of the system into an ODE-type system by means of imposing that σfρf\partial_{\sigma}f\propto\partial_{\rho}f-. We keep the partial derivative notation, to recall that the system truly depends on two spatial variables. We can then extend these relations into second order derivatives, then we will look for a family of ansätze that satisfy the following relations.

σfa=jaρfawitha=1,2,3.\displaystyle\partial_{\sigma}f_{a}=j_{a}\partial_{\rho}f_{a}\quad\textrm{with}\quad a=1,2,3. (26)

The membrane is closed and therefore all single-valued functions must be periodic on the coordinates (σ,ρ\sigma,\rho) of the torus. For simplicity we are assuming a rectangular 2-torus with the moduli described by the two radii (R9,R10R_{9},R_{10}). For the Ar(σ,ρ,τ)A_{r}(\sigma,\rho,\tau) scalar field, we impose the same type of dependence σArρAr\partial_{\sigma}A_{r}\propto\partial_{\rho}A_{r} at the level of the first derivatives. We impose the following relation of proportionality

σAr=jrρAr,r,s=9,10.\displaystyle\partial_{\sigma}A_{r}=j_{r}\partial_{\rho}A_{r},\,\quad r,s=9,10. (27)

The proportionality in the first derivatives imposes the following second derivatives relation

σ2fa=ja2ρ2fa,σρ2fa=jaρ2fa;\displaystyle\partial_{\sigma}^{2}f_{a}=j_{a}^{2}\partial_{\rho}^{2}f_{a},\quad\partial_{\sigma\rho}^{2}f_{a}=j_{a}\partial_{\rho}^{2}f_{a}; (28)
σ2Ar=jr2ρ2Ar,σρ2Ar=jrρ2Ar.\displaystyle\partial_{\sigma}^{2}A_{r}=j_{r}^{2}\partial_{\rho}^{2}A_{r},\quad\partial_{\sigma\rho}^{2}A_{r}=j_{r}\partial_{\rho}^{2}A_{r}\,.

We can observe that the EOM is greatly simplified under these assumptions. For ZcZ_{c},

ωc2fc=\displaystyle\omega_{c}^{2}f_{c}= r=910[Rr(jcmrnr)]2ρ2fc\displaystyle-\sum_{r=9}^{10}\left[R_{r}(j_{c}m_{r}-n_{r})\right]^{2}\partial_{\rho}^{2}f_{c} (29)
2r=910Rr(jcmrnr)(jcjr)ρArρ2fc+\displaystyle-2\sum_{r=9}^{10}R_{r}(j_{c}m_{r}-n_{r})(j_{c}-j_{r})\partial_{\rho}A_{r}\partial_{\rho}^{2}f_{c}+
r=910(jcjr)2(ρAr)2ρ2fc+\displaystyle-\sum_{r=9}^{10}(j_{c}-j_{r})^{2}(\partial_{\rho}A_{r})^{2}\partial_{\rho}^{2}f_{c}+
+rRr(mrjrnr)(jrjc)ρfcρ2Ar\displaystyle+\sum_{r}R_{r}\left(m_{r}j_{r}-n_{r}\right)(j_{r}-j_{c})\partial_{\rho}f_{c}\partial_{\rho}^{2}A^{r}
a(jcja)2ρ2fc(ρfa)2.\displaystyle-\sum_{a}(j_{c}-j_{a})^{2}\partial_{\rho}^{2}f_{c}(\partial_{\rho}f_{a})^{2}\ .

For ArA_{r},

A¨r=\displaystyle\ddot{A}_{r}= a=13(jrja)2(ρfa)2ρ2Ar+\displaystyle\sum_{a=1}^{3}\left(j_{r}-j_{a}\right)^{2}\left(\partial_{\rho}f_{a}\right)^{2}\partial_{\rho}^{2}A_{r}+ (30)
+\displaystyle+ s=910ρ2Ar[Rs(jrmsns)+(jrjs)ρAr]2\displaystyle\sum_{s=9}^{10}\partial_{\rho}^{2}A_{r}\left[R_{s}\left(j_{r}m_{s}-n_{s}\right)+\left(j_{r}-j_{s}\right)\partial_{\rho}A_{r}\right]^{2}
\displaystyle- s=910ρ2AsRr(jsmrnr)Rs(jsmsns)\displaystyle\sum_{s=9}^{10}\partial_{\rho}^{2}A_{s}R_{r}\left(j_{s}m_{r}-n_{r}\right)R_{s}\left(j_{s}m_{s}-n_{s}\right)
+\displaystyle+ s=910ρ2Ass=910(jsjr)ρArRs(jsmsns),\displaystyle\sum_{s=9}^{10}\partial_{\rho}^{2}A_{s}\sum_{s=9}^{10}\left(j_{s}-j_{r}\right)\partial_{\rho}A_{r}R_{s}\left(j_{s}m_{s}-n_{s}\right)\,,

where rsr\neq s. The APD constraint (24) with the above ansatz (27) becomes:

0\displaystyle 0 =\displaystyle= r=910Rr(jrmrnr)ρA˙r,\displaystyle\sum_{r=9}^{10}R_{r}(j_{r}m_{r}-n_{r})\partial_{\rho}\dot{A}_{r}\,, (31)

and the topological constraint (25) remains invariant. In the following we propose two different forms to obtain a family of solutions.

First family of analytic solutions

We propose the symplectic field 𝔸=dA\mathbb{A}=dA defined on the compact sector of the theory is related with the QBL solution of the noncompact sector, (20) Za=fa(σ,ρ)eiωτZ_{a}=f_{a}(\sigma,\rho)e^{i\omega\tau} through the following ansatz for the real scalar field

Ar=arafa(σ,ρ)+crτ.A_{r}=a_{r}\sum_{a}f_{a}(\sigma,\rho)+c_{r}\tau. (32)

We assume the isotropic case fa=f,af_{a}=f,\,\forall a, and impose the same proportionality constant for partial derivatives of ArA_{r} and faf_{a} such that ja=j=jrj_{a}=j=j_{r} with a=1,2,3a=1,2,3 and r=9,10r=9,10. The two equations of motion for ArA_{r} become again simplified,

0=\displaystyle 0= ar[Rs(jrmsns)]2+\displaystyle a_{r}\left[R_{s}\left(j_{r}m_{s}-n_{s}\right)\right]^{2}+ (33)
as[Rr(jsmrnr)Rs(jsmsns)].\displaystyle-a_{s}\left[R_{r}\left(j_{s}m_{r}-n_{r}\right)R_{s}\left(j_{s}m_{s}-n_{s}\right)\right]\,.

By analyzing both equations we arrive to the following relation,

a102R92(jm9n9)2=a92R102(jm10n10)2,a_{10}^{2}{R_{9}^{2}\left(jm_{9}-n_{9}\right)^{2}}=a_{9}^{2}{R_{10}^{2}\left(jm_{10}-n_{10}\right)}^{2}\,, (34)

which may be used to establish a relation between the amplitudes ara_{r} of the two fields Ar(σ,ρ,τ)A_{r}(\sigma,\rho,\tau) given by

a=a9/a10=±R9(km9ln9)R10(km10ln10).a=a_{9}/a_{10}=\pm\frac{R_{9}(km_{9}-ln_{9})}{R_{10}(km_{10}-ln_{10})}. (35)

The EOM associated with Z becomes reduced to

ω2f+ρ2frRr2(nrjmr)2=0.\omega^{2}f+\partial_{\rho}^{2}f\sum_{r}R_{r}^{2}\left(n_{r}-j\ m_{r}\right)^{2}=0. (36)

It is an eigenvalue-like equation for ff, ρ2f(σ,ρ)=λρ2f(σ,ρ)\partial_{\rho}^{2}f(\sigma,\rho)=-\lambda_{\rho}^{2}f(\sigma,\rho) with

λρ2=ω2rRr2(nrjmr)2.\lambda_{\rho}^{2}=\frac{\omega^{2}}{\sum_{r}R_{r}^{2}\left(n_{r}-j\ m_{r}\right)^{2}}\,. (37)

Observe that in order to guarantee the periodicity condition on ff, then λρ\lambda_{\rho} must be integer which imposes ω\omega to be discrete,

ω2=λρ2rRr2(nrjmr)2,\omega^{2}=\lambda_{\rho}^{2}\sum_{r}R_{r}^{2}\left(n_{r}-j\ m_{r}\right)^{2}\,, (38)

being proportional to the previous solution already identified, whose solution is

f(σ,ρ)=c2(σ)sin(λρρ)+c1(σ)cos(λρρ).f(\sigma,\rho)=c_{2}(\sigma)\sin\left(\lambda_{\rho}\rho\right)+c_{1}(\sigma)\cos\left(\lambda_{\rho}\rho\right)\,. (39)

By considering ρf=j1σf\partial_{\rho}f=j^{-1}\partial_{\sigma}f , derive

σ2f(σ,ρ)=λσ2f(σ,ρ),\partial_{\sigma}^{2}f(\sigma,\rho)=-\lambda_{\sigma}^{2}f(\sigma,\rho)\,, (40)

and the solution is

f(σ,ρ)=c1(ρ)sin(λσσ)+c2(ρ)cos(λσσ).f(\sigma,\rho)=c_{1}(\rho)\sin\left(\lambda_{\sigma}\sigma\right)+c_{2}(\rho)\cos\left(\lambda_{\sigma}\sigma\right)\,. (41)

Hence, the relation between the eingenvalue constants is λσ=jλρ\lambda_{\sigma}=j\lambda_{\rho}. From (39) and (41) it is possible to obtain two general solutions,

f1(σ,ρ)=\displaystyle f_{1}(\sigma,\rho)= c11sin(jλρσ)cos(λρρ)+c21cos(jλρσ)sin(λρρ)\displaystyle c_{1}^{1}\sin\left(j\lambda_{\rho}\sigma\right)\cos\left(\lambda_{\rho}\rho\right)+c_{2}^{1}\cos(j\lambda_{\rho}\sigma)\sin\left(\lambda_{\rho}\rho\right) (42)
f2(σ,ρ)=\displaystyle f_{2}(\sigma,\rho)= c12cos(jλρσ)cos(λρρ)+c22sin(jλρσ)sin(λρρ).\displaystyle c_{1}^{2}\cos\left(j\lambda_{\rho}\sigma\right)\cos\left(\lambda_{\rho}\rho\right)+c_{2}^{2}\sin(j\lambda_{\rho}\sigma)\sin\left(\lambda_{\rho}\rho\right).

In order to find a solution of the complete system (26) we fix the proportionality constant jj\in\mathbb{Q} such that jλρj\lambda_{\rho}\in\mathbb{Z}. For the case when the arbitrary constants cab=rbc_{a}^{b}=r_{b} for different a=1,2a=1,2 , the solutions become reduced to

f1(σ,ρ)=\displaystyle f_{1}(\sigma,\rho)= r1sin(λρ(jσ+ρ))\displaystyle r_{1}\sin\left(\lambda_{\rho}(j\sigma+\rho)\right) (43)
f2(σ,ρ)=\displaystyle f_{2}(\sigma,\rho)= r2cos(λρ(jσ+ρ)).\displaystyle r_{2}\cos\left(\lambda_{\rho}(j\sigma+\rho)\right)\,.

and they become string-like solutions to the membrane equations. For j=k/lj=k/l with k,lk,l\in\mathbb{Z}, and since the problem is symmetric in the worldvolume coordinates (σ,ρ)(\sigma,\rho), the eigenvalues of the function ff can be re-expressed as λσ2=k2λ2\lambda_{\sigma}^{2}=k^{2}\lambda^{2} and λρ2=l2λ2\lambda_{\rho}^{2}=l^{2}\lambda^{2} with λ\lambda integer or rational and r1r_{1} and r2r_{2} arbitrary constants representing the amplitudes

f(σ,ρ)=r1sin(λ(kσ+lρ))+r2cos(λ(kσ+lρ)).f(\sigma,\rho)=r_{1}\sin\left(\lambda(k\sigma+l\rho)\right)+r_{2}\cos\left(\lambda(k\sigma+l\rho)\right)\,. (44)

Finally, the string-like solutions to the equations of motion for the M2-brane with the (32) ansatz are :

Z(σ,ρ)=\displaystyle Z(\sigma,\rho)= (r1sin(λ(kσ+lρ))+r2cos(λ(kσ+lρ)))eiωτ,\displaystyle(r_{1}\sin(\lambda(k\sigma+l\rho))+r_{2}\cos(\lambda(k\sigma+l\rho)))e^{i\omega\tau}\,, (45)
Ar(σ,ρ,τ)=\displaystyle A_{r}(\sigma,\rho,\tau)= ar(r1sin(λ(kσ+lρ))+r2cos(λ(kσ+lρ))+crτ,\displaystyle a_{r}(r_{1}\sin(\lambda(k\sigma+l\rho))+r_{2}\cos(\lambda(k\sigma+l\rho))+c_{r}\tau\,,

when the breathing modes ω\omega have the following dispersion relation (38),

ω=λrRr2(kmrlnr)2,\displaystyle\omega=\lambda\sqrt{\sum_{r}R_{r}^{2}(km_{r}-ln_{r})^{2}}, (46)

and using lλ=kl\lambda=k, and the amplitudes ratio aa of ArA_{r} determined through equation (35). See that for each pair of Fourier modes there is a given relation between the amplitudes associated to the ArA_{r} allowed. See figure 1 that illustrate the behaviour of the ZaZ_{a} solution. The APD constraint is identically satisfied and the central charge condition can be also satisfied for a proper choice of the winding numbers.

