Introduction
Cosmic strings are solitonic defects that arise in many extensions of the Standard Model and may have been formed during symmetry breaking phase transitions in the Early Universe [1 ] . Their cosmological signatures probe energy scales inaccessible to terrestrial accelerators and provide a complementary test of fundamental physics [2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 ] . Extracting information about the underlying high energy theory from these signatures requires, however, a controlled connection between the microscopic field theory and the resulting string dynamics.
Cosmic string evolution is commonly described by the Nambu–Goto (NG) action, which governs the massless Goldstone modes associated with broken transverse translations. Spacetime symmetries fix this leading description. A field theory string, however, has finite width and can support massive excitations localized in its core. These modes can be long lived and exchange energy with transverse motion and bulk radiation [14 , 15 , 16 , 17 ] . Understanding their influence requires an Effective Field Theory (EFT) that incorporates the internal structure of the string. Such a description should identify how string motion excites the localized modes, how these excitations backreact on the worldsheet, and under which conditions the Nambu–Goto approximation remains reliable.
Our approach is closely analogous to the philosophy of worldline EFT for compact objects where the response of microscopic degrees of freedom is encoded in worldline operators coupled to external fields [18 , 19 , 20 ] . Such a framework provides a systematic description of finite-size effects, resonant excitations, and dissipation within a local EFT on the object’s worldline [21 , 22 ] , and has proven highly successful in describing the response of compact objects to external gravitational fields and the dynamics of compact binaries [23 , 24 ] .
Here we implement the same idea for string defects, replacing the worldline by a worldsheet and introducing explicit worldsheet fields associated with bound states localized on the string. The lowest order interactions are dictated by symmetries and couple these degrees of freedom to geometric invariants of the induced worldsheet. However, symmetry alone does not determine the localized spectrum or the value of coupling coefficients. This requires matching to the underlying four dimensional theory. A microscopic derivation is therefore essential for identifying the relevant degrees of freedom and their interactions beyond the Nambu–Goto approximation. When localized massive modes are dynamically accessible, including them explicitly allows the resulting effective action to describe their production and subsequent energy exchange with the translational sector, providing a framework for assessing their phenomenological implications.
The Abelian-Higgs (AH) model provides a concrete setting for this program through its Nielsen–Olesen vortex strings [25 ] . Their localized spectrum includes a scalar shape mode [16 , 26 , 27 ] and a distinct vector excitation associated with the gauge field polarization along the string [17 , 28 , 29 ] . At critical coupling, these modes have equal masses but different dynamical effects. Both lie below the bulk mass threshold, so their decay into bulk radiation proceeds through nonlinear interactions and is suppressed at small amplitudes allowing us to write an effective field theory for them in this regime. Previous work derived the shape mode interactions with the worldsheet geometry, including couplings to its Ricci scalar [16 ] . The vector mode produces a qualitatively different response in the translational sector [17 ] , motivating a detailed and unified description that keeps both massive excitations and their couplings to string motion.
In this paper, we derive a covariant worldsheet effective action for the translational, scalar, and vector modes of the Abelian-Higgs string at critical coupling. In order to do that, we express the microscopic fields in coordinates adapted to the string worldsheet, expand in the amplitudes of the internal modes, and integrate over the transverse directions. Matching directly to the microscopic theory determines their mutual interactions and couplings to the worldsheet geometry at the order considered. All coefficients are expressed as transverse integrals involving the vortex background solution and normalized bound state profiles. This construction relates the internal structure of the string to departures from Nambu–Goto evolution and hence provides a framework for assessing the role of localized excitations in cosmic string dynamics.
The effective action considered here describes the conservative dynamics of the worldsheet degrees of freedom and omits radiation into bulk fields. Furthermore, its application requires evolution times short compared with the characteristic radiative damping times of the excited internal modes. Because their radiative decay arises from nonlinear interactions, these damping times increase as the excitation amplitudes decrease, allowing a controlled description over many oscillation periods in the weak excitation regime.
The resulting action reveals an important distinction between the two massive bound modes present in the Abelian-Higgs model. The shape scalar couples to the intrinsic Ricci scalar of the worldsheet, while the vector field strength couples to the curvature of the normal bundle, which describes the rotation of the transverse frame along the string. In two spacetime dimensions, the massive vector mode carries one propagating degree of freedom and admits an equivalent pseudoscalar description. In this dual formulation, the two internal excitations couple to the intrinsic and normal curvature densities, respectively. The pseudoscalar interaction has the same geometric structure as the worldsheet axion coupling proposed for confining QCD flux tubes [30 ] . The Abelian-Higgs vortex thus supplies an explicit microscopic realization of this interaction, with its mass and coupling fixed by the underlying field theory.
We analyze the resulting dynamical consequences and test the effective theory predictions using lattice simulations with adaptive mesh refinement. The geometric couplings account for the distinct parametric instabilities previously observed when the scalar and vector modes are excited [16 , 17 ] . Their expansion about a straight string reproduces the corresponding collective coordinate interactions and explains the different polarization patterns of the induced transverse motion.
Collisions of counterpropagating transverse waves provide a complementary test in which the incoming string carries no internal excitation. Nambu–Goto dynamics determines the leading motion and scattering time delay, while the curvature generated during the collision sources the massive modes. The intrinsic and normal curvature couplings predict specific dependences on the relative response of the shape and vector mode on the polarization of the incoming waves. Comparing these predictions with the internal amplitudes extracted from the lattice simulations tests how worldsheet geometry transfers energy into the microscopic degrees of freedom of the string.
A direct interaction between the two internal modes also predicts the shape excitation generated by a coherent vector oscillation. We isolate this process numerically and compare its amplitude and time dependence with the effective action description.
The outline of the paper is the following. We begin by reviewing the vortex background and its localized excitation spectrum. We then derive the covariant worldsheet action from the microscopic theory and dualize the vector mode into a pseudoscalar. Next, we compare the predictions of the effective theory with lattice simulations in three settings: the parametric excitation of the Goldstone modes by a homogeneous bound mode, the excitation of the bound modes by colliding transverse waves, and the direct coupling between the vector and shape modes. We conclude by summarizing our results and discussing their implications for string dynamics. The appendices contain the technical details: Appendix A provides the detailed microscopic derivation of the effective action, Appendix B compares it with the collective coordinate approach, Appendix C provides an in depth discussion on the Nambu–Goto scattering of the Goldstone modes analytically, and derives the bound mode response it induces. Finally, Appendix D describes the numerical methodology and convergence tests.
String background and excitation spectrum
In this paper, we consider the Abelian-Higgs model described by the 4 d 4d 4 d action
S = − ∫ d 4 x [ 1 4 F μ ν F μ ν + 1 2 ( D μ ϕ ) ∗ D μ ϕ + λ 8 ( ∣ ϕ ∣ 2 − η 2 ) 2 ] , (1) S=-\int d^4x\left[\frac{1}{4}F_{\mu\nu}F^{\mu\nu}+\frac{1}{2}(D_\mu\phi)^*D^\mu\phi+\frac{\lambda}{8}\left(|\phi|^2-\eta^2\right)^2\right], \tag*{(1)} S = − ∫ d 4 x [ 4 1 F μν F μν + 2 1 ( D μ ϕ ) ∗ D μ ϕ + 8 λ ( ∣ ϕ ∣ 2 − η 2 ) 2 ] , (1)
where D μ = ∂ μ − i e A μ D_\mu= \partial_\mu- ieA_\mu D μ = ∂ μ − i e A μ , F μ ν = ∂ μ A ν − ∂ ν A μ F_{\mu\nu} = \partial_\mu A_\nu- \partial_\nu A_\mu F μν = ∂ μ A ν − ∂ ν A μ , and we will use η μ ν = diag ( − , + , + , + ) \eta_{\mu\nu} = \operatorname{diag}(-,+,+,+) η μν = diag ( − , + , + , + ) . The symmetry breaking vacuum ∣ ϕ ∣ = η |\phi| = \eta ∣ ϕ ∣ = η has Higgs and gauge masses given by m H = λ η m_H = \sqrt{\lambda}\eta m H = λ η and m A = e η m_A = e\eta m A = eη . For simplicity, we introduce the dimensionless quantities x ~ μ = e η x μ \tilde{x}^{\mu} = e\eta x^\mu x ~ μ = eη x μ , ϕ ~ = ϕ / η \tilde{\phi} = \phi/\eta ϕ ~ = ϕ / η , and A ~ μ = A μ / η \tilde{A}_\mu= A_\mu/\eta A ~ μ = A μ / η , and we will suppress tildes below. Thus the action carries an overall factor e − 2 e^{-2} e − 2 and the classical equations depend on β = λ / e 2 = ( m H / m A ) 2 \beta= \lambda/e^2 = (m_H/m_A)^2 β = λ / e 2 = ( m H / m A ) 2 where masses and frequencies are measured in units of e η e\eta eη . One can show that within this theory, a static, straight, unit winding Nielsen–Olesen string along the z z z direction can be described by the following ansatz
ϕ ˉ = f ( ρ ) e i θ , A ˉ i = α ( ρ ) ∂ i θ , A ˉ t = A ˉ z = 0 , (2) \bar{\phi} = f(\rho)e^{i\theta}, \qquad\bar{A}_i = \alpha(\rho)\partial_i\theta, \qquad\bar{A}_t = \bar{A}_z = 0, \tag*{(2)} ϕ ˉ = f ( ρ ) e i θ , A ˉ i = α ( ρ ) ∂ i θ , A ˉ t = A ˉ z = 0 , (2)
where i = x , y i = x,y i = x , y , ρ 2 = x 2 + y 2 \rho^2 = x^2+y^2 ρ 2 = x 2 + y 2 , are the transverse directions to the string, and θ \theta θ is the polar angle [25 ] . Regularity and finite tension require f ( 0 ) = α ( 0 ) = 0 f(0) = \alpha(0) = 0 f ( 0 ) = α ( 0 ) = 0 and f ( ∞ ) = α ( ∞ ) = 1 f(\infty) = \alpha(\infty) = 1 f ( ∞ ) = α ( ∞ ) = 1 . At critical coupling, β = 1 \beta= 1 β = 1 , the string saturates the Bogomolny-Prasad-Sommerfield (BPS) bound and the profile functions obey
f ′ = 1 − α ρ f , α ′ = ρ 2 ( 1 − f 2 ) . (3) f' = \frac{1-\alpha}{\rho}f, \qquad\alpha' = \frac{\rho}{2}(1-f^2). \tag*{(3)} f ′ = ρ 1 − α f , α ′ = 2 ρ ( 1 − f 2 ) . (3)
where primes denote derivatives with respect to the radial coordinate perpendicular to the string, ρ \rho ρ .
Physical linear perturbations around the string solution and proportional to e − i Ω t + i k z e^{-i\Omega t+ikz} e − i Ω t + ik z satisfy the dispersion relation Ω 2 = k 2 + ω I 2 \Omega^2 = k^2 + \omega_I^2 Ω 2 = k 2 + ω I 2 , where ω I \omega_I ω I denotes the worldsheet rest mass for the three different kinds of normal modes present in the Abelian-Higgs model string. The two translational zero modes (Goldstone modes) correspond to displacements of the vortex in the transverse plane. They have ω 0 = 0 \omega_0 = 0 ω 0 = 0 and propagate as massless transverse waves. In addition, the BPS string in the Abelian-Higgs model supports only two other localized massive modes.
The scalar shape mode that describes an axially symmetric deformation of the form [16 , 26 , 27 , 31 ]
δ ϕ = χ ( t , z ) φ ( ρ ) e i θ , δ A i = χ ( t , z ) ρ a ( ρ ) ∂ i θ , (4) \delta\phi= \chi(t,z)\varphi(\rho)e^{i\theta}, \qquad\delta A_i = \chi(t,z)\rho a(\rho)\partial_i\theta, \tag*{(4)} δ ϕ = χ ( t , z ) φ ( ρ ) e i θ , δ A i = χ ( t , z ) ρ a ( ρ ) ∂ i θ , (4)
with δ A t = δ A z = 0 \delta A_t = \delta A_z = 0 δ A t = δ A z = 0 . Here, the real functions φ ( ρ ) \varphi(\rho) φ ( ρ ) and a ( ρ ) a(\rho) a ( ρ ) specify its transverse profiles, while the real worldsheet massive scalar χ ( t , z ) \chi(t,z) χ ( t , z ) describes its amplitude and propagation along the string.
The vector mode has components along the worldsheet coordinates σ A = ( t , z ) \sigma^A = (t,z) σ A = ( t , z ) ,
δ A A = C A ( t , z ) ψ ( ρ ) , (5) \delta A_A = C_A(t,z)\psi(\rho), \tag*{(5)} δ A A = C A ( t , z ) ψ ( ρ ) , (5)
with δ ϕ = δ A i = 0 \delta\phi= \delta A_i = 0 δ ϕ = δ A i = 0 . Here ψ ( ρ ) \psi(\rho) ψ ( ρ ) is the radial profile, and the worldsheet vector C A ( t , z ) C_A(t,z) C A ( t , z ) satisfies the constraint ∂ A C A = 0 \partial^A C_A = 0 ∂ A C A = 0 .
This mode carries one physical massive vector polarization whose worldsheet vector mode is described by C A C_A C A and its radial profile satisfies [17 ]
− ψ ′ ′ − 1 ρ ψ ′ + f 2 ψ = ω v 2 ψ . (6) -\psi'' - \frac{1}{\rho}\psi' + f^2\psi= \omega_v^2\psi. \tag*{(6)} − ψ ′′ − ρ 1 ψ ′ + f 2 ψ = ω v 2 ψ . (6)
The bound solution is regular and square integrable so we can normalize it by choosing
2 π ∫ 0 ∞ ρ d ρ ψ 2 = 1. (7) 2\pi\int_0^\infty\rho\,d\rho\,\psi^2 = 1. \tag*{(7)} 2 π ∫ 0 ∞ ρ d ρ ψ 2 = 1. (7)
At critical coupling, factorization of the fluctuation operators relates the shape and vector mode profiles and establishes their mass degeneracy [17 ] :
φ = f ψ ω b , a = − ψ ′ ω b , ω s = ω v = ω b . (8) \varphi= \frac{f\psi}{\omega_b}, \qquad a = -\frac{\psi'}{\omega_b}, \qquad\omega_s = \omega_v = \omega_b. \tag*{(8)} φ = ω b f ψ , a = − ω b ψ ′ , ω s = ω v = ω b . (8)
Here ω s \omega_s ω s and ω v \omega_v ω v denote the worldsheet rest masses of the shape and vector modes, respectively, and ω b \omega_b ω b denotes their common value throughout the remainder of this work. The mode relations in Eq. (8) , together with the eigenvalue equation (6), imply, after integration by parts,
2 π ∫ 0 ∞ ρ d ρ ( φ 2 + a 2 ) = 2 π ∫ 0 ∞ ρ d ρ ψ 2 = 1 , (9) 2\pi\int_0^\infty\rho\,d\rho\,(\varphi^2+a^2) = 2\pi\int_0^\infty\rho\,d\rho\,\psi^2 = 1, \tag*{(9)} 2 π ∫ 0 ∞ ρ d ρ ( φ 2 + a 2 ) = 2 π ∫ 0 ∞ ρ d ρ ψ 2 = 1 , (9)
where the boundary terms vanish by regularity and normalizability, and the final equality follows from Eq. (7) . Numerically, ω b ≃ 0.882 \omega_b \simeq0.882 ω b ≃ 0.882 [17 ] . Restoring dimensions, the localized modes have masses m 0 = 0 m_0 = 0 m 0 = 0 and m s = m v ≃ 0.882 e η m_s = m_v \simeq0.882\,e\eta m s = m v ≃ 0.882 eη . Both massive modes lie below the bulk threshold m H = m A = e η m_H = m_A = e\eta m H = m A = eη and therefore cannot radiate into the bulk at linear order. Their classical decay is mediated by nonlinear interactions and is suppressed at small amplitudes, allowing these excitations to persist for many oscillation periods [17 , 27 ] .
Microscopic derivation of the worldsheet effective action
To separate the motion of the string from changes in its internal structure, we introduce coordinates adapted to the string worldsheet, following the method introduced in Ref. [16 , 32 ] . The embedding X μ ( σ A ) X^\mu(\sigma^A) X μ ( σ A ) , with A = { 0 , 1 } A = \{0,1\} A = { 0 , 1 } , specifies the string position and defines a local decomposition into tangent and normal directions. In the local rest frame of the string, we approximate the fields in each normal plane by the straight BPS vortex profiles supplemented by the retained bound modes. Substituting this ansatz into the microscopic four dimensional action and integrating over the transverse coordinates yields the effective worldsheet action and its geometric couplings. We outline the construction of this effective action here and present the detailed derivation of each term in Appendix A.
Let us start by introducing some geometric quantities related to the string worldsheet, in particular the tangent vectors e A μ = ∂ A X μ e^\mu_A = \partial_A X^\mu e A μ = ∂ A X μ and the oriented spacelike normals satisfying n i μ n j μ = δ i j n^\mu_i n_{j\mu} = \delta_{ij} n i μ n j μ = δ ij and e A μ n i μ = 0 e^\mu_A n_{i\mu} = 0 e A μ n i μ = 0 , where Greek indices refer to four adopted coordinates, A , B = { 0 , 1 } A,B = \{0,1\} A , B = { 0 , 1 } to the worldsheet, and i , j = { 1 , 2 } i,j = \{1,2\} i , j = { 1 , 2 } to the normal directions. Using these vectors, we can describe the position of points near the string by
y μ ( σ , u ) = X μ ( σ ) + u i n i μ ( σ ) , ρ 2 = δ i j u i u j . (10) y^\mu(\sigma,u)=X^\mu(\sigma)+u^i n_i^\mu(\sigma),\qquad\rho^2=\delta_{ij}u^i u^j. \tag*{(10)} y μ ( σ , u ) = X μ ( σ ) + u i n i μ ( σ ) , ρ 2 = δ ij u i u j . (10)
This construction applies when the curvature radii are large compared with the transverse widths of the specified modes. The induced metric is γ A B = e A μ e B μ \gamma_{AB}=e_A^\mu e_{B\mu} γ A B = e A μ e B μ . Bending of the worldsheet and rotation of its normal frame (the so-called twist ) are described by K A B i = − n i μ ∂ A e B μ K^i_{AB}=-n_i^\mu\partial_A e_B^\mu K A B i = − n i μ ∂ A e B μ and ω A = n 1 μ ∂ A n 2 μ \omega_A=n_{1\mu}\partial_A n_2^\mu ω A = n 1 μ ∂ A n 2 μ , respectively. Writing K i = γ A B K A B i K^i=\gamma^{AB}K^i_{AB} K i = γ A B K A B i , we define the intrinsic Ricci scalar and the curvature of the normal connection by
R = K i K i − K A B i K i A B , Ω A B = ∇ A ω B − ∇ B ω A . (11) \mathcal{R}=K^iK_i-K^i_{AB}K_i^{AB},\qquad\Omega_{AB}=\nabla_A\omega_B-\nabla_B\omega_A. \tag*{(11)} R = K i K i − K A B i K i A B , Ω A B = ∇ A ω B − ∇ B ω A . (11)
To compute the effective action, we allow the amplitudes of the shape and vector modes to vary along the string. Choosing the Higgs phase to be the polar angle θ ( u ) \theta(u) θ ( u ) , we write the fields in adapted coordinates as
ϕ = [ f ( ρ ) + φ ( ρ ) χ ( σ ) ] e i θ , A i = [ α ( ρ ) + ρ a ( ρ ) χ ( σ ) ] ∂ i θ , A A = [ 1 − α ( ρ ) − ρ a ( ρ ) χ ( σ ) ] ω A + ψ ( ρ ) C A ( σ ) , (12) \begin{aligned}
\phi&=[f(\rho)+\varphi(\rho)\chi(\sigma)]e^{i\theta},\\
A_i&=[\alpha(\rho)+\rho a(\rho)\chi(\sigma)]\partial_i\theta,\\
A_A&=[1-\alpha(\rho)-\rho a(\rho)\chi(\sigma)]\omega_A+\psi(\rho)C_A(\sigma), \tag*{(12)}
\end{aligned} ϕ A i A A = [ f ( ρ ) + φ ( ρ ) χ ( σ )] e i θ , = [ α ( ρ ) + ρ a ( ρ ) χ ( σ )] ∂ i θ , = [ 1 − α ( ρ ) − ρ a ( ρ ) χ ( σ )] ω A + ψ ( ρ ) C A ( σ ) , (12)
where the radial profiles keep the normalizations of the preceding section, while χ \chi χ and C A C_A C A describe the internal worldsheet excitations. Even in the absence of these excitations, the background contains the component A ˉ A = ( 1 − α ) ω A \bar{A}_A=(1-\alpha)\omega_A A ˉ A = ( 1 − α ) ω A . This component accounts for the rotation of the normal frame and ensures covariance under the accompanying U ( 1 ) U(1) U ( 1 ) gauge transformation.
Inserting Eq. (12) into the microscopic four dimensional action and integrating over the transverse directions, u i u^i u i , determines the effective couplings [16 , 17 ] . We organize the calculation of the effective action through successive field variations at a fixed embedding. The contribution of degree n n n in χ \chi χ and C A C_A C A is S ( n ) = δ n S / n ! S^{(n)}=\delta^nS/n! S ( n ) = δ n S / n ! , evaluated on the background vortex configuration. The successive variations and their reduction are presented in Appendix A.
At zeroth order in the internal amplitudes, the background energy gives the Nambu–Goto action with dimensionless tension μ = π \mu=\pi μ = π , corresponding to physical tension π η 2 \pi\eta^2 π η 2 . It also generates an Euler term with a constant coefficient, which does not affect the local equations of motion and is only included below for completeness.
The first variation need not vanish because the vortex ansatz is not an exact solution for a general curved worldsheet. At linear order, the straight vortex equations cancel the terms independent of curvature. The remaining terms describe how the string geometry excites its internal modes. The transverse measure produces the scalar coupling χ R \chi\mathcal{R} χ R , while the cross term between F ˉ A B = ( 1 − α ) Ω A B \bar{F}_{AB}=(1-\alpha)\Omega_{AB} F ˉ A B = ( 1 − α ) Ω A B and the vector fluctuation δ F A B = ψ G A B \delta F_{AB}=\psi G_{AB} δ F A B = ψ G A B produces a term of the form G A B Ω A B G_{AB}\Omega^{AB} G A B Ω A B . Here G A B = ∇ A C B − ∇ B C A G_{AB}=\nabla_A C_B-\nabla_B C_A G A B = ∇ A C B − ∇ B C A is the worldsheet field strength of the vector mode. Note that this coupling respects the local rotation symmetry of the normal frame, since the twist enters through the invariant normal bundle curvature Ω A B \Omega_{AB} Ω A B .
At quadratic order, the transverse eigenvalue equations combine the radial gradient and potential contributions into worldsheet mass terms. This gives a massive scalar and a Proca field, with kinetic coefficients fixed by the profile normalizations and a common mass ω b \omega_b ω b fixed by the transverse eigenvalue problem.
At cubic order, a direct interaction arises from the scalar kinetic term. The shape deformation changes the Higgs profile and hence the mass term experienced by the vector mode, producing an interaction proportional to χ C A C A \chi C_A C^A χ C A C A . At the same order, the scalar kinetic term and the Higgs potential generate a scalar self interaction proportional to χ 3 \chi^3 χ 3 , which we also include for consistency of the cubic truncation. Note that the BPS profile equations relate the cubic scalar self interaction to the mixed scalar-vector coupling g S V g_{SV} g S V .
