Wave-Klein–Gordon Coupled Systems

• Wave–Klein–Gordon coupled systems are nonlinear PDEs combining a massless wave component and a massive Klein–Gordon field with distinct dispersive rates and resonance structures.
• They exhibit varied coupling forms—such as null forms, divergence structures, and derivative-free interactions—that are key in models like the Einstein–Klein–Gordon equations.
• Analytical frameworks including hyperboloidal foliation, ghost weights, and high-order energy estimates are employed to address global well-posedness, modified scattering, and asymptotic behavior.

Wave–Klein–Gordon coupled systems are nonlinear PDE systems in which a massless wave field and a massive Klein–Gordon field interact through semilinear or quasilinear terms. In the model problems studied in recent work, the wave component satisfies a wave equation of the form $\Box u=\cdots$, while the massive component satisfies $(\Box+1)v=\cdots$, so the two unknowns have different dispersive behavior, different characteristic geometry, and different resonance structure. This class includes simplified models for the Einstein–Klein–Gordon equations in harmonic coordinates, where the wave variable represents a metric perturbation and the Klein–Gordon variable represents a massive scalar field (Ionescu et al., 2017), as well as the wave-coordinate reduction of the full Einstein–Klein–Gordon system itself (Ionescu et al., 2019).

1. PDE architecture and coupling mechanisms

A representative $3+1$-dimensional model is

$$\Box u = A^{\alpha\beta}\partial_\alpha v\,\partial_\beta v + D\,v^2,\qquad (\Box+1)v = u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v + E\,u\,v,$$

with real constant coefficients and symmetric tensors $A^{\alpha\beta},B^{\alpha\beta}$ (Ionescu et al., 2017). In this model, the wave equation is driven by quadratic Klein–Gordon terms, while the Klein–Gordon equation feels the wave field through a variable-coefficient second-order term $u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v$, which is the most delicate term in the analysis.

In two space dimensions, strong quasilinear coupling appears in systems such as

$$\left\{ \begin{aligned} &(\partial_t^2-\Delta_x)u=\mathbf N_1(v,\partial v)+\mathbf N_2(u,\partial v),\ &(\partial_t^2-\Delta_x+1)v=\mathbf N_1(v,\partial u)+\mathbf N_2(u,\partial u), \end{aligned} \right.$$

where the nonlinearities are linear combinations of the classical quadratic null forms $Q_{ij}$, $Q_{0i}$, and $Q_0$ (Ifrim et al., 2019). In this setting the coupling is described as strong because the same nonlinear mechanisms appear with derivatives on the unknowns in both equations, so the interaction is not merely semilinear.

The literature distinguishes several structurally important subclasses. One is the null-form class, where every quadratic interaction is written in terms of $(\Box+1)v=\cdots$0 or $(\Box+1)v=\cdots$1; this is central in low dimensions because wave–wave resonances and near-resonant wave–Klein–Gordon interactions must be canceled algebraically (Dong et al., 2022). Another is the divergence-form class, for example

$(\Box+1)v=\cdots$2

or

$(\Box+1)v=\cdots$3

where the wave forcing can be represented through shifted primitives (Duan et al., 2020). A further class contains derivative-free wave factors, such as

$(\Box+1)v=\cdots$4

where the difficulty is precisely that there are no derivatives on the massless wave component on the right-hand side (Dong et al., 2018).

These examples show that the phrase “wave–Klein–Gordon coupled system” does not designate a single normal form. It designates a family of multiple-speed systems whose analytic difficulty depends on whether the coupling is semilinear or quasilinear, whether it carries null structure, whether it is in divergence form, and whether the most slowly decaying wave factor appears with or without derivatives.

2. Dimensional regimes and well-posedness theory

The contrast between two and three spatial dimensions is decisive. In $(\Box+1)v=\cdots$5 dimensions, the wave component has the sharp $(\Box+1)v=\cdots$6 decay rate and the Klein–Gordon component has the faster $(\Box+1)v=\cdots$7 decay rate after the appropriate renormalization (Ionescu et al., 2017). This decay gap makes global small-data theory feasible for several semilinear and quasilinear models. The three-dimensional global regularity theorem of Ionescu and Pausader proves global existence and modified scattering for small smooth decaying data for the model above, with high-order energy control and scattering of the wave part in a weak norm (Ionescu et al., 2017). Other $(\Box+1)v=\cdots$8-dimensional results treat systems with non-compactly supported data and no null condition, while still proving global nonlinear stability, sharp pointwise decay, and linear scattering by combining exterior flat-slice estimates with interior hyperboloidal estimates (Zhang, 2023).