\animategraphics[loop,autoplay,width=0.23]5GiffF1/F1C-049 \animategraphics[loop,autoplay,width=0.23]5GiffF1/F1T-049
Figure 1: Maple animated solutions of the first two families analyzed, representing the behaviour of ZaZ_{a}, plotted on cartesian coordinates and on the 2-torus toroidal coordinates, for the following choice of parameters k=2,k=2, l=3,l=3, r1=r2=1,r_{1}=r_{2}=1, n1=m2=1,m1=n2=0,n_{1}=m_{2}=1,m_{1}=n_{2}=0, R1=3,R2=2,R_{1}=3,R_{2}=2, λ=1,\lambda=1, ω=5\omega=5. Use Adobe Acrobat Reader. for a correct visualization.

These solutions trivialize the symplectic bracket, however they do not correspond to trivial solutions. They represent the lowest energy mass operator excitations. These type of solutions, in spite of its simplicity are solitonic due to the non-linear behaviour of the system, as we will see. An important point in order to characterize the solitonic behaviour of these solutions is to analyze the existence of dispersion. It serves to distinguish linear from nonlinear behaviors and the dynamics of the solutions. Standard solitonic solutions are those in which the dispersion relations are compensated by the nonlinearities of the system, resulting in stable solutions. It is important to note that the presence of nonlinearity does not necessarily imply any loss of energy. This is the situation that we analyze here. In all of these solutions the energy is preserved. This fact can be understood since we are describing stable excitations of a single membrane propagating along the M9×T2M_{9}\times T^{2} space time. Since the M2-brane worldvolume has two spatial dimensions, these quantities must be defined in terms of their associated velocity vectors. An interesting property is that the frequency ω\omega takes discrete values for a given moduli. This is not an obvious property. The membrane worldvolume is compact, however this does not automatically imply that the spectrum will be automatically discrete, a very famous counter-example is the 11D compact supermembrane whose mass operator which has continuous spectrum from [0,+)[0,+\infty) [dwln]. In [mpgm8] the authors rigorously prove for the bosonic membrane that the spectrum is discrete from the monopole charge contribution induced by the central charge (C-fluxes) until infinity.

Dispersion

In the following we will see the dispersion relations for the Q-ball-like ansatzs. Following [book:AtmosphericGeoffrey, book:aguero2020Introduccion] the group velocity vector is defined as,

vg=ωk(ωk,ωl)=kω.\textbf{v}_{g}=\frac{\partial\omega}{\partial\textbf{k}}\equiv\left(\frac{\partial\omega}{\partial k},\frac{\partial\omega}{\partial l}\right)=\nabla_{\textbf{k}}\omega\,. (47)

For the dispersion relation (46), the group velocity vector is

vg=λ2ωrRr2((kmrlnr)(mr,nr)\textbf{v}_{g}=\frac{{\lambda}^{2}}{\omega}\sum_{r}R_{r}^{2}\left((km_{r}-ln_{r}\right)(m_{r},-n_{r}) (48)

and its magnitude is given by,

|vg|=λ2ω[(rnrRr2(kmrlnr))2\displaystyle|\textbf{v}_{g}|=\frac{{\lambda}^{2}}{\omega}[\,(\sum_{r}n_{r}R_{r}^{2}\left(km_{r}-ln_{r}\right))^{2} (49)
+(rmrRr2(kmrlnr))2]1/2.\displaystyle+(\sum_{r}m_{r}R_{r}^{2}\left(km_{r}-ln_{r}\right))^{2}\,]^{1/2}\,.

The phase speed is defined as

c(k)=ω|k|,c(\textbf{k})=\frac{\omega}{|\textbf{k}|}, (50)

where k=k2+l2\textbf{k}=\sqrt{k^{2}+l^{2}}. It is possible to define also the phase speed along each of the directions as, (cσ=ωk,cρ=ωl)(c^{\sigma}=\frac{\omega}{k},c^{\rho}=\frac{\omega}{l}). See, that they are not the components of a vector, i.e. c(k)(cbσ)2+(cbρ)2c(\textbf{k})\neq\sqrt{(c_{b}^{\sigma})^{2}+(c_{b}^{\rho})^{2}}.

The phase speed for the dispersion relation (46)is given by

c(k)=λrRr2(kmrlnr)2k2+l2.c(\textbf{k})=\lambda\frac{\sqrt{\sum_{r}R_{r}^{2}(km_{r}-ln_{r})^{2}}}{\sqrt{k^{2}+l^{2}}}\,. (51)
Refer to caption
Refer to caption Refer to caption Refer to caption
ω\omega vfv_{f} vgv_{g}
Figure 2: In these graphics we represent the dispersion relation, phase speed and group speed for a choice of parameters R9=1R_{9}=1, R10=2R_{10}=2, m1=n1=n2=1m_{1}=n_{1}=n_{2}=1, m2=2m_{2}=2 . In the top pictures we represent them in 2D and in the bottom the same quantities in 3D, since both complement visually the description. In the left column we observe dispersion relation ω(k,l)\omega(k,l) (Top contour plot and bottom as 3D plot), the central column represents those values for the phase speed. The right column corresponds to the group speed.

When we impose the group velocity and the phase speed to be equal, generically the system exhibit dispersion. However, it also admits solutions without dispersion and non vanishing central charge. To illustrate it we give an example: for a choice of j=nr/mrj=n_{r}/m_{r} with a fixed rr, and winding charges mr=ms;nr=nsm_{r}=m_{s};n_{r}=-n_{s} with rsr\neq s the two velocities coincide and the system does not exhibit dispersion. There are other more particular winding charge configurations that also satisfy the identity of velocities for the jj considered, like for example n1=3,m1=4,n=2=12,m2=16n_{1}=3,m_{1}=4,n=2=12,m_{2}=16 with non-vanishing central charge.

Second family of solutions

Since we are interested in allowing a more general dependence on time of the real scalar fields ArA_{r}, in order to obtain a generic field strength \mathcal{F}, we assume the following ansatz

Za=(sin(λ(kσ+lρ))+cos(λ(kσ+lρ)))aaeiτω,Z_{a}=\left(\sin(\lambda(k\sigma+l\rho))+\cos(\lambda(k\sigma+l\rho))\right)a_{a}e^{i\tau\omega}\,, (52)
Ar=ar(asin(λ(kσ+lρ)+τΩ)+bcos(λ(kσ+lρ)+τΩ)).A_{r}=a_{r}\left(a\sin(\lambda(k\sigma+l\rho)+\tau\Omega)+b\cos(\lambda(k\sigma+l\rho)+\tau\Omega)\right)\,. (53)

See figure 1, that illustrates the ZaZ_{a} string QBL over the 2-torus. Proceeding as before, we obtain from the EOM for ZZ, the frequency ω\omega is not affected by the modification of the ArA_{r} ansatz. This is due to the fact that in the isotropic case considered in this section, both EOM decouple and its value is,

ω2=λ2rRr2(kmrlnr)2,\omega^{2}=\lambda^{2}\sum_{r}R_{r}^{2}\left(km_{r}-ln_{r}\right)^{2}\,, (54)

and from the EOM for ArA_{r}, one obtains a different frequency Ω\Omega. It depends on the moduli of the torus, the winding numbers and the amplitudes of the ArA_{r} field,

Ω2=\displaystyle\Omega^{2}= λ2(kmslns)2Rs2\displaystyle\lambda^{2}\left(km_{s}-ln_{s}\right)^{2}R_{s}^{2} (55)
λ2asar(kmrlnr)(kmslns)RrRs,\displaystyle-\lambda^{2}\frac{a_{s}}{a_{r}}\left(km_{r}-ln_{r}\right)\left(km_{s}-ln_{s}\right)R_{r}R_{s}\,,

with rs,r=1,2r\neq s,r=1,2. In distinction with the previous case analyzed, the APD constraint is not automatically satisfied and leaves the following restriction,

arRr(kmrlnr)+asRs(kmslns)=0.a_{r}R_{r}\left(km_{r}-ln_{r}\right)+a_{s}R_{s}\left(km_{s}-ln_{s}\right)=0\,. (56)

The constraint can be used to fix the relation a=as/asa=a_{s}/a_{s}. Imposing it in (55) the frequency Ω\Omega becomes

Ω2=2λ2Rs2(kmslns)2,\Omega^{2}=2\lambda^{2}R_{s}^{2}\left(km_{s}-ln_{s}\right)^{2}\,, (57)

for ss fixed. By imposing the constraint on (54),

ω2=λ2Rs2(kmslns)2(1+a2),\omega^{2}=\lambda^{2}R_{s}^{2}\left(km_{s}-ln_{s}\right)^{2}(1+a^{2})\,, (58)

with no sum is intended. Due to the APD constraint and for this particular ansatz, both frequencies become related by the amplitudes ratio of the ArA_{r} field,

Ω2=2ω21+a2,\Omega^{2}=\frac{2\omega^{2}}{1+a^{2}}\,, (59)

This family of solutions allows for a nontrivial time dependence of ArA_{r}, resulting in a general symplectic curvature \mathcal{F}. Under the ansatz considered, the frequencies of the ArA_{r} and ZZ solutions become related, for a=1a=1, both determine a unique excitation for a pair (k,l)(k,l). In this case, for a given moduli parametrizing the compact dimensions, there is a unique pair of Fourier modes for each set of winding numbers with a given central charge. The APD constraint represents the main restriction to the allowed solutions.

Dispersion

In order to analyze the existence of solutions with vanishing dispersion relations, let us consider the case with a=1a=1. It implies that the two components of the ArA_{r} become equal and that Ω=ω\Omega=\omega. In distinction with the first family case, the relation between the Fourier modes of the non-compact sector j=k/lj=k/l gets restricted by the APD constraint to a unique value, fixed by the topological content, i.e. a fixed moduli and winding charges of the compact sector,

j=k/l=(Rrnr+Rsns)(Rrmr+Rsms).j=k/l=\frac{(R_{r}n_{r}+R_{s}n_{s})}{(R_{r}m_{r}+R_{s}m_{s})}\,. (60)

The integer character of the Fourier modes forces, on general grounds, the radii to be integer.

Now, the velocity group associated to the dispersion relation is given by,

vg=\displaystyle\textbf{v}_{g}= ωkλRr2(mr,nr);\displaystyle\frac{\partial\omega}{\partial\textbf{k}}\equiv\lambda R_{r}\sqrt{2}\left(m_{r}\ ,-\ n_{r}\right);\quad\Rightarrow (61)
|vg|=\displaystyle|\textbf{v}_{g}|= 2λRr(mr2+nr2).\displaystyle\sqrt{2}\lambda R_{r}\sqrt{\left(m_{r}^{2}+n_{r}^{2}\right)}\ .

It has has a constant magnitude and the index rr is fixed, there is no summation.

Their phase speed is given by

c(k)\displaystyle c(\textbf{k}) =2λRr(kmrlnr)k2+l2.\displaystyle=\sqrt{2}\lambda R_{r}\frac{\left(km_{r}-ln_{r}\right)}{\sqrt{k^{2}+l^{2}}}\,. (62)

It is straightforward to see, that their phase speed along each of the directions verify jcσ=cρjc_{\sigma}=c_{\rho}. We can observe that although the system presents generically dispersion, there exists areas which are dispersion-free. In order to see if there exists points in the phase space moduli free of dispersion, we impose c(k)=|vg|,c(\textbf{k})=|\textbf{v}_{g}|,

(kmrlnr)=k2+l2(mr2+nr2),(km_{r}-ln_{r})=\sqrt{k^{2}+l^{2}}\sqrt{\left(m_{r}^{2}+n_{r}^{2}\right)}\,, (63)

which implies,

l=nrmrk.l=-\frac{n_{r}}{m_{r}}k\ . (64)

for rr fixed. Hence, for central charge different from zero the system in general exhibits dispersion but it also allows solutions with no dispersion satisfying (64).

A particular subfamily of solutions corresponds to those that appear by imposing 2ff\partial^{2}f\propto f. This solution subfamily corresponds to the λ=1\lambda=1 choice. The main constraints on this first family of solutions are imposed by the periodicity condition required since they propagate on the world volume of a compact membrane. Solving this system of equations on a flat plane would admit much more general solutions than those considered here, for the same given ansatz.

We have found two string-like families with a non-vanishing cost in energy solutions that satisfy the Q-ball-ansatz to the nonlinear system. The non-linearity is present in spite of the vanishing symplectic bracket due to the ansatz choice of the symplectic field 𝔸\mathbb{A} and to the complex scalar field ZZ which become determined in terms of a same real single function f(σ,ρ)f(\sigma,\rho). The system is highly simplified but still non-linear, and it admits solutions sinusoidal type. It is important to mention that these solutions are not present for the case without worldvolume fluxes, they exist due to them. They correspond to excitations with minimal energy. In addition, as we can observe when represented over the 2-torus, they generate patterns associated with twists over the surface. They admit solutions with and without dispersion.