Collecting the leading quadratic terms, the geometric sources, and the cubic interaction gives
S e f f = − 1 e 2 ∫ d 2 σ − γ [ μ + ν R − κ Ω Ω A B Ω A B + 1 2 ∇ A χ ∇ A χ + ω b 2 2 χ 2 + 1 4 G A B G A B + ω b 2 2 C A C A − κ S χ R + κ V 2 G A B Ω A B + g S V χ C A C A + 1 2 g S V χ 3 + ⋯ ] . (13) \begin{aligned}
S_{\mathrm{eff}}={}&-\frac{1}{e^2}\int d^2\sigma\sqrt{-\gamma}\left[\mu+\nu\mathcal{R}-\kappa_\Omega\Omega_{AB}\Omega^{AB}\right.\\
&\left.+\frac{1}{2}\nabla_A\chi\nabla^A\chi+\frac{\omega_b^2}{2}\chi^2\right.\\
&\left.+\frac{1}{4}G_{AB}G^{AB}+\frac{\omega_b^2}{2}C_A C^A\right.\\
&\left.-\kappa_S\chi\mathcal{R}+\frac{\kappa_V}{2}G_{AB}\Omega^{AB}\right.\\
&\left.+g_{SV}\chi C_A C^A+\frac{1}{2}g_{SV}\chi^3+\cdots\right]. \tag*{(13)}
\end{aligned} S eff = − e 2 1 ∫ d 2 σ − γ [ μ + ν R − κ Ω Ω A B Ω A B + 2 1 ∇ A χ ∇ A χ + 2 ω b 2 χ 2 + 4 1 G A B G A B + 2 ω b 2 C A C A − κ S χ R + 2 κ V G A B Ω A B + g S V χ C A C A + 2 1 g S V χ 3 + ⋯ ] . (13)
where the coefficients are determined by the background and bound state profiles,
μ = 2 π ∫ 0 ∞ d ρ ( ρ f ′ 2 + α ′ 2 ρ ) = π , \mu=2\pi\int_0^\infty d\rho\left(\rho f'^2+\frac{\alpha'^2}{\rho}\right)=\pi, μ = 2 π ∫ 0 ∞ d ρ ( ρ f ′2 + ρ α ′2 ) = π ,
ν = π 2 ∫ 0 ∞ d ρ ( ρ 3 f ′ 2 + ρ α ′ 2 ) = π 2 , \nu=\frac{\pi}{2}\int_0^\infty d\rho\left(\rho^3f'^2+\rho\alpha'^2\right)=\frac{\pi}{2}, ν = 2 π ∫ 0 ∞ d ρ ( ρ 3 f ′2 + ρ α ′2 ) = 2 π ,
κ S = π ∫ 0 ∞ d ρ ( ρ 2 f ′ φ ′ + ρ α ′ a ′ ) ≈ 1.2005 , \kappa_S=\pi\int_0^\infty d\rho\left(\rho^2f'\varphi'+\rho\alpha'a'\right)\approx1.2005, κ S = π ∫ 0 ∞ d ρ ( ρ 2 f ′ φ ′ + ρ α ′ a ′ ) ≈ 1.2005 ,
κ V = 2 π ∫ 0 ∞ ρ d ρ ( 1 − α ) ψ ≈ 2.7228 , \kappa_V=2\pi\int_0^\infty\rho\,d\rho(1-\alpha)\psi\approx2.7228, κ V = 2 π ∫ 0 ∞ ρ d ρ ( 1 − α ) ψ ≈ 2.7228 ,
κ Ω = − π 2 ∫ 0 ∞ ρ d ρ ( 1 − α ) 2 ≈ − 1.9503 , \kappa_\Omega=-\frac{\pi}{2}\int_0^\infty\rho\,d\rho(1-\alpha)^2\approx-1.9503, κ Ω = − 2 π ∫ 0 ∞ ρ d ρ ( 1 − α ) 2 ≈ − 1.9503 ,
g S V = 2 π ∫ 0 ∞ ρ d ρ f φ ψ ≈ 0.0864. (14) g_{SV}=2\pi\int_0^\infty\rho\,d\rho f\varphi\psi\approx0.0864. \tag*{(14)} g S V = 2 π ∫ 0 ∞ ρ d ρ f φ ψ ≈ 0.0864. (14)
The ellipsis in Eq. (13) includes additional geometric corrections and interactions among the internal modes. In a simultaneous weak field expansion about a straight string, the transverse displacement, χ \chi χ , and C A C_A C A are all counted as first order, whereas R \mathcal{R} R and Ω A B \Omega_{AB} Ω A B begin at second order. The displayed geometric couplings and the direct mixed interaction therefore enter at cubic order in this counting. The detailed derivation of this effective action, including the additional amplitude and curvature terms, is given in Appendix A.
Dualization of the vector mode
In two spacetime dimensions, a massive vector field carries a single propagating degree of freedom and admits an equivalent pseudoscalar description. We derive this dual formulation and describe its geometric interpretation, keeping the dependence on the shape mode amplitude explicit.
We begin with the terms in the effective action involving the worldsheet vector,
S V = − 1 e 2 ∫ d 2 σ − γ [ 1 4 G A B G A B + 1 2 M 2 ( χ ) C A C A + κ V 2 G A B Ω A B ] . (15) S_V=-\frac{1}{e^2}\int d^2\sigma\sqrt{-\gamma}\left[\frac{1}{4}G_{AB}G^{AB}+\frac{1}{2}M^2(\chi)C_AC^A+\frac{\kappa_V}{2}G_{AB}\Omega^{AB}\right]. \tag*{(15)} S V = − e 2 1 ∫ d 2 σ − γ [ 4 1 G A B G A B + 2 1 M 2 ( χ ) C A C A + 2 κ V G A B Ω A B ] . (15)
where
M 2 ( χ ) = ω b 2 + 2 g S V χ (16) M^2(\chi)=\omega_b^2+2g_{SV}\chi \tag*{(16)} M 2 ( χ ) = ω b 2 + 2 g S V χ (16)
includes the mixed interaction between the shape and vector modes, and as we mentioned earlier, G A B = ∇ A C B − ∇ B C A G_{AB}=\nabla_AC_B-\nabla_BC_A G A B = ∇ A C B − ∇ B C A .
Now, in order to dualize this theory, we treat G A B G_{AB} G A B as an independent antisymmetric tensor and introduce a pseudoscalar Lagrange multiplier ϑ \vartheta ϑ through the term
Δ S = ω b 2 e 2 ∫ d 2 σ − γ ε A B ( G A B − ∇ A C B + ∇ B C A ) . (17) \Delta S=\frac{\omega_b}{2e^2}\int d^2\sigma\sqrt{-\gamma}\,\varepsilon^{AB}\left(G_{AB}-\nabla_AC_B+\nabla_BC_A\right). \tag*{(17)} Δ S = 2 e 2 ω b ∫ d 2 σ − γ ε A B ( G A B − ∇ A C B + ∇ B C A ) . (17)
Varying with respect to ϑ \vartheta ϑ enforces the original relation between G A B G_{AB} G A B and C A C_A C A , since an antisymmetric tensor in two dimensions has only one independent component.
Using the antisymmetry of ε A B \varepsilon^{AB} ε A B and integrating the terms containing derivatives of C A C_A C A by parts gives, up to a boundary term,
S V + Δ S = − 1 e 2 ∫ d 2 σ − γ [ 1 4 G A B G A B + 1 2 M 2 ( χ ) C A C A + κ V 2 G A B Ω A B − ω b 2 ϑ ε A B G A B − ω b ε A B ( ∇ A ϑ ) C B ] . (18) S_V+\Delta S=-\frac{1}{e^2}\int d^2\sigma\sqrt{-\gamma}\left[\frac{1}{4}G_{AB}G^{AB}+\frac{1}{2}M^2(\chi)C_AC^A+\frac{\kappa_V}{2}G_{AB}\Omega^{AB}-\frac{\omega_b}{2}\vartheta\varepsilon^{AB}G_{AB}-\omega_b\varepsilon^{AB}(\nabla_A\vartheta)C_B\right]. \tag*{(18)} S V + Δ S = − e 2 1 ∫ d 2 σ − γ [ 4 1 G A B G A B + 2 1 M 2 ( χ ) C A C A + 2 κ V G A B Ω A B − 2 ω b ϑ ε A B G A B − ω b ε A B ( ∇ A ϑ ) C B ] . (18)
Both G A B G_{AB} G A B and C A C_A C A now appear without derivatives. Their algebraic equations give
G A B = ω b ϑ ε A B − κ V Ω A B , C B = ω b M 2 ( χ ) ε A B ∇ A ϑ . (19) G_{AB}=\omega_b\vartheta\varepsilon_{AB}-\kappa_V\Omega_{AB},
\qquad
C^B=\frac{\omega_b}{M^2(\chi)}\varepsilon^{AB}\nabla_A\vartheta. \tag*{(19)} G A B = ω b ϑ ε A B − κ V Ω A B , C B = M 2 ( χ ) ω b ε A B ∇ A ϑ . (19)
Contracting the first relation with ε A B \varepsilon^{AB} ε A B yields
ϑ = − 1 2 ω b ε A B ( G A B + κ V Ω A B ) . (20) \vartheta=-\frac{1}{2\omega_b}\varepsilon^{AB}\left(G_{AB}+\kappa_V\Omega_{AB}\right). \tag*{(20)} ϑ = − 2 ω b 1 ε A B ( G A B + κ V Ω A B ) . (20)
The contraction with the Levi-Civita tensor identifies ϑ \vartheta ϑ as a pseudoscalar. We define the dual normal curvature by
T ≡ 1 2 ε A B Ω A B = 1 2 ε A B ϵ i j K A C i K B j C , (21) T\equiv\frac{1}{2}\varepsilon^{AB}\Omega_{AB}=\frac{1}{2}\varepsilon^{AB}\epsilon_{ij}K^i_{AC}K^{jC}_{\ \ B}, \tag*{(21)} T ≡ 2 1 ε A B Ω A B = 2 1 ε A B ϵ ij K A C i K B j C , (21)
where the second equality follows from the Ricci equation in flat spacetime. On the algebraic solutions (19), the terms involving G A B G_{AB} G A B in the bracket of (18) combine as
1 4 G A B G A B + κ V 2 G A B Ω A B − ω b 2 ϑ ε A B G A B = − 1 4 G A B G A B = ω b 2 2 ϑ 2 + ω b κ V ϑ T − κ V 2 4 Ω A B Ω A B . (22) \frac{1}{4}G_{AB}G^{AB}+\frac{\kappa_V}{2}G_{AB}\Omega^{AB}-\frac{\omega_b}{2}\vartheta\varepsilon^{AB}G_{AB}
=-\frac{1}{4}G_{AB}G^{AB}
=\frac{\omega_b^2}{2}\vartheta^2+\omega_b\kappa_V\vartheta T-\frac{\kappa_V^2}{4}\Omega_{AB}\Omega^{AB}. \tag*{(22)} 4 1 G A B G A B + 2 κ V G A B Ω A B − 2 ω b ϑ ε A B G A B = − 4 1 G A B G A B = 2 ω b 2 ϑ 2 + ω b κ V ϑT − 4 κ V 2 Ω A B Ω A B . (22)
Similarly, the terms involving C A C_A C A give
1 2 M 2 ( χ ) C A C A − ω b ε A B ( ∇ A ϑ ) C B = − 1 2 M 2 ( χ ) C A C A = ω b 2 2 M 2 ( χ ) ∇ A ϑ ∇ A ϑ , (23) \frac{1}{2}M^2(\chi)C_AC^A-\omega_b\varepsilon^{AB}(\nabla_A\vartheta)C_B
=-\frac{1}{2}M^2(\chi)C_AC^A
=\frac{\omega_b^2}{2M^2(\chi)}\nabla_A\vartheta\nabla^A\vartheta, \tag*{(23)} 2 1 M 2 ( χ ) C A C A − ω b ε A B ( ∇ A ϑ ) C B = − 2 1 M 2 ( χ ) C A C A = 2 M 2 ( χ ) ω b 2 ∇ A ϑ ∇ A ϑ , (23)
where again we have used (19) .
Substituting these expressions into the first order action gives the dual vector sector,
S d u a l , V = − 1 e 2 ∫ d 2 σ − γ [ ω b 2 2 M 2 ( χ ) ∇ A ϑ ∇ A ϑ + ω b 2 2 ϑ 2 + ω b κ V ϑ T − κ V 2 4 Ω A B Ω A B ] . (24) S_{\mathrm{dual},V}=-\frac{1}{e^2}\int d^2\sigma\sqrt{-\gamma}\left[\frac{\omega_b^2}{2M^2(\chi)}\nabla_A\vartheta\nabla^A\vartheta+\frac{\omega_b^2}{2}\vartheta^2+\omega_b\kappa_V\vartheta T-\frac{\kappa_V^2}{4}\Omega_{AB}\Omega^{AB}\right]. \tag*{(24)} S dual , V = − e 2 1 ∫ d 2 σ − γ [ 2 M 2 ( χ ) ω b 2 ∇ A ϑ ∇ A ϑ + 2 ω b 2 ϑ 2 + ω b κ V ϑT − 4 κ V 2 Ω A B Ω A B ] . (24)
This expression is exact for the vector terms displayed in (15) . We restrict ourselves to the regime where M 2 ( χ ) > 0 M^2(\chi)>0 M 2 ( χ ) > 0 , for which the pseudoscalar kinetic coefficient is positive. About the unexcited straight string, where χ = 0 \chi=0 χ = 0 , the chosen normalization gives ϑ \vartheta ϑ the same quadratic kinetic normalization as χ \chi χ and mass ω b \omega_b ω b .
To recover the expansion through cubic order in the worldsheet fields, we assume ∣ 2 g S V χ ∣ ≪ ω b 2 \lvert2g_{SV}\chi\rvert\ll\omega_b^2 ∣ 2 g S V χ ∣ ≪ ω b 2 and expand
ω b 2 2 M 2 ( χ ) = 1 2 − g S V ω b 2 χ + O ( χ 2 ) . (25) \frac{\omega_b^2}{2M^2(\chi)}=\frac{1}{2}-\frac{g_{SV}}{\omega_b^2}\chi+O(\chi^2). \tag*{(25)} 2 M 2 ( χ ) ω b 2 = 2 1 − ω b 2 g S V χ + O ( χ 2 ) . (25)
This expansion is consistent with the small amplitude approximation used in constructing the effective action. The term linear in χ \chi χ generates the mixed derivative interaction − g S V χ ∇ A ϑ ∇ A ϑ / ω b 2 -g_{SV}\chi\nabla_A\vartheta\nabla^A\vartheta/\omega_b^2 − g S V χ ∇ A ϑ ∇ A ϑ / ω b 2 inside the action integrand. Restoring the remaining terms in (13) , including the cubic scalar self-interaction 1 2 g S V χ 3 \frac{1}{2}g_{SV}\chi^3 2 1 g S V χ 3 , gives
S d u a l = − 1 e 2 ∫ d 2 σ − γ [ μ + ν R + 1 2 ∇ A χ ∇ A χ + 1 2 ∇ A ϑ ∇ A ϑ + ω b 2 2 ( χ 2 + ϑ 2 ) + 1 2 g S V χ 3 − κ S χ R + ω b κ V ϑ T − g S V ω b 2 χ ∇ A ϑ ∇ A ϑ − ( κ Ω + κ V 2 4 ) Ω A B Ω A B + ⋯ ] . (26) S_{\mathrm{dual}}=-\frac{1}{e^2}\int d^2\sigma\sqrt{-\gamma}\left[\mu+\nu\mathcal{R}
+\frac{1}{2}\nabla_A\chi\nabla^A\chi+\frac{1}{2}\nabla_A\vartheta\nabla^A\vartheta
+\frac{\omega_b^2}{2}\left(\chi^2+\vartheta^2\right)+\frac{1}{2}g_{SV}\chi^3
-\kappa_S\chi\mathcal{R}+\omega_b\kappa_V\vartheta\mathcal{T}
-\frac{g_{SV}}{\omega_b^2}\chi\nabla_A\vartheta\nabla^A\vartheta
-\left(\kappa_\Omega+\frac{\kappa_V^2}{4}\right)\Omega_{AB}\Omega^{AB}+\cdots\right]. \tag*{(26)} S dual = − e 2 1 ∫ d 2 σ − γ [ μ + ν R + 2 1 ∇ A χ ∇ A χ + 2 1 ∇ A ϑ ∇ A ϑ + 2 ω b 2 ( χ 2 + ϑ 2 ) + 2 1 g S V χ 3 − κ S χ R + ω b κ V ϑ T − ω b 2 g S V χ ∇ A ϑ ∇ A ϑ − ( κ Ω + 4 κ V 2 ) Ω A B Ω A B + ⋯ ] . (26)
The contribution proportional to κ V 2 / 4 \kappa_V^2/4 κ V 2 /4 is the local contact term generated by eliminating G A B G_{AB} G A B . It shifts the coefficient of the Ω A B Ω A B \Omega_{AB}\Omega^{AB} Ω A B Ω A B term already present in Eq. (13) , whose κ Ω \kappa_\Omega κ Ω contribution is inherited from the microscopic reduction. The ellipsis denotes additional contributions at this geometric order and interactions beyond the order considered in the amplitude expansion.
Geometric interpretation of the couplings
The dual formulation identifies the geometric origin of the two sources. The scalar χ \chi χ couples to the intrinsic curvature R \mathcal{R} R , while the pseudoscalar ϑ \vartheta ϑ couples to the normal curvature T = 1 2 ϵ A B Ω A B \mathcal{T}=\frac{1}{2}\epsilon^{AB}\Omega_{AB} T = 2 1 ϵ A B Ω A B . These quantities represent, up to normalization, the Euler densities of the tangent and normal bundles, respectively. On a closed, oriented Euclidean worldsheet, their integrals yield the Euler characteristic and the normal Euler number, the latter being related to the signed self-intersection number [33 ] . With constant coefficients, both contributions are topological and leave the local classical equations of motion unchanged, although they may affect boundary terms and the relative weighting of topological sectors. Their coupling to dynamical fields generates local interactions that mediate energy exchange between the string geometry and its internal excitations.
At critical coupling, the geometric coefficients satisfy ω b κ V = 2 κ S \omega_b\kappa_V=2\kappa_S ω b κ V = 2 κ S , as derived in Sec. A 9 of Appendix A. With the field and curvature normalizations adopted here, the magnitude of the pseudoscalar coupling is therefore twice that of the scalar coupling1 . Together with the mass degeneracy, this relation suggests a common origin in the BPS structure of the microscopic vortex.
The pseudoscalar interaction has the same geometric form as the coupling introduced for the worldsheet axion of confining QCD flux tubes [30 , 34 ] . In that context, a massive pseudoscalar coupled to the transverse Goldstone modes accounts for features of the lattice excitation spectrum that are not described by the Nambu–Goto action.
In the present construction, this interaction follows from microscopic matching to the four dimensional Abelian-Higgs theory. The pseudoscalar is dual to the single physical polarization of the localized massive vector mode identified in Ref. [17 ] , with its mass and couplings determined by transverse integrals involving the vortex background and normalized mode profiles. The vector and pseudoscalar formulations thus describe the same localized degree of freedom, providing an explicit microscopic realization of the geometric interaction employed in effective theories of QCD flux tubes.
The topological character of the curvature density also clarifies the axion interpretation. A constant shift of ϑ \vartheta ϑ changes the geometric coupling only by a topological contribution, whereas the mass term explicitly breaks the continuous shift symmetry. The resulting theory provides a microscopic framework for studying axion interactions and energy transfer between transverse motion and internal string excitations. Its geometric structure connects this theory to effective descriptions of QCD flux tubes. The BPS specific coupling relations and scalar spectrum follow from the present model, although similar features may arise in other field theories supporting flux tubes.
Numerical tests of the effective action
At leading order, the effective action reduces to the Nambu–Goto description of the transverse Goldstone modes. Field theory simulations of local strings have tested this description in several settings, including traveling disturbances and loop evolution [35 , 36 , 37 , 38 , 39 , 40 ] . Even when Nambu–Goto dynamics provides the leading approximation, the excitation of internal modes can induce subleading corrections to the string trajectory. Recent simulations with adaptive mesh refinement confirm that Nambu–Goto dynamics governs circular loop collapse at leading order, while resolving subleading corrections arising from the backreaction of the shape excitation, accurately described by its coupling to the intrinsic worldsheet curvature [16 ] .
Localized internal modes can also affect string dynamics through parametric instabilities that transfer energy from internal excitations to transverse motion. This mechanism has been identified in domain wall strings [41 ] , axionic strings [14 ] , and local strings [16 , 17 ] . These findings suggest that such instabilities may be a generic feature of extended solitons supporting bound modes localized in their cores.
For local strings, both the shape and vector modes can amplify transverse Goldstone modes, but with different polarization patterns. A shape excitation amplifies a planar disturbance while preserving its polarization, whereas a vector excitation couples orthogonal transverse directions and produces helical deformations. Our effective action explains this difference through the couplings of the two internal modes to intrinsic and normal curvature, respectively.
In this section we first examine these instabilities and show that their leading growth rates coincide with the predictions of our effective theory. We then consider the complementary process in which colliding transverse disturbances excite the bound modes. Finally, we isolate the direct interaction between the vector and shape modes. Throughout these comparisons, the interaction coefficients take the values obtained from the microscopic calculation outlined earlier. We evolve the Abelian-Higgs equations of motion using GRDzhadzha [42 ] , developed by the GRTL Collaboration (www.grtlcollaboration.org ). Adaptive mesh refinement allows us to resolve the relevant scales, from the string core to the full string configuration.
Parametric excitation of the Goldstone modes
To demonstrate this instability, we consider a straight string with a small, spatially homogeneous excitation of one internal mode and a weak transverse perturbation that seeds the instability. During the initial growth stage, we neglect depletion of the homogeneous excitation and radiation, keeping terms quadratic in the transverse displacement X i ( t , z ) X^i(t,z) X i ( t , z ) and linear in the internal mode amplitude. The relevant interactions are
S i n t = 1 e 2 ∫ d 2 σ − γ [ κ S χ R − κ V 2 G A B Ω A B ] . (27) S_{\mathrm{int}}=\frac{1}{e^2}\int d^2\sigma\sqrt{-\gamma}\left[\kappa_S\chi\mathcal{R}-\frac{\kappa_V}{2}G_{AB}\Omega^{AB}\right]. \tag*{(27)} S int = e 2 1 ∫ d 2 σ − γ [ κ S χ R − 2 κ V G A B Ω A B ] . (27)
For both internal excitations, we consider the transverse Fourier component
X 1 ( t , z ) = X ( t ) cos ( k z ) , X 2 ( t , z ) = Y ( t ) sin ( k z ) , (28) X^1(t,z)=X(t)\cos(kz),\qquad X^2(t,z)=Y(t)\sin(kz), \tag*{(28)} X 1 ( t , z ) = X ( t ) cos ( k z ) , X 2 ( t , z ) = Y ( t ) sin ( k z ) , (28)
where k = 2 π n / L z k=2\pi n/L_z k = 2 π n / L z , with n > 0 n>0 n > 0 , on a periodic interval of length L z L_z L z . This ansatz allows us to compare the evolution of the two transverse amplitudes and their coupling to each internal mode.
For a homogeneous shape oscillation, χ ( t ) = χ 0 cos ( ω b t ) \chi(t)=\chi_0\cos(\omega_b t) χ ( t ) = χ 0 cos ( ω b t ) , the interaction is determined by the quadratic intrinsic curvature,
R ( 2 ) = 2 [ ( ∂ t ∂ z X i ) 2 − ( ∂ t 2 X i ) ( ∂ z 2 X i ) ] . (29) \mathcal{R}^{(2)}=2\left[(\partial_t\partial_zX^i)^2-(\partial_t^2X^i)(\partial_z^2X^i)\right]. \tag*{(29)} R ( 2 ) = 2 [ ( ∂ t ∂ z X i ) 2 − ( ∂ t 2 X i ) ( ∂ z 2 X i ) ] . (29)
Substituting Eq. (28) and integrating over the periodic interval gives
∫ 0 L z d z R ( 2 ) = L z k 2 2 d 2 d t 2 ( X 2 + Y 2 ) . (30) \int_0^{L_z} dz\,\mathcal{R}^{(2)}=\frac{L_zk^2}{2}\frac{d^2}{dt^2}(X^2+Y^2). \tag*{(30)} ∫ 0 L z d z R ( 2 ) = 2 L z k 2 d t 2 d 2 ( X 2 + Y 2 ) . (30)
After two integrations by parts in time, and including the quadratic Nambu–Goto contribution, the transverse
Lagrangian becomes
L ⊥ , S ( 2 ) = L z 4 e 2 { μ [ X ˙ 2 + Y ˙ 2 − k 2 ( X 2 + Y 2 ) ] + 2 κ S k 2 χ ¨ ( X 2 + Y 2 ) } . (31) L_{\perp,S}^{(2)}=\frac{L_z}{4e^2}\left\{\mu\left[\dot{X}^2+\dot{Y}^2-k^2(X^2+Y^2)\right]+2\kappa_S k^2\ddot{\chi}(X^2+Y^2)\right\}. \tag*{(31)} L ⊥ , S ( 2 ) = 4 e 2 L z { μ [ X ˙ 2 + Y ˙ 2 − k 2 ( X 2 + Y 2 ) ] + 2 κ S k 2 χ ¨ ( X 2 + Y 2 ) } . (31)
where boundary terms have been omitted. The two amplitudes are independent and obey identical equations. Writing ξ = X \xi=X ξ = X or Y Y Y , we obtain
ξ ¨ + k 2 [ 1 + 2 κ S ω b 2 χ 0 μ cos ( ω b t ) ] ξ = 0. (32) \ddot{\xi}+k^2\left[1+\frac{2\kappa_S\omega_b^2\chi_0}{\mu}\cos(\omega_b t)\right]\xi=0. \tag*{(32)} ξ ¨ + k 2 [ 1 + μ 2 κ S ω b 2 χ 0 cos ( ω b t ) ] ξ = 0. (32)
This Mathieu equation describes a periodic modulation of the restoring force, with a principal resonance near 2 k = ω b 2k=\omega_b 2 k = ω b . Since the shape excitation does not mix the transverse amplitudes, an initially planar seed with Y = Y ˙ = 0 Y=\dot{Y}=0 Y = Y ˙ = 0 remains planar as it grows. This polarization dependence follows from the contraction of transverse indices with δ i j \delta_{ij} δ ij in R ( 2 ) \mathcal{R}^{(2)} R ( 2 ) .