In two space dimensions the situation is much more delicate because decay is slower. For semilinear null systems, global existence and optimal decay have been proved for non-compactly supported data. One result treats

$(\Box+1)v=\cdots$9

and obtains global smooth solutions together with the optimal pointwise decay

$3+1$0

for sufficiently small weighted Sobolev data (Dong et al., 2022). Another null-form result in $3+1$1 dimensions proves global existence without compact support by applying Alinhac’s ghost weight method to both the wave and the Klein–Gordon equations (Dong, 2020).

The two-dimensional strong-coupling regime requires additional structure. For a class of quasilinear strongly coupled systems with null-form quadratic terms, almost global well-posedness has been proved for small localized data with only mild decay at infinity and minimal regularity, with lifespan

$3+1$2

and dyadic energy growth $3+1$3 on $3+1$4 (Ifrim et al., 2019). The same work states that, after adding asymptotic analysis in a second paper, this control is upgraded to full global well-posedness with $3+1$5 (Ifrim et al., 2019).

Other $3+1$6-dimensional results isolate special borderline structures. One paper proves small-data global existence when the wave equation carries a critical semilinear source and the Klein–Gordon equation carries a below-critical semilinear source; the same paper emphasizes that swapping the nonlinearities would cause finite-time blow-up in the wave equation (Dong et al., 2020). Another treats a prototype strong-coupling model and develops a method to exchange one type of decay for another in order to close the bootstrap (Ma, 2020). A related divergence-form theory proves global well-posedness for strong quadratic couplings in the wave equation and applies directly to the $3+1$7-dimensional Klein–Gordon–Zakharov system (Duan et al., 2020).

This body of work makes clear that global theory is not determined by the presence of a wave equation and a Klein–Gordon equation alone. In practice it is determined by the interaction between dimension, decay, resonance, and the exact algebraic form of the quadratic coupling.

3. Structural mechanisms and analytical frameworks

A major organizing principle is the hyperboloidal foliation method. Foundational works in $3+1$8 dimensions develop energy estimates on hyperboloids

$3+1$9

and use the hyperboloidal tangent fields $$\Box u = A^{\alpha\beta}\partial_\alpha v\,\partial_\beta v + D\,v^2,\qquad (\Box+1)v = u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v + E\,u\,v,$$0 to treat wave and Klein–Gordon equations in a common framework (Ma, 2011, Ma, 2011). These works emphasize two advantages: no need for the scaling vector field $$\Box u = A^{\alpha\beta}\partial_\alpha v\,\partial_\beta v + D\,v^2,\qquad (\Box+1)v = u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v + E\,u\,v,$$1, which is problematic for Klein–Gordon equations, and no need for delicate $$\Box u = A^{\alpha\beta}\partial_\alpha v\,\partial_\beta v + D\,v^2,\qquad (\Box+1)v = u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v + E\,u\,v,$$2-$$\Box u = A^{\alpha\beta}\partial_\alpha v\,\partial_\beta v + D\,v^2,\qquad (\Box+1)v = u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v + E\,u\,v,$$3 estimates, since hyperboloidal Sobolev inequalities convert energy bounds directly into decay.

In low dimensions, vector-field arguments are frequently combined with ghost weights and localized spacetime decompositions. For the two-dimensional quasilinear strong-coupling problem, the proof is built from a corrected quasilinear energy, dyadic localization relative to the light cone, and an Alinhac-type ghost weight $$\Box u = A^{\alpha\beta}\partial_\alpha v\,\partial_\beta v + D\,v^2,\qquad (\Box+1)v = u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v + E\,u\,v,$$4 that yields coercive spacetime control of tangential derivatives $$\Box u = A^{\alpha\beta}\partial_\alpha v\,\partial_\beta v + D\,v^2,\qquad (\Box+1)v = u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v + E\,u\,v,$$5 (Ifrim et al., 2019). The same analysis uses the key null-form identity

$$\Box u = A^{\alpha\beta}\partial_\alpha v\,\partial_\beta v + D\,v^2,\qquad (\Box+1)v = u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v + E\,u\,v,$$6

which is the main algebraic mechanism behind improved cone estimates (Ifrim et al., 2019).

A different mechanism appears in divergence-form systems. There the wave unknown is rewritten through shifted primitives: if $$\Box u = A^{\alpha\beta}\partial_\alpha v\,\partial_\beta v + D\,v^2,\qquad (\Box+1)v = u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v + E\,u\,v,$$7 solves $$\Box u = A^{\alpha\beta}\partial_\alpha v\,\partial_\beta v + D\,v^2,\qquad (\Box+1)v = u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v + E\,u\,v,$$8, then one sets $$\Box u = A^{\alpha\beta}\partial_\alpha v\,\partial_\beta v + D\,v^2,\qquad (\Box+1)v = u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v + E\,u\,v,$$9 or $A^{\alpha\beta},B^{\alpha\beta}$0, so the coupling entering the Klein–Gordon equation becomes Hessian-type and decays better on hyperboloids (Duan et al., 2020). This is a structural reduction rather than a mere estimate.