Localized dynamic QBL solutions over the worldvolume

Now, we would like to show that the M2-brane with worldvolume fluxes also admits a family of string-like solutions that correspond to localized spinning Q-ball-like solutions propagating over the worldvolume that rotate and have the characteristic Q-ball breathing modes. By localization we mean a localized pulse over the torus worldvolume defined by f(σ,ρ,τ)f(\sigma,\rho,\tau) for a fixed time τ0\tau_{0}. We illustrate a particular example but there can be found many more. The solutions evolve in time as shown in figure 3. It also exhibits in the phase the characteristic dependence of a spinning Q-ball. This solution may be also seen as the Lorentz of a boost of a static QBL solution.
In order to find these solutions we generalize the ansatz to the following one:

Za=fa(σ,ρ,τ)ei(τω+kσ+lρ)Z_{a}=f_{a}(\sigma,\rho,\tau)e^{i(\tau\omega+k\sigma+l\rho)} (65)

with faf_{a} being real, acting in the general equations given by ( 17,18,19) and subject to the topological constraint (25) . An admissible string-like family of localized solutions of solitonic profile is the following one,

Za=fa(σ,ρ,τ)eiβ=raArcotan(aebsin(β)+bcos(β))eiβ,Z_{a}=f_{a}(\sigma,\rho,\tau)e^{i\beta}=r_{a}Arcotan(ae^{b\sin(\beta)+b\cos(\beta)})e^{i\beta}\,, (66)
Ar=ar(sin(kσ+lρ+τΩ)+cos(kσ+lρ+τΩ)).A_{r}=a_{r}\left(\sin(k\sigma+l\rho+\tau\Omega)+\cos(k\sigma+l\rho+\tau\Omega)\right)\,. (67)

where

β=kσ+lρ+ωτ\beta=k\sigma+l\rho+\omega\tau (68)

This string-like solution correspond to a dynamic rotating Q-ball with non-vanishing breathing modes, that propagates along the worldvolume. For a visual representation, see Figures 3 and 4. In order to illustrate better the excitation, only the dependence of the pulse on one of the variables is shown.

\animategraphics[loop,autoplay,width=0.25]5giffLocalSolution-Rho2D/LocalRho2d099
\animategraphics[loop,autoplay,width=0.25]5giffLocalSolution-Rho2D/LocalRho3d-099
\animategraphics[loop,autoplay,width=0.25]12.5giffLocalSolution-Rho2D/LocalRhoToro-0249
Figure 3: Using the following coefficients n1=1n2=0m1=0m2=1R1=1R2=2n=1,k=0l=1ω=1ϕ=(1/4)π,ra=1,a=0.0001andb=20n_{1}=1\ n_{2}\ =0\,m_{1}=0\,m_{2}=1\,R_{1}=1\,R_{2}=2\,n=1,\ k=0\,l=1\,\omega=1\,\phi=-(1/4)\pi,\ r_{a}=1,a=0.0001\ and\ b=20
\animategraphics[loop,autoplay,width=0.25]5GiffLocalSolution/Local2d099
\animategraphics[loop,autoplay,width=0.25]12.5GiffLocalSolution/LocalToro0249
Figure 4: The Cartesian coordinates solution is displayed at the top, with the solution plotted on the torus shown at the bottom. Using the following coefficients n1=1n2=0m1=0m2=1R1=1R2=2n=1k=1l=1ω=5ϕ=(1/4)π,ra=1,a=0.0001andb=20n_{1}=1\ n_{2}\ =0\,m_{1}=0\,m_{2}=1\,R_{1}=1\,R_{2}=2\,n=1\ k=1\,l=1\,\omega=\sqrt{5}\,\phi=-(1/4)\pi,\ r_{a}=1,a=0.0001\ and\ b=20

It’s worth noting that the frequency remains unchanged compared to the previously discussed solutions. This is due to the decoupling of equations of motion (EOM) in the isotropic cases considered in this section. By following the same analysis done for the second family of solutions the frequencies ω\omega and Ω\Omega can be obtained and they correspond to the same values previously obtained, i.e. for a=1a=1 and fixed ss.

Ω2=2Rs2(kmslns)2=ω2,\Omega^{2}=2R_{s}^{2}\left(km_{s}-ln_{s}\right)^{2}=\omega^{2}\,, (69)

The dispersion relation is the same as the one previously analyzed, hence also admits dispersion-free solutions by imposing nsn_{s} multiple of msm_{s} for a fixed ss.

IV Numerical Membrane Q-ball-like solutions

In this section we consider the non-isotropic case with the nonlinearities associated with the complex scalars ZaZ_{a} present, and not just induced because of the coupling to the ArA_{r} scalar. They still satisfy the ansatz σf(σ,ρ)ρf(σ,ρ)\partial_{\sigma}f(\sigma,\rho)\propto\partial_{\rho}f(\sigma,\rho). These solutions are numerical. For simplicity, we will assume a vanishing symplectic field 𝔸\mathbb{A}, i.e. ArA_{r} is assumed to be constant or with a linear dependence only on τ\tau, focusing exclusively in the role of the scalars ZaZ_{a}.

IV.1 Q-ball-like ansatz for the isotropic case

As a warm-up exercise to test the numerical computation error, in the isotropic case, we compare it with previously obtained analytical results. For the isotropic case and constant ArA_{r} we consider the system of equations given by (138) and (139) subject to the central charge condition 2 discussed in Section 4. The Q-ball ansatz automatically satisfy the Diffeomorphism area preserving constraint. The solutions of the equation (138) was analyzed in [SpinningSolutionsBosonicM2-brane2022] and a discrete set of frequencies was obtained. From a numerical point of view the M2-brane system of partial differential equations that we consider is: i) Overdetermined. In the sense that the unknown function must satisfy two different differential equations ii) Nonlinear. Although only to the second equation, where in this case the non-linearities are represented by products of different derivatives (in order and variable) of the unknown function and iii)Rare. The most striking feature is that this form of nonlinearity is unusual and, as far to our knowledge, there are no reported numerical methods for solving it. Neither in the finite dimensional (ODEs) or infinite dimensional (PDEs) case. The three steps algorithm we will follow in order to a find numerical solution to the (138-139) system will be

  • i.

    Set the boundary condition. In principle boundary (Γ\Gamma) condition can be fixed as: Dirichlet or periodic, though in this case we will restrict ourselves to periodic boundary conditions on Γ\Gamma:

    f(0,ρ)=f(2π,ρ),f(σ,0)=f(σ,2π).\displaystyle f(0,\rho)=f(2\pi,\rho),\quad f(\sigma,0)=f(\sigma,2\pi)\,. (70)
  • ii.

    Solve (138 ) as an eigenvalue problem:

    pφn(σ,ρ)=ωnφn(σ,ρ).{\cal L}_{p}\varphi_{n}(\sigma,\rho)=\omega_{n}~{}\varphi_{n}(\sigma,\rho). (71)
  • iii.

    Write the solution of (139) as

    fa(σ,ρ)=n=0γnφn(σ,ρ).f_{a}(\sigma,\rho)=\sum_{n=0}^{\infty}\gamma_{n}\varphi_{n}(\sigma,\rho). (72)

    and solve an optimization problem to find the best set of γn\gamma_{n}, so that (72) is solution of (139).

IV.1.1 Finite differences approach for the solution

The system (138)-(139) can be re-written using centered Euler scheme:

𝒟σφn\displaystyle{\cal D}_{\sigma}\varphi_{n} =\displaystyle= φni+1,jφni1,j2Δσ,\displaystyle\frac{\varphi_{n}^{i+1,j}-\varphi_{n}^{i-1,j}}{2\Delta\sigma}, (73)
𝒟ρφn\displaystyle{\cal D}_{\rho}\varphi_{n} =\displaystyle= φni,j+1φni,j12Δρ,\displaystyle\frac{\varphi_{n}^{i,j+1}-\varphi_{n}^{i,j-1}}{2\Delta\rho}, (74)
𝒟σ2φn\displaystyle{\cal D}^{2}_{\sigma}\varphi_{n} =\displaystyle= φni+1,j2φni,j+φni1,jΔσ2\displaystyle\frac{\varphi_{n}^{i+1,j}-2\varphi_{n}^{i,j}+\varphi_{n}^{i-1,j}}{\Delta\sigma^{2}} (75)
𝒟ρ2φn\displaystyle{\cal D}^{2}_{\rho}\varphi_{n} =\displaystyle= φni,j+12φni,j+φni,j1Δρ2,\displaystyle\frac{\varphi_{n}^{i,j+1}-2\varphi_{n}^{i,j}+\varphi_{n}^{i,j-1}}{\Delta\rho^{2}}, (76)

with the mixed second derivative given:

𝒟σ,ρ2φn=φni+1,j+1φni1,j+1φni+1,j1+φni1,j14ΔσΔρ{\cal D}^{2}_{\sigma,\rho}\varphi_{n}=\frac{\varphi_{n}^{i+1,j+1}-\varphi_{n}^{i-1,j+1}-\varphi_{n}^{i+1,j-1}+\varphi_{n}^{i-1,j-1}}{4\Delta\sigma\Delta\rho} (77)

It is easy to verify that:

𝒟ρρ,𝒟σσ,𝒟σ2σ2,\displaystyle{\cal D}_{\rho}\approx{\partial}_{\rho},\quad{\cal D}_{\sigma}\approx{\partial}_{\sigma},\quad{\cal D}^{2}_{\sigma}\approx{\partial}^{2}_{\sigma}, (78)
𝒟σ,ρ2σ,ρ2,𝒟ρ2ρ2.\displaystyle{\cal D}^{2}_{\sigma,\rho}\approx{\partial}^{2}_{\sigma,\rho},\quad{\cal D}^{2}_{\rho}\approx{\partial}^{2}_{\rho}\,.

Since equation (138) is linear, it can be solved relatively easily using finite differences or built-in functions from Mathematica or Python. In the case of the second equation we must use a different methodology due to its intrinsic nonlinearities.

IV.1.2 A finite differences representation for the nonlinear equation

Re-writing the equation (139) in terms of finite differences become,

(mrσ2n=0γnφn(σ,ρ)nrσρ2n=0γnφn(σ,ρ))ρn=0γnφn(σ,ρ)+\displaystyle\left(m_{r}\partial_{\sigma}^{2}\sum_{n=0}^{\infty}\gamma_{n}\varphi_{n}(\sigma,\rho)-n_{r}\partial^{2}_{\sigma\rho}\sum_{n=0}^{\infty}\gamma_{n}\varphi_{n}(\sigma,\rho)\right)\partial_{\rho}\sum_{n=0}^{\infty}\gamma_{n}\varphi_{n}(\sigma,\rho)+ (79)
(mrρσ2n=0γnφn(σ,ρ)nrρ2n=0γnφn(σ,ρ))σn=0γnφn(σ,ρ)=0.\displaystyle-\left(m_{r}\partial^{2}_{\rho\sigma}\sum_{n=0}^{\infty}\gamma_{n}\varphi_{n}(\sigma,\rho)-n_{r}\partial_{\rho}^{2}\sum_{n=0}^{\infty}\gamma_{n}\varphi_{n}(\sigma,\rho)\right)\partial_{\sigma}\sum_{n=0}^{\infty}\gamma_{n}\varphi_{n}(\sigma,\rho)=0.
(mrn=0γnσ2φn(σ,ρ)nrn=0γnσρ2φn(σ,ρ))n=0γnρφn(σ,ρ)\displaystyle\left(m_{r}\sum_{n=0}^{\infty}\gamma_{n}\partial_{\sigma}^{2}\varphi_{n}(\sigma,\rho)-n_{r}\sum_{n=0}^{\infty}\gamma_{n}\partial^{2}_{\sigma\rho}\varphi_{n}(\sigma,\rho)\right)\sum_{n=0}^{\infty}\gamma_{n}\partial_{\rho}\varphi_{n}(\sigma,\rho) \displaystyle-
(mrn=0γnρσ2φn(σ,ρ)nrn=0γnρ2φn(σ,ρ))n=0γnσφn(σ,ρ)\displaystyle\left(m_{r}\sum_{n=0}^{\infty}\gamma_{n}\partial^{2}_{\rho\sigma}\varphi_{n}(\sigma,\rho)-n_{r}\sum_{n=0}^{\infty}\gamma_{n}\partial_{\rho}^{2}\varphi_{n}(\sigma,\rho)\right)\sum_{n=0}^{\infty}\gamma_{n}\partial_{\sigma}\varphi_{n}(\sigma,\rho) =\displaystyle= 0.\displaystyle 0.
[n=0γn(mrσ2φn(σ,ρ)nrσρ2φn(σ,ρ))].n=0γnρφn(σ,ρ)\displaystyle\left[\sum_{n=0}^{\infty}\gamma_{n}\left(m_{r}\partial_{\sigma}^{2}\varphi_{n}(\sigma,\rho)-n_{r}\partial^{2}_{\sigma\rho}\varphi_{n}(\sigma,\rho)\right)\right].\sum_{n=0}^{\infty}\gamma_{n}\partial_{\rho}\varphi_{n}(\sigma,\rho) \displaystyle-
[n=0γn(mrρσ2φn(σ,ρ)nrρ2φn(σ,ρ))].n=0γnσφn(σ,ρ)\displaystyle\left[\sum_{n=0}^{\infty}\gamma_{n}\left(m_{r}\partial^{2}_{\rho\sigma}\varphi_{n}(\sigma,\rho)-n_{r}\partial_{\rho}^{2}\varphi_{n}(\sigma,\rho)\right)\right].\sum_{n=0}^{\infty}\gamma_{n}\partial_{\sigma}\varphi_{n}(\sigma,\rho) =\displaystyle= 0.\displaystyle 0.