For a homogeneous vector oscillation, C z = C ( t ) = C 0 cos ( ω b t ) C_z=C(t)=C_0\cos(\omega_b t) C z = C ( t ) = C 0 cos ( ω b t ) and C t = 0 C_t=0 C t = 0 , one has G t z = C ˙ G_{tz}=\dot{C} G t z = C ˙ , and the interaction becomes κ V C Ω t z / e 2 \kappa_V C\Omega_{tz}/e^2 κ V C Ω t z / e 2 . The quadratic normal curvature contains the antisymmetric contraction
Ω A B ( 2 ) = ϵ i j ( ∂ A ∂ C X i ) ( ∂ B ∂ C X j ) , (33) \Omega_{AB}^{(2)}=\epsilon_{ij}(\partial_A\partial^C X^i)(\partial_B\partial_C X^j), \tag*{(33)} Ω A B ( 2 ) = ϵ ij ( ∂ A ∂ C X i ) ( ∂ B ∂ C X j ) , (33)
which couples the two transverse directions. Using the same Fourier ansatz gives
∫ 0 L z d z Ω t z ( 2 ) = − L z k 2 d d t ( X ˙ Y − k 2 X Y ) . (34) \int_0^{L_z} dz\,\Omega_{tz}^{(2)}=-\frac{L_zk}{2}\frac{d}{dt}\left(\dot{X}Y-k^2XY\right). \tag*{(34)} ∫ 0 L z d z Ω t z ( 2 ) = − 2 L z k d t d ( X ˙ Y − k 2 X Y ) . (34)
After an integration by parts in time, the transverse Lagrangian is
L ⊥ , V ( 2 ) = L z 4 e 2 { μ [ X ˙ 2 + Y ˙ 2 − k 2 ( X 2 + Y 2 ) ] + 2 κ V k C ˙ ( X ˙ Y − k 2 X Y ) } . (35) L_{\perp,V}^{(2)}=\frac{L_z}{4e^2}\left\{\mu\left[\dot{X}^2+\dot{Y}^2-k^2(X^2+Y^2)\right]+2\kappa_V k\dot{C}\left(\dot{X}Y-k^2XY\right)\right\}. \tag*{(35)} L ⊥ , V ( 2 ) = 4 e 2 L z { μ [ X ˙ 2 + Y ˙ 2 − k 2 ( X 2 + Y 2 ) ] + 2 κ V k C ˙ ( X ˙ Y − k 2 X Y ) } . (35)
The combinations h ± = ( X ± Y ) / 2 h_\pm=(X\pm Y)/\sqrt{2} h ± = ( X ± Y ) / 2 diagonalize this Lagrangian, with time dependent kinetic coefficients
a ± ( t ) = μ ∓ κ V k ω b 2 C ( t ) . (36) a_\pm(t)=\mu\mp\kappa_V k\omega_b^2C(t). \tag*{(36)} a ± ( t ) = μ ∓ κ V k ω b 2 C ( t ) . (36)
Defining Q ± = a ± / μ h ± Q_\pm=\sqrt{a_\pm/\mu}\,h_\pm Q ± = a ± / μ h ± gives
Q ¨ ± + [ k 2 − a ¨ ± 2 a ± + a ˙ ± 2 4 a ± 2 ] Q ± = 0. (37) \ddot{Q}_\pm+\left[k^2-\frac{\ddot{a}_\pm}{2a_\pm}+\frac{\dot{a}_\pm^2}{4a_\pm^2}\right]Q_\pm=0. \tag*{(37)} Q ¨ ± + [ k 2 − 2 a ± a ¨ ± + 4 a ± 2 a ˙ ± 2 ] Q ± = 0. (37)
To first order in C 0 C_0 C 0 , this becomes
Q ¨ ± + [ k 2 ∓ κ V k ω b 4 C 0 2 μ cos ( ω b t ) ] Q ± = 0. (38) \ddot{Q}_\pm+\left[k^2\mp\frac{\kappa_V k\omega_b^4 C_0}{2\mu}\cos(\omega_b t)\right]Q_\pm=0. \tag*{(38)} Q ¨ ± + [ k 2 ∓ 2 μ κ V k ω b 4 C 0 cos ( ω b t ) ] Q ± = 0. (38)
Both channels have a principal resonance near 2 k = ω b 2k=\omega_b 2 k = ω b . Their opposite modulation signs correspond to different phases relative to the homogeneous oscillation and give equal leading growth rates. The corresponding spatial polarizations are proportional to ( cos k z , ± sin k z ) (\cos kz, \pm\sin kz) ( cos k z , ± sin k z ) and describe helical deformations. A seed in one transverse direction is a superposition of these channels and generically develops an orthogonal component with a relative spatial phase. In the dual formulation, this behavior follows from the pseudoscalar coupling to T T T .
To compare the growth rates, one can write both Equations (32) and (38) as
q ¨ + [ k 2 + b cos ( ω b t ) ] q = 0. (39) \ddot{q} + \left[k^2 + b\cos(\omega_b t)\right]q = 0. \tag*{(39)} q ¨ + [ k 2 + b cos ( ω b t ) ] q = 0. (39)
For weak modulation, the leading Floquet exponent at the center of the principal band, expressed in worldsheet time t t t , is Γ = ∣ b ∣ / ( 2 ω b ) \Gamma= |b|/(2\omega_b) Γ = ∣ b ∣/ ( 2 ω b ) . Equations (32) and (38) therefore give
Γ S = ∣ κ S χ 0 ∣ ω b 3 4 μ , Γ V = ∣ κ V C 0 ∣ ω b 4 8 μ . (40) \Gamma_{\mathrm{S}} = \frac{|\kappa_{\mathrm{S}}\chi_0|\omega_b^3}{4\mu}, \qquad\Gamma_{\mathrm{V}} = \frac{|\kappa_{\mathrm{V}}C_0|\omega_b^4}{8\mu}. \tag*{(40)} Γ S = 4 μ ∣ κ S χ 0 ∣ ω b 3 , Γ V = 8 μ ∣ κ V C 0 ∣ ω b 4 . (40)
Thus, the parametric instabilities driven by the internal mode excitations generate oscillatory transverse displacements with envelopes that grow exponentially as ∼ e Γ S , V t \sim e^{\Gamma_{\mathrm{S,V}}t} ∼ e Γ S , V t .
To test the EFT predictions against field theory simulations, we initialize a straight string with a small, spatially homogeneous excitation of either the shape or the vector mode. The instability is seeded by a transverse displacement initially at rest,
X ( 0 ) = X 0 , X ˙ ( 0 ) = Y ( 0 ) = Y ˙ ( 0 ) = 0 , (41) X(0) = X_0,\qquad\dot{X}(0) = Y(0) = \dot{Y}(0) = 0, \tag*{(41)} X ( 0 ) = X 0 , X ˙ ( 0 ) = Y ( 0 ) = Y ˙ ( 0 ) = 0 , (41)
with wavenumber k = ω b / 2 k = \omega_b/2 k = ω b /2 , corresponding to the center of the principal resonance band at leading order.
For the phases of the homogeneous oscillations specified above, the leading resonant solutions obtained from Eq. (39) in the weak-modulation regime, ∣ b ∣ / k 2 ≪ 1 |b|/k^2 \ll1 ∣ b ∣/ k 2 ≪ 1 , are
X ( t ) ≃ X 0 [ cosh ( Γ S t ) cos ( k t ) − sinh ( Γ S t ) sin ( k t ) ] , Y ( t ) = 0 , (42) \begin{aligned}
X(t) &\simeq X_0\left[\cosh(\Gamma_{\mathrm{S}}t)\cos(kt)-\sinh(\Gamma_{\mathrm{S}}t)\sin(kt)\right],\\
Y(t) &= 0, \tag*{(42)}
\end{aligned} X ( t ) Y ( t ) ≃ X 0 [ cosh ( Γ S t ) cos ( k t ) − sinh ( Γ S t ) sin ( k t ) ] , = 0 , (42)
for a homogeneous shape excitation, and
X ( t ) ≃ X 0 cosh ( Γ V t ) cos ( k t ) , Y ( t ) ≃ X 0 sinh ( Γ V t ) sin ( k t ) , (43) \begin{aligned}
X(t) &\simeq X_0\cosh(\Gamma_{\mathrm{V}}t)\cos(kt),\\
Y(t) &\simeq X_0\sinh(\Gamma_{\mathrm{V}}t)\sin(kt), \tag*{(43)}
\end{aligned} X ( t ) Y ( t ) ≃ X 0 cosh ( Γ V t ) cos ( k t ) , ≃ X 0 sinh ( Γ V t ) sin ( k t ) , (43)
for a homogeneous vector excitation. These expressions retain the cumulative resonant growth while neglecting subleading periodic corrections and higher order corrections to the growth rates.
To quantify the amplification, we define the combined transverse displacement envelope
Env ( X , Y ) ≡ [ X 2 + Y 2 + X ˙ 2 + Y ˙ 2 k 2 ] 1 / 2 ≃ ∣ X 0 ∣ cosh ( 2 Γ t ) . (44) \begin{aligned}
\operatorname{Env}(X,Y) \equiv\left[X^2+Y^2+\frac{\dot{X}^2+\dot{Y}^2}{k^2}\right]^{1/2} \\
&\simeq|X_0|\sqrt{\cosh(2\Gamma t)}. \tag*{(44)}
\end{aligned} Env ( X , Y ) ≡ [ X 2 + Y 2 + k 2 X ˙ 2 + Y ˙ 2 ] 1/2 ≃ ∣ X 0 ∣ cosh ( 2Γ t ) . (44)
The second line follows from Eqs. (42) and (43) at leading order in Γ / k \Gamma/k Γ/ k . At critical coupling, equal normalized
internal-mode amplitudes, ∣ χ 0 ∣ = ∣ C 0 ∣ |\chi_0| = |C_0| ∣ χ 0 ∣ = ∣ C 0 ∣ , imply Γ = Γ S = Γ V \Gamma= \Gamma_{\mathrm{S}} = \Gamma_{\mathrm{V}} Γ = Γ S = Γ V , with the growth rates given by Eq. (40) .
We perform separate simulations with initial shape amplitude χ 0 = 0.1 \chi_0 = 0.1 χ 0 = 0.1 and initial vector amplitude C 0 = 0.1 C_0 = 0.1 C 0 = 0.1 , using the same transverse seed amplitude X 0 = 10 − 3 X_0 = 10^{-3} X 0 = 1 0 − 3 . The lattice simulations reproduce the predicted polarization patterns, as illustrated by the string configurations in Figure 1 . The shape excitation amplifies the initial planar perturbation, while the vector excitation also generates a displacement in the orthogonal transverse direction.
Figure 1. Transverse deformations induced by homogeneous internal mode excitations. The upper and lower panels show the string core for initial shape and vector amplitudes χ 0 = 0.1 \chi_0 = 0.1 χ 0 = 0.1 and C 0 = 0.1 C_0 = 0.1 C 0 = 0.1 , respectively. The shape excitation preserves the seeded planar polarization, whereas the vector excitation generates an orthogonal transverse component, which results in a time dependent helical configuration. A movie is available at https://youtu.be/tI1DMFg6DY .
Figure 2 compares the transverse amplitudes and their combined envelopes with the leading EFT prediction in Eq. (44) . The comparison uses the microscopically determined couplings, without adjusting the predicted growth rates. The observed agreement tests both the polarization dependence and the strength of the geometric interactions at the chosen resonant wavenumber. Note, however, that all comparisons are restricted to the weak excitation regime in which the leading resonant approximation remains applicable. Beyond this regime, additional interactions and the backreaction between the transverse and internal modes must be included, allowing for energy transfer back into the internal excitations (See Refs. [16 , 17 ] for further details.).
Figure 2. Growth of transverse Goldstone excitations seeded by X 0 = 10 − 3 X_0 = 10^{-3} X 0 = 1 0 − 3 at k = ω b / 2 k = \omega_b/2 k = ω b /2 , for initial shape amplitude χ 0 = 0.1 \chi_0 = 0.1 χ 0 = 0.1 and initial vector amplitude C 0 = 0.1 C_0 = 0.1 C 0 = 0.1 in separate simulations. The upper panel shows the transverse amplitudes, while the lower panel compares the logarithms of their combined envelopes with the leading EFT prediction in Eq. (44) . For these equal normalized internal-mode amplitudes, the BPS relation implies Γ S = Γ V \Gamma_S = \Gamma_V Γ S = Γ V , with the common growth rate fixed by Eq. (40) .
In Appendix B, we also show that these cubic vertices reproduce the interactions obtained in the earlier collective coordinate reductions [17 ] . The analytic matching and numerical comparisons thus provide independent checks of the couplings identified in the covariant effective action.
Bound mode excitation by colliding transverse waves
We next consider collisions of counterpropagating transverse disturbances on a straight string, providing a complementary test of the same geometric interactions. The incoming string has no internal excitation, and the overlap of the two waves generates the curvature sources that excite the bound modes. The coefficients have already been determined by microscopic matching and tested by the instabilities described above. The collision experiment therefore uses the same coefficients and tensorial couplings, with no further adjustment. Agreement with both numerical experiments using a single set of microscopically determined coefficients provides a nontrivial validation of the worldsheet effective theory in the regime considered.
Field theory initial data and mode extraction
We initialize each incoming disturbance using an exact Vachaspati-Vachaspati traveling wave solution [43 ] . Each isolated null traveling wave is an exact solution of the field theory equations. Two nonoverlapping disturbances of compact support consequently provide exact separated incoming data. Here we use Gaussian profiles with an initial separation sufficiently large that their overlap is exponentially small.
In the incoming region, the transverse displacement is
X i ( t , z ) ≃ e + i F ( t + z ) + e − i G ( t − z ) , (45) X^i(t,z) \simeq e_+^i F(t+z) + e_-^i G(t-z), \tag*{(45)} X i ( t , z ) ≃ e + i F ( t + z ) + e − i G ( t − z ) , (45)
where the unit polarization vectors satisfy e + ⋅ e − = cos δ e_+ \cdot e_- = \cos\delta e + ⋅ e − = cos δ and ϵ i j e + i e − j = sin δ \epsilon_{ij}e_+^i e_-^j = \sin\delta ϵ ij e + i e − j = sin δ . The F F F wave moves toward decreasing z z z , and the G G G wave toward increasing z z z . For simplicity we take
F ( s ) = G ( s ) = A exp [ − s 2 2 L 2 ] , ϵ ≡ A / L ≪ 1. (46) F(s) = G(s) = A \exp\left[-\frac{s^2}{2L^2}\right], \qquad\epsilon\equiv A/L \ll1. \tag*{(46)} F ( s ) = G ( s ) = A exp [ − 2 L 2 s 2 ] , ϵ ≡ A / L ≪ 1. (46)
Here A A A is the transverse amplitude of each individual incoming wave and L L L is its width. The leading collision is centered at t = z = 0 t = z = 0 t = z = 0 . We choose A = 0.35 A = 0.35 A = 0.35 and L = 4 L = 4 L = 4 , so that the incoming waves are well within the perturbative regime.
We locate the string core from the zeros of the Higgs field on each z z z slice and extract the internal mode amplitudes by projecting the scalar and gauge perturbations onto the normalized straight string eigenmodes. The shape projection includes both the Higgs and transverse gauge contributions, while the vector projection uses the longitudinal gauge field. The reference vortex and the perturbations are expressed in the same gauge. In the regular gauge used for the background, the projections on each transverse slice are
χ ( t , z ) = ∫ d 2 x [ φ ( ρ ) δ ∣ ϕ ∣ + a ( ρ ) δ A θ ] , C ≡ C z ( t , z ) = ∫ d 2 x ψ ( ρ ) A z . (47) \begin{aligned}
\chi(t,z) &= \int d^2x \left[\varphi(\rho)\,\delta|\phi| + a(\rho)\,\delta A_\theta\right], \\
C \equiv C_z(t,z) &= \int d^2x\,\psi(\rho) A_z . \tag*{(47)}
\end{aligned} χ ( t , z ) C ≡ C z ( t , z ) = ∫ d 2 x [ φ ( ρ ) δ ∣ ϕ ∣ + a ( ρ ) δ A θ ] , = ∫ d 2 x ψ ( ρ ) A z . (47)
where δ ∣ ϕ ∣ = ∣ ϕ ∣ − f ( ρ ) \delta|\phi| = |\phi| - f(\rho) δ ∣ ϕ ∣ = ∣ ϕ ∣ − f ( ρ ) and δ A θ ≡ A θ − α ( ρ ) / ρ \delta A_\theta\equiv A_\theta- \alpha(\rho)/\rho δ A θ ≡ A θ − α ( ρ ) / ρ is the perturbation of the orthonormal azimuthal component. The mode normalizations ensure that these overlaps recover the corresponding amplitudes at linear order. More details about this computing these quantities numerically can be found in Appendix § 2.2 .
Geometric sources and the bound mode response
During the collision, Eq. (45) is the leading term in the small slope expansion of the Nambu–Goto solution. The detailed perturbative solution and an exact solution in null coordinates are given in Appendix § 1.4.3 . At leading order, the induced curvature sources according to our effective theory are,
R = − 8 F ′ ′ ( t + z ) G ′ ′ ( t − z ) cos δ + ⋯ , T = 4 F ′ ′ ( t + z ) G ′ ′ ( t − z ) sin δ + ⋯ . (48) \begin{aligned}
\mathcal{R} &= -8F''(t+z)G''(t-z)\cos\delta+ \cdots, \\
\mathcal{T} &= 4F''(t+z)G''(t-z)\sin\delta+ \cdots. \tag*{(48)}
\end{aligned} R T = − 8 F ′′ ( t + z ) G ′′ ( t − z ) cos δ + ⋯ , = 4 F ′′ ( t + z ) G ′′ ( t − z ) sin δ + ⋯ . (48)
An isolated traveling wave has neither source, consistently with its exact field theory realization. The internal modes are therefore sourced where the two incoming disturbances overlap.
Figure 3. Excitation of the shape and vector modes by colliding transverse waves. The upper panels show the incoming waves, and the lower panels show the internal excitation between the outgoing waves. The color represents the shape amplitude for δ = 45 ∘ \delta= 45^\circ δ = 4 5 ∘ , A = 0.35 A = 0.35 A = 0.35 , and L = 4 L = 4 L = 4 . The transverse directions are stretched for visibility. A movie is available at https://youtu.be/O7oJLk80Ce4 .
Writing D = ∂ t 2 − ∂ z 2 + ω b 2 \mathcal{D} = \partial_t^2 - \partial_z^2 + \omega_b^2 D = ∂ t 2 − ∂ z 2 + ω b 2 , the leading response equations are
D χ = κ S R , D ϑ = − ω b κ V T . (49) \mathcal{D}\chi= \kappa_S\mathcal{R}, \qquad\mathcal{D}\vartheta= -\omega_b\kappa_V\mathcal{T}. \tag*{(49)} D χ = κ S R , D ϑ = − ω b κ V T . (49)
Let U U U be the retarded solution of
D U = J ( t , z ) , J ( t , z ) = F ′ ′ ( t + z ) G ′ ′ ( t − z ) , (50) \mathcal{D}U = J(t,z), \qquad J(t,z) = F^{\prime\prime}(t+z)G^{\prime\prime}(t-z), \tag*{(50)} D U = J ( t , z ) , J ( t , z ) = F ′′ ( t + z ) G ′′ ( t − z ) , (50)
with no incoming massive excitation. Then, one can write the solution for the fields as
χ = − 8 κ S U cos δ , \chi= -8\kappa_S U\cos\delta, χ = − 8 κ S U cos δ ,
ϑ = − 4 ω b κ V U sin δ . (51) \vartheta= -4\omega_b\kappa_V U\sin\delta. \tag*{(51)} ϑ = − 4 ω b κ V U sin δ . (51)
At critical coupling, ω b κ V = 2 κ S \omega_b\kappa_V = 2\kappa_S ω b κ V = 2 κ S , so the two responses have the same prefactor and complementary cosine and sine dependences on the relative polarization. Reversing δ \delta δ leaves the scalar response unchanged and reverses the pseudoscalar response. This provides a direct test of the different geometric couplings. At fixed L L L , the leading internal response is of order ϵ 2 \epsilon^2 ϵ 2 , with corrections beginning at order ϵ 4 \epsilon^4 ϵ 4 within the truncated worldsheet theory.
For quantitative comparisons, the response can be evaluated within the late time approximation where we assume that the sources have already shut down. For the Gaussian profiles, the late time response at the collision center is
U ( t , 0 ) ≃ 2 π 32 A 2 L 2 e − ω b 2 L 2 4 ω b 7 / 2 sin ( ω b t + π / 4 ) t . (52) U(t,0) \simeq\frac{\sqrt{2\pi}}{32} A^2 L^2 e^{-\frac{\omega_b^2L^2}{4}}\omega_b^{7/2}\frac{\sin(\omega_b t+\pi/4)}{\sqrt{t}}. \tag*{(52)} U ( t , 0 ) ≃ 32 2 π A 2 L 2 e − 4 ω b 2 L 2 ω b 7/2 t sin ( ω b t + π /4 ) . (52)
The derivation and the full outgoing integral solution are given in Appendix § 1.4.3 . This asymptotic expression requires t ≫ max { L , ω b L 2 , ω b − 1 } t \gg\max\{L,\omega_bL^2,\omega_b^{-1}\} t ≫ max { L , ω b L 2 , ω b − 1 } , while decay and backreaction remain small.
The exponential factor shows that residual bound mode production in these Gaussian collisions is suppressed when L L L is large compared with the inverse bound-state mass. It measures the spectral overlap of the curvature source with the massive worldsheet dispersion relation. The t − 1 / 2 t^{-1/2} t − 1/2 envelope describes dispersive propagation along the string. These features are consistent with the shape excitation found in collisions of disturbances on domain wall strings [41 ] .
The lattice vector observable is C z C_z C z . At leading order, the duality relation gives
C z = ∂ t ϑ ω b , C t = ∂ z ϑ ω b . (53) C_z = \frac{\partial_t\vartheta}{\omega_b}, \qquad C_t = \frac{\partial_z\vartheta}{\omega_b}. \tag*{(53)} C z = ω b ∂ t ϑ , C t = ω b ∂ z ϑ . (53)
We therefore compare the projected vector signal with ∂ t ϑ / ω b \partial_t\vartheta/\omega_b ∂ t ϑ / ω b . At late times, C z C_z C z is in temporal quadrature with ϑ \vartheta ϑ , while its leading envelope has the same normalization. Figure 4 shows excellent agreement between the internal mode amplitudes extracted from the field theory simulations and the effective theory predictions. This agreement extends to both the absolute amplitudes produced by the collision and their dependence on the relative polarization angle δ \delta δ , providing a simultaneous test of the coupling strength κ S \kappa_S κ S and the geometric structure of the interactions.
Figure 4. Vector and shape amplitudes at the collision center, compared with the effective-theory predictions. The vector observable is C z C_z C z , obtained from the pseudoscalar response through Eq. (53) . Time is measured from the collision. The displayed t \sqrt{t} t rescaling highlights the dispersive envelope.
Nambu–Goto time delay
The preceding analysis assumes that the transverse wave packets follow Nambu–Goto dynamics during the collision. We now test this assumption independently by measuring their time delay after they separate. This delay provides a direct probe of the nonlinear interaction between the two disturbances. The Nambu–Goto equations admit an exact solution for the collision of counterpropagating transverse profiles [44 , 45 ] , allowing a quantitative comparison between the predicted delay in NG and the full field theory evolution.
The exact NG solution shows that after the waves separate, each retains its incoming shape and acquires a delay determined by the opposite profile, namely
Δ t G = ∫ − ∞ ∞ d s [ F ′ ( s ) ] 2 , Δ t F = ∫ − ∞ ∞ d s [ G ′ ( s ) ] 2 . (54) \Delta t_G = \int_{-\infty}^{\infty} ds\,[F'(s)]^2,\qquad\Delta t_F = \int_{-\infty}^{\infty} ds\,[G'(s)]^2. \tag*{(54)} Δ t G = ∫ − ∞ ∞ d s [ F ′ ( s ) ] 2 , Δ t F = ∫ − ∞ ∞ d s [ G ′ ( s ) ] 2 . (54)
For the Gaussian convention in Eq. (46) , this gives
Δ t F = Δ t G = π A 2 2 L . (55) \Delta t_F = \Delta t_G = \frac{\sqrt{\pi} A^2}{2L}. \tag*{(55)} Δ t F = Δ t G = 2 L π A 2 . (55)
The result is independent of δ \delta δ . In physical units, the delay of each wave equals the excess energy of the counterpropagating wave divided by the string tension. Appendix C derives these statements both in perturbative static gauge and from the exact null coordinate solution.
We measure the delay by tracking the center z w ( t ) z_w(t) z w ( t ) of an outgoing Gaussian disturbance, obtained by fitting the appropriate transverse component of the core displacement. For a wave moving toward positive z z z , its lag relative to free propagation from an initial time t i t_i t i is
Δ t ( t ) = ( t − t i ) − [ z w ( t ) − z w ( t i ) ] . (56) \Delta t(t) = (t-t_i)-[z_w(t)-z_w(t_i)]. \tag*{(56)} Δ t ( t ) = ( t − t i ) − [ z w ( t ) − z w ( t i )] . (56)
This lag approaches a constant after the waves separate. The fitting procedure does not assign a reliable individual wave position during the overlap interval. Figure 5 shows the agreement of the outgoing delay with the Nambu–Goto prediction. Together with the bound mode comparison, this supports using the leading Nambu–Goto solution to compute the geometric sources of the internal response.
Figure 5. Upper panels show the shape and vector amplitudes in spacetime after the collision of two transverse waves. The lower panel compares the measured outgoing delay with Eq. (55) . The overlap interval is shaded, and the mode amplitudes are rescaled by t \sqrt{t} t for t > 0 t>0 t > 0 .
The comparison assumes small incoming slopes and times over which nonlinear decay of the bound modes and their backreaction remain perturbatively small. The worldsheet description also requires negligible emission into bulk fields, which depends on both the amplitude and the width relative to the transverse core scale. We furthermore restrict attention to smooth configurations for which cusp formation2 and other strongly curved regions are absent [36 , 37 , 48 ] . The convergence study reported in Appendix D using three resolutions, supports fourth-order convergence of the collision induced vector signal.
For the amplitudes and widths considered, the simulations support a description in which Nambu–Goto dynamics governs the leading string motion and the geometric sources excite the localized modes with subleading backreaction. Increasing the amplitude or decreasing the width enhances the importance of additional interactions and finite width corrections. An appreciable bulk flux requires the corresponding bulk degrees of freedom in the dynamical description.
Finally, let us also briefly discuss the possibility of integrating out the bound modes at low energies. When characteristic frequencies are well below the bound state masses and on-shell production is negligible, integrating out the massive modes instead yields a local derivative expansion. The shape mode exchange generates an R 2 \mathcal{R}^2 R 2 interaction. On the other hand, the pseudoscalar exchange contributes to Ω A B Ω A B \Omega_{AB}\Omega^{AB} Ω A B Ω A B and cancels the contact term generated by dualization at leading derivative order, leaving the independent contribution inherited from the vortex background. Other terms omitted from our truncation may contribute at the same derivative order.
However, the production and parametric amplification processes studied here require explicit dynamical internal fields.
Direct coupling between the vector and shape modes
Finally, the interaction g S V χ C A C A g_{\mathrm{SV}}\chi C_A C^A g SV χ C A C A can be isolated in a straight string simulation. We initialize a small, uniform vector excitation and choose the periodic length L z L_z L z so that the allowed nonzero wavenumbers 2 π n / L z 2\pi n/L_z 2 π n / L z lie outside the principal vector induced resonance band [17 ] . This suppresses transverse growth during the comparison, allowing the induced shape excitation to be identified more clearly.
For C z ≃ C 0 cos ( ω b t ) C_z \simeq C_0 \cos(\omega_b t) C z ≃ C 0 cos ( ω b t ) , C t = 0 C_t = 0 C t = 0 , and χ ( 0 ) = χ ˙ ( 0 ) = 0 \chi(0) = \dot{\chi}(0) = 0 χ ( 0 ) = χ ˙ ( 0 ) = 0 , the leading prediction for the shape mode is described by the equation,
χ ¨ + ω b 2 χ = − g S V C 0 2 cos 2 ( ω b t ) , (57) \ddot{\chi} + \omega_b^2 \chi= -g_{\mathrm{SV}} C_0^2 \cos^2(\omega_b t), \tag*{(57)} χ ¨ + ω b 2 χ = − g SV C 0 2 cos 2 ( ω b t ) , (57)
which can be solved exactly at this order to yield,
χ ( t ) = g S V C 0 2 6 ω b 2 [ cos ( 2 ω b t ) + 2 cos ( ω b t ) − 3 ] . (58) \chi(t) = \frac{g_{\mathrm{SV}} C_0^2}{6\omega_b^2}\left[\cos(2\omega_b t) + 2\cos(\omega_b t) - 3\right]. \tag*{(58)} χ ( t ) = 6 ω b 2 g SV C 0 2 [ cos ( 2 ω b t ) + 2 cos ( ω b t ) − 3 ] . (58)
The forcing contains a constant part and a second harmonic. Since the induced shape amplitude is O ( C 0 2 ) \mathcal{O}(C_0^2) O ( C 0 2 ) , the scalar self-interaction 3 / 2 g S V χ 2 3/2g_{\mathrm{SV}}\chi^2 3/2 g SV χ 2 first contributes at order C 0 4 C_0^4 C 0 4 to its equation of motion so we neglect it here.
Figure 6 compares the projected field theory amplitudes with this prediction. The amplitude and harmonic content test the direct mixed interaction, independently of the geometric couplings. The effective theory also predicts quadratic scaling of the induced shape amplitude with C 0 C_0 C 0 . The agreement between the effective action and the field theory results is remarkable.
Figure 6. Shape excitation induced by a homogeneous vector mode with initial amplitude C 0 = 0.1 C_0 = 0.1 C 0 = 0.1 , compared with Eq. (58) using the microscopically determined value of g S V g_{\mathrm{SV}} g SV , Eq. (14) .
Together with the instability and collision experiments, this comparison tests energy exchange among the translational, shape, and vector degrees of freedom using one set of microscopically determined coefficients.
Conclusions
We have derived a covariant worldsheet action that connects the microscopic structure of Abelian-Higgs strings to their dynamics beyond the Nambu–Goto approximation. At critical coupling, the theory includes the transverse Goldstone modes, a massive scalar shape mode, and a massive vector mode, together with their geometric and direct interactions. All coefficients retained at this order are determined by transverse integrals of the vortex background and normalized mode profiles. The resulting description is therefore predictive, with no additional phenomenological parameters at this order in the effective action expansion.
The dynamics has a simple geometric interpretation. The shape mode couples to intrinsic worldsheet curvature, while the pseudoscalar dual to the vector couples to the curvature of the normal bundle. These interactions explain why shape excitation amplifies planar transverse motion, whereas vector excitation produces helical deformations. Despite their different polarization structures, the two instabilities have equal leading growth rates for equal normalized internal amplitudes at critical coupling. The action also identifies the direct interaction responsible for vector induced shape excitation and reproduces the earlier collective coordinate description through cubic order.
Lattice field theory simulations test this description through three complementary processes: parametric amplification of transverse motion, excitation of the internal modes by colliding transverse waves, and direct energy transfer between the vector and shape modes. The same microscopically determined coefficients describe all three processes without further adjustment. In particular, the scattering experiments reproduce the predicted complementary dependence of the internal response on the relative polarization of the incoming waves. Their collision time delay also agrees with the Nambu–Goto prediction, supporting a description in which the leading string trajectory supplies the geometric sources for the internal excitations. Together, these comparisons provide nontrivial tests of both the interaction strengths and their geometric structure in the regime studied.
The construction also suggests applications beyond cosmic strings. The dual pseudoscalar has the same geometric interaction as the worldsheet axion proposed for confining QCD flux tubes [30 ] , providing an explicit realization of this coupling through a localized gauge excitation in a four dimensional field theory. More generally, the separation of transverse motion, internal excitations, and their geometric couplings offers a useful starting point for studying other flux tubes and vortex lines, including those in condensed matter. Their spectra and symmetries will differ, but the microscopic matching strategy provides a common way to investigate how internal structure affects their dynamics.
For cosmology, these results provide a first step toward incorporating the effects of internal string excitations into future studies of network evolution beyond the Nambu–Goto description. Energy transfer into these modes constitutes an additional dynamical channel that may contribute to the discrepancies between Nambu–Goto and field theory simulations of string networks as suggested in [49 ] . On the other hand, the exponential suppression of residual excitation in wide Gaussian collisions suggests that smooth, slowly varying string segments are inefficient at exciting the bound modes through this mechanism. Appreciable production requires geometric sources with spectral support on the massive worldsheet spectrum, whose threshold lies close to that of the bulk continuum in the present model. This suggests that the production of worldsheet excitations may be concentrated in localized regions undergoing rapid deformation. Such conditions may arise in regions of large curvature, including those observed in the field theory simulations of Ref. [37 ] .
In a cosmological network, intercommutation may provide an additional source of internal excitations. Reconnection between distinct strings, or between separated segments of the same string, involves a rapid rearrangement of the vortex cores that may populate the localized bound modes. Although local in spacetime, this process couples distinct worldsheets or intrinsically separated points on the same worldsheet and changes string connectivity. It is therefore nonlocal from the perspective of the worldsheet description used here and is not captured by the present effective action. This excitation mechanism may be related to the energy exchange between translational motion and internal modes studied in multivortex collisions [50 , 51 , 52 ] . Similar studies of string intercommutation could clarify the role of reconnection in exciting localized internal modes.
These configurations provide a natural setting for investigating whether internal excitations can influence string evolution and its cosmological signatures. Establishing their role will require extending the present analysis beyond the smooth configurations studied here.
In this work, we have focused on bound modes localized on the string worldsheet. The adapted coordinate construction also provides a systematic route to incorporating radiation by retaining the relevant bulk fluctuations and deriving their couplings to the string and its internal modes. This extension would strengthen the analogy with worldline effective theories of compact objects, providing a unified description of conservative worldsheet dynamics, resonant excitation of localized degrees of freedom, and dissipation through radiation into the bulk.
The same techniques could also be adapted to global strings to include their internal bound modes [14 , 53 ] and their couplings to the standard effective action of the global string [54 , 55 , 56 ] . (See also some recent results on this topic in [57 , 58 ] .)
These extensions would help establish when microscopic string structure has observable consequences for cosmological evolution. Work in this direction is underway.
Acknowledgments
We are grateful to Maximilian Bachmaier and Andrzej Wereszczynski for stimulating discussions and comments on previous versions of the manuscript. We would also like to thank Katy Clough, Thomas Helfer, Eugene Lim, Dina Traykova and the GRTL collaboration (www.grtlcollaboration.org ) for their support and code development work. This work has been supported in part by the PID2024-156016NB-I00 grants funded by MCIN/AEI/10.13039/501100011033/and by ERDF; “A way of making Europe”; the Basque Government grant (IT-1977-26) and the Basque Foundation for Science (IKERBASQUE). This publication is part of the R&D&i project PID2025-172448NB-I00, funded by MICIU/AEI/10.13039/501100011033. AGMC acknowledges support from the Xunta de Galicia (Consellería de Educación, Ciencia, Universidades e Formación Profesional) through Grant No. ED431B-2024/42 and the postdoctoral fellowship program [Grant No. ED481B-2025/059]. JCA acknowledges funding from the Department of Physics at MIT through a CTP Postdoctoral Fellowship. JQ has been supported in part by the Spanish Ministerio de Ciencia e Innovación (MCIN) with funding from the grant PID2023-148409NB-I00 MTM.
This work used Stampede3 at the Texas Advanced Computing Center through allocations PHY260066 and PHY0900003 from the Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program [59 ] , which is supported by U.S. National Science Foundation grants 2138259, 2138286, 2138307, 2137603, and 2138296. The authors acknowledge the Texas Advanced Computing Center (TACC) at the University of Texas at Austin for providing computational resources that have contributed to the research results reported within this paper. URL: www.tacc.utexas.edu .
We acknowledge the use of Claude (by Anthropic) for assistance with text editing, checking of mathematical derivations, and the development of analysis and plotting scripts. The authors reviewed all such material and take full responsibility for the content of this work.
Computing the Effective Action from the top down
General procedure and expansion of the action
We obtain the worldsheet action by expressing the microscopic fields in coordinates adapted to a prescribed string embedding, expanding in the amplitudes of the retained bound modes, and integrating over the transverse coordinates. In the dimensionless variables of the main text, the Abelian-Higgs action at critical coupling is
S = − 1 e 2 ∫ d 2 σ d 2 u − g [ 1 4 F α β F α β + 1 2 ( D α ϕ ) ∗ D α ϕ + 1 8 ( ∣ ϕ ∣ 2 − 1 ) 2 ] , (59) S=-\frac{1}{e^2}\int d^2\sigma\,d^2u\,\sqrt{-g}\left[\frac{1}{4}F_{\alpha\beta}F^{\alpha\beta}+\frac{1}{2}(D_\alpha\phi)^*D^\alpha\phi+\frac{1}{8}(|\phi|^2-1)^2\right], \tag*{(59)} S = − e 2 1 ∫ d 2 σ d 2 u − g [ 4 1 F α β F α β + 2 1 ( D α ϕ ) ∗ D α ϕ + 8 1 ( ∣ ϕ ∣ 2 − 1 ) 2 ] , (59)
where D α = ∂ α − i A α D_\alpha=\partial_\alpha-iA_\alpha D α = ∂ α − i A α and the spacetime signature is ( − , + , + , + ) (-,+,+,+) ( − , + , + , + ) . Greek indices refer to all four adapted coordinates, A , B = { 0 , 1 } A,B=\{0,1\} A , B = { 0 , 1 } to the worldsheet, and i , j = { 1 , 2 } i,j=\{1,2\} i , j = { 1 , 2 } to the normal directions.
The adapted coordinates are
y μ ( σ , u ) = X μ ( σ ) + u i n i μ ( σ ) , ρ 2 = δ i j u i u j . (60) y^\mu(\sigma,u)=X^\mu(\sigma)+u^i n_i^\mu(\sigma),\qquad\rho^2=\delta_{ij}u^i u^j. \tag*{(60)} y μ ( σ , u ) = X μ ( σ ) + u i n i μ ( σ ) , ρ 2 = δ ij u i u j . (60)
where σ A \sigma^A σ A are the coordinates on the worldsheet and e A μ = ∂ A X μ e_A^\mu=\partial_A X^\mu e A μ = ∂ A X μ , γ A B = e A μ e B μ \gamma_{AB}=e_A^\mu e_{B\mu} γ A B = e A μ e B μ , and n i μ n_i^\mu n i μ are orthonormal spacelike normals. We use K i A B = − n i μ ∂ A e B μ K_{iAB}=-n_{i\mu}\partial_A e_B^\mu K i A B = − n i μ ∂ A e B μ , K i = γ A B K i A B K_i=\gamma^{AB}K_{iAB} K i = γ A B K i A B , and ω A = n 1 μ ∂ A n 2 μ \omega_A=n_{1\mu}\partial_A n_2^\mu ω A = n 1 μ ∂ A n 2 μ , with
R = K i K i − K i A B K i A B , Ω A B = ∇ A ω B − ∇ B ω A . (61) \mathcal{R}=K_iK_i-K_{iAB}K_i^{AB},\qquad\Omega_{AB}=\nabla_A\omega_B-\nabla_B\omega_A. \tag*{(61)} R = K i K i − K i A B K i A B , Ω A B = ∇ A ω B − ∇ B ω A . (61)
Indices on worldsheet fields are raised with γ A B \gamma^{AB} γ A B and normal indices with δ i j \delta^{ij} δ ij . Components of microscopic fields are instead raised with the full inverse metric g α β g^{\alpha\beta} g α β . The construction applies within a nonsingular tubular neighborhood, with curvature radii large compared with the transverse extent of the profiles of the fields we consider.
We place the straight string profiles in each local normal plane. In the gauge where the Higgs phase is θ ( u ) \theta(u) θ ( u ) , this gives
ϕ ˉ = f ( ρ ) e i θ , A ˉ i = α ( ρ ) ∂ i θ , A ˉ A = [ 1 − α ( ρ ) ] ω A , (62) \bar{\phi}=f(\rho)e^{i\theta},\qquad\bar{A}_i=\alpha(\rho)\partial_i\theta,\qquad\bar{A}_A=[1-\alpha(\rho)]\omega_A, \tag*{(62)} ϕ ˉ = f ( ρ ) e i θ , A ˉ i = α ( ρ ) ∂ i θ , A ˉ A = [ 1 − α ( ρ )] ω A , (62)
where a overbar denotes evaluation on this background, with the internal amplitudes set to zero. In particular, F ˉ α β = ∂ α A ˉ β − ∂ β A ˉ α \bar{F}_{\alpha\beta}=\partial_\alpha\bar{A}_\beta-\partial_\beta\bar{A}_\alpha F ˉ α β = ∂ α A ˉ β − ∂ β A ˉ α and D ˉ α = ∂ α − i A ˉ α \bar{D}_\alpha=\partial_\alpha-i\bar{A}_\alpha D ˉ α = ∂ α − i A ˉ α . The component A ˉ A \bar{A}_A A ˉ A accounts for the rotation of the normal frame and is required by the accompanying gauge covariance. It is present even when the independent vector excitation vanishes.
We then set the total fields as the sum of these background fields plus the internal modes, namely, ϕ = ϕ ˉ + δ ϕ \phi=\bar{\phi}+\delta\phi ϕ = ϕ ˉ + δ ϕ and A α = A ˉ α + δ A α A_\alpha=\bar{A}_\alpha+\delta A_\alpha A α = A ˉ α + δ A α , with
δ ϕ = χ ( σ ) φ ( ρ ) e i θ , δ A i = ρ a ( ρ ) χ ( σ ) ∂ i θ , \delta\phi=\chi(\sigma)\varphi(\rho)e^{i\theta},\qquad
\delta A_i=\rho a(\rho)\chi(\sigma)\partial_i\theta, δ ϕ = χ ( σ ) φ ( ρ ) e i θ , δ A i = ρ a ( ρ ) χ ( σ ) ∂ i θ ,
δ A A = − ρ a ( ρ ) χ ( σ ) ω A + ψ ( ρ ) C A ( σ ) . (63) \delta A_A=-\rho a(\rho)\chi(\sigma)\omega_A+\psi(\rho)C_A(\sigma). \tag*{(63)} δ A A = − ρ a ( ρ ) χ ( σ ) ω A + ψ ( ρ ) C A ( σ ) . (63)
Here the real functions φ ( ρ ) \varphi(\rho) φ ( ρ ) and a ( ρ ) a(\rho) a ( ρ ) specify the radial profiles of the Higgs and azimuthal gauge perturbations that together constitute the shape mode. These profiles are determined by the linearized fluctuation equations about the straight vortex. Their common coefficient χ ( σ ) \chi(\sigma) χ ( σ ) is a real worldsheet scalar that describes the local amplitude and propagation of this internal excitation.
Similarly, ψ ( ρ ) \psi(\rho) ψ ( ρ ) specifies the transverse localization of the vector mode, while C A ( σ ) C_A(\sigma) C A ( σ ) describes its amplitude and dynamics along the worldsheet. Thus, the radial profiles encode the microscopic structure of the bound modes, whereas χ \chi χ and C A C_A C A are the dynamical fields of the worldsheet effective theory.
The term − ρ a ( ρ ) χ ( σ ) ω A -\rho a(\rho)\chi(\sigma)\omega_A − ρ a ( ρ ) χ ( σ ) ω A is the geometric completion of the shape perturbation, where ω A \omega_A ω A is the normal bundle connection introduced earlier, which describes the rotation of the local normal basis along the worldsheet. This contribution ensures that the shape mode ansatz transforms consistently under local rotations of that basis. The worldsheet field strength associated with the vector excitation is
G A B = ∇ A C B − ∇ B C A . (64) G_{AB}=\nabla_A C_B-\nabla_B C_A. \tag*{(64)} G A B = ∇ A C B − ∇ B C A . (64)
Using this approach, our action can be obtained using the following formal Taylor expansion
S = ∑ n = 0 ∞ S ( n ) , S ( n ) = 1 n ! δ n S ∣ ϕ ˉ , A ˉ . (65) S=\sum_{n=0}^{\infty}S^{(n)},\qquad S^{(n)}=\left.\frac{1}{n!}\delta^n S\right|_{\bar{\phi},\bar{A}}. \tag*{(65)} S = n = 0 ∑ ∞ S ( n ) , S ( n ) = n ! 1 δ n S ϕ ˉ , A ˉ . (65)
The superscript counts powers of χ \chi χ and C A C_A C A . The background solves the straight string equations, but generally not the full field equations for a curved embedding. Its first variation can therefore be nonzero and provides geometric sources for the internal modes. The contributions given in the main text action are
S e f f = S ( 0 ) + S S ( 1 ) + S V ( 1 ) S_{\mathrm{eff}}=S^{(0)}+S_S^{(1)}+S_V^{(1)} S eff = S ( 0 ) + S S ( 1 ) + S V ( 1 )
+ S S ( 2 ) + S V ( 2 ) + S S V ( 3 ) + ⋯ . (66) \qquad+S_S^{(2)}+S_V^{(2)}+S_{SV}^{(3)}+\cdots. \tag*{(66)} + S S ( 2 ) + S V ( 2 ) + S S V ( 3 ) + ⋯ . (66)
Here S and V label the shape and vector modes. We calculate their leading kinetic and mass terms, their leading linear geometric couplings, and the direct mixed cubic interaction. The ellipsis includes further interaction and curvature corrections, including mixed terms quadratic in the internal amplitudes. All these contributions are computed separately below.
For radial transverse integrands, the geometric reduction follows from
− g = − γ [ 1 + u i K i + 1 2 u i u j ( K i K j − K i A B K j A B ) ] . (67) \sqrt{-g}=\sqrt{-\gamma}\left[1+u^iK_i+\frac{1}{2}u^i u^j(K_iK_j-K_{iAB}K_j^{AB})\right]. \tag*{(67)} − g = − γ [ 1 + u i K i + 2 1 u i u j ( K i K j − K i A B K j A B ) ] . (67)
Using ⟨ u i u j ⟩ θ = ρ 2 δ i j / 2 \langle u^i u^j\rangle_\theta=\rho^2\delta^{ij}/2 ⟨ u i u j ⟩ θ = ρ 2 δ ij /2 gives
1 2 π ∫ 0 2 π d θ − g = − γ ( 1 + ρ 2 4 R ) . (68) \frac{1}{2\pi}\int_0^{2\pi}d\theta\,\sqrt{-g}
=\sqrt{-\gamma}\left(1+\frac{\rho^2}{4}\mathcal{R}\right). \tag*{(68)} 2 π 1 ∫ 0 2 π d θ − g = − γ ( 1 + 4 ρ 2 R ) . (68)
This factor fixes both Ricci couplings. In the leading quadratic operators we also use g A B = γ A B + O ( u K ) g^{AB}=\gamma^{AB}+O(uK) g A B = γ A B + O ( u K ) ; odd transverse moments vanish, and the remaining curvature corrections belong to the ellipsis. Finally, we will consider that localized perturbations are regular at the origin and decay at infinity.
Zeroth order: tension and Euler term
The transverse Higgs gradient, magnetic Maxwell term, and Higgs potential in Eq. (59) give, respectively,
S ( 0 ) = − 1 e 2 ∫ d 2 σ d 2 u − g [ 1 2 f ′ 2 + ( 1 − α ) 2 f 2 2 ρ 2 + α ′ 2 2 ρ 2 + 1 8 ( f 2 − 1 ) 2 ] + ⋯ . (69) S^{(0)}=-\frac{1}{e^2}\int d^2\sigma\,d^2u\,\sqrt{-g}\left[\frac{1}{2}f^{\prime2}+\frac{(1-\alpha)^2f^2}{2\rho^2}+\frac{\alpha^{\prime2}}{2\rho^2}+\frac{1}{8}(f^2-1)^2\right]+\cdots. \tag*{(69)} S ( 0 ) = − e 2 1 ∫ d 2 σ d 2 u − g [ 2 1 f ′2 + 2 ρ 2 ( 1 − α ) 2 f 2 + 2 ρ 2 α ′2 + 8 1 ( f 2 − 1 ) 2 ] + ⋯ . (69)
where primes denote d / d ρ d/d\rho d / d ρ . The BPS equations and boundary conditions are
f ′ = 1 − α ρ f , α ′ = ρ 2 ( 1 − f 2 ) , (70) f^{\prime}=\frac{1-\alpha}{\rho}f,\qquad\alpha^{\prime}=\frac{\rho}{2}(1-f^2), \tag*{(70)} f ′ = ρ 1 − α f , α ′ = 2 ρ ( 1 − f 2 ) , (70)
with f ( 0 ) = α ( 0 ) = 0 f(0)=\alpha(0)=0 f ( 0 ) = α ( 0 ) = 0 and f ( ∞ ) = α ( ∞ ) = 1 f(\infty)=\alpha(\infty)=1 f ( ∞ ) = α ( ∞ ) = 1 . They reduce the bracket in Eq. (69) to f ′ 2 + α ′ 2 / ρ 2 f^{\prime2}+\alpha^{\prime2}/\rho^2 f ′2 + α ′2 / ρ 2 . Integrating with Eq. (68) yields
S ( 0 ) = − 1 e 2 ∫ d 2 σ − γ ( μ + ν R ) + ⋯ , (71) S^{(0)}=-\frac{1}{e^2}\int d^2\sigma\sqrt{-\gamma}\left(\mu+\nu\mathcal{R}\right)+\cdots, \tag*{(71)} S ( 0 ) = − e 2 1 ∫ d 2 σ − γ ( μ + ν R ) + ⋯ , (71)
where
μ = 2 π ∫ 0 ∞ d ρ ( ρ f ′ 2 + α ′ 2 ρ ) = π , (72) \mu=2\pi\int_0^\infty d\rho\left(\rho f^{\prime2}+\frac{\alpha^{\prime2}}{\rho}\right)=\pi, \tag*{(72)} μ = 2 π ∫ 0 ∞ d ρ ( ρ f ′2 + ρ α ′2 ) = π , (72)
ν = π 2 ∫ 0 ∞ d ρ ( ρ 3 f ′ 2 + ρ α ′ 2 ) = π 2 . (73) \nu=\frac{\pi}{2}\int_0^\infty d\rho\left(\rho^3f^{\prime2}+\rho\alpha^{\prime2}\right)=\frac{\pi}{2}. \tag*{(73)} ν = 2 π ∫ 0 ∞ d ρ ( ρ 3 f ′2 + ρ α ′2 ) = 2 π . (73)
Both evaluations follow from
ρ ( f ′ 2 + α ′ 2 ρ 2 ) = 1 2 d d ρ [ α + ( 1 − α ) f 2 ] . (74) \rho\left(f^{\prime2}+\frac{\alpha^{\prime2}}{\rho^2}\right)=\frac{1}{2}\frac{d}{d\rho}\left[\alpha+(1-\alpha)f^2\right]. \tag*{(74)} ρ ( f ′2 + ρ 2 α ′2 ) = 2 1 d ρ d [ α + ( 1 − α ) f 2 ] . (74)
For the second moment, write the bracket as 1 − ( 1 − α ) ( 1 − f 2 ) 1-(1-\alpha)(1-f^2) 1 − ( 1 − α ) ( 1 − f 2 ) before integrating by parts. This gives ν = π ∫ 0 ∞ ( 1 − α ) α ′ d ρ = π / 2 \nu=\pi\int_0^\infty(1-\alpha)\alpha^{\prime}d\rho=\pi/2 ν = π ∫ 0 ∞ ( 1 − α ) α ′ d ρ = π /2 . Thus μ \mu μ gives the Nambu–Goto tension, equal to π η 2 \pi\eta^2 π η 2 in physical units, whereas the constant coefficient Euler term does not affect the worldsheet equations of motion.