This can be re-expressed in a more compact form as:

n=0γn(AnBn)n=0ωnCn+\displaystyle\sum_{n=0}^{\infty}\gamma_{n}\left(A_{n}-B_{n}\right)\cdot\sum_{n=0}^{\infty}\omega_{n}C_{n}+
n=0γn(BnDn)n=0ωnEn\displaystyle-\sum_{n=0}^{\infty}\gamma_{n}\left(B_{n}-D_{n}\right)\cdot\sum_{n=0}^{\infty}\omega_{n}E_{n} =\displaystyle= 0.\displaystyle 0. (82)

where:

An\displaystyle A_{n} =\displaystyle= mrσ2φn(σ,ρ),\displaystyle m_{r}\partial_{\sigma}^{2}\varphi_{n}(\sigma,\rho),
Bn\displaystyle B_{n} =\displaystyle= nrσρ2φn(σ,ρ),\displaystyle n_{r}\partial^{2}_{\sigma\rho}\varphi_{n}(\sigma,\rho), (83)
Cn\displaystyle C_{n} =\displaystyle= ρφn(σ,ρ)\displaystyle\partial_{\rho}\varphi_{n}(\sigma,\rho)

and

Dn=ρ2φn(σ,ρ),En=σφn(σ,ρ).D_{n}=\partial_{\rho}^{2}\varphi_{n}(\sigma,\rho),\quad E_{n}=\partial_{\sigma}\varphi_{n}(\sigma,\rho)\,. (84)

Using the Cauchy product in (82), we obtain:

n=0m=0nγn(AnBn)γnmCnm\displaystyle\sum_{n=0}^{\infty}\sum_{m=0}^{n}\gamma_{n}\left(A_{n}-B_{n}\right)~{}\odot\gamma_{n-m}C_{n-m}-
n=0m=0nγn(BnDn)γnmEnm=0.\displaystyle\sum_{n=0}^{\infty}\sum_{m=0}^{n}\gamma_{n}\left(B_{n}-D_{n}\right)~{}\odot\gamma_{n-m}E_{n-m}=0\,. (85)
n=0γn(m=0nγnmSn,m)=0,\sum_{n=0}^{\infty}~{}\gamma_{n}\left(\sum_{m=0}^{n}\gamma_{n-m}S_{n,m}\right)=0, (86)

with,

Sn,m=(AnBn)Cnm(BnDn)Enm,S_{n,m}=\left(A_{n}-B_{n}\right)\odot C_{n-m}-\left(B_{n}-D_{n}\right)\odot E_{n-m}\,, (87)

and \odot represents the pointwise product. Thus, the solution of the problem can be reduced to find a solution for this nonlinear algebraic equation with matrix coefficients and γn\gamma_{n} as unknowns.

Refer to caption
Refer to caption
Refer to caption
Refer to caption
Refer to caption
Refer to caption
Refer to caption
Refer to caption
Figure 5: Approximate solution of (138)-(139) and approximation error.

We define the error function in the following way,

E=(mrσ2fa(ρij,σij)nrσρ2fa(ρij,σij))ρfa(ρij,σij)\displaystyle E=\left(m_{r}\partial_{\sigma}^{2}f_{a}(\rho_{ij},\sigma_{ij})-n_{r}\partial^{2}_{\sigma\rho}f_{a}(\rho_{ij},\sigma_{ij})\right)\partial_{\rho}f_{a}(\rho_{ij},\sigma_{ij})- (88)
(mrρσ2fa(ρij,σij)nrρ2fa(ρij,σij))σfa(ρij,σij),\displaystyle\left(m_{r}\partial^{2}_{\rho\sigma}f_{a}(\rho_{ij},\sigma_{ij})-n_{r}\partial_{\rho}^{2}f_{a}(\rho_{ij},\sigma_{ij})\right)\partial_{\sigma}f_{a}(\rho_{ij},\sigma_{ij})\,,

with EE(ρij,σij)E\equiv E(\rho_{ij},\sigma_{ij}). Figura 5 shows graphs of the possible solutions (note that the numerical scheme allows for many, unless the number of bases equals the number of lattices squared) and the errors made in those approximations at each lattice point.

We obtain two kinds of solutions: some with large amplitudes and others with small amplitudes. This may be because the parameters nn, mm, and RR may be in very different ranges.

IV.2 Membrane Q-ball-like solution: The non-isotropic case

In section 3. we have obtained the first exact analytical families of solutions for a Q-ball ansatz of the complex scalar field. They correspond to isotropic solutions and with the nonlinear behaviour encoded in the ArA_{r} field which is expressed in terms of ff, and hence makes the complex scalar field ZZ to be nontrivial. However, the family of solutions obtained trivializes the non-isotropic contribution of the brackets to the ZZ EOM. In order to capture their contribution, in this section we consider the non-isotropic case in the absence of a symplectic gauge field 𝔸\mathbb{A}.

Imposing Ar=ConstantA_{r}=Constant in the E.O.M (29) they become the following nonlinear but simplified system of equations:

ωc2fc=r=910[Rr(jcmrnr)]2ρ2fc+\displaystyle\omega_{c}^{2}f_{c}=-\sum_{r=9}^{10}\left[R_{r}(j_{c}m_{r}-n_{r})\right]^{2}\partial_{\rho}^{2}f_{c}+
a(jcja)2ρ2fc(ρfa)2.\displaystyle-\sum_{a}(j_{c}-j_{a})^{2}\partial_{\rho}^{2}f_{c}(\partial_{\rho}f_{a})^{2}\ . (89)

with a,c=1,2,3a,c=1,2,3 and r,s=1,2r,s=1,2. In order to simplify the notation, let us define the constants κa\kappa_{a},

κa=r=910[Rr(jamrnr)]2,\kappa_{a}=\sum_{r=9}^{10}\left[R_{r}(j_{a}m_{r}-n_{r})\right]^{2}\,, (90)

in addition to the matrices λ\lambda and μ\mu, with elements:

λc,a=(jcja)2,μr,a=[Rr(jamrnr)]2,\lambda_{c,a}=(j_{c}-j_{a})^{2},\qquad\mu_{r,a}=\left[R_{r}\left(j_{a}m_{r}-n_{r}\right)\right]^{2}\,, (91)

with rr, cc y aa, defined as before. With this notation, the above equations can be written as:

κ1ρ2f1ρ2f1aλ1,a(ρfa)2\displaystyle-\kappa_{1}~{}\partial_{\rho}^{2}f_{1}-\partial_{\rho}^{2}f_{1}~{}\sum_{a}\lambda_{1,a}(\partial_{\rho}f_{a})^{2} =\displaystyle= ω12f1,\displaystyle\omega_{1}^{2}f_{1}\,,
κ2ρ2f2ρ2f2aλ2,a(ρfa)2\displaystyle-\kappa_{2}~{}\partial_{\rho}^{2}f_{2}-\partial_{\rho}^{2}f_{2}~{}\sum_{a}\lambda_{2,a}(\partial_{\rho}f_{a})^{2} =\displaystyle= ω22f2,\displaystyle\omega_{2}^{2}f_{2}\,,
κ3ρ2f3ρ2f3aλ3,a(ρfa)2\displaystyle-\kappa_{3}~{}\partial_{\rho}^{2}f_{3}-\partial_{\rho}^{2}f_{3}~{}\sum_{a}\lambda_{3,a}(\partial_{\rho}f_{a})^{2} =\displaystyle= ω32f3.\displaystyle\omega_{3}^{2}f_{3}\,. (92)

IV.2.1 A numerical approach to the solution

In order to write our problem in a convenient way for its numerical integration, x=𝐟(x)x^{\prime}={\mathbf{f}}(x) con x3x\in\Re^{3}, let’s start by defining ρfa=ga\partial_{\rho}f_{a}=g_{a}, so that the equations (IV.2) can be written as:

ρf1=g1;κ11ρg1ρg1aλ1,aga2=ω12f1,\displaystyle\partial_{\rho}f_{1}=g_{1};\quad-\kappa_{11}~{}\partial_{\rho}g_{1}-\partial_{\rho}g_{1}~{}\sum_{a}\lambda_{1,a}g_{a}^{2}=\omega_{1}^{2}f_{1}\,,
ρf2=g2;κ22ρg2ρg2aλ2,aga2=ω22f2,\displaystyle\partial_{\rho}f_{2}=g_{2};\quad-\kappa_{22}~{}\partial_{\rho}g_{2}-\partial_{\rho}g_{2}~{}\sum_{a}\lambda_{2,a}g_{a}^{2}=\omega_{2}^{2}f_{2}\,,
ρf3=g3;κ33ρg3ρg3aλ3,aga2=ω32f3.\displaystyle\partial_{\rho}f_{3}=g_{3};\quad-\kappa_{33}~{}\partial_{\rho}g_{3}-\partial_{\rho}g_{3}~{}\sum_{a}\lambda_{3,a}g_{a}^{2}=\omega_{3}^{2}f_{3}\,. (93)

So, if we include the boundary conditions, our problem will look like:

ρfb=gb;ρgb=ωb2fbκbb+aλb,aga2,\displaystyle\partial_{\rho}f_{b}=g_{b};\qquad\partial_{\rho}g_{b}=\frac{\omega_{b}^{2}f_{b}}{\kappa_{bb}~{}+~{}\sum_{a}\lambda_{b,a}g_{a}^{2}}\,, (94)

with fi(0)=fi(2π)i=1,2,3f_{i}(0)=f_{i}(2\pi)\quad i=1,2,3. Thus, our problem is to find a function:

x(ρ)=(f1(ρ),g1(ρ),f2(ρ),g2(ρ),f3(ρ),g3(ρ)),x(\rho)=(f_{1}(\rho),g_{1}(\rho),f_{2}(\rho),g_{2}(\rho),f_{3}(\rho),g_{3}(\rho)), (95)

that satisfies:

x(ρ)ρ=𝐟(x),\frac{\partial x(\rho)}{\partial\rho}={\mathbf{f}}(x)\,, (96)

where 𝐟{\mathbf{f}} is a vector-valued function given by:

𝐟(x)=(g1(ρ)ω12f1(ρ)κ1+aλ1,aga2(ρ)g2(ρ)ω22f2(ρ)κ2+aλ2,aga2(ρ)g3(ρ)ω32f3(ρ)κ3+aλ3,aga2(ρ)),{\mathbf{f}}(x)=\left(\begin{array}[]{c}g_{1}(\rho)\\ \\ \frac{\omega_{1}^{2}f_{1}(\rho)}{\kappa_{1}~{}+~{}\sum_{a}\lambda_{1,a}g_{a}^{2}(\rho)}\\ \\ g_{2}(\rho)\\ \\ \frac{\omega_{2}^{2}f_{2}(\rho)}{\kappa_{2}~{}+~{}\sum_{a}\lambda_{2,a}g_{a}^{2}(\rho)}\\ \\ g_{3}(\rho)\\ \\ \frac{\omega_{3}^{2}f_{3}(\rho)}{\kappa_{3}~{}+~{}\sum_{a}\lambda_{3,a}g_{a}^{2}(\rho)}\end{array}\right), (97)

with the boundary condition x(0)=x(2π)x(0)=x(2\pi).

IV.2.2 The proposed solution

Since this problem can be assimilated to the class of ”two point boundary value problems”, we propose to use a customized version of the ”shooting method”, to obtain numerical solutions. In these methods, since only part of the initial condition (fi(0)f_{i}(0)) is known, the remaining part is assumed to be known and the initial value problem is solved, varying this last part, until the solution complies with the remaining boundary condition.

A more or less efficient way to implement this idea, is to find the roots (g1,g2,g3g_{1}^{*},g_{2}^{*},g_{3}^{*}) of the function:

F(x(0),ρ)=Φ(x(0),2π)x(2π),F(x(0),\rho)=\Phi(x(0),2\pi)-x(2\pi)\,, (98)

where Φ(x(0),ρ)\Phi(x(0),\rho) is the solution of the initial value problem:

x(ρ)ρ=𝐟(x),\frac{\partial x(\rho)}{\partial\rho}={\mathbf{f}}(x)\,, (99)

with initial conditions

x(0)=(f1(0),g1(0),f2(0),g2(0),f3(0),g3(0))x(0)=(f_{1}(0),g_{1}(0),f_{2}(0),g_{2}(0),f_{3}(0),g_{3}(0))

.

The most popular way to solve the problem (98), is using Newton’s method. This type of methods can be extended to the case of problems of the type that concern us (”two point boundary values”) in many ways. One of them, given in [Aprille], states that:

xi+1(0)=xi(0)+(JΦI)1[Φ(xi(0),2π)xi(0)],x^{i+1}(0)=x^{i}(0)+(J_{\Phi}-I)^{-1}\left[\Phi(x^{i}(0),2\pi)-x^{i}(0)\right]\,, (100)

where JΦJ_{\Phi} is the derivative of the solution (Φ\Phi) evaluated 2π2\pi, also called ”sensitivity matrix”,

JΦ=Φ(x(0),2π)x(0)|xi(0).J_{\Phi}=\frac{\partial\Phi(x(0),2\pi)}{\partial x(0)}|_{x^{i}(0)}\,. (101)

and II is the identity matrix.