There is also a tangential Maxwell contribution from F ˉ A B = ( 1 − α ) Ω A B \bar{F}_{AB}=(1-\alpha)\Omega_{AB} F ˉ A B = ( 1 − α ) Ω A B . Its leading reduction is
S ( 0 ) ⊃ κ Ω e 2 ∫ d 2 σ − γ Ω A B Ω A B . (75) S^{(0)}\supset\frac{\kappa_{\Omega}}{e^2}\int d^2\sigma\sqrt{-\gamma}\,\Omega_{AB}\Omega^{AB}. \tag*{(75)} S ( 0 ) ⊃ e 2 κ Ω ∫ d 2 σ − γ Ω A B Ω A B . (75)
where
κ Ω = − π 2 [ ∫ 0 ∞ ρ d ρ ( 1 − α ) 2 ] . (76) \kappa_{\Omega}=-\frac{\pi}{2}\left[\int_0^\infty\rho\,d\rho\,(1-\alpha)^2\right]. \tag*{(76)} κ Ω = − 2 π [ ∫ 0 ∞ ρ d ρ ( 1 − α ) 2 ] . (76)
One can easily show that this term is fourth order in extrinsic curvature and belongs to the ellipsis in Eq. (71) . We keep it here and in the action in the main text, however, to compare with other similar terms that appear in the subsequent dualization derivations. This term was first introduced in the literature in [60 ] .
First order in the shape mode
The linear shape term receives contributions from all three microscopic terms in Eq. (59) . Their transverse parts give
S S ( 1 ) = − 1 e 2 ∫ d 2 σ d 2 u − g χ [ f ′ φ ′ + ( 1 − α ) 2 ρ 2 f φ − 1 − α ρ f 2 a + α ′ ρ ( a ′ + a ρ ) + 1 2 ( f 2 − 1 ) f φ + ⋯ ] . (77) S^{(1)}_{S}=-\frac{1}{e^2}\int d^2\sigma\,d^2u\,\sqrt{-g}\,\chi\left[f^{\prime}\varphi^{\prime}+\frac{(1-\alpha)^2}{\rho^2}f\varphi-\frac{1-\alpha}{\rho}f^2a+\frac{\alpha^{\prime}}{\rho}\left(a^{\prime}+\frac{a}{\rho}\right)+\frac{1}{2}(f^2-1)f\varphi+\cdots\right]. \tag*{(77)} S S ( 1 ) = − e 2 1 ∫ d 2 σ d 2 u − g χ [ f ′ φ ′ + ρ 2 ( 1 − α ) 2 f φ − ρ 1 − α f 2 a + ρ α ′ ( a ′ + ρ a ) + 2 1 ( f 2 − 1 ) f φ + ⋯ ] . (77)
The first three terms arise from the Higgs gradient, the fourth from the magnetic energy, and the last from the Higgs potential. The straight string equations combine this bracket into a total radial derivative:
S S ( 1 ) = − 1 e 2 ∫ d 2 σ d 2 u − g χ d d ρ ( ρ f ′ φ + α ′ a ) + ⋯ . (78) S^{(1)}_{S}=-\frac{1}{e^2}\int d^2\sigma\,d^2u\,\sqrt{-g}\,\chi\frac{d}{d\rho}\left(\rho f^{\prime}\varphi+\alpha^{\prime}a\right)+\cdots. \tag*{(78)} S S ( 1 ) = − e 2 1 ∫ d 2 σ d 2 u − g χ d ρ d ( ρ f ′ φ + α ′ a ) + ⋯ . (78)
With a flat transverse measure this is a vanishing boundary term. For a curved worldsheet, Eq. (68) introduces the weight ρ 2 R / 4 \rho^2\mathcal{R}/4 ρ 2 R /4 . One radial integration by parts then gives
S S ( 1 ) = κ S e 2 ∫ d 2 σ − γ χ R + ⋯ , (79) S^{(1)}_{S}=\frac{\kappa_S}{e^2}\int d^2\sigma\sqrt{-\gamma}\,\chi\mathcal{R}+\cdots, \tag*{(79)} S S ( 1 ) = e 2 κ S ∫ d 2 σ − γ χ R + ⋯ , (79)
κ S = π ∫ 0 ∞ d ρ ( ρ 2 f ′ φ + ρ α ′ a ) . \kappa_S=\pi\int_0^\infty d\rho\,(\rho^2 f^{\prime}\varphi+\rho\alpha^{\prime}a). κ S = π ∫ 0 ∞ d ρ ( ρ 2 f ′ φ + ρ α ′ a ) .
The leading scalar source thus comes from the curvature dependence of the transverse volume element.
First order in the vector mode
For this variation, δ ϕ = δ A i = 0 \delta\phi=\delta A_i=0 δ ϕ = δ A i = 0 and δ A A = ψ C A \delta A_A=\psi C_A δ A A = ψ C A . The relevant microscopic contribution is the Maxwell cross term
S V ( 1 ) = − 1 2 e 2 ∫ d 2 σ d 2 u − g F ˉ α β δ F α β . (80) S^{(1)}_{V}=-\frac{1}{2e^2}\int d^2\sigma\,d^2u\,\sqrt{-g}\,\bar{F}^{\alpha\beta}\delta F_{\alpha\beta}. \tag*{(80)} S V ( 1 ) = − 2 e 2 1 ∫ d 2 σ d 2 u − g F ˉ α β δ F α β . (80)
To see why the Higgs kinetic term does not contribute, it is essential to keep the mixed inverse metric. With ϵ 12 = 1 \epsilon_{12}=1 ϵ 12 = 1 and ξ i = ϵ j i u j \xi^i=\epsilon^j{}_i u^j ξ i = ϵ j i u j , one has
g A i = − ω A ξ i , g A i = g A B ω B ξ i , ξ i ∂ i θ = 1. (81) g_{Ai}=-\omega_A\xi_i,\qquad g^{Ai}=g^{AB}\omega_B\xi^i,\qquad\xi^i\partial_i\theta=1. \tag*{(81)} g A i = − ω A ξ i , g A i = g A B ω B ξ i , ξ i ∂ i θ = 1. (81)
Together with Eq. (62) , these relations give
D ˉ B ˉ ϕ ˉ = 0 , \bar{D}^{\bar{B}}\bar{\phi}=0, D ˉ B ˉ ϕ ˉ = 0 ,
F ˉ A B = ( 1 − α ) g A C g B D Ω C D , \bar{F}^{AB}=(1-\alpha)g^{AC}g^{BD}\Omega_{CD}, F ˉ A B = ( 1 − α ) g A C g B D Ω C D ,
F ˉ i B = ξ i ω A F ˉ A B . (82) \bar{F}^{iB}=\xi^i\omega_A\bar{F}^{AB}. \tag*{(82)} F ˉ i B = ξ i ω A F ˉ A B . (82)
The first identity eliminates the linear Higgs current. Moreover, δ F A B = ψ G A B \delta F_{AB}=\psi G_{AB} δ F A B = ψ G A B and δ F i A = ψ ′ u i C A / ρ \delta F_{iA}=\psi^{\prime}u_iC_A/\rho δ F i A = ψ ′ u i C A / ρ . Since ξ i u i = 0 \xi^i u_i=0 ξ i u i = 0 , the mixed Maxwell contraction vanishes. The remaining term is therefore
S V ( 1 ) = − 1 2 e 2 ∫ d 2 σ d 2 u − g ψ G A B F ˉ A B = − κ V 2 e 2 ∫ d 2 σ − γ G A B Ω A B + ⋯ , (83) \begin{aligned}
S_V^{(1)}=-\frac{1}{2e^2}\int d^2\sigma\,d^2u\,\sqrt{-g}\,\psi G_{AB}\bar{F}^{AB} \\
&=-\frac{\kappa_V}{2e^2}\int d^2\sigma\sqrt{-\gamma}\,G_{AB}\Omega^{AB}+\cdots, \tag*{(83)}
\end{aligned} S V ( 1 ) = − 2 e 2 1 ∫ d 2 σ d 2 u − g ψ G A B F ˉ A B = − 2 e 2 κ V ∫ d 2 σ − γ G A B Ω A B + ⋯ , (83)
κ V = 2 π ∫ 0 ∞ ρ d ρ ( 1 − α ) ψ . \kappa_V=2\pi\int_0^\infty\rho\,d\rho\,(1-\alpha)\psi. κ V = 2 π ∫ 0 ∞ ρ d ρ ( 1 − α ) ψ .
The factor 1 / 2 1/2 1/2 is the cross-term coefficient in the expansion of F α β F α β / 4 F_{\alpha\beta}F^{\alpha\beta}/4 F α β F α β /4 . The source depends on the normal curvature.
Quadratic action for the shape mode
Worldsheet derivatives of χ \chi χ enter through the Higgs kinetic term and the mixed Maxwell components. Combining the coordinate components with the inverse metric gives
S S ( 2 ) ∣ ∇ χ = − 1 2 e 2 ∫ d 2 σ d 2 u − g g A B ( φ 2 + a 2 ) ∂ A χ ∂ B χ . (84) S_S^{(2)}\big|_{\nabla\chi}=-\frac{1}{2e^2}\int d^2\sigma\,d^2u\,\sqrt{-g}\,g^{AB}(\varphi^2+a^2)\partial_A\chi\partial_B\chi. \tag*{(84)} S S ( 2 ) ∇ χ = − 2 e 2 1 ∫ d 2 σ d 2 u − g g A B ( φ 2 + a 2 ) ∂ A χ ∂ B χ . (84)
The contributions φ 2 \varphi^2 φ 2 and a 2 a^2 a 2 come from the Higgs and Maxwell terms, respectively. At leading order in the curvature expansion, we replace − g g A B \sqrt{-g}\,g^{AB} − g g A B by − γ γ A B \sqrt{-\gamma}\,\gamma^{AB} − γ γ A B . The normalization
2 π ∫ 0 ∞ ρ d ρ ψ 2 = 2 π ∫ 0 ∞ ρ d ρ ( φ 2 + a 2 ) = 1 (85) 2\pi\int_0^\infty\rho\,d\rho\,\psi^2=2\pi\int_0^\infty\rho\,d\rho\,(\varphi^2+a^2)=1 \tag*{(85)} 2 π ∫ 0 ∞ ρ d ρ ψ 2 = 2 π ∫ 0 ∞ ρ d ρ ( φ 2 + a 2 ) = 1 (85)
then gives the canonical scalar kinetic term.
The transverse gradients and potential determine the mass. A convenient way to obtain all their relative factors is to complete the transverse energy into BPS squares before expanding. At leading order in curvature, the coefficient of χ 2 / 2 \chi^2/2 χ 2 /2 becomes
2 π ∫ 0 ∞ ρ d ρ [ ( φ ′ − 1 − α ρ φ + f a ) 2 + ( a ′ + a ρ + f φ ) 2 ] . (86) 2\pi\int_0^\infty\rho\,d\rho\left[\left(\varphi'-\frac{1-\alpha}{\rho}\varphi+fa\right)^2+\left(a'+\frac{a}{\rho}+f\varphi\right)^2\right]. \tag*{(86)} 2 π ∫ 0 ∞ ρ d ρ [ ( φ ′ − ρ 1 − α φ + f a ) 2 + ( a ′ + ρ a + f φ ) 2 ] . (86)
The radial boundary term from the BPS completion does not contribute, since the perturbations preserve the vortex boundary conditions.
The BPS bound modes satisfy
− ψ ′ ′ − ψ ′ ρ + f 2 ψ = ω b 2 ψ , φ = f ψ ω b , a = − ψ ′ ω b . (87) -\psi''-\frac{\psi'}{\rho}+f^2\psi=\omega_b^2\psi,\qquad
\varphi=\frac{f\psi}{\omega_b},\qquad
a=-\frac{\psi'}{\omega_b}. \tag*{(87)} − ψ ′′ − ρ ψ ′ + f 2 ψ = ω b 2 ψ , φ = ω b f ψ , a = − ω b ψ ′ . (87)
Together with the background BPS equation f ′ = ( 1 − α ) f / ρ f'=(1-\alpha)f/\rho f ′ = ( 1 − α ) f / ρ , these relations imply
φ ′ − 1 − α ρ φ + f a = 0 , \varphi'-\frac{1-\alpha}{\rho}\varphi+fa=0, φ ′ − ρ 1 − α φ + f a = 0 ,
a ′ + a ρ + f φ = ω b ψ . (88) a'+\frac{a}{\rho}+f\varphi=\omega_b\psi. \tag*{(88)} a ′ + ρ a + f φ = ω b ψ . (88)
Substituting these identities into eq:A27 , the first square vanishes and the second becomes ω b 2 ψ 2 \omega_b^2\psi^2 ω b 2 ψ 2 . The coefficient of χ 2 / 2 \chi^2/2 χ 2 /2 is therefore
2 π ∫ 0 ∞ ρ d ρ ω b 2 ψ 2 = ω b 2 ( 2 π ∫ 0 ∞ ρ d ρ ψ 2 ) = ω b 2 , (89) 2\pi\int_0^\infty\rho\,d\rho\,\omega_b^2\psi^2=\omega_b^2\left(2\pi\int_0^\infty\rho\,d\rho\,\psi^2\right)=\omega_b^2, \tag*{(89)} 2 π ∫ 0 ∞ ρ d ρ ω b 2 ψ 2 = ω b 2 ( 2 π ∫ 0 ∞ ρ d ρ ψ 2 ) = ω b 2 , (89)
where the final equality follows from eq:A26 . Thus, with the kinetic term canonically normalized, the worldsheet scalar has mass ω b \omega_b ω b .
The resulting quadratic terms in the action are
S S ( 2 ) = − 1 2 e 2 ∫ d 2 σ − γ ( ∇ A χ ∇ A χ + ω b 2 χ 2 ) + ⋯ , (90) S_S^{(2)}=-\frac{1}{2e^2}\int d^2\sigma\sqrt{-\gamma}\left(\nabla_A\chi\nabla^A\chi+\omega_b^2\chi^2\right)+\cdots, \tag*{(90)} S S ( 2 ) = − 2 e 2 1 ∫ d 2 σ − γ ( ∇ A χ ∇ A χ + ω b 2 χ 2 ) + ⋯ , (90)
where the ellipsis denotes curvature dependent corrections to the quadratic action.
Quadratic action for the vector mode
The worldsheet Maxwell components provide the vector kinetic term, the mixed Maxwell components provide a radial gradient contribution to its mass, and the Higgs kinetic term supplies the remaining mass contribution. At leading order in curvature these are
S V ( 2 ) = − 2 π e 2 ∫ d 2 σ − γ ∫ 0 ∞ ρ d ρ [ 1 4 ψ 2 G A B G A B + 1 2 ( ψ ′ 2 + f 2 ψ 2 ) C A C A ] + ⋯ . (91) S_V^{(2)}=-\frac{2\pi}{e^2}\int d^2\sigma\sqrt{-\gamma}\int_0^\infty\rho\,d\rho\left[\frac{1}{4}\psi^2G_{AB}G^{AB}+\frac{1}{2}(\psi'^2+f^2\psi^2)C_A C^A\right]+\cdots. \tag*{(91)} S V ( 2 ) = − e 2 2 π ∫ d 2 σ − γ ∫ 0 ∞ ρ d ρ [ 4 1 ψ 2 G A B G A B + 2 1 ( ψ ′2 + f 2 ψ 2 ) C A C A ] + ⋯ . (91)
The Higgs potential is independent of A α A_\alpha A α and contributes nothing to this variation. Multiplying the radial eigenvalue equation by ρ ψ \rho\psi ρ ψ and integrating by parts gives
2 π ∫ 0 ∞ ρ d ρ ( ψ ′ 2 + f 2 ψ 2 ) = ω b 2 2 π ∫ 0 ∞ ρ d ρ ψ 2 = ω b 2 . (92) 2\pi\int_0^\infty\rho\,d\rho\,(\psi'^2+f^2\psi^2)=\omega_b^2\,2\pi\int_0^\infty\rho\,d\rho\,\psi^2=\omega_b^2. \tag*{(92)} 2 π ∫ 0 ∞ ρ d ρ ( ψ ′2 + f 2 ψ 2 ) = ω b 2 2 π ∫ 0 ∞ ρ d ρ ψ 2 = ω b 2 . (92)
The same normalization therefore fixes both kinetic and mass coefficients, yielding the Proca action
S V ( 2 ) = − 1 e 2 ∫ d 2 σ − γ ( 1 4 G A B G A B + ω b 2 2 C A C A ) + ⋯ . (93) S_V^{(2)}=-\frac{1}{e^2}\int d^2\sigma\sqrt{-\gamma}\left(\frac{1}{4}G_{AB}G^{AB}+\frac{\omega_b^2}{2}C_A C^A\right)+\cdots. \tag*{(93)} S V ( 2 ) = − e 2 1 ∫ d 2 σ − γ ( 4 1 G A B G A B + 2 ω b 2 C A C A ) + ⋯ . (93)
No transverse integral or additional normalization factor remains. The equal scalar and vector masses follow from the BPS relations in eq:A28 .
Direct mixed cubic interaction
The action in (13) of the main text also displays the interaction linear in χ \chi χ and quadratic in C A C_A C A . This term comes entirely from the Higgs kinetic term. Its part quadratic in the vector amplitude is
S ⊃ − 1 2 e 2 ∫ d 2 σ d 2 u − g ( f + χ φ ) 2 ψ 2 g A B C A C B . (94) S \supset-\frac{1}{2e^2}\int d^2\sigma\,d^2u\,\sqrt{-g}\,(f+\chi\varphi)^2\psi^2 g^{AB}C_A C_B . \tag*{(94)} S ⊃ − 2 e 2 1 ∫ d 2 σ d 2 u − g ( f + χ φ ) 2 ψ 2 g A B C A C B . (94)
Expanding the squared Higgs profile and integrating over the normal plane gives
S V ( 3 ) = − g S V e 2 ∫ d 2 σ − γ χ C A C A + ⋯ , (95) S_{V}^{(3)}=-\frac{g_{SV}}{e^2}\int d^2\sigma\sqrt{-\gamma}\,\chi C_A C^A+\cdots, \tag*{(95)} S V ( 3 ) = − e 2 g S V ∫ d 2 σ − γ χ C A C A + ⋯ , (95)
g S V = 2 π ∫ 0 ∞ ρ d ρ f φ ψ 2 . (96) g_{SV}=2\pi\int_0^\infty\rho\,d\rho\,f\varphi\psi^2. \tag*{(96)} g S V = 2 π ∫ 0 ∞ ρ d ρ f φ ψ 2 . (96)
The factor of two in ( f + χ φ ) 2 (f+\chi\varphi)^2 ( f + χ φ ) 2 cancels the microscopic factor 1 / 2 1/2 1/2 . The Maxwell action is quadratic in A α A_\alpha A α , and the Higgs potential contains no vector field, so neither produces this direct cubic interaction.
Collected action and order of the omitted terms
Adding the preceding contributions gives
S e f f = − 1 e 2 ∫ d 2 σ − γ [ μ + ν R − κ Ω Ω A B Ω A B + 1 2 ∇ A χ ∇ A χ + ω b 2 2 χ 2 + 1 4 G A B G A B + ω b 2 2 C A C A − κ S χ R + κ V 2 G A B Ω A B + g S V χ C A C A ] + ⋯ . (97) \begin{aligned}
S_{\mathrm{eff}}={}&-\frac{1}{e^2}\int d^2\sigma\sqrt{-\gamma}\left[\mu+\nu\mathcal{R}-\kappa_{\Omega}\Omega_{AB}\Omega^{AB}\right.\\
&\left.+\frac{1}{2}\nabla_A\chi\nabla^A\chi+\frac{\omega_b^2}{2}\chi^2\right.\\
&\left.+\frac{1}{4}G_{AB}G^{AB}+\frac{\omega_b^2}{2}C_A C^A\right.\\
&\left.-\kappa_S\chi\mathcal{R}+\frac{\kappa_V}{2}G_{AB}\Omega^{AB}\right.\\
&\left.+g_{SV}\chi C_A C^A\right]+\cdots. \tag*{(97)}
\end{aligned} S eff = − e 2 1 ∫ d 2 σ − γ [ μ + ν R − κ Ω Ω A B Ω A B + 2 1 ∇ A χ ∇ A χ + 2 ω b 2 χ 2 + 4 1 G A B G A B + 2 ω b 2 C A C A − κ S χ R + 2 κ V G A B Ω A B + g S V χ C A C A ] + ⋯ . (97)
This reproduces all displayed terms, signs, and normalizations of Eq. (13) in the main text. The coefficient integrals are given in Eqs. (A14), (A20), (A24), and (A36).
Finally, the linear profile ansatz also produces a scalar cubic selfinteraction. The second BPS square has the argument ω b ψ χ + φ 2 χ 2 / 2 \omega_b\psi\chi+\varphi^2\chi^2/2 ω b ψ χ + φ 2 χ 2 /2 , so its cubic energy coefficient is π ω b ∫ 0 ∞ ρ d ρ ψ φ 2 = g S V / 2 \pi\omega_b\int_0^\infty\rho\,d\rho\,\psi\varphi^2=g_{SV}/2 π ω b ∫ 0 ∞ ρ d ρ ψ φ 2 = g S V /2 . Hence
S S ( 3 ) ⊃ − g S V 2 e 2 ∫ d 2 σ − γ χ 3 . (98) S_{S}^{(3)}\supset-\frac{g_{SV}}{2e^2}\int d^2\sigma\sqrt{-\gamma}\,\chi^3. \tag*{(98)} S S ( 3 ) ⊃ − 2 e 2 g S V ∫ d 2 σ − γ χ 3 . (98)
This term is also included in the displayed main text action. If the transverse displacement, χ \chi χ , and C A C_A C A are all counted as first order about a straight string, R \mathcal{R} R and Ω A B \Omega_{AB} Ω A B begin at second order. Under that counting, Eq. (98) must be added to obtain the complete cubic action within the stated profile ansatz; further terms not displayed here begin therefore at fourth order.
Relation between V and S
In this subsection, we explain how the two geometric coupling constants discussed in the previous subsections, κ V \kappa_V κ V and κ S \kappa_S κ S , are related. We begin with the definition of the scalar geometric coupling,
κ S = π ω b ∫ 0 ∞ d ρ ( ρ 2 f ′ φ + ρ α ′ a ) . (99) \kappa_S=\frac{\pi}{\omega_b}\int_0^\infty d\rho\,\left(\rho^2 f'\varphi+\rho\alpha'a\right). \tag*{(99)} κ S = ω b π ∫ 0 ∞ d ρ ( ρ 2 f ′ φ + ρ α ′ a ) . (99)
Substituting the BPS mode relations from Eq. (A28) expresses both shape mode profiles in terms of ψ \psi ψ :
κ S = π ω b ∫ 0 ∞ d ρ ( ρ 2 f f ′ ψ − ρ α ′ ψ ′ ) . (100) \kappa_S=\frac{\pi}{\omega_b}\int_0^\infty d\rho\,\left(\rho^2ff'\psi-\rho\alpha'\psi'\right). \tag*{(100)} κ S = ω b π ∫ 0 ∞ d ρ ( ρ 2 f f ′ ψ − ρ α ′ ψ ′ ) . (100)
We integrate the second term by parts to remove the derivative from ψ \psi ψ . The boundary term [ ρ α ′ ψ ] 0 ∞ [\rho\alpha'\psi]_0^\infty [ ρ α ′ ψ ] 0 ∞ vanishes by regularity at the origin and decay at infinity. Using the background BPS equations in the form ( ρ α ′ ) ′ = 2 α ′ − ρ 2 f f ′ (\rho\alpha')'=2\alpha'-\rho^2ff' ( ρ α ′ ) ′ = 2 α ′ − ρ 2 f f ′ , we obtain
κ S = π ω b ∫ 0 ∞ d ρ [ ρ 2 f f ′ + ( ρ α ′ ) ′ ] ψ = 2 π ω b ∫ 0 ∞ d ρ α ′ ψ . (101) \begin{aligned}
\kappa_S&=\frac{\pi}{\omega_b}\int_0^\infty d\rho\,\left[\rho^2ff'+(\rho\alpha')'\right]\psi\\
&=\frac{2\pi}{\omega_b}\int_0^\infty d\rho\,\alpha'\psi. \tag*{(101)}
\end{aligned} κ S = ω b π ∫ 0 ∞ d ρ [ ρ 2 f f ′ + ( ρ α ′ ) ′ ] ψ = ω b 2 π ∫ 0 ∞ d ρ α ′ ψ . (101)
To relate this expression to κ V \kappa_V κ V , let us denote by H H H the radial operator appearing in Eq. (A28). The background BPS equations then imply H ( 1 − α ) = 2 α ′ / ρ H(1-\alpha)=2\alpha'/\rho H ( 1 − α ) = 2 α ′ / ρ . Multiplying by ρ ψ \rho\psi ρ ψ and integrating by parts twice transfers the radial derivatives from 1 − α 1-\alpha 1 − α to ψ \psi ψ , without boundary contributions. The eigenvalue equation for ψ \psi ψ then gives
2 ∫ 0 ∞ d ρ α ′ ψ = ∫ 0 ∞ ρ d ρ ψ H ( 1 − α ) = ∫ 0 ∞ ρ d ρ ( 1 − α ) H ψ = ω b 2 ∫ 0 ∞ ρ d ρ ( 1 − α ) ψ . (102) \begin{aligned}
2\int_0^\infty d\rho\,\alpha'\psi&=\int_0^\infty\rho\,d\rho\,\psi H(1-\alpha)\\
&=\int_0^\infty\rho\,d\rho\,(1-\alpha)H\psi\\
&=\omega_b^2\int_0^\infty\rho\,d\rho\,(1-\alpha)\psi. \tag*{(102)}
\end{aligned} 2 ∫ 0 ∞ d ρ α ′ ψ = ∫ 0 ∞ ρ d ρ ψ H ( 1 − α ) = ∫ 0 ∞ ρ d ρ ( 1 − α ) H ψ = ω b 2 ∫ 0 ∞ ρ d ρ ( 1 − α ) ψ . (102)
Substituting this result into Eq. (101) and recognizing the radial integral defining κ V \kappa_V κ V , we conclude that
κ S = ω b 2 κ V . (103) \kappa_S=\frac{\omega_b}{2}\kappa_V. \tag*{(103)} κ S = 2 ω b κ V . (103)
Effective Action versus Collective Coordinate Approach
In [17 ] , the dynamics of a straight string was approximated by expanding the microscopic fields in a finite set of fluctuation profiles with time dependent amplitudes. Substituting this ansatz into the microscopic action and integrating over the spatial coordinates yields a reduced mechanical Lagrangian for these amplitudes, following a sort of collective coordinate approach widely used in soliton dynamics. Here we establish its connection to the covariant worldsheet action, incorporating the shape mode description of [16 ] within a unified framework.
We demonstrate analytically that the cubic interaction coefficients agree. The equivalence between the two descriptions is understood through cubic order, modulo total derivatives and terms proportional to the free equations of motion. These interactions govern the leading parametric instabilities and the direct energy exchange between the two internal modes. This agreement supports interpreting the worldsheet action as a covariant nonlinear extension of the collective coordinate description, within the particular set of modes that we are considering.
Shape mode couplings
The shape mode coupling in the EFT approach
We start by writing the relevant terms in the effective action
S e f f = − ∫ d 2 σ − γ [ μ + 1 2 ∇ A χ ∇ A χ + ω b 2 2 χ 2 − κ S χ R ] + ⋯ . (104) S_{\rm eff}=-\int d^{2}\sigma\sqrt{-\gamma}\left[\mu+\frac{1}{2}\nabla_A\chi\nabla^A\chi+\frac{\omega_b^2}{2}\chi^2-\kappa_S\chi R\right]+\cdots. \tag*{(104)} S eff = − ∫ d 2 σ − γ [ μ + 2 1 ∇ A χ ∇ A χ + 2 ω b 2 χ 2 − κ S χ R ] + ⋯ . (104)
In particular, the curvature interaction has a positive sign in the action itself:
S χ R = κ S ∫ d 2 σ − γ χ R , (105) S_{\chi R}=\kappa_S\int d^{2}\sigma\sqrt{-\gamma}\chi R, \tag*{(105)} S χ R = κ S ∫ d 2 σ − γ χ R , (105)
For a small transverse displacement in one direction X ( t , z ) X(t,z) X ( t , z ) from a straight string along the z-direction, the Ricci scalar, to quadratic order is given by
R ≃ ( □ X ) 2 − ( ∂ a ∂ b X ) ( ∂ a ∂ b X ) , (106) R\simeq(\Box X)^2-(\partial_a\partial_bX)(\partial^a\partial^bX), \tag*{(106)} R ≃ ( □ X ) 2 − ( ∂ a ∂ b X ) ( ∂ a ∂ b X ) , (106)
where indices are raised with the flat worldsheet metric η a b = diag ( − , + ) \eta_{ab}=\operatorname{diag}(-,+) η ab = diag ( − , + ) and □ = − ∂ t 2 + ∂ z 2 \Box=-\partial_t^2+\partial_z^2 □ = − ∂ t 2 + ∂ z 2 .
On the other hand, the linearized transverse equation for the zero modes of this transverse displacement is
□ X = 0. (107) \Box X=0. \tag*{(107)} □ X = 0. (107)
Using this equation we obtain,
R ≃ − ( ∂ a ∂ b X ) ( ∂ a ∂ b X ) = − 1 4 □ 2 ( X 2 ) . (108) R\simeq-(\partial_a\partial_bX)(\partial^a\partial^bX)=-\frac{1}{4}\Box^2(X^2). \tag*{(108)} R ≃ − ( ∂ a ∂ b X ) ( ∂ a ∂ b X ) = − 4 1 □ 2 ( X 2 ) . (108)
Consequently, the leading mixed interaction is
S χ R ≃ − κ S 4 ∫ d t d z χ □ 2 ( X 2 ) . (109) S_{\chi R}\simeq-\frac{\kappa_S}{4}\int dt\,dz\,\chi\Box^2(X^2). \tag*{(109)} S χ R ≃ − 4 κ S ∫ d t d z χ □ 2 ( X 2 ) . (109)
Integrating by parts twice and discarding boundary terms yields
S χ R ≃ − κ S 4 ∫ d t d z χ □ 2 ( X 2 ) = − κ S 4 ∫ d t d z X 2 □ 2 χ . (110) S_{\chi R}\simeq-\frac{\kappa_S}{4}\int dt\,dz\,\chi\Box^2(X^2)=-\frac{\kappa_S}{4}\int dt\,dz\,X^2\Box^2\chi. \tag*{(110)} S χ R ≃ − 4 κ S ∫ d t d z χ □ 2 ( X 2 ) = − 4 κ S ∫ d t d z X 2 □ 2 χ . (110)
The free bound-state equation,
( □ − ω b 2 ) χ = 0 , (111) (\Box-\omega_b^2)\chi=0, \tag*{(111)} ( □ − ω b 2 ) χ = 0 , (111)
implies □ 2 χ = ω b 4 χ \Box^2\chi=\omega_b^4\chi □ 2 χ = ω b 4 χ . Thus,
S χ R ≃ − κ S ω b 4 4 ∫ d t d z χ X 2 . (112) S_{\chi R}\simeq-\frac{\kappa_S\omega_b^4}{4}\int dt\,dz\,\chi X^2. \tag*{(112)} S χ R ≃ − 4 κ S ω b 4 ∫ d t d z χ X 2 . (112)
The shape mode coupling in the Collective Coordinate approach
To compare to the collective coordinate description, consider a periodic interval 0 ≤ z < L 0\leq z<L 0 ≤ z < L and write
X ( t , z ) = X ( t ) cos ( k z ) , χ ( t , z ) = S ( t ) . (113) X(t,z)=X(t)\cos(kz),\qquad\chi(t,z)=S(t). \tag*{(113)} X ( t , z ) = X ( t ) cos ( k z ) , χ ( t , z ) = S ( t ) . (113)
Using
∫ 0 L d z cos 2 ( k z ) = L 2 , (114) \int_0^L dz\,\cos^2(kz)=\frac{L}{2}, \tag*{(114)} ∫ 0 L d z cos 2 ( k z ) = 2 L , (114)
we obtain
S c o u p l i n g ≃ − κ S L ω b 4 8 ∫ d t S ( t ) X ( t ) 2 . (115) S_{\rm coupling}\simeq-\frac{\kappa_S L\omega_b^4}{8}\int dt\,S(t)X(t)^2. \tag*{(115)} S coupling ≃ − 8 κ S L ω b 4 ∫ d t S ( t ) X ( t ) 2 . (115)
The corresponding mixed cubic term in the reduced Lagrangian is therefore
L E F T ( 3 ) = − κ S L ω b 4 8 S X 2 . (116) L_{\rm EFT}^{(3)}=-\frac{\kappa_S L\omega_b^4}{8}SX^2. \tag*{(116)} L EFT ( 3 ) = − 8 κ S L ω b 4 S X 2 . (116)
Another approach one can follow is to consider the straight string and allow for a perturbation of the shape mode with time dependent amplitude given by S ( t ) S(t) S ( t ) and a zero mode parametrized by X ( t ) X(t) X ( t ) . Using the radial profiles of these modes already discussed in the main part of the text, one can introduce these perturbations into the 4d action and integrate the transverse directions. Following this procedure we arrive to an mechanical Lagrangian of the form,
L C C ⊃ L 2 ( S ˙ 2 − ω b 2 S 2 ) − g s L 2 S X 2 + ⋯ . (117) \mathcal{L}_{\rm CC}\supset\frac{L}{2}(\dot{S}^{2}-\omega_b^{2}S^{2})-\frac{g_sL}{2}SX^{2}+\cdots. \tag*{(117)} L CC ⊃ 2 L ( S ˙ 2 − ω b 2 S 2 ) − 2 g s L S X 2 + ⋯ . (117)
so that the relevant coupling term is given by,
L C C ( 3 ) = − L g s 2 S X 2 . (118) L_{\rm CC}^{(3)}=-\frac{Lg_s}{2}SX^2. \tag*{(118)} L CC ( 3 ) = − 2 L g s S X 2 . (118)
and where the coupling between the shape and the zero mode is obtained by computing
g s = π ∫ 0 ∞ d ρ ρ f ′ 2 [ f φ + a ′ + a ρ ] . (119) g_s=\pi\int_0^\infty d\rho\,\rho f'^2\left[f\varphi+a'+\frac{a}{\rho}\right]. \tag*{(119)} g s = π ∫ 0 ∞ d ρ ρ f ′2 [ f φ + a ′ + ρ a ] . (119)
Matching the coefficients
We now show that the collective coordinate coefficient g s g_s g s and the geometric coupling κ S \kappa_S κ S satisfy
g s = ω b 4 4 κ S . (120) g_s=\frac{\omega_b^4}{4}\kappa_S. \tag*{(120)} g s = 4 ω b 4 κ S . (120)
Their radial expressions are
g s = π ∫ 0 ∞ d ρ ρ ( f ′ ) 2 ( f φ + a ′ + a ρ ) (121) g_s = \pi\int_0^\infty d\rho\,\rho(f')^2\left(f\varphi+a'+\frac{a}{\rho}\right) \tag*{(121)} g s = π ∫ 0 ∞ d ρ ρ ( f ′ ) 2 ( f φ + a ′ + ρ a ) (121)
κ S = π ∫ 0 ∞ d ρ ( ρ 2 f ′ φ + ρ α ′ a ) . (122) \kappa_S = \pi\int_0^\infty d\rho\,\left(\rho^2 f'\varphi+\rho\alpha'a\right). \tag*{(122)} κ S = π ∫ 0 ∞ d ρ ( ρ 2 f ′ φ + ρ α ′ a ) . (122)
We use the normalized profiles of the main text,
φ = f ψ ω b , a = − ψ ′ ω b , (123) \varphi=\frac{f\psi}{\omega_b},\qquad a=-\frac{\psi'}{\omega_b}, \tag*{(123)} φ = ω b f ψ , a = − ω b ψ ′ , (123)
where ψ \psi ψ satisfies
H ψ = ω b 2 ψ , H = − d 2 d ρ 2 − 1 ρ d d ρ + f 2 . (124) H\psi=\omega_b^2\psi,\qquad H=-\frac{d^2}{d\rho^2}-\frac{1}{\rho}\frac{d}{d\rho}+f^2. \tag*{(124)} H ψ = ω b 2 ψ , H = − d ρ 2 d 2 − ρ 1 d ρ d + f 2 . (124)
Consequently,
f φ + a ′ + a ρ = 1 ω b ( f 2 ψ − ψ ′ ′ − ψ ′ ρ ) = ω b ψ , (125) f\varphi+a'+\frac{a}{\rho}=\frac{1}{\omega_b}\left(f^2\psi-\psi''-\frac{\psi'}{\rho}\right)=\omega_b\psi, \tag*{(125)} f φ + a ′ + ρ a = ω b 1 ( f 2 ψ − ψ ′′ − ρ ψ ′ ) = ω b ψ , (125)
and the collective coordinate coefficient becomes
g s = π ω b ∫ 0 ∞ d ρ ρ ( f ′ ) 2 ψ . (126) g_s=\pi\omega_b\int_0^\infty d\rho\,\rho(f')^2\psi. \tag*{(126)} g s = π ω b ∫ 0 ∞ d ρ ρ ( f ′ ) 2 ψ . (126)
To simplify κ S \kappa_S κ S , define the background magnetic field
B ( ρ ) ≡ α ′ ρ = 1 − f 2 2 , B ′ = − f f ′ , (127) B(\rho)\equiv\frac{\alpha'}{\rho}=\frac{1-f^2}{2},\qquad B'=-ff', \tag*{(127)} B ( ρ ) ≡ ρ α ′ = 2 1 − f 2 , B ′ = − f f ′ , (127)
where the second equality follows from the BPS equations. Substituting Eq. (123) into the expression for κ S \kappa_S κ S gives
κ S = π ω b ∫ 0 ∞ d ρ ρ 2 ( f f ′ ψ − B ψ ′ ) = − π ω b ∫ 0 ∞ d ρ ρ 2 ( B ψ ) ′ = 2 π ω b ∫ 0 ∞ d ρ ρ B ψ . (128) \begin{aligned}
\kappa_S&=\frac{\pi}{\omega_b}\int_0^\infty d\rho\,\rho^2(ff'\psi-B\psi')\\
&=-\frac{\pi}{\omega_b}\int_0^\infty d\rho\,\rho^2(B\psi)'\\
&=\frac{2\pi}{\omega_b}\int_0^\infty d\rho\,\rho B\psi. \tag*{(128)}
\end{aligned} κ S = ω b π ∫ 0 ∞ d ρ ρ 2 ( f f ′ ψ − B ψ ′ ) = − ω b π ∫ 0 ∞ d ρ ρ 2 ( B ψ ) ′ = ω b 2 π ∫ 0 ∞ d ρ ρB ψ . (128)
The boundary term [ ρ 2 B ψ ] 0 ∞ [\rho^2B\psi]_0^\infty [ ρ 2 B ψ ] 0 ∞ vanishes by regularity at the origin and decay at infinity. The two remaining overlap integrals are related by the identity
H B = 2 ( f ′ ) 2 . (129) HB=2(f')^2. \tag*{(129)} H B = 2 ( f ′ ) 2 . (129)
To establish it, differentiate the first BPS equation, f ′ = ( 1 − α ) f / ρ f'=(1-\alpha)f/\rho f ′ = ( 1 − α ) f / ρ , obtaining
f ′ ′ + f ′ ρ = 1 − α ρ f ′ − B f . (130) f''+\frac{f'}{\rho}=\frac{1-\alpha}{\rho}f'-Bf. \tag*{(130)} f ′′ + ρ f ′ = ρ 1 − α f ′ − B f . (130)
Together with B ′ = − f f ′ B'=-ff' B ′ = − f f ′ , this gives
B ′ ′ + B ′ ρ = − ( f ′ ) 2 − f ( f ′ ′ + f ′ ρ ) = − 2 ( f ′ ) 2 + B f 2 . (131) \begin{aligned}
B''+\frac{B'}{\rho}&=-(f')^2-f\left(f''+\frac{f'}{\rho}\right)\\
&=-2(f')^2+Bf^2. \tag*{(131)}
\end{aligned} B ′′ + ρ B ′ = − ( f ′ ) 2 − f ( f ′′ + ρ f ′ ) = − 2 ( f ′ ) 2 + B f 2 . (131)
which immediately yields Eq. (129) .
Vector mode couplings
Here we will show the analogous situation for the coupling of the vector mode to the transverse string displacements.
The vector mode coupling in the EFT approach
The relevant terms in the effective action are
S e f f = − ∫ d 2 σ − γ [ μ + 1 4 G A B G A B + ω b 2 2 C A C A + κ V 2 G A B Ω A B ] + ⋯ . (132) S_{\mathrm{eff}}=-\int d^2\sigma\sqrt{-\gamma}\left[\mu+\frac{1}{4}G_{AB}G^{AB}+\frac{\omega_b^2}{2}C_A C^A+\frac{\kappa_V}{2}G_{AB}\Omega^{AB}\right]+\cdots. \tag*{(132)} S eff = − ∫ d 2 σ − γ [ μ + 4 1 G A B G A B + 2 ω b 2 C A C A + 2 κ V G A B Ω A B ] + ⋯ . (132)
where
κ V = 2 π ∫ 0 ∞ ρ d ρ ( 1 − α ) ψ . (133) \kappa_V=2\pi\int_0^\infty\rho\,d\rho\,(1-\alpha)\psi. \tag*{(133)} κ V = 2 π ∫ 0 ∞ ρ d ρ ( 1 − α ) ψ . (133)
For the convention ω A = n 1 μ ∂ A n μ \omega_A=n_1\mu\partial_A n^\mu ω A = n 1 μ ∂ A n μ of the main text, the Ricci equation gives, to quadratic order in the static gauge displacements X i ( t , z ) X^i(t,z) X i ( t , z ) ,
Ω A B ( 2 ) = ϵ i j ( ∂ A ∂ C X i ) ( ∂ B ∂ C X j ) ≃ 1 2 ϵ i j □ ( ∂ A X i ∂ B X j ) . (134) \Omega_{AB}^{(2)}=\epsilon_{ij}(\partial_A\partial_CX^i)(\partial_B\partial^CX^j)\simeq\frac{1}{2}\epsilon_{ij}\Box(\partial_AX^i\partial_BX^j). \tag*{(134)} Ω A B ( 2 ) = ϵ ij ( ∂ A ∂ C X i ) ( ∂ B ∂ C X j ) ≃ 2 1 ϵ ij □ ( ∂ A X i ∂ B X j ) . (134)
The boundary term vanishes because B B B and ψ \psi ψ are regular at the origin and decay at infinity. Using H B = 2 ( f ′ ) 2 HB=2(f')^2 H B = 2 ( f ′ ) 2 and H ψ = ω b 2 ψ H\psi=\omega_b^2\psi H ψ = ω b 2 ψ , we therefore find
2 ∫ 0 ∞ d ρ ρ ( f ′ ) 2 ψ = ω b 2 ∫ 0 ∞ d ρ ρ B ψ . (135) 2\int_0^\infty d\rho\,\rho(f')^2\psi=\omega_b^2\int_0^\infty d\rho\,\rho B\psi. \tag*{(135)} 2 ∫ 0 ∞ d ρ ρ ( f ′ ) 2 ψ = ω b 2 ∫ 0 ∞ d ρ ρB ψ . (135)
Substituting this identity into Eq. (126) and comparing with Eq. (128) gives
g s = π ω b 3 2 ∫ 0 ∞ d ρ ρ B ψ = ω b 4 4 κ S . (136) g_s=\frac{\pi\omega_b^3}{2}\int_0^\infty d\rho\,\rho B\psi=\frac{\omega_b^4}{4}\kappa_S. \tag*{(136)} g s = 2 π ω b 3 ∫ 0 ∞ d ρ ρB ψ = 4 ω b 4 κ S . (136)
Thus the two expressions for the mixed cubic coupling agree analytically at critical coupling, with the mode normalizations used in the main text. This provides an explicit consistency check between the effective action and the collective coordinate description for this interaction, supporting their agreement on the associated leading parametric instability within their common regime of validity. The last step uses □ X i = 0 \Box X^i = 0 □ X i = 0 . Here and below, ≃ \simeq ≃ denotes equivalence through the mixed cubic order, up to total derivatives and terms proportional to the free equations of motion.
The free Proca equation is
∂ A G A B = ω b 2 C B . (137) \partial_A G^{AB} = \omega_b^2 C^B . \tag*{(137)} ∂ A G A B = ω b 2 C B . (137)
Its divergence implies ∂ A C A = 0 \partial_A C^A = 0 ∂ A C A = 0 , and consequently □ G A B = ω b 2 G A B \Box G^{AB} = \omega_b^2 G^{AB} □ G A B = ω b 2 G A B . Moving the wave operator in Eq. (134) onto G A B G^{AB} G A B gives
S G Ω ( 3 ) = − κ V 2 ∫ d t d z G A B Ω A B ( 2 ) ≃ − κ V ω b 2 4 ∫ d t d z ϵ i j G A B ∂ A X i ∂ B X j ≃ κ V ω b 4 4 ∫ d t d z C B ϵ i j X i ∂ B X j . (138) \begin{aligned}
S_{\mathrm{G}\Omega}^{(3)} &= -\frac{\kappa_V}{2}\int dt\,dz\,G^{AB}\Omega_{AB}^{(2)} \\
&\simeq-\frac{\kappa_V\omega_b^2}{4}\int dt\,dz\,\epsilon_{ij}G^{AB}\partial_A X^i\partial_B X^j \\
&\simeq\frac{\kappa_V\omega_b^4}{4}\int dt\,dz\,C^B\epsilon_{ij}X^i\partial_B X^j . \tag*{(138)}
\end{aligned} S G Ω ( 3 ) = − 2 κ V ∫ d t d z G A B Ω A B ( 2 ) ≃ − 4 κ V ω b 2 ∫ d t d z ϵ ij G A B ∂ A X i ∂ B X j ≃ 4 κ V ω b 4 ∫ d t d z C B ϵ ij X i ∂ B X j . (138)
In the last step we integrated by parts once and used Eq. (137) ; the term containing G A B ∂ A ∂ B X j G^{AB}\partial_A\partial_B X^j G A B ∂ A ∂ B X j vanishes by antisymmetry.