In this case (101) it is estimated using the simplest of the strategies, i. e., the problem system is solved in the case of two close initial conditions xi(0)x^{i}(0) and xi(0)+hx^{i}(0)+h Φ(xi(0),2π)\Phi(x^{i}(0),2\pi) and Φ(xi(0)+δ,2π)\Phi(x^{i}(0)+\delta,2\pi) and JΦJ_{\Phi} is approximated as:

JΦΦ(xi(0),2π)Φ(xi(0)+δ,2π)δ.J_{\Phi}\approx\frac{\Phi(x^{i}(0),2\pi)-\Phi(x^{i}(0)+\delta,2\pi)}{\delta}. (102)

On the other hand, the solution Φ(xi(0),2π)\Phi(x^{i}(0),2\pi) is estimated using Euler’s method, that is, a discretization (the simplest) of the form:

xn+1=xn+h𝐟(xn)x_{n+1}=x_{n}+h\,\mathbf{f}(x_{n})
Refer to caption
Figure 6: EnE_{n} represents the difference x(0)Φ(x(0),2π)x(0)-\Phi(x(0),2\pi), in each iteration of Newton’s method.
Refer to caption
Refer to caption
Figure 7: Numerical solution of (IV.2). Left: the different faf_{a}’s are represented in different colors for a unit of the torus lattice. Right: The compatibility between the three faf_{a} solutions is depicted. The red dot in the right figure points out the boundary conditions.

with x(0)=x0x(0)=x_{0}, h=δρh=\delta\rho y δρ=102\delta\rho=10^{-2} . Due to the way the Newton’s method is designed, it is desirable to start the process with some initial value as close to zero as possible. But, in the absence of any intuition about where the zero might be in this case, we propose a exhaustive search over a coarse grid of the values of ωi\omega_{i} (a ”free” parameter) and faf_{a} that produce a good approach to the fulfillment of the boundary conditions. From these values, Newton method is applied. The Figure 6 show the difference x(0)Φ(x(0),2π)x(0)-\Phi(x(0),2\pi), in each iteration of Newton’s method.

To summarize this section, we obtain a numerical solution for the non-isotropic Q-ball-like soliton. The shape of the individuals faf_{a} are depicted in Figure 7a. Note that the ZaZ_{a} solutions must be thought with respect to both (σ,ρ)(\sigma,\rho) variables although only the dependence on ρ\rho is depicted. It can be seen that the amplitudes and the frequencies of the solutions are not constant and they do not correspond to trigonometric functions. Note that they vary over their period. In figure 7b we can observe the matching of the three different solutions to the system of equations of motion satisfying all of the three constraints with non-vanishing central charge condition, providing an admissible solution.

V Dynamics of the solutions

In this section we study the dynamics of the analytic solutions previously found under a Lorentz Boost and under a Galilean transformation. Since we are studying the propagation of excitation over the membrane it is not mandatory that their propagation to be relativistic, hence we will analyze the two cases. We will see that the Galilean one is more restrictive. A particularity of our approach is that we work in the LCG and not in a covariant formulation, hence, it is not enough to compute the transformed solutions to guarantee that they still remain being solutions of the system, hence, it is need to study them with care. In first place we will make general considerations valid for all the solutions considered , alaytic and numerical, that satisfy the ansatz σfρf\partial_{\sigma}f\propto\partial_{\rho}f.

In the case of a Lorentz transformation of a generic family of solutions, we define a group velocity 𝐯=(vσ,vρ){\bf{v}}=(v_{\sigma},v_{\rho}) with v2=|𝐯|2=vσ2+vρ2v^{2}=|{\bf{v}}|^{2}=v_{\sigma}^{2}+v_{\rho}^{2}. The worldvolume coordinate transformations imply

σ\displaystyle\sigma\prime =γvστ+σ+(vσv)2(γ1)σ+vσvρv2(γ1)ρ,\displaystyle=-\gamma v_{\sigma}\tau+\sigma+\left(\frac{v_{\sigma}}{v}\right)^{2}(\gamma-1)\sigma+\frac{v_{\sigma}v_{\rho}}{v^{2}}(\gamma-1)\rho\,, (103)
ρ\displaystyle\rho\prime =γvρτ+ρ+(vρv)2(γ1)ρ+vρvσv2(γ1)σ,\displaystyle=-\gamma v_{\rho}\tau+\rho+\left(\frac{v_{\rho}}{v}\right)^{2}(\gamma-1)\rho+\frac{v_{\rho}v_{\sigma}}{v^{2}}(\gamma-1)\sigma\,,
τ\displaystyle\tau\prime =γvσc2σγvρc2ρ+γτ,\displaystyle=-\gamma\frac{v_{\sigma}}{c^{2}}\sigma-\gamma\frac{v_{\rho}}{c^{2}}\rho+\gamma\tau\,,

with γ=11v2c2\gamma=\frac{1}{\sqrt{1-\frac{v^{2}}{c^{2}}}} . The function Z=f(σ,ρ)eiωτZ=f(\sigma,\rho)e^{i\omega\tau} becomes modified under the above transformation as

Z(σ,ρ,τ)\displaystyle Z^{\prime}(\sigma\prime,\rho\prime,\tau\prime) =f(kσ+lρ+ωτ)ei(ασ+Λρ+ητ),\displaystyle=f(k^{\prime}\sigma+l^{\prime}\rho+\omega^{\prime}\tau)e^{i(\alpha\sigma+\Lambda\rho+\eta\tau)}\,, (104)

where

k\displaystyle k^{\prime} :=k(1+(vσv)2(γ1))+lvρvσv2(γ1),\displaystyle:=k(1+\left(\frac{v_{\sigma}}{v}\right)^{2}(\gamma-1))+l\frac{v_{\rho}v_{\sigma}}{v^{2}}(\gamma-1)\,, (105)
l\displaystyle l^{\prime} :=l(1+(vρv)2(γ1))+kvρvσv2(γ1),\displaystyle:=l(1+\left(\frac{v_{\rho}}{v}\right)^{2}(\gamma-1))+k\frac{v_{\rho}v_{\sigma}}{v^{2}}(\gamma-1)\,,
ω\displaystyle\omega^{\prime} :=γ(kvσ+lvρ),\displaystyle:=-\gamma\left(-kv_{\sigma}+lv_{\rho}\right)\,,
α\displaystyle\alpha :=γωvσc2,Λ:=γωvρc2,η:=γω.\displaystyle:=-\gamma\omega\frac{v_{\sigma}}{c^{2}},\quad\Lambda:=-\gamma\omega\frac{v_{\rho}}{c^{2}},\quad\eta:=\gamma\omega\,.

It describes a travelling soliton. Clearly the transformed solution acquires a spinning Q-ball ansatz when ff does not depend on time, i.e.

ω=0=kvσ+lvρ,\omega^{{}^{\prime}}=0=kv_{\sigma}+lv_{\rho}, (106)

which imposes a relation between the two components of the group velocity vector. As we have already explained, a transformed solution is not automatically a solution to the equations of motion. On one side, the functions ff must be periodic on the M2-brane worldvolume since ZZ and ArA_{r} are single-valued functions. Also, in order to guarantee that the transformed solutions remain in the set of allowed solutions we restrict ourselves to impose them to be in the family of solutions analyzed in section 3. The first condition implies that the Fourier modes (k,l)(k^{\prime},l^{\prime}) must be integers222In these examples we have fixed the eigenvalue λ\lambda to one. For arbitrary λ\lambda the periodicity condition is satisfied imposing λk,λl.\lambda k,\lambda l\in\mathbb{Z}. and requires the following necessary conditions to be satisfied:

(vσv)2(γ1)\displaystyle\left(\frac{v_{\sigma}}{v}\right)^{2}(\gamma-1)\in Z,\displaystyle Z, (107)
(vρv)2(γ1)\displaystyle\left(\frac{v_{\rho}}{v}\right)^{2}(\gamma-1)\in Z,\displaystyle Z,
vρvσv2(γ1)\displaystyle\frac{v_{\rho}v_{\sigma}}{v^{2}}(\gamma-1)\in Z.\displaystyle Z.

They restrict enormously the possible values of the allowed gamma-factor and group velocities. In order to illustrate it let us consider the following example: assume that v2/c2=ϵ2=P/Qv^{2}/c^{2}=\epsilon^{2}=P/Q with P<QP<Q and vσ=1/Nvv_{\sigma}=1/Nv, vρ=1/Mvv_{\rho}=1/Mv with P,Q,N,MZP,Q,N,M\in Z, this would imply that the following condition is satisfied for example by requiring that ϵ2=n21n2\epsilon^{2}=\frac{n^{2}-1}{n^{2}} which leads to γ=nZ{0}\gamma=n\in Z-\{0\} and if n1=M2N2n-1=M^{2}N^{2} and then the group velocities becomes vσ=MNvρv_{\sigma}=\frac{M}{N}v_{\rho} and the necessary conditions are automatically satisfied. Hence, the Lorentz transformed Fourier frequencies are

k=k(1+N2)+lMN,l=l(1+M2)+kMN.k^{\prime}=k(1+N^{2})+lMN,\quad l^{\prime}=l(1+M^{2})+kMN. (108)

In the example considered, one can see that the allowed γ\gamma coefficients are discrete and rapidly grows towards the speed of light as nn grows. It implies that for this choice of the v/cv/c factor, that the group velocity of the propagating excitations is close to the speed of light, though it does not exclude the existence of other solutions with less group velocity.

These are necessary condition but they do not automatically guarantee the existence of solutions. In the case of the family of solutions verifying (26) and (27) we find that those whose transformed Lorentz ZZ function must also verify that

α=k,Λ=l,η=ω,\alpha=k^{\prime},\quad\Lambda=l^{\prime},\quad\eta=\omega^{\prime}\,, (109)

to remain as solutions of the M2-brane with worldvolume fluxes EOM. See that only (α,Λ)(\alpha,\Lambda) impose requirements on

(k,l)=ω(n+1n)(M2vσ,N2vρ).(k^{\prime},l^{\prime})=\omega\left(\frac{n+1}{n}\right)(\frac{M^{2}}{v_{\sigma}},\frac{N^{2}}{v_{\rho}}).

In the first family case considered in section 3, for the breathing modes ω\omega given by (46) and k/l=jk/l=j, it is possible to obtain a solution by assuming ω=Unvσ=Tnvρ\omega=Unv_{\sigma}=Tnv_{\rho} with U,TZU,T\in Z. For example there exists a solution for M=N=2M=N=2, T=5tT=5t, U=5MNtU=5\frac{M}{N}t with m1=0,m2n2m_{1}=0,m_{2}\neq n_{2} and m2,n2,n1>0m_{2},n_{2},n_{1}>0 and R12R22=(m2jn2jn2)(jn2m2(1U)jn1)>0\frac{R_{1}^{2}}{R_{2}^{2}}=\left(\frac{m_{2}-jn_{2}}{jn_{2}}\right)\left(\frac{jn_{2}-m_{2}(1-U)}{jn_{1}}\right)>0 with t<0t<0. The transformed Fourier modes become shifted by two different discrete values.

In the case when the ArA_{r} real scalar field is included, the gauge field also gets transformed. For the ansatz previously previously considered, with Z(σ,ρ)=f(σ,ρ)ei(λ(lρ+kσ)+τω)Z(\sigma,\rho)=f(\sigma,\rho)\ e^{i(\lambda(l\rho+k\sigma)+\tau\omega)} its Lorentz transformation corresponds to

fa(σ,ρ)=ra(cos[λ(lρ+kσ)]+sin[λ(lρ+kσ)]),f^{\prime}_{a}(\sigma^{\prime},\rho^{\prime})=r_{a}\left(\operatorname{cos}[\lambda(l\rho+k\sigma)]+\operatorname{sin}[\lambda(l\rho+k\sigma)]\right)\,, (110)

with the transformed frequency given by

ω2\displaystyle\omega^{\prime 2} =\displaystyle= λ2Rr2(kmrlnr)2+λ2Rs2(kmslns)2\displaystyle\lambda^{2}R_{r}^{2}\left(k^{\prime}m_{r}-l^{\prime}n_{r}\right)^{2}+\lambda^{2}R_{s}^{2}\left(k^{\prime}m_{s}-l^{\prime}n_{s}\right)^{2} (111)
=\displaystyle= λ2rRr2(vσcmrvρcnr)2.\displaystyle\lambda^{2}\sum_{r}R_{r}^{2}\left(\frac{v_{\sigma}}{c}m_{r}-\frac{v_{\rho}}{c}n_{r}\right)^{2}\,.