To allow for a general relative spatial phase, choose two orthogonal transverse standing waves of the form
X 1 = X ( t ) cos ( k z ) , X 2 = Y ( t ) cos ( k z + δ ) (139) X^1 = X(t)\cos(kz), \qquad X^2 = Y(t)\cos(kz+\delta) \tag*{(139)} X 1 = X ( t ) cos ( k z ) , X 2 = Y ( t ) cos ( k z + δ ) (139)
together with
C z = C ( t ) (140) C_z = C(t) \tag*{(140)} C z = C ( t ) (140)
where δ \delta δ is constant and k = 2 π n / L k = 2\pi n/L k = 2 π n / L with n ≥ 1 n \ge1 n ≥ 1 . For 0 ≤ z < L 0 \le z < L 0 ≤ z < L , their antisymmetric product is
ϵ i j X i ∂ z X j = k X Y [ sin ( k z ) cos ( k z + δ ) − cos ( k z ) sin ( k z + δ ) ] = − k X Y sin δ . (141) \begin{aligned}
\epsilon_{ij}X^i\partial_z X^j &= kXY\left[\sin(kz)\cos(kz+\delta)-\cos(kz)\sin(kz+\delta)\right] \\
&= -kXY\sin\delta. \tag*{(141)}
\end{aligned} ϵ ij X i ∂ z X j = k X Y [ sin ( k z ) cos ( k z + δ ) − cos ( k z ) sin ( k z + δ ) ] = − k X Y sin δ . (141)
Thus the mixed cubic interaction takes the form
L E F T ( 3 ) = − L k ω b 4 κ V 4 C X Y sin δ . (142) L_{\mathrm{EFT}}^{(3)} = -Lk\frac{\omega_b^4\kappa_V}{4}CXY\sin\delta. \tag*{(142)} L EFT ( 3 ) = − L k 4 ω b 4 κ V C X Y sin δ . (142)
The vector mode coupling in the collective coordinate approach
For the microscopic reduction [17 ] , use the gauge-compensated translational profiles and the vector profile:
ϕ = ϕ ˉ − q i D ˉ i ϕ ˉ , A j = A ˉ j − q i F ˉ i j , A z = C ( t ) ψ ( ρ ) , A t = 0 , (143) \begin{aligned}
\phi&= \bar{\phi}-q^i\bar{D}_i\bar{\phi}, \\
A_j &= \bar{A}_j-q^i\bar{F}_{ij}, \\
A_z &= C(t)\psi(\rho), \\
A_t &= 0, \tag*{(143)}
\end{aligned} ϕ A j A z A t = ϕ ˉ − q i D ˉ i ϕ ˉ , = A ˉ j − q i F ˉ ij , = C ( t ) ψ ( ρ ) , = 0 , (143)
together with,
q 1 = X ( t ) cos ( k z ) , q 2 = Y ( t ) cos ( k z + δ ) , (144) q^1 = X(t)\cos(kz), \qquad q^2 = Y(t)\cos(kz+\delta), \tag*{(144)} q 1 = X ( t ) cos ( k z ) , q 2 = Y ( t ) cos ( k z + δ ) , (144)
where i , j = x , y i,j=x,y i , j = x , y and D ˉ i = ∂ i − i A ˉ i \bar{D}_i=\partial_i-i\bar{A}_i D ˉ i = ∂ i − i A ˉ i . The BPS relations give D ˉ x ϕ ˉ = f ′ \bar{D}_x\bar{\phi}=f' D ˉ x ϕ ˉ = f ′ and D ˉ y ϕ ˉ = i f ′ \bar{D}_y\bar{\phi}=if' D ˉ y ϕ ˉ = i f ′ , so
δ ϕ = − f ′ ( q 1 + i q 2 ) . (145) \delta\phi=-f'(q^1+iq^2). \tag*{(145)} δ ϕ = − f ′ ( q 1 + i q 2 ) . (145)
The required cubic term comes entirely from the longitudinal Higgs kinetic term − ∣ D z ϕ ∣ 2 / ( 2 e 2 ) -|D_z\phi|^2/(2e^2) − ∣ D z ϕ ∣ 2 / ( 2 e 2 ) , where D z = ∂ z − i C ψ D_z=\partial_z-iC\psi D z = ∂ z − i C ψ . Its part linear in C C C and quadratic in δ ϕ \delta\phi δ ϕ is
L 4 d , C q q ( 3 ) = C ψ Im ( δ ϕ ∗ ∂ z δ ϕ ) = C ψ ( f ′ ) 2 ( q 1 ∂ z q 2 − q 2 ∂ z q 1 ) . (146) \begin{aligned}
L_{4d,Cqq}^{(3)} &= C\psi\,\operatorname{Im}(\delta\phi^*\partial_z\delta\phi) \\
&= C\psi(f')^2(q^1\partial_zq^2-q^2\partial_zq^1). \tag*{(146)}
\end{aligned} L 4 d , C q q ( 3 ) = C ψ Im ( δ ϕ ∗ ∂ z δ ϕ ) = C ψ ( f ′ ) 2 ( q 1 ∂ z q 2 − q 2 ∂ z q 1 ) . (146)
The Abelian-Maxwell action is quadratic in the linear field ansatz, and the Higgs potential contains no vector field; neither supplies an additional C X Y CXY C X Y term. Integrating over the transverse plane and the periodic interval gives
L C C ( 3 ) = − L k g v C X Y sin δ , (147) L_{\mathrm{CC}}^{(3)}=-Lk g_v CXY\sin\delta, \tag*{(147)} L CC ( 3 ) = − L k g v C X Y sin δ , (147)
where
g v = 2 π ∫ 0 ∞ ρ d ρ ( f ′ ) 2 ψ . (148) g_v=2\pi\int_0^\infty\rho\,d\rho\,(f')^2\psi. \tag*{(148)} g v = 2 π ∫ 0 ∞ ρ d ρ ( f ′ ) 2 ψ . (148)
Matching the coefficients
Using the already defined radial operator, H H H ,
H = − d 2 d ρ 2 − 1 ρ d d ρ + f 2 , H ψ = ω b 2 ψ , (149) H=-\frac{d^2}{d\rho^2}-\frac{1}{\rho}\frac{d}{d\rho}+f^2,\qquad H\psi=\omega_b^2\psi, \tag*{(149)} H = − d ρ 2 d 2 − ρ 1 d ρ d + f 2 , H ψ = ω b 2 ψ , (149)
and the background magnetic field,
B = α ′ ρ = 1 − f 2 2 , (150) B=\frac{\alpha'}{\rho}=\frac{1-f^2}{2}, \tag*{(150)} B = ρ α ′ = 2 1 − f 2 , (150)
as well as f ′ = ( 1 − α ) f / ρ f'=(1-\alpha)f/\rho f ′ = ( 1 − α ) f / ρ and B ′ = − f f ′ B'=-ff' B ′ = − f f ′ , direct differentiation yields
H ( 1 − α ) = 2 B , H B = 2 ( f ′ ) 2 . (151) H(1-\alpha)=2B,\qquad HB=2(f')^2. \tag*{(151)} H ( 1 − α ) = 2 B , H B = 2 ( f ′ ) 2 . (151)
The second identity is the same one used in the shape mode calculation. The operator H H H is self-adjoint with radial measure ρ d ρ \rho\,d\rho ρ d ρ :
∫ 0 ∞ ρ d ρ ( u H v − v H u ) = [ ρ ( v u ′ − u v ′ ) ] 0 ∞ . (152) \int_0^\infty\rho\,d\rho\,(uHv-vHu)=\left[\rho(vu'-uv')\right]_0^\infty. \tag*{(152)} ∫ 0 ∞ ρ d ρ ( u H v − v H u ) = [ ρ ( v u ′ − u v ′ ) ] 0 ∞ . (152)
For ψ \psi ψ , B B B , and 1 − α 1-\alpha 1 − α , the relevant boundary terms vanish by regularity at the origin and decay at infinity. Applying this identity twice gives
∫ 0 ∞ ρ d ρ ( f ′ ) 2 ψ = 1 2 ∫ 0 ∞ ρ d ρ ψ H B = ω b 2 2 ∫ 0 ∞ ρ d ρ B ψ = ω b 2 4 ∫ 0 ∞ ρ d ρ ψ H ( 1 − α ) = ω b 4 4 ∫ 0 ∞ ρ d ρ ( 1 − α ) ψ . (153) \begin{aligned}
\int_0^\infty\rho\,d\rho\,(f')^2\psi
&=\frac{1}{2}\int_0^\infty\rho\,d\rho\,\psi HB
=\frac{\omega_b^2}{2}\int_0^\infty\rho\,d\rho\,B\psi\\
&=\frac{\omega_b^2}{4}\int_0^\infty\rho\,d\rho\,\psi H(1-\alpha) \\
&=\frac{\omega_b^4}{4}\int_0^\infty\rho\,d\rho\,(1-\alpha)\psi. \tag*{(153)}
\end{aligned} ∫ 0 ∞ ρ d ρ ( f ′ ) 2 ψ = 2 1 ∫ 0 ∞ ρ d ρ ψ H B = 2 ω b 2 ∫ 0 ∞ ρ d ρ B ψ = 4 ω b 2 ∫ 0 ∞ ρ d ρ ψ H ( 1 − α ) = 4 ω b 4 ∫ 0 ∞ ρ d ρ ( 1 − α ) ψ . (153)
Multiplying by 2 π 2\pi 2 π establishes
g v = ω b 4 4 κ V , (154) g_{\mathrm{v}}=\frac{\omega_{\mathrm{b}}^{4}}{4}\kappa_{\mathrm{V}}, \tag*{(154)} g v = 4 ω b 4 κ V , (154)
and therefore
L C C ( 3 ) = L E F T ( 3 ) = − L k ω b 4 κ V 4 C X Y sin δ . (155) L_{\mathrm{CC}}^{(3)}=L_{\mathrm{EFT}}^{(3)}=-\frac{L k\omega_{\mathrm{b}}^{4}\kappa_{\mathrm{V}}}{4}C_{XY}\sin\delta. \tag*{(155)} L CC ( 3 ) = L EFT ( 3 ) = − 4 L k ω b 4 κ V C X Y sin δ . (155)
The calculations above establish agreement between the cubic on-shell interactions obtained from the covariant effective action and those derived within the collective coordinate approach. This agreement provides an independent consistency check of the couplings and demonstrates that, after matching the amplitude definitions, the collective coordinate description reproduces the corresponding weak field interactions of the effective action through cubic order.
Let us finally note that the correspondence between the EFT and the collective coordinate approach becomes considerably more straightforward for contributions involving exclusively massive internal modes. This is expected, as in both approaches these modes are introduced as fluctuations of the original fields around the background solution, using the same transverse profiles. At leading order in curvature corrections, this leads to the same coupling coefficients and dependence on mode amplitudes. This applies, for instance, to the couplings between the shape and vector modes, as well as to the self-interaction terms of the shape mode.
Nambu–Goto scattering and the induced bound mode response
We derive the Nambu–Goto solution used to evaluate the leading geometric sources in the main text. We first construct a perturbative solution in static gauge, allowing a direct comparison with the field theory simulations, and then present an exact solution in null worldsheet coordinates. From this solution, we calculate the time delay produced by the nonlinear interaction of counterpropagating transverse wave packets, providing an analytic prediction for comparison with the lattice results. We subsequently derive the retarded response of the internal modes to an arbitrary source and specialize to the Gaussian profiles used in the numerical experiment. Coordinates and masses are expressed in the dimensionless units adopted in the main text.
Perturbative solution in static gauge
We choose the physical coordinates ( t , z ) (t,z) ( t , z ) on the worldsheet and write the embedding of the string in flat spacetime as
X μ ( t , z ) = ( t , X 1 ( t , z ) , X 2 ( t , z ) , z ) . (156) X^\mu(t,z)=(t,X^1(t,z),X^2(t,z),z). \tag*{(156)} X μ ( t , z ) = ( t , X 1 ( t , z ) , X 2 ( t , z ) , z ) . (156)
The Nambu–Goto equations of motion in this gauge can be written as
( ∂ t 2 − ∂ z 2 ) X i = − ( ∂ z X j ∂ z X j ) ∂ t 2 X i + 2 ( ∂ t X j ∂ z X j ) ∂ t ∂ z X i − ( ∂ t X j ∂ t X j ) ∂ z 2 X i . (157) (\partial_t^2-\partial_z^2)X^i=-(\partial_zX^j\partial_zX^j)\partial_t^2X^i+2(\partial_tX^j\partial_zX^j)\partial_t\partial_zX^i-(\partial_tX^j\partial_tX^j)\partial_z^2X^i. \tag*{(157)} ( ∂ t 2 − ∂ z 2 ) X i = − ( ∂ z X j ∂ z X j ) ∂ t 2 X i + 2 ( ∂ t X j ∂ z X j ) ∂ t ∂ z X i − ( ∂ t X j ∂ t X j ) ∂ z 2 X i . (157)
An isolated traveling wave X i = F i ( t ± z ) X^i=F^i(t\pm z) X i = F i ( t ± z ) solves this equation exactly. Corresponding solutions also exist in the full nonlinear field theory [43 ] , providing initial data for the scattering experiment.
For two counterpropagating disturbances, we define
q = t + z , r = t − z , (158) q=t+z,\qquad r=t-z, \tag*{(158)} q = t + z , r = t − z , (158)
and take the leading solution
X ( 1 ) i = e + i F ( q ) + e − i G ( r ) , (159) X^i_{(1)}=e^i_{+}F(q)+e^i_{-}G(r), \tag*{(159)} X ( 1 ) i = e + i F ( q ) + e − i G ( r ) , (159)
where
e + i e + i = 1 , e + i e − i = cos δ , e^i_{+}e^i_{+}=1,\qquad e^i_{+}e^i_{-}=\cos\delta, e + i e + i = 1 , e + i e − i = cos δ ,
ϵ i j e + i e − j = sin δ . (160) \epsilon_{ij}e^i_{+}e^j_{-}=\sin\delta. \tag*{(160)} ϵ ij e + i e − j = sin δ . (160)
Thus F F F propagates toward decreasing z z z and G G G toward increasing z z z . The functions F F F and G G G are the physical incoming profiles in static gauge.
For profiles of amplitude A A A and characteristic width L L L , we introduce ϵ = A / L ≪ 1 \epsilon=A/L\ll1 ϵ = A / L ≪ 1 and expand
X i = X ( 1 ) i + X ( 3 ) i + X ( 5 ) i + ⋯ , X ( n ) i = O ( ϵ n L ) . (161) X^i=X^i_{(1)}+X^i_{(3)}+X^i_{(5)}+\cdots,\qquad X^i_{(n)}=\mathcal{O}(\epsilon^nL). \tag*{(161)} X i = X ( 1 ) i + X ( 3 ) i + X ( 5 ) i + ⋯ , X ( n ) i = O ( ϵ n L ) . (161)
Here, F , G = O ( ϵ L ) F,G=\mathcal{O}(\epsilon L) F , G = O ( ϵ L ) , F ′ , G ′ = O ( ϵ ) F',G'=\mathcal{O}(\epsilon) F ′ , G ′ = O ( ϵ ) , and F ′ ′ , G ′ ′ = O ( ϵ / L ) F'',G''=\mathcal{O}(\epsilon/L) F ′′ , G ′′ = O ( ϵ / L ) . Substituting the leading solution into the nonlinear terms in Eq. (C2) gives
( ∂ t 2 − ∂ z 2 ) X ( 3 ) i = − 4 e + i F ′ ′ ( q ) G ′ 2 ( r ) − 4 e − i G ′ ′ ( r ) F ′ 2 ( q ) . (162) (\partial_t^2-\partial_z^2)X^i_{(3)}=-4e^i_{+}F''(q)G'^2(r)-4e^i_{-}G''(r)F'^2(q). \tag*{(162)} ( ∂ t 2 − ∂ z 2 ) X ( 3 ) i = − 4 e + i F ′′ ( q ) G ′2 ( r ) − 4 e − i G ′′ ( r ) F ′2 ( q ) . (162)
The terms depending on e + ⋅ e − e_{+}\cdot e_{-} e + ⋅ e − cancel. Define
I + ( q ) = ∫ − ∞ q d s F ′ 2 ( s ) , I_{+}(q)=\int_{-\infty}^{q}ds\,F'^2(s), I + ( q ) = ∫ − ∞ q d s F ′2 ( s ) ,
I − ( r ) = ∫ − ∞ r d s G ′ 2 ( s ) . (163) I_{-}(r)=\int_{-\infty}^{r}ds\,G'^2(s). \tag*{(163)} I − ( r ) = ∫ − ∞ r d s G ′2 ( s ) . (163)
Since ∂ t 2 − ∂ z 2 = 4 ∂ q ∂ r \partial_t^2-\partial_z^2=4\partial_q\partial_r ∂ t 2 − ∂ z 2 = 4 ∂ q ∂ r , we integrate Eq. (C7) with boundary conditions that recover the prescribed incoming profiles F F F and G G G before the collision. The resulting solution, including the leading nonlinear correction generated by their interaction, is
X i ( t , z ) = e + i F ( q ) + e − i G ( r ) − e + i F ′ ( q ) I − ( r ) − e − i G ′ ( r ) I + ( q ) + O ( ϵ 5 L ) . (164) X^i(t,z)=e^i_{+}F(q)+e^i_{-}G(r)
-e^i_{+}F'(q)I_{-}(r)-e^i_{-}G'(r)I_{+}(q)
+\mathcal{O}(\epsilon^5L). \tag*{(164)} X i ( t , z ) = e + i F ( q ) + e − i G ( r ) − e + i F ′ ( q ) I − ( r ) − e − i G ′ ( r ) I + ( q ) + O ( ϵ 5 L ) . (164)
For initially separated profiles of compact support, the correction and its first time derivative vanish before the collision. Gaussian profiles satisfy this condition asymptotically, with exponentially small corrections at finite initial separation.
For completeness, the next contribution is
X ( 5 ) i = e + i [ F ′ G ′ 2 I + + 1 2 F ′ ′ I − 2 ] + e − i [ G ′ F ′ 2 I − + 1 2 G ′ ′ I + 2 ] . (165) X^{i}_{(5)}=e^{i}_{+}\left[F'G'^{2}I_{+}+\frac{1}{2}F''I_{-}^{2}\right]+e^{i}_{-}\left[G'F'^{2}I_{-}+\frac{1}{2}G''I_{+}^{2}\right]. \tag*{(165)} X ( 5 ) i = e + i [ F ′ G ′ 2 I + + 2 1 F ′′ I − 2 ] + e − i [ G ′ F ′ 2 I − + 2 1 G ′′ I + 2 ] . (165)
where F F F , its derivatives, and I + I_{+} I + are evaluated at q q q , while G G G , its derivatives, and I − I_{-} I − are evaluated at r r r .
Along the outgoing F F F disturbance, the G G G disturbance has passed and I − ( r ) = I − ( + ∞ ) I_{-}(r)=I_{-}(+\infty) I − ( r ) = I − ( + ∞ ) . Equation (C9) becomes
X i = e + i [ F ( q ) − I − ( + ∞ ) F ′ ( q ) ] + O ( ϵ 5 L ) . (166) X^{i}=e^{i}_{+}\left[F(q)-I_{-}(+\infty)F'(q)\right]+O(\epsilon^{5}L). \tag*{(166)} X i = e + i [ F ( q ) − I − ( + ∞ ) F ′ ( q ) ] + O ( ϵ 5 L ) . (166)
This is the expansion of a delayed profile e + i F ( t + z − Δ t F ) e^{i}_{+}F(t+z-\Delta t_{F}) e + i F ( t + z − Δ t F ) , with
Δ t F = ∫ − ∞ ∞ d s G ′ 2 ( s ) , Δ t G = ∫ − ∞ ∞ d s F ′ 2 ( s ) . (167) \begin{aligned}
\Delta t_{F}&=\int_{-\infty}^{\infty}ds\,G'^{2}(s),\\
\Delta t_{G}&=\int_{-\infty}^{\infty}ds\,F'^{2}(s). \tag*{(167)}
\end{aligned} Δ t F Δ t G = ∫ − ∞ ∞ d s G ′ 2 ( s ) , = ∫ − ∞ ∞ d s F ′ 2 ( s ) . (167)
The perturbative calculation determines these delays at leading order O ( ϵ 2 L ) O(\epsilon^{2}L) O ( ϵ 2 L ) . The exact solution below shows that Eq. (167) is in fact exact for the specified incoming profiles.
Exact solution in null coordinates
The exact scattering solution can be expressed in null worldsheet coordinates [44 , 45 ] . Introduce coordinates ( u , v ) (u,v) ( u , v ) and define the embedding implicitly by
t + z = u + I − ( v ) , t − z = v + I + ( u ) , X i = e + i F ( u ) + e − i G ( v ) . (168) \begin{aligned}
t+z&=u+I_{-}(v),\\
t-z&=v+I_{+}(u),\\
X^{i}&=e^{i}_{+}F(u)+e^{i}_{-}G(v). \tag*{(168)}
\end{aligned} t + z t − z X i = u + I − ( v ) , = v + I + ( u ) , = e + i F ( u ) + e − i G ( v ) . (168)
Its tangent vectors are
∂ u X μ = ( 1 + F ′ 2 2 , e + i F ′ , 1 − F ′ 2 2 ) , ∂ v X μ = ( 1 + G ′ 2 2 , e − i G ′ , − 1 − G ′ 2 2 ) . (169) \begin{aligned}
\partial_{u}X^{\mu}&=\left(\frac{1+F'^{2}}{2},e^{i}_{+}F',\frac{1-F'^{2}}{2}\right),\\
\partial_{v}X^{\mu}&=\left(\frac{1+G'^{2}}{2},e^{i}_{-}G',-\frac{1-G'^{2}}{2}\right). \tag*{(169)}
\end{aligned} ∂ u X μ ∂ v X μ = ( 2 1 + F ′ 2 , e + i F ′ , 2 1 − F ′ 2 ) , = ( 2 1 + G ′ 2 , e − i G ′ , − 2 1 − G ′ 2 ) . (169)
They satisfy
∂ u X μ ∂ u X μ = 0 , ∂ v X μ ∂ v X μ = 0 , ∂ u ∂ v X μ = 0. (170) \partial_{u}X^{\mu}\partial_{u}X_{\mu}=0,\qquad\partial_{v}X^{\mu}\partial_{v}X_{\mu}=0,\qquad\partial_{u}\partial_{v}X^{\mu}=0. \tag*{(170)} ∂ u X μ ∂ u X μ = 0 , ∂ v X μ ∂ v X μ = 0 , ∂ u ∂ v X μ = 0. (170)
These are the Nambu–Goto equations and constraints in null coordinates. Our solution given by Eq. (168) satisfies these equations in the small slope regime expressing a single valued transverse position for the string in the static gauge3 .
Expanding the implicit relations gives
u = q − I − ( r ) + G ′ 2 ( r ) I + ( q ) + O ( ϵ 6 L ) , v = r − I + ( q ) + F ′ 2 ( q ) I − ( r ) + O ( ϵ 6 L ) . (171) \begin{aligned}
u&=q-I_{-}(r)+G'^{2}(r)I_{+}(q)+O(\epsilon^{6}L),\\
v&=r-I_{+}(q)+F'^{2}(q)I_{-}(r)+O(\epsilon^{6}L). \tag*{(171)}
\end{aligned} u v = q − I − ( r ) + G ′ 2 ( r ) I + ( q ) + O ( ϵ 6 L ) , = r − I + ( q ) + F ′ 2 ( q ) I − ( r ) + O ( ϵ 6 L ) . (171)
Substitution into F ( u ) F(u) F ( u ) and G ( v ) G(v) G ( v ) reproduces Eqs. (164) and (C10).
To determine the delay exactly, follow a feature u = u 0 u=u_{0} u = u 0 of the F F F disturbance. Along this characteristic,
d t d v = 1 + G ′ 2 2 , d z d v = − 1 − G ′ 2 2 , (172) \frac{dt}{dv}=\frac{1+G'^{2}}{2},\qquad\frac{dz}{dv}=-\frac{1-G'^{2}}{2}, \tag*{(172)} d v d t = 2 1 + G ′ 2 , d v d z = − 2 1 − G ′ 2 , (172)
d ( t + z ) d v = G ′ 2 . (173) \frac{d(t+z)}{dv}=G'^{2}. \tag*{(173)} d v d ( t + z ) = G ′ 2 . (173)
Its incoming trajectory t + z = u 0 t+z=u_{0} t + z = u 0 becomes t + z = u 0 + I − ( + ∞ ) t+z=u_{0}+I_{-}(+\infty) t + z = u 0 + I − ( + ∞ ) after the collision. At a fixed detector, its arrival is therefore delayed by I − ( + ∞ ) I_{-}(+\infty) I − ( + ∞ ) , establishing Eq. (167) . Interchanging the two profiles gives the delay of G G G .
The outgoing profiles retain their shapes within the NG approximation. Their delays are independent of the relative polarization, although the geometry during the collision depends on δ \delta δ .
Gaussian profiles and time delay
For Gaussian profiles of the form,
F ( s ) = G ( s ) = A exp [ − s 2 2 L 2 ] , (174) F(s)=G(s)=A\exp\left[-\frac{s^{2}}{2L^{2}}\right], \tag*{(174)} F ( s ) = G ( s ) = A exp [ − 2 L 2 s 2 ] , (174)
the quantity I ( s ) = ∫ d s F ′ 2 ( s ) I(s)=\int ds\,F'^{2}(s) I ( s ) = ∫ d s F ′ 2 ( s ) becomes
I ( s ) = A 2 L [ π 4 ( 1 + erf s L ) − s 2 L e − s 2 / L 2 ] . (175) I(s)=\frac{A^{2}}{L}\left[\frac{\sqrt{\pi}}{4}\left(1+\operatorname{erf}\frac{s}{L}\right)-\frac{s}{2L}e^{-s^{2}/L^{2}}\right]. \tag*{(175)} I ( s ) = L A 2 [ 4 π ( 1 + erf L s ) − 2 L s e − s 2 / L 2 ] . (175)
Hence the time delay from Eq. (167) is given by,
Δ t F = Δ t G = π A 2 2 L . (176) \Delta t_{F}=\Delta t_{G}=\frac{\sqrt{\pi}A^{2}}{2L}. \tag*{(176)} Δ t F = Δ t G = 2 L π A 2 . (176)
For an isolated traveling wave, the excess Nambu–Goto energy above that of the straight string is
E F = μ e 2 ∫ d s F ′ 2 ( s ) , (177) E_{F}=\frac{\mu}{e^{2}}\int ds\,F'^{2}(s), \tag*{(177)} E F = e 2 μ ∫ d s F ′ 2 ( s ) , (177)
The delays can therefore be written as
Δ t F = e 2 E G μ . (178) \Delta t_{F}=\frac{e^{2}E_{G}}{\mu}. \tag*{(178)} Δ t F = μ e 2 E G . (178)
Restoring physical units gives
Δ t F p h y s = E G p h y s μ p h y s , μ p h y s = π η 2 . (179) \Delta t_{F}^{\mathrm{phys}}=\frac{E_{G}^{\mathrm{phys}}}{\mu^{\mathrm{phys}}},
\qquad
\mu^{\mathrm{phys}}=\pi\eta^2. \tag*{(179)} Δ t F phys = μ phys E G phys , μ phys = π η 2 . (179)
with identical expressions for the G G G wiggle. This was originally identify in [44 , 45 ] . The same relation between delay and incoming excess energy was obtained for domain wall strings in field theory simulations in Ref. [41 ] .
The bound mode response
To the order we are considering here the dynamics of the worldsheet fields obey, according to our effective theory, the equations,
( − □ γ + ω b 2 ) χ = κ S R , ( − □ γ + ω b 2 ) ϑ = − ω b κ V T , (180) \left(-\Box_\gamma+\omega_b^2\right)\chi=\kappa_S\mathcal{R},
\qquad
\left(-\Box_\gamma+\omega_b^2\right)\vartheta=-\omega_b\kappa_V\mathcal{T}, \tag*{(180)} ( − □ γ + ω b 2 ) χ = κ S R , ( − □ γ + ω b 2 ) ϑ = − ω b κ V T , (180)
where □ γ = ( − γ ) − 1 / 2 ∂ A ( − γ γ A B ∂ B ) \Box_\gamma=(-\gamma)^{-1/2}\partial_A(\sqrt{-\gamma}\gamma^{AB}\partial_B) □ γ = ( − γ ) − 1/2 ∂ A ( − γ γ A B ∂ B ) . In the static gauge we get,
( ∂ t 2 − ∂ z 2 + ω b 2 ) χ = κ S R + O ( ϵ 4 ) , ( ∂ t 2 − ∂ z 2 + ω b 2 ) ϑ = − ω b κ V T + O ( ϵ 4 ) , (181) \left(\partial_t^2-\partial_z^2+\omega_b^2\right)\chi=\kappa_S\mathcal{R}+O(\epsilon^4),
\qquad
\left(\partial_t^2-\partial_z^2+\omega_b^2\right)\vartheta=-\omega_b\kappa_V\mathcal{T}+O(\epsilon^4), \tag*{(181)} ( ∂ t 2 − ∂ z 2 + ω b 2 ) χ = κ S R + O ( ϵ 4 ) , ( ∂ t 2 − ∂ z 2 + ω b 2 ) ϑ = − ω b κ V T + O ( ϵ 4 ) , (181)
where the curvature sources are evaluated on the leading transverse solution given by Eq. (159) , namely,
R = − 8 F ′ ′ ( t + z ) G ′ ′ ( t − z ) cos δ + O ( ϵ 4 / L 2 ) , T = 4 F ′ ′ ( t + z ) G ′ ′ ( t − z ) sin δ + O ( ϵ 4 / L 2 ) . (182) \mathcal{R}=-8F^{\prime\prime}(t+z)G^{\prime\prime}(t-z)\cos\delta+O(\epsilon^4/L^2),
\qquad
\mathcal{T}=4F^{\prime\prime}(t+z)G^{\prime\prime}(t-z)\sin\delta+O(\epsilon^4/L^2). \tag*{(182)} R = − 8 F ′′ ( t + z ) G ′′ ( t − z ) cos δ + O ( ϵ 4 / L 2 ) , T = 4 F ′′ ( t + z ) G ′′ ( t − z ) sin δ + O ( ϵ 4 / L 2 ) . (182)
Note that both sources vanish for an isolated traveling wave and become nonzero when the two disturbances overlap. Their polarization dependence determines which internal mode is excited: parallel polarizations source the shape mode, while orthogonal polarizations source the pseudoscalar mode.
On the other hand, the two bound modes have the same mass and their spacetime profile sources during the collision are proportional to one another. Therefore, their responses can be expressed in terms of a single function U \mathcal{U} U , which satisfies
D U = J ( t , z ) , J ( t , z ) = F ′ ′ ( t + z ) G ′ ′ ( t − z ) , (183) \mathcal{D}\mathcal{U}=J(t,z),
\qquad
J(t,z)=F^{\prime\prime}(t+z)G^{\prime\prime}(t-z), \tag*{(183)} DU = J ( t , z ) , J ( t , z ) = F ′′ ( t + z ) G ′′ ( t − z ) , (183)
where
D = ∂ t 2 − ∂ z 2 + ω b 2 . (184) \mathcal{D}=\partial_t^2-\partial_z^2+\omega_b^2. \tag*{(184)} D = ∂ t 2 − ∂ z 2 + ω b 2 . (184)
We choose the retarded solution, so that the internal excitation is generated entirely by the collision. The coupling strengths and relative polarization then determine how this common response appears in each mode:
χ = − 8 κ S U cos δ + O ( ϵ 4 ) , ϑ = − 4 ω b κ V U sin δ + O ( ϵ 4 ) . (185) \chi=-8\kappa_S\mathcal{U}\cos\delta+O(\epsilon^4),
\qquad
\vartheta=-4\omega_b\kappa_V\mathcal{U}\sin\delta+O(\epsilon^4). \tag*{(185)} χ = − 8 κ S U cos δ + O ( ϵ 4 ) , ϑ = − 4 ω b κ V U sin δ + O ( ϵ 4 ) . (185)
As we showed in Appendix subsection A 9, at critical coupling, ω b κ V = 2 κ S \omega_b\kappa_V=2\kappa_S ω b κ V = 2 κ S , so both responses have the same coefficient multiplying their respective polarization factors. Reversing δ \delta δ preserves the scalar response and reverses the pseudoscalar response. The leading duality relations give
C z = ∂ t ϑ ω b + O ( ϵ 4 ) , C t = ∂ z ϑ ω b + O ( ϵ 4 ) . (186) C_z=\frac{\partial_t\vartheta}{\omega_b}+O(\epsilon^4),
\qquad
C_t=\frac{\partial_z\vartheta}{\omega_b}+O(\epsilon^4). \tag*{(186)} C z = ω b ∂ t ϑ + O ( ϵ 4 ) , C t = ω b ∂ z ϑ + O ( ϵ 4 ) . (186)
To solve Eq. (183) , define the spacetime Fourier transform
J ~ ( ω , k ) = ∫ d t d z e i ω t − i k z J ( t , z ) . (187) \widetilde{J}(\omega,k)=\int dt\,dz\,e^{i\omega t-ikz}J(t,z). \tag*{(187)} J ( ω , k ) = ∫ d t d z e iω t − ik z J ( t , z ) . (187)
Then, the retarded solution can be written in terms of this function as,
U ( t , z ) = lim ϵ R → 0 + ∫ d ω 2 π d k 2 π e − i ω t + i k z J ~ ( ω , k ) k 2 + ω b 2 − ( ω + i ϵ R ) 2 , (188) \mathcal{U}(t,z)=\lim_{\epsilon_R\to0^+}\int\frac{d\omega}{2\pi}\frac{dk}{2\pi}\frac{e^{-i\omega t+ikz}\widetilde{J}(\omega,k)}{k^2+\omega_b^2-(\omega+i\epsilon_R)^2}, \tag*{(188)} U ( t , z ) = ϵ R → 0 + lim ∫ 2 π d ω 2 π d k k 2 + ω b 2 − ( ω + i ϵ R ) 2 e − iω t + ik z J ( ω , k ) , (188)
where ϵ R \epsilon_R ϵ R is the retarded regulator. Equivalently, the spatial Fourier modes satisfy
U ¨ k + ω k 2 U k = J k ( t ) , ω k = k 2 + ω b 2 , (189) \ddot{U}_k+\omega_k^2U_k=J_k(t),
\qquad
\omega_k=\sqrt{k^2+\omega_b^2}, \tag*{(189)} U ¨ k + ω k 2 U k = J k ( t ) , ω k = k 2 + ω b 2 , (189)
with solution
U k ( t ) = 1 ω k ∫ − ∞ t d t ′ sin [ ω k ( t − t ′ ) ] J k ( t ′ ) . (190) U_k(t)=\frac{1}{\omega_k}\int_{-\infty}^{t}dt'\,\sin[\omega_k(t-t')]J_k(t'). \tag*{(190)} U k ( t ) = ω k 1 ∫ − ∞ t d t ′ sin [ ω k ( t − t ′ )] J k ( t ′ ) . (190)
For a source that vanishes after a finite time, extending the upper limit to + ∞ +\infty + ∞ gives the exact outgoing response after that time,
U o u t ( t , z ) = ∫ d k 2 π e i k z 2 i ω k [ e i ω k t J ~ ( − ω k , k ) − e − i ω k t J ~ ( ω k , k ) ] . (191) \mathcal{U}_{\mathrm{out}}(t,z)=\int\frac{dk}{2\pi}\frac{e^{ikz}}{2i\omega_k}\left[e^{i\omega_k t}\widetilde{J}(-\omega_k,k)-e^{-i\omega_k t}\widetilde{J}(\omega_k,k)\right]. \tag*{(191)} U out ( t , z ) = ∫ 2 π d k 2 i ω k e ik z [ e i ω k t J ( − ω k , k ) − e − i ω k t J ( ω k , k ) ] . (191)
For a source with decaying temporal tails, this expression approximates the response up to the contribution of the remaining tail. The outgoing excitation is determined by the source transform on the massive dispersion relation ω 2 = k 2 + ω b 2 \omega^2=k^2+\omega_b^2 ω 2 = k 2 + ω b 2 .
For the source in Eq. (183) , we obtain
J ~ ( ω , k ) = 1 2 F ~ ′ ′ ( ω − k 2 ) G ~ ′ ′ ( ω + k 2 ) , (192) \widetilde{J}(\omega,k)=\frac{1}{2}\widetilde{F}^{\prime\prime}\left(\frac{\omega-k}{2}\right)\widetilde{G}^{\prime\prime}\left(\frac{\omega+k}{2}\right), \tag*{(192)} J ( ω , k ) = 2 1 F ′′ ( 2 ω − k ) G ′′ ( 2 ω + k ) , (192)
where F ~ ′ ′ ( p ) = ∫ d s e i p s F ′ ′ ( s ) \widetilde{F}^{\prime\prime}(p)=\int ds\,e^{ips}F^{\prime\prime}(s) F ′′ ( p ) = ∫ d s e i p s F ′′ ( s ) and similarly for G G G .
Gaussian source and late time response
For the profiles in Eq. (174) ,
F ′ ′ ( s ) = A L 2 ( s 2 L 2 − 1 ) e − s 2 / ( 2 L 2 ) . (193) F^{\prime\prime}(s)=\frac{A}{L^2}\left(\frac{s^2}{L^2}-1\right)e^{-s^2/(2L^2)}. \tag*{(193)} F ′′ ( s ) = L 2 A ( L 2 s 2 − 1 ) e − s 2 / ( 2 L 2 ) . (193)
The common source is therefore
J ( t , z ) = A 2 L 4 [ ( t 2 − z 2 ) 2 L 4 − 2 ( t 2 + z 2 ) L 2 + 1 ] e − ( t 2 + z 2 ) / L 2 . (194) J(t,z)=\frac{A^2}{L^4}\left[\frac{(t^2-z^2)^2}{L^4}-\frac{2(t^2+z^2)}{L^2}+1\right]e^{-(t^2+z^2)/L^2}. \tag*{(194)} J ( t , z ) = L 4 A 2 [ L 4 ( t 2 − z 2 ) 2 − L 2 2 ( t 2 + z 2 ) + 1 ] e − ( t 2 + z 2 ) / L 2 . (194)
Since
F ~ ′ ′ ( p ) = − 2 π A L p 2 e − L 2 p 2 / 2 , (195) \widetilde{F}''(p)=-\sqrt{2\pi}\,ALp^2e^{-L^2p^2/2}, \tag*{(195)} F ′′ ( p ) = − 2 π A L p 2 e − L 2 p 2 /2 , (195)
Eq. (C37) gives
J ~ ( ω , k ) = π A 2 L 2 16 ( ω 2 − k 2 ) 2 e − L 2 ( ω 2 + k 2 ) / 4 . (196) \widetilde{J}(\omega,k)=\frac{\pi A^2L^2}{16}(\omega^2-k^2)^2e^{-L^2(\omega^2+k^2)/4}. \tag*{(196)} J ( ω , k ) = 16 π A 2 L 2 ( ω 2 − k 2 ) 2 e − L 2 ( ω 2 + k 2 ) /4 . (196)
Evaluating this expression on the massive dispersion relation gives
J ~ ( ω k , k ) = π A 2 L 2 ω b 4 16 e − ω b 2 L 2 / 4 e − L 2 k 2 / 2 . (197) \widetilde{J}(\omega_k,k)=\frac{\pi A^2L^2\omega_b^4}{16}e^{-\omega_b^2L^2/4}e^{-L^2k^2/2}. \tag*{(197)} J ( ω k , k ) = 16 π A 2 L 2 ω b 4 e − ω b 2 L 2 /4 e − L 2 k 2 /2 . (197)
Substituting Eq. (C42) into Eq. (C36) yields
U o u t ( t , z ) ≃ π A 2 L 2 ω b 4 16 e − ω b 2 L 2 / 4 × ∫ d k 2 π e − L 2 k 2 / 2 cos ( k z ) sin ( ω k t ) ω k . (198) \begin{aligned}
U_{\mathrm{out}}(t,z)&\simeq\frac{\pi A^2L^2\omega_b^4}{16}e^{-\omega_b^2L^2/4}\\
&\quad\times\int\frac{dk}{2\pi}e^{-L^2k^2/2}\cos(kz)\frac{\sin(\omega_k t)}{\omega_k}. \tag*{(198)}
\end{aligned} U out ( t , z ) ≃ 16 π A 2 L 2 ω b 4 e − ω b 2 L 2 /4 × ∫ 2 π d k e − L 2 k 2 /2 cos ( k z ) ω k sin ( ω k t ) . (198)
At this stage, the approximation consists of neglecting the remaining Gaussian forcing. The integral retains the full massive dispersion relation and can be evaluated before taking a large time limit.
At the collision center, z = 0 z=0 z = 0 , the phase has a stationary point at k = 0 k=0 k = 0 . Expanding
ω k = ω b + k 2 2 ω b + O ( k 4 ) (199) \omega_k=\omega_b+\frac{k^2}{2\omega_b}+\mathcal{O}(k^4) \tag*{(199)} ω k = ω b + 2 ω b k 2 + O ( k 4 ) (199)
and using
∫ − ∞ ∞ d k 2 π e i t k 2 / ( 2 ω b ) = ω b 2 π t e i π / 4 ( t > 0 ) (200) \int_{-\infty}^{\infty}\frac{dk}{2\pi}e^{itk^2/(2\omega_b)}
=\sqrt{\frac{\omega_b}{2\pi t}}e^{i\pi/4}\qquad(t>0) \tag*{(200)} ∫ − ∞ ∞ 2 π d k e i t k 2 / ( 2 ω b ) = 2 π t ω b e iπ /4 ( t > 0 ) (200)
gives
U ( t , 0 ) ≃ 2 π 32 A 2 L 2 ω b 7 / 2 e − ω b 2 L 2 / 4 sin ( ω b t + π / 4 ) t . (201) U(t,0)\simeq\frac{\sqrt{2\pi}}{32}A^2L^2\omega_b^{7/2}e^{-\omega_b^2L^2/4}\frac{\sin(\omega_b t+\pi/4)}{\sqrt{t}}. \tag*{(201)} U ( t , 0 ) ≃ 32 2 π A 2 L 2 ω b 7/2 e − ω b 2 L 2 /4 t sin ( ω b t + π /4 ) . (201)
This asymptotic form requires
t ≫ max { L , ω b L 2 , ω b − 1 } , (202) t\gg\max\{L,\omega_bL^2,\omega_b^{-1}\}, \tag*{(202)} t ≫ max { L , ω b L 2 , ω b − 1 } , (202)
while nonlinear decay and backreaction must remain perturbatively small over the comparison interval. The factor t − 1 / 2 t^{-1/2} t − 1/2 describes dispersive spreading along the string. The neglected source tail and the stationary-phase corrections are separate from the amplitude expansion, whose leading internal response is O ( ϵ 2 ) \mathcal{O}(\epsilon^2) O ( ϵ 2 ) at fixed L L L .
Details of the numerical simulations and convergence testing
In order to solve for the field theory dynamics in flat space, we implement the equations of motion obtained from Eq. (1) in GRChombo [42 ] , as described in [38 , 62 ] . This is a code by the GRTL Collaboration that incorporates Adaptive Mesh Refinement (AMR) [63 ] to accurately resolve the different scales in the problem: the string core and the length of the string4 . We extract the values of the scalar and gauge fields and compute the shape and vector modes by projecting these values onto the bound modes of a straight BPS string along the z z z -axis. We follow the procedure of Ref. [17 ] , but instead apply it separately on each slice z = c o n s t z=\mathrm{const} z = const , rather than to the whole volume. We use normalized modes
2 π ∫ ρ d ρ ( φ 2 + a 2 ) = 2 π ∫ ρ d ρ ψ 2 = 1 (203) 2\pi\int\rho\,d\rho\,(\varphi^2+a^2)=2\pi\int\rho\,d\rho\,\psi^2=1 \tag*{(203)} 2 π ∫ ρ d ρ ( φ 2 + a 2 ) = 2 π ∫ ρ d ρ ψ 2 = 1 (203)
with ρ 2 = x 2 + y 2 \rho^2=x^2+y^2 ρ 2 = x 2 + y 2 , so that the expressions for the shape and vector modes are
χ ( t , z ) = ∫ 0 2 π d θ ∫ 0 ∞ ρ d ρ ( δ ∣ φ ∣ φ + δ A θ a ) , C ( t , z ) = ∫ 0 2 π d θ ∫ 0 ∞ ρ d ρ A z ψ . (204) \begin{aligned}
\chi(t,z)&=\int_0^{2\pi}d\theta\int_0^\infty\rho\,d\rho\,(\delta|\varphi|\varphi+\delta A_\theta a),\\
C(t,z)&=\int_0^{2\pi}d\theta\int_0^\infty\rho\,d\rho A_z\psi. \tag*{(204)}
\end{aligned} χ ( t , z ) C ( t , z ) = ∫ 0 2 π d θ ∫ 0 ∞ ρ d ρ ( δ ∣ φ ∣ φ + δ A θ a ) , = ∫ 0 2 π d θ ∫ 0 ∞ ρ d ρ A z ψ . (204)
where δ ∣ φ ∣ ≡ ∣ φ ∣ − f ( ρ ) \delta|\varphi|\equiv|\varphi|-f(\rho) δ ∣ φ ∣ ≡ ∣ φ ∣ − f ( ρ ) and δ A θ ≡ A θ − α ( ρ ) / ρ \delta A_\theta\equiv A_\theta-\alpha(\rho)/\rho δ A θ ≡ A θ − α ( ρ ) / ρ . In practice, on the AMR grid the discretised string is not exactly the continuum BPS solution we start from, and even with no perturbation, (204) induces some small, grid-dependent χ \chi χ . For very precise measurements of the shape mode, this noise can be problematic as it propagates from the AMR refinement boundaries. But it can be removed by subtracting, slice by slice and time by time, the χ ( t , z ) \chi(t,z) χ ( t , z ) of a run of the unperturbed string on the same grid. This is not necessary for the vector mode, as the background has A z = 0 A_z=0 A z = 0 .
In addition, note that for the initial traveling waves, the projection onto the shape mode leads to an artificial χ a r t \chi_{\mathrm{art}} χ art where the waves are located: a displaced core projects onto the shape mode even if it is not deformed. This artificial mode propagates along the string with the traveling waves at the speed of light, as shown in the top left panel of Figure 7 . We can remove it by subtracting χ ( t , z ) → χ ( t , z ) − χ a r t ( t , z ) \chi(t,z)\to\chi(t,z)-\chi_{\mathrm{art}}(t,z) χ ( t , z ) → χ ( t , z ) − χ art ( t , z ) at each slice z z z . Here χ a r t \chi_{\mathrm{art}} χ art is the projection of an undeformed string whose core is displaced by d ( t , z ) \mathbf{d}(t,z) d ( t , z ) , so that it is given by Eq. (D2) with ∣ φ ∣ ( t , x − d ) |\varphi|(t,\mathbf{x}-\mathbf{d}) ∣ φ ∣ ( t , x − d ) and A θ ( t , x − d ) A_\theta(t,\mathbf{x}-\mathbf{d}) A θ ( t , x − d ) . The displacement of the string d ( t , z ) \mathbf{d}(t,z) d ( t , z ) can be measured by finding the zeros of the scalar field ϕ \phi ϕ at each time t t t and slice z z z .
Figure 7. Cleaning of the artificial shape mode of the traveling waves. Top: the shape mode in spacetime as projected (left), and after subtracting χ a r t \chi_{\rm art} χ art at each slice (right) due to the displacement of the core. Bottom: the shape mode at z = 0 z=0 z = 0 as projected (blue) and cleaned (black). Time is measured from the collision, t ′ = t − t c t'=t-t_{\rm c} t ′ = t − t c , and the modes are rescaled by t ′ \sqrt{t'} t ′ for t ′ > 1 t'>1 t ′ > 1 .
We compute numerically the time delay caused by the non-linear interaction of the traveling waves by tracking the position of one of them along the string, and compare it with the Nambu–Goto prediction Eq. (55) . On each slice z z z , we obtain the transverse displacement of the string d = ( x c , y c ) \mathbf{d}=(x_{\rm c},y_{\rm c}) d = ( x c , y c ) . At each time, we fit y c y_{\rm c} y c to a Gaussian ∼ exp [ − ( z − z w ) 2 ] \sim\exp[-(z-z_{\rm w})^2] ∼ exp [ − ( z − z w ) 2 ] , and take its centre z w z_{\rm w} z w as the position of the wave. The time delay is the lag of z w z_{\rm w} z w behind a light ray leaving the wave’s initial position. For the wave traveling towards positive z z z the time delay is then
Δ t ( t ) = t − ∣ z w ( t ) − z w ( 0 ) ∣ (205) \Delta t(t)=t-\left|z_{\rm w}(t)-z_{\rm w}(0)\right| \tag*{(205)} Δ t ( t ) = t − ∣ z w ( t ) − z w ( 0 ) ∣ (205)
The delay vanishes while the wave travels freely, and settles to a constant once the waves have separated, see the bottom panel of Figure 5 . While the two waves overlap, the fit does not give any meaningful position, so we do not interpret the time delay in this interval.
To validate our numerical simulations and results, we perform a convergence test of the excitation of the vector during the collision of traveling waves with ( A , L , z w ) = ( 0.35 , 4 , 17 ) (A,L,z_{\rm w})=(0.35,4,17) ( A , L , z w ) = ( 0.35 , 4 , 17 ) and δ = 45 ∘ \delta=45^\circ δ = 4 5 ∘ . We choose three different resolutions: ( N L R , N M R , N H R ) = ( 32 , 64 , 128 ) (N_{\rm LR},N_{\rm MR},N_{\rm HR})=(32,64,128) ( N LR , N MR , N HR ) = ( 32 , 64 , 128 ) number of points per side in the coarsest level, with three levels of refinement. We use a cubic box of side 128 2 128\sqrt{2} 128 2 , which leads to the finest grid spacings ( Δ x L R , Δ x M R , Δ x H R ) ≈ ( 0.71 , 0.35 , 0.18 ) (\Delta x_{\rm LR},\Delta x_{\rm MR},\Delta x_{\rm HR})\approx(0.71,0.35,0.18) ( Δ x LR , Δ x MR , Δ x HR ) ≈ ( 0.71 , 0.35 , 0.18 ) . In Figure 8 we compare the errors between resolutions, which show the results converging at 4th order.
Figure 8. Convergence test of the excitation of the vector mode C ( t ) C(t) C ( t ) after the collision of two traveling waves of A = 0.35 A=0.35 A = 0.35 , L = 4 L=4 L = 4 and δ = 45 ∘ \delta=45^\circ δ = 4 5 ∘ . Low, medium and high resolution correspond to N = 32 N=32 N = 32 , 64, 128 number of coarse grid points per side. We add three levels of refinement, and our results converge at 4th order.
References