See that due to the ArA_{r} EOM gives a frequency

arΩr2=arω2\displaystyle a_{r}\Omega_{r}^{2}=a_{r}\omega^{\prime 2} =\displaystyle= λ2(ar(kmslns)2Rs2+\displaystyle\lambda^{\prime 2}\left(a_{r}\left(k^{\prime}m_{s}-l^{\prime}n_{s}\right)^{2}R_{s}^{2}+\right.
as(kmrlnr)(kmslns)RrRs),\displaystyle-a_{s}\left(k^{\prime}m_{r}-l^{\prime}n_{r}\right)\left(k^{\prime}m_{s}-l^{\prime}n_{s}\right)R_{r}R_{s}\Big{)}\,,

with the amplitude of ArA_{r} given by

ar=\displaystyle a_{r}= λ2c4(ar(vσmsvρns)2Rs2+\displaystyle\frac{\lambda^{2}}{c^{4}}(a_{r}\left(v_{\sigma}m_{s}-v_{\rho}n_{s}\right)^{2}R_{s}^{2}+ (113)
as(vσmrvρnr)(vσmsvρns)RrRs),\displaystyle-{a_{s}}\left(v_{\sigma}m_{r}-v_{\rho}n_{r}\right)\left(v_{\sigma}m_{s}-v_{\rho}n_{s}\right)R_{r}R_{s})\,,

and verifying the APD transformed constraint.

rar(kmrlnr)Rr=rar(vσmrvρnr)Rr=0.\sum_{r}a_{r}\left(k^{\prime}m_{r}-l^{\prime}n_{r}\right){R}_{r}=\sum_{r}a_{r}\left(v_{\sigma}m_{r}-v_{\rho}n_{r}\right){R}_{r}=0\,. (114)

The previous equations and the ansatz, impose that the frequencies of the Q-ball-like breathing modes and the temporal dependence of ArA_{r} to be equal, ω=Ω=Ωr\omega^{\prime}=\Omega^{\prime}=\Omega_{r}^{\prime} with r=1,2,rsr=1,2,r\neq s. Again, it acquires a spinning Q-ball ansatz when a relation between the components of the group vector velocity is imposed,

kvσ+lvρ=0,kv_{\sigma}+lv_{\rho}=0, (115)

Let us remind that these solitonic solutions are defined over the M2-brane worldvolume and describe the traveling excitations over the M2-brane with worldvolume fluxes. The center of mass of the M2-brane at this stage propagate as a free particle and its mode is decoupled from the excitations ones [23], (as it should happen in case of a particle interpretation). From that point of view, there is not any prerequisite that oblige the excitations to be relativistic, they can behave as classical ones. Galileo transformations are obtained for γ=1\gamma=1, satisfying,

σ\displaystyle\sigma\prime =σγvστ,ρ\displaystyle=\sigma-\gamma v_{\sigma}\tau,\quad\rho\prime =ργvρτ,τ\displaystyle=\rho-\gamma v_{\rho}\tau,\quad\tau\prime =τ,\displaystyle=\tau\,, (116)

implies a modification in the Q-ball expression that acquires a temporal dependence as a traveling wave,

Z(σ,ρ,τ)f(kσ+lρ+ωτ)eiωτ,\displaystyle Z^{\prime}(\sigma\prime,\rho\prime,\tau\prime)\equiv f(k\sigma+l\rho+\omega^{\prime}\tau)e^{i\omega\tau}\,, (117)

where

ω\displaystyle\omega^{\prime} :=(kvσ+lvρ),\displaystyle:=-\left(kv_{\sigma}+lv_{\rho}\right)\,, (118)

In order to keep the solution stationary inside the Q-ball ansatz it is required that ω=0\omega^{\prime}=0, then the same relation between the group speeds hold

kvσ+lvρ=0.kv_{\sigma}+lv_{\rho}=0\,. (119)

See, that this requirement, needs to be fulfilled for every single function ff that satisfies the equations. Hence, it holds not only of the analytic family of solutions found in section 3 but for the entire range of solutions analytically or numerically modeled. Indeed, from the numerical point of view, the Galilean transformation can be directly imposed and it does not change the picture of the Q-ball solution already obtained. It can be straightforwardly verified that the dynamics of the solutions preserve the the topological central charge condition in all cases, as expected.

VI Topological and non-topological charges of the M2-brane

The supermembrane with worldvolume fluxes contains a topological magnetic monopole associated with the topological constraint central charge. We have also seen that some of the solutions that also admits a non-topological Q-ball-like in the non-compact variables. A natural check is to study the dependence of the breathing frequency with the energy.

The Q-ball-like solutions have an associated U(1)U(1) Noether charge defined as

Qa=ωad2σfa2(σ,ρ).Q_{a}=\omega_{a}\int d^{2}\sigma f^{2}_{a}(\sigma,\rho). (120)

in terms of its density charge qaq_{a} can be expressed,

Qa=𝑑σ𝑑ρqa(σ,ρ),withqa(σ,ρ)=ωafa2(σ,ρ).Q_{a}=\int d\sigma d\rho q_{a}(\sigma,\rho),\quad\textrm{with}\quad q_{a}(\sigma,\rho)=\omega_{a}f_{a}^{2}(\sigma,\rho)\,. (121)

In order to simplify the analysis at the maximum, to see th enon-vanishing contribution we will assume the Hamiltonian (6), restricted to the Q-ball-like ansatz, for the isotropic case with the ArA_{r} field constant. Then, it becomes,

H=T2/3Σd2σW12[Pa2+T22(Drfa)2].H=T^{-2/3}\int_{\Sigma}d^{2}\sigma\sqrt{W}\frac{1}{2}[P_{a}^{2}+\frac{T^{2}}{2}(D_{r}f_{a})^{2}]\,. (122)

For the family of solutions considered in this paper (analytical and numerical) satisfying (26), ρfa=ja1σfa\partial_{\rho}f_{a}=j_{a}^{-1}\partial_{\sigma}f_{a}, the quadratic contributions become

r(Drfa)2=14ωaqarRr2(mrja1nr)2(σqa)2.\sum_{r}(D_{r}f_{a})^{2}=\frac{1}{4\omega_{a}q_{a}}\sum_{r}R_{r}^{2}(m_{r}j_{a}^{-1}-n_{r})^{2}(\partial_{\sigma}q_{a})^{2}\,. (123)

Substituting the momentum associated with the ZZ coordinates in terms of the Noether charge, then the Hamiltonian can be bounded by below by the following terms

H32T2/3Qaωa+T4/32Σd2σWr(Drfa)2.H\geq\frac{3}{2}T^{2/3}Q_{a}\omega_{a}+\frac{T^{4/3}}{2}\int_{\Sigma}d^{2}\sigma\sqrt{W}\sum_{r}(D_{r}f_{a})^{2}\,. (124)

For the particular case of solutions of the family, the Hamiltonian lower bound expressed in terms of the Noether density charge gets simplified to

T2/3H\displaystyle T^{2/3}H T4/332Qaω+\displaystyle\geq T^{4/3}\frac{3}{2}Q_{a}\omega+ (125)
+T28ωarRr2(mrja1nr)2Σd2σW(σqa)2qa.\displaystyle+\frac{T^{2}}{8\omega_{a}}\sum_{r}R_{r}^{2}(m_{r}-j_{a}^{-1}n_{r})^{2}\int_{\Sigma}d^{2}\sigma\sqrt{W}\frac{(\partial_{\sigma}q_{a})^{2}}{q_{a}}\,.

The mass operator for the M2-brane with fluxes [SpinningSolutionsBosonicM2-brane2022], is bounded by below by the contribution of the topological charges: the central charge (or equivalently CC_{-} flux associated with the presence of a monopole), the C+C_{+} flux contribution and the Kaluza Klein charges. The non-topological ones associated with the presence of Q-ball-like solutions and the interaction terms between both types of solitons contained in the mass term (Drfa)2(D_{r}f_{a})^{2}

2\displaystyle\mathcal{M}^{2} Qtopological2+T4/332Qaω+\displaystyle\geq Q_{topological}^{2}+T^{4/3}\frac{3}{2}Q_{a}\omega+ (126)
+T28ωarRr2(mrja1nr)2Σd2σW(σqa)2qa,\displaystyle+\frac{T^{2}}{8\omega_{a}}\sum_{r}R_{r}^{2}(m_{r}-j_{a}^{-1}n_{r})^{2}\int_{\Sigma}d^{2}\sigma\sqrt{W}\frac{(\partial_{\sigma}q_{a})^{2}}{q_{a}}\,,

where

Qtopological2=\displaystyle Q_{topological}^{2}= Qfluxes2+QKK2\displaystyle Q_{fluxes}^{2}+Q_{KK}^{2} (127)
=\displaystyle= ΣC+ΣC++(m|qτ~p|R[Imτ~])2,\displaystyle\int_{\Sigma}C_{-}+\int_{\Sigma}C_{+}+\left(\frac{m|q\tilde{\tau}-p|}{R[I_{m}\tilde{\tau}]}\right)^{2}\,,

with

ΣC=4π2R1R2T2n2,\int_{\Sigma}C_{-}=4\pi^{2}R_{1}R_{2}T^{2}n^{2}\,, (128)

the units of flux, corresponding to the value of the central charge (25) as originally shown in [23, 22]. See, that the interaction term contains the Noether charge density of the noncompact sector and the winding numbers required to satisfy the central charge condition that implies the presence of the magnetic monopole. The interaction term is due to the presence of the worldvolume fluxes. Due to it, (responsible for the change in the spectral properties), the characteristic relation between the energy and the frequency of the breathing modes in a standard Q-ball, acquires an extra term associated with the non-vanishing mass terms that add a new dependence on the frequency, that leads a relation between the Energy, the Noether charge, the topological winding numbers and the density Noether charge relations. For T=1T=1 it is,

Eω=32aQa+18ωa2ΘΣd2σW(σqa)2qa,\frac{\partial E}{\partial\omega}=\frac{3}{2}\sum_{a}Q_{a}+\frac{1}{8\omega_{a}^{2}}\Theta\int_{\Sigma}d^{2}\sigma\sqrt{W}\frac{(\partial_{\sigma}q_{a})^{2}}{q_{a}}\,, (129)

with Θ=rRr2(mrunr)2\Theta=\sum_{r}R_{r}^{2}(m_{r}-un_{r})^{2}. In this analysis faf_{a} is an arbitrary function of the family. This type of dependence has been also found in the recent paper [3] in which they discuss Q-balls beyond thin-wall limit.

If now we restrict to the analytic family of solutions considered in section 3, we can observe that for ja=1j_{a}=1 we can integrate the function and obtain the standard dependence for a Q-ball Eω=32aQa\frac{\partial E}{\partial\omega}=\frac{3}{2}\sum_{a}Q_{a} shifted by a constant term.

Stability of the Solutions

An important point that has not been addressed so far is the stability of the solutions found. The monopole charge induced by the fluxes acting on the Hamiltonian through the non-vanishing covariant derivative ”mass term”, associated to the monopole and Q-ball interaction term, protects the worldvolume of the membrane from fission at the classical level, since there are no string-like solutions with vanishing energy that can split and merge the membrane without energy cost when the area of the worldvolume is conserved [gmr]. It is well known that this instability classically occurs over generically for the membrane, [dwhn]. In that scenario, a Q-ball excitation would not make sense, if defined on an object that can split and merge without any energy cost as is the case for the worldvolume of the flux-free supermembrane, i.e. vanishing central charge. Therefore, the flux condition is a necessary condition for the stability of the Q-ball against fission.

On the other hand, we expect that the fact that the interaction term (Covariant derivative mass terms) is non-vanishing, will provide a sufficient condition for the stability against fission of the Q-ball-like excitations that propagates on the membrane. Indeed, this is the case for the analytic solutions of the string Q-ball-like excitations considered in this paper, as it can be straightforwardly checked. Given a fixed time t, those Q-ball-like solutions require an addition of energy to become splitted. Furthermore, since the functions considered are periodic, the stability in the propagation of those solutions is also guaranteed. We expect to holfd this property generically. However, to ensure it, this needs to be proved for arbitrary functions. This question remains unanswered so far and is beyond the scope of this paper.

VII Superposition of isotropic solutions: Vectorization

An important aspect of solitonic solutions is the fact that on very general grounds, solitons cannot be superposed linearly. In fact, this is a characteristic property, however, there are some exceptions, that for a restricted range of frequencies, admit linear superpositions like it is the case for optical solitons, [SuperpositionsBrightanddarksolitons, LinearSuperpositionPonomarenko2002, ZHOU20171697:LinearSuperpositionPrinciple]. We show in this section that this is also the case of the Q-ball-like solutions analyzed in section 3. We will see that in the absence of a symplectic gauge field, we are able to obtain multi-soliton solutions that naively seems to corresponds to a linear superposition but only when restricted to a very specific discrete set of frequencies given by the monopole contribution. In order to understand it better we will do it in several steps.

VII.1 Fourier series solution

Although we are interested in Q-ball-like solutions, we will start by analyzing the behaviour of the EOM of the M2-branes with worldvolume fluxes for the case of a superposition of the complexified analytic solutions found previously in section 3. We can observe that they correspond to a Fourier serie. Since this is a sum, with matrix entries, it is not automatically guaranteed that the superposition of solutions will be also a solution due to the non-linearities of the differential equations. Indeed we are interested in finding the conditions under which this may happen. Let us consider fa(σ,ρ)f_{a}(\sigma,\rho) represented by a Fourier series:

fa=k,l=0aklfakl(σ,ρ).f_{a}^{\mathbb{C}}=\sum_{k,l=0}^{\infty}a_{kl}f^{kl}_{a}(\sigma,\rho)\,. (130)

The klkl-th element of the orthogonal basis is given by:

fakl(σ,ρ)=cos(kσ+lρ)+isin(kσ+lρ).f^{kl}_{a}(\sigma,\rho)=\cos(k\sigma+l\rho)+i\sin(k\sigma+l\rho)\,. (131)

It is easy to verify that, with a proper normalization:

02π02πfakl(σ,ρ)fauv(σ,ρ)𝑑σ𝑑ρ={0uvkl1uv=kl\int_{0}^{2\pi}\int_{0}^{2\pi}f^{kl}_{a}(\sigma,\rho)~{}f^{uv}_{a}(\sigma,\rho)~{}d\sigma~{}d\rho=\begin{cases}0&uv\neq kl\\ 1&uv=kl\end{cases} (132)

with,

anm=02π02πfa(σ,ρ)fakl(σ,ρ)𝑑σ𝑑ρ.a_{nm}=\int_{0}^{2\pi}\int_{0}^{2\pi}f_{a}(\sigma,\rho)~{}f^{kl}_{a}(\sigma,\rho)~{}d\sigma d\rho\,. (133)

Imposing smoothness on ff and differentiability term-by-term, see for example section 3.43.4 in [Haberman]. The series (130) can be written as:

fa=ξu=0aξueiξu.u,f_{a}^{\mathbb{C}}=\sum_{\xi_{u}=0}^{\infty}a_{\xi_{u}}e^{i~{}\xi_{u}.u}\,, (134)

and derivatives

ufa(u)\displaystyle\partial_{u}f_{a}(u) =ξu=0aξuiξueiξu.u,u2fa(σ,ρ),=\displaystyle=\sum_{\xi_{u}=0}^{\infty}a_{\xi_{u}}i~{}\xi_{u}~{}e^{i~{}\xi_{u}.u},\ \ \partial_{u}^{2}f_{a}(\sigma,\rho),= (135)
=ξu=0aξuξu2eiξuu,uv2fa(ξu)=ξu=0aξuξuξveiξuu,\displaystyle=-\sum_{\xi_{u}=0}^{\infty}a_{\xi_{u}}\xi_{u}^{2}~{}e^{i\xi_{u}u},\ \ \partial_{uv}^{2}f_{a}(\xi_{u})=-\sum_{\xi_{u}=0}^{\infty}a_{\xi_{u}}\xi_{u}\xi_{v}~{}e^{i~{}\xi_{u}u},

with uvu\not=v. We have defined the vector u=(σ,ρ)u=(\sigma,\rho) labeling the spatial coordinates and the vector ξu=(k,l)\xi_{u}=(k,l) labeling the Fourier modes of the expansion, and axiua_{xi_{u}} the constant amplitude of faf_{a}^{\mathbb{C}}.

To implement the expansion numerically, we use a truncated version of the series (130) and define, the vectors:

𝐤\displaystyle{\bf k} =(0,1,2,3,N,1,2,3,,N,),\displaystyle=\left(0,1,2,3,\cdots N,1,2,3,\cdots,N,\cdots\right)\,, (136)
𝐥\displaystyle{\bf l} =(0,0,0,0,1,1,1,1,,N,N,N,),\displaystyle=\left(0,0,0,0,\cdots 1,1,1,1,\cdots,N,N,N,\cdots\right),
𝐤2\displaystyle{\bf k}^{2} =(0,1,4,9,,k2,N2),\displaystyle=\left(0,1,4,9,\cdots,k^{2},\cdots N^{2}\right),
𝐥2\displaystyle{\bf l}^{2} =(0,0,0,0,,l2,N2),\displaystyle=\left(0,0,0,0,\cdots,l^{2},\cdots N^{2}\right),
𝐤𝐥\displaystyle{\bf kl} =(0,0,0,0;0,1,2,3,;0,2,4,6,,kl,,),\displaystyle=\left(0,0,0,0;0,1,2,3,\dots;0,2,4,6,\cdots,kl,\cdots,\right),
𝐟a(σ,ρ)\displaystyle{\bf f}_{a}(\sigma,\rho) =(a11fa11(σ,ρ),a12fa12(σ,ρ),,\displaystyle=(a_{11}~{}f_{a}^{11}(\sigma,\rho),a_{12}~{}f_{a}^{12}(\sigma,\rho),\cdots,
,a1Nfa1N(σ,ρ),a21fa21aNNfaNN),\displaystyle\cdots,a_{1N}~{}f_{a}^{1N}(\sigma,\rho),a_{21}~{}f_{a}^{21}\cdots a_{NN}~{}f_{a}^{NN}...),
ω2\displaystyle\mathbf{\omega}^{2} =(ω11,ω12,ω13,ωkl,,ωNN).\displaystyle=(\omega_{11},\omega_{12},\omega_{13},\omega_{kl},\cdots,\omega_{NN}).

See also that 𝐥{\bf l} must be defined in a different way than the others indices. With the above notation 333There, the vectors kk and ll are a concatenation of NN vectors of integers from 0 to NN and 𝐟a(σ,ρ){\bf f}_{a}(\sigma,\rho) is a flattened version of the matrix faklf_{a}^{kl}. defined and considering only SineSine or CosineCosine series, we can write:

σfa(σ,ρ)\displaystyle\partial_{\sigma}f_{a}(\sigma,\rho) =\displaystyle= i𝐤fa(σ,ρ),ρfa(σ,ρ)=i𝐥f(σ,ρ),\displaystyle i~{}{\bf k}\cdot\textbf{f}_{a}(\sigma,\rho),\quad\partial_{\rho}f_{a}(\sigma,\rho)=i~{}{\bf l}\cdot\textbf{f}(\sigma,\rho),
σ2fa(σ,ρ),\displaystyle\partial_{\sigma}^{2}f_{a}(\sigma,\rho), =\displaystyle= 𝐤2f(σ,ρ),ρ2fa(σ,ρ)=𝐥2f(σ,ρ),\displaystyle-{\bf k}^{2}\cdot\textbf{f}(\sigma,\rho),\quad\partial_{\rho}^{2}f_{a}(\sigma,\rho)=-{\bf l}^{2}\cdot\textbf{f}(\sigma,\rho),
σρ2fa(σ,ρ)\displaystyle\partial^{2}_{\sigma\rho}f_{a}(\sigma,\rho) =\displaystyle= 𝐤𝐥f(σ,ρ),\displaystyle-{\bf kl}\cdot\textbf{f}(\sigma,\rho), (137)

where, ()(\cdot) denotes a inner product. See that obviously this is not a general property but a property of the Lie bracket and the base chosen that we know that satisfies the equations in the case of a single solution. The Figure (8) compares in a graphic fashion, the ”analytic” derivative and ”inner product version” of the derivative in the case of σρ2fa(σ,ρ)\partial^{2}_{\sigma\rho}f_{a}(\sigma,\rho). See that the agreement is very good.

Refer to caption
Refer to caption
Figure 8: The Figure of left represents the analytic derivative behaviour and the Figure of the right ”inner product version” of the derivative in the case of σρ2fa(σ,ρ)\partial^{2}_{\sigma\rho}f_{a}(\sigma,\rho). As we can observe both graphics are indistinguibles.

The system that we intend to solve corresponds to the EOM (21) and (II) with Ar=constantA_{r}=constant

C1(σ2fa)+2C2(σρ2fa)C3ρ2fa=ω2fa,-C_{1}\left(\partial_{\sigma}^{2}f_{a}\right)+2~{}C_{2}\left(\partial^{2}_{\sigma\rho}f_{a}\right)-C_{3}\partial_{\rho}^{2}f_{a}=\omega^{2}f_{a}, (138)
(mrσ2fanrσρ2fa)ρfa(mrρσ2fanrρ2fa)σfa=0,\left(m_{r}\partial_{\sigma}^{2}f_{a}-n_{r}\partial^{2}_{\sigma\rho}f_{a}\right)\partial_{\rho}f_{a}-\left(m_{r}\partial^{2}_{\rho\sigma}f_{a}-n_{r}\partial_{\rho}^{2}f_{a}\right)\partial_{\sigma}f_{a}=0\,, (139)

where C1=m92R92+m102R102C_{1}=m_{9}^{2}R_{9}^{2}+m_{10}^{2}R_{10}^{2}, C2=m9n9R92+n10m10R102C_{2}=m_{9}n_{9}R_{9}^{2}+n_{10}m_{10}R_{10}^{2}, C3=n92R92+n102R102,C_{3}=n_{9}^{2}R_{9}^{2}+n_{10}^{2}R_{10}^{2}, with r=9,10r=9,10, nr,mrn_{r},m_{r}, the winding numbers, RrR_{r} the torus radii and ω\omega\in\mathbb{R}. The APD constraint is identically verified. It is subject to the central charge condition (25).

We impose now the ansatz of the family (26) for the isotropic case. The equation (138) written in the inner product version results in:

(C1𝐤22C2𝐤𝐥+C3𝐥2)f(σ,ρ)\displaystyle\left(C_{1}~{}{\bf k}^{2}-2~{}C_{2}~{}{\bf kl}+C_{3}{\bf l}^{2}\right)\cdot\textbf{f}(\sigma,\rho) =\displaystyle= ω2f(σ,ρ),\displaystyle\mathbf{\omega}^{2}\cdot\textbf{f}(\sigma,\rho), (140)

where each component, it can be expressed directly in terms of their breathing modes:

C1k22C2kl+C3l2=ωkl2.C_{1}~{}k^{2}-2~{}C_{2}~{}kl+C_{3}~{}l^{2}=\omega_{kl}^{2}\,. (141)

The second equation (139), since it contains the nonlinearities is more involved, but still can be expressed in this simple algebraic way:

[(mr𝐤2nr𝐤𝐥)𝐟(σ,ρ)]×[𝐥f(σ,ρ)]\displaystyle-[\left(m_{r}{\bf k}^{2}-n_{r}{\bf kl}\right)\cdot\mathbf{f}(\sigma,\rho)]\times[~{}{\bf l}\cdot\textbf{f}(\sigma,\rho)] +\displaystyle+ (142)
+[(mr𝐤𝐥nr𝐥2)𝐟(σ,ρ)]×[𝐤f(σ,ρ)]\displaystyle+[\left(m_{r}{\bf kl}-n_{r}{\bf l}^{2}\right)\cdot\mathbf{f}(\sigma,\rho)]\times[~{}{\bf k}\cdot\textbf{f}(\sigma,\rho)] =0.\displaystyle=0.

Here ×\times represents the product between scalars or scalar and vectors. In order to solve this equation, let us define the vectors:

𝜶r=mr𝐤2nr𝐤𝐥,𝜷r=mr𝐤𝐥nr𝐥2,{\bm{\alpha}}_{r}=m_{r}{\bf k}^{2}-n_{r}{\bf kl},\quad{\bm{\beta}}_{r}=m_{r}{\bf kl}-n_{r}{\bf l}^{2}\,, (143)

with components αm\alpha_{m} and βm\beta_{m}, respectively, and being 𝐟mn{\bf f}_{mn} represents the matrix elements of 𝐟{\bf f}. The equation (142) expressed in terms of 𝜶,𝜷\bm{\alpha},\bm{\beta} becomes

(𝜶r𝐟(σ,ρ))×(𝐥f(σ,ρ))+(𝜷r𝐟(σ,ρ))×(𝐤f(σ,ρ))=0.-(\bm{\alpha}_{r}\cdot\mathbf{f}(\sigma,\rho))\times({\bf l}\cdot\textbf{f}(\sigma,\rho))+(\bm{\beta}_{r}\cdot\mathbf{f}(\sigma,\rho))\times({\bf k}\cdot\textbf{f}(\sigma,\rho))=0\,. (144)

Before continuing, it is interesting to note that if the series (130) has only one term, the equation (139) is identically satisfied, since,

nrl2kmrlk2+mrlk2nrkl2\displaystyle n_{r}~{}l^{2}~{}k-m_{r}~{}l~{}k^{2}+m_{r}~{}l~{}k^{2}-n_{r}~{}kl^{2} =\displaystyle= 0,\displaystyle 0\,, (145)

which agrees with the results of Section 3. therefore, The eigenvalues and eigenvectors obtained between the analytical and numerical methods are very similar to those found in [SpinningSolutionsBosonicM2-brane2022] although in that case they were numerical solutions to the approximate system, (i.e. without considering the restriction of (139)). This means that the simplest solutions obtained is in the kernel of the monopole restriction.

VII.2 Solving an algebraic problem

In order to find a larger set of solutions, we have transformed our initial problem into another one: find a 𝐟\bf f that fulfill (144). Since (144) can be written as,

[𝜶r𝐟(σ,ρ)×𝐥+𝜷r𝐟(σ,ρ)×𝐤]f(σ,ρ)=0.\left[-\bm{\alpha}_{r}\cdot\mathbf{f}(\sigma,\rho)\times~{}{\bf l}+\bm{\beta}_{r}\cdot\mathbf{f}(\sigma,\rho)\times~{}{\bf k}\right]\cdot\textbf{f}(\sigma,\rho)=0. (146)

There are two alternatives: one can linearly transform 𝐟\bf f (selecting the set of coefficients amna_{mn}), so that obtain a vector

𝜶r𝐟(σ,ρ)×𝐥+𝜷r𝐟(σ,ρ)×𝐤,-\bm{\alpha}_{r}\cdot\mathbf{f}(\sigma,\rho)\times~{}{\bf l}+\bm{\beta}_{r}\cdot\mathbf{f}(\sigma,\rho)\times~{}{\bf k}, (147)

orthogonal to 𝐟\bf f (this seems to be an interesting, but complicated problem), or more easily, we can suppose that 𝐟0{\bf f}\neq 0, and find 𝐟{\bf f} (the set of coefficients amna_{mn} that satisfy the equation), that is:

𝜶r𝐟(σ,ρ)×𝐥=𝜷r𝐟(σ,ρ)×𝐤.\bm{\alpha}_{r}\cdot\mathbf{f}(\sigma,\rho)\times~{}{\bf l}=\bm{\beta}_{r}\cdot\mathbf{f}(\sigma,\rho)\times~{}{\bf k}\,. (148)

In order to find a solution to ((139)), we denote the position of an element by ljl_{j}, in the corresponding vector, say for example 𝐥{\bf{l}}. Then the solution with the allowed values is found when kj=ljk_{j}=l_{j} for j=i+(i1)Nj=i+(i-1)N being NN the order of the truncation and ii the position inside it, i=1,,Ni=1,\dots,N. The coefficient values in those positions are given by lj=kj=i1l_{j}=k_{j}=i-1. The solution then arises for the subfamily of functions ff^{\mathbb{C}} with akjlj=1a_{k_{j}l_{j}}=1 and zero otherwise. It corresponds to the diagonal terms of f in a matrix representation.

Hence, we obtain that the superposition of arbitrary functions with equal Fourier modes is solution to the EOM (138) and (139) whenever those modes are equal. This condition, constrains the set of solutions of equation (138), and reflects the nonlinearity in behavior to the equation (139), associated with the presence of a monopole solution constructed in terms of the X^r\widehat{X}_{r}.

In short, a general solution for the EOM (138) and (139) system with a QBL ansatz and nonvanishing central charge can be written in terms of the real parts of ff^{\mathbb{C}}. It is straightforward to see that the same result holds for

fa(σ,ρ,w)=k=1Makk[cos(k(ρ+σ))+sin(k(ρ+σ)].f_{a}(\sigma,\rho,w)=\sum_{k=1}^{M}a_{kk}\left[\cos(k(\rho+\sigma))+\sin(k(\rho+\sigma)\right]\,. (149)

The sum runs over the number of pairs MM that fulfill the solution. It represents a Q-ball-like multisoliton solution for the isotropic case with a non vanishing central charge.

In summary, the analytic family of QBL solutions for the isotropic case and constant gauge field and with central charge different from zero admit a linear superposition law but only when the frequencies become strongly constrained to be equal. We emphasize that this is not true otherwise. All of these case require a central charge different from zero -which we recall it is associated to the presence of a magnetic monopole over the worldvolume [24]. This represents the superposition law for this type of solutions.

In the following we illustrate an example of superposition of solutions.

Here, it can be seen how the proposed solution complies with the equation (138-139), with {kj,lj}j=0M={{0,0},{1,1},{2,2},{3,3},{4,4}}\{k_{j},l_{j}\}_{j=0}^{M}=\{\{0,0\},\{1,1\},\{2,2\},\{3,3\},\{4,4\}\} and m9=9m_{9}=9, m10=2m_{10}=2, n9=9n_{9}=9, n10=8n_{10}=8, R9=1R_{9}=1 and R10=11R_{10}=11 with non-vanishing central charge.

Refer to caption
Figure 9: Analytic solution of (138)-(139)

We observe the following: i) The equation (144), is very interesting since it allows to reduce the order of non-linearity in this type of equations, ii) The condition (139) seems to act as a pruning to the basis (solutions) generated by the eigenvalue (138) ad iii) The linear superposition is not possible on general grounds due to the nonlinearities of the equation (139). However, it can be obtained when the values of the Fourier modes becomes strongly constrained to specific pairs that determine the frequencies due to the nonlinear behaviour of (139). There are in the literature other examples of nonlinear EPD that admit restricted linear superposition, see [Han2022] .

VIII Discussion and Conclusions

In this paper we obtain Q-ball-like solitonic solutions to the dynamics of the bosonic part of a supermembrane compactified on a target space M9×T2M_{9}\times T^{2} with a nontrivial C±C_{\pm} fluxes. We have focused in obtaining solutions with a Q-ball-like ansatz defined on the M2-brane worldvolume. These are solutions to the complex scalar fields ZaZ_{a} that determine the embedding of the supermembrane on the non-compact dimensions. We have analyzed them satisfying the ansatz Za=fa(σ,ρ)eiωτZ_{a}=f_{a}(\sigma,\rho)e^{i\omega\tau}. The name of Q-ball has been maintained, though the solutions characterize solitons over a torus, -since the spatial part of the worldvolume of the supermembrane is assumed to be Riemann surface of genus one Σ1\Sigma_{1}. The system is in the presence of a symplectic gauge field 𝔸\mathbb{A} on the M2-brane worldvolume whose components are defined in terms of derivatives of a single-valued real scalar field ArA_{r}. Both fields ZaZ_{a} and ArA_{r} define the dynamics of the M2-brane with a system of nine coupled nonlinear equations of motion. The dynamics is restricted by two constraints, the area preserving diffeomorphisms (APD) that is the residual first class constraint of the membrane theory, and the flux condition. This last topological condition implies the so-called central charge condition [24], which implies a restriction on the M2-brane wrapping numbers and induces the presence of a monopole charge.

We obtain analytical solutions to the isotropic case Za=ZZ_{a}=Z for all a=1,2,3a=1,2,3, and numerical solutions in the non-isotropic case ZaZbZ_{a}\neq Z_{b}. The analytic solutions are obtained in the presence of a non-vanishing ArA_{r} field and attending to the type of time dependence imposed on the ArA_{r} fields, they are organized into different families. All the cases generate a symplectic gauge field 𝔸=dA\mathbb{A}=dA. In the first case, the temporal dependence on ArA_{r} is linear. The frequencies characterizing the Q-ball breathing modes are discrete and they are parametrized in terms of the compactified target space moduli, the nontrivial wrapping of the M2-brane, and the amplitudes of the ArA_{r} field. The Area Preserving Diffeomorphisms (APD) constraint is automatically satisfied. We obtain dynamical solutions to the system that satisfy the central charge restriction and hence the worldvolume flux condition. In the second family, the dependence of ArA_{r} with time, τ\tau, is generalized. The associated symplectic field strength depends on all of the variables. Due to the ansatz chosen and the APD constraint, a restriction to the EOM appear that leads to a relation between the dispersion relation of the Q-ball breathing modes ω\omega and the dispersion relation Ω\Omega associated with the ArA_{r} field. Both frequencies remain discrete. The APD constraint impose a restriction in the amplitudes of the ArA_{r}. For a fixed relation between them, a ratio between the Fourier modes is determined. The type of solutions considered satisfy the central charge condition for winding charges suitably chosen. We also provide an example of a localized family of solutions propagating over the worldvolume of the membrane. Other examples even more sofisticated of localized solutions can be constructed.

We also study their dispersion relations. Generically, the solutions exhibit dispersion although for certain combinations of parameters they also admit solutions dispersion-free. Usually, solitons are associated with solutions in which the dispersion effects are compensated by the non-linearities of the equation. In our case we prove that these type of solitonic solutions free of dispersion effects exist, but also a second type of solitons, that in spite of the dispersion effects, and due to the periodicity of the functions on the M2-brane worldvolume remain being solitons. Examples of periodic solitons with dispersion effects have been considered in the context of optic solitons [Dispersion-Managed:TURITSYN2012]. The dispersion effects in this context do not mean that the system present any loose in energy of the system. Energy is conserved since it corresponds to the excitations on a single M2-brane that propagates freely in the space.

We find also solutions to the non-isotropic case ZaZbZ_{a}\neq Z_{b} by performing a numerical analysis. From the viewpoint of numerical computation, we have proposed and implemented strategies to find approximate solutions of the Q-ball in the isotropic and non-isotropic cases. We have used the isotropic case to compare the numerical method with the results previously obtained in the analytical case. We use a fine-difference scheme is used to solve the eigenvalue problem (138), and then the resulting eigenfunctions are used to represent the solution of (139). This transforms the equation (139) into a system of nonlinear algebraic equations, whose solution can be obtained using methods such as Newton’s method. The non-isotropic case for constant ArA_{r} introduces the non-linearities of the ZZ equation (not only due to the ArA_{r} EOM). In this case, the solution of the system (IV.2), is approximated using a shooting method where the gradient of the solution is approximated using finite differences.
In both cases, very simple approximation schemes are used with solutions that present reasonable approximation errors, so we suppose that more sophisticated schemes could produce better solutions even though they have a higher computational cost. Hence, for the isotropic and non-isotropic case, with a vanishing symplectic gauge field, we are able to obtain numerically a Q-ball soliton solutions to the M2-brane with worldvolume fluxes.

The dynamics of the analytic families of solutions is also investigated. We consider Lorentz transformations. Since we work in the LCG, the transformed solutions are not automatically solutions of the EOM. In this sense, we have shown some examples of where we have obtained the necessary restrictions to satisfy the EOM. Since we are considering solutions that represent solitons defined over the worldvolume, -they correspond to excitations on a M2-brane which propagates freely on the space-, it is not necessary that the excitations will be relativistic. Hence, we also study their Galilean transformations. For this case we find, for the examples considered, that the Galilean transformations restrict further the solutions allowed. We find that for the ansatz considered that only those that preserve their Q-ball shape are allowed, i.e. only contain breathing modes but the pulse does not propagate. However, it does not exclude that more general solutions could be obtained by choosing different ansatzs. Each of the dynamical solutions preserve the flux condition.

The M2-brane mass operator is bounded by below by the topological contribution due to the flux condition and the Kaluza-Klein terms. For the type of solutions considered in this paper, the kinetic term generates a Q-ball contribution, that also contributes to the lower bound. It also appears a new interaction term between the topological and non-topological sectors in the mass term contribution. The M2-brane with worldvolume fluxes induces a central charge condition that describes a magnetic monopole [24], and at the same time it may also describe a non-topological Q-ball soliton. We show that there are nontrivial interaction terms between these two, through the mass terms of the potential containing the winding charges that generate the monopole charge and the Noether charge. They describe a relation between the energy EE and the frequency ω\omega that contains on top of the standard term characterizing the Q-ball relation, and a new term that represents a correction to the Q-ball relation reflecting the interaction among those terms and the compactified and non compactified sectors. This type of correction, that here has a topological origin, has also appeared in the literature, in a different context, when going beyond the thin-wall limit [Heeck_2022:ExcitedQ-balls, 3, Nugaev2020:ReviewNontopologicalSolitons]. Recently, it has been discovered a new type of solitons which share these two types of properties. They have been named as monopole-Q-ball soliton and their Q-ball solutions show an increased stability due to the effect of the topological charge. It would be interesting if both models are related.

In this case the stability of the membrane against fission is guaranted because of the discreteness of the mass operator spectrum of the bosonic membrane that at classical level forbids the existence of flat directions associated to the presence of string-like spikes without energy cost that could induce the non-conservation of the topology nor the number of particles. In the present model, the string-like spikes contribute with a positive energy to the spectrum mass operator of the membrane. This is a special property that for the supersymmetric case is not guaranteed for the target space considered unless the nontrivial worldvolume fluxes are present. The interaction term is responsible for that change in the nature of the spectrum providing more stability to the Q-ball. In the case of the analytical solutions it is easy to prove that also the stability of the QBL excitation is also guaranteed since its split is energetically disfavoured. We believe that this is a general property, however a more general statement for arbitrary functions requires a study beyond the scope of the present paper and are left for a future work.

We also investigate the superposition law for the analytic family of solutions found. To this end we find a new method that we denote vectorization, that allows to convert the non-linear differential equations in terms of algebraic expressions, much simpler to solve. We find that, in spite that the EOM associated with the field ArA_{r} is identically satisfied for a single soliton, due to the nonlinearities of the system of equations, this equation has to be taken into account in the superposition of solutions. We obtain a multi-solitonic solution characterized by a linear superposition, but only when the Fourier modes k,lk,l are equal. This is consistent with the fact that they are solution to the system of nonlinear equations. In the literature it has appeared other optical soliton cases that under restricted conditions admit linear superpositions. See for example, [LinearSuperpositionPonomarenko2002, ZHOU20171697:LinearSuperpositionPrinciple, SuperpositionsBrightanddarksolitons].

In summary, we have obtained Q-ball-like solitons solutions of the dynamics of the supermembrane propagating on M9×T2M_{9}\times T^{2} once that the flux condition is imposed. See, that the non-linearities are present even in the case when the symplectic gauge field vanishes and this is due to the residual equation associated with the presence of multivalued modes that generate the topological magnetic monopole soliton and from those inherited from the quartic potential contribution.

Acknowledgements.
MPGM thanks to G. Alvarez, A. Bellorin and A. Restuccia for helpful conversations. MPGM is partially supported by the PID2021-125700NB-C21 MCI Spanish Grant. R.P. acknowledge to the Doctorado en Física, mención Física Matemática (Ph.D. in Physics) program of the Universidad de Antofagasta, Chile and thank to the projects ANT20992, ANT1956 y ANT1955 of the U. Antofagasta. M.P.G.M., and R.P. want to thank to SEM18-02 project of the U. Antofagasta.

References