For quasilinear $A^{\alpha\beta},B^{\alpha\beta}$1 systems with mildly decaying data, another line of work combines para-differential symmetrization, quasi-linear normal forms, a second energy-level normal-form correction, and semi-classical microlocal analysis (Stingo, 2018). The purpose of the para-differential step is to symmetrize the principal operator and prevent derivative loss, while the microlocal step recovers the optimal $A^{\alpha\beta},B^{\alpha\beta}$2 behavior by reducing the Klein–Gordon component to an ODE near its characteristic manifold and the wave component to a transport equation.

Three-dimensional Fourier-analytic theory uses a different language but addresses the same underlying issues. In the Ionescu–Pausader framework, the proof combines high-order energies, weighted profile estimates, and $A^{\alpha\beta},B^{\alpha\beta}$3-norms built from a space-frequency decomposition $A^{\alpha\beta},B^{\alpha\beta}$4, so that different frequency regimes and the geometry of space-time resonances can be treated separately (Ionescu et al., 2017). This suggests a broad methodological division within the subject: hyperboloidal and ghost-weight methods are especially natural for geometric and localized analyses, whereas profile and resonance methods are especially natural when the asymptotic mechanism itself must be identified precisely.

4. Resonance, asymptotics, and scattering

A recurring correction to linear intuition is that the Klein–Gordon component does not, in general, scatter linearly. In the three-dimensional model of Ionescu and Pausader, the wave field scatters linearly in a weak norm, but the Klein–Gordon profile acquires a real phase correction

$A^{\alpha\beta},B^{\alpha\beta}$5

and only after this renormalization does the Klein–Gordon component converge to a scattering state (Ionescu et al., 2017). The source of this long-range effect is the low-frequency wave field inside the Klein–Gordon characteristic region.

Later work develops this picture further. The modified wave operator theory shows that the large-time dynamics are dictated by a small set of resonant interactions: a low-frequency wave bulk is generated by Klein–Gordon self-interaction, and that wave bulk feeds back into the Klein–Gordon equation as a slowly varying nonlinear phase (Ouyang, 2020). In that setting, asymptotic completeness is formulated as a modified scattering statement: every sufficiently small admissible asymptotic pair $A^{\alpha\beta},B^{\alpha\beta}$6 is realized by a unique global solution (Ouyang, 2020).

Hyperboloidal asymptotics reveal the same mechanism in physical space. In $A^{\alpha\beta},B^{\alpha\beta}$7 dimensions, one writes

$A^{\alpha\beta},B^{\alpha\beta}$8

and derives an interior asymptotic form for the Klein–Gordon field,

$A^{\alpha\beta},B^{\alpha\beta}$9

together with an interior wave profile $u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v$0 and a radiation field at null infinity (Chen et al., 2023). The same work also treats the inverse problem of scattering from infinity: given asymptotic Klein–Gordon profiles $u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v$1 and a free radiation field $u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v$2 satisfying the compatibility condition dictated by the interior profile, one constructs an exact global solution with those asymptotics (Chen et al., 2023).

In two dimensions, asymptotic questions are tied more tightly to null structure. For semilinear null systems, the main achievement is global existence together with the optimal linear decay rates for both components (Dong et al., 2022). For the quasilinear strong-coupling problem, the almost-global result is explicitly presented as the robust local-to-almost-global mechanism, while the conversion to full global control is delegated to a separate asymptotic analysis (Ifrim et al., 2019).

5. Geometric origin and related physical models

The Einstein–Klein–Gordon system is a central geometric source of wave–Klein–Gordon models. In wave coordinates, the covariant system reduces to a quasilinear wave system for the metric components coupled to a quasilinear Klein–Gordon equation for the massive scalar field (Ionescu et al., 2019). In schematic form one obtains

$u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v$3

so the geometry becomes a multiple-speed quasilinear wave–Klein–Gordon system (Ionescu et al., 2019).

The simplified flat-background model studied by Wang and LeFloch-Ma and then by Ionescu and Pausader isolates the dominant nonlinear mechanism of the Einstein–Klein–Gordon equations in harmonic gauge: the wave variable is sourced by $u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v$4 and $u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v$5, while the Klein–Gordon variable is affected by a quasilinear wave coefficient $u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v$6 (Ionescu et al., 2017). The full Einstein–Klein–Gordon monograph proves global regularity, precise asymptotics, modified scattering profiles, weak peeling estimates, ADM and Bondi energy statements, and geodesic asymptotics for small localized perturbations of Minkowski space (Ionescu et al., 2019).

A different physical regime appears in the low-field, non-relativistic limit of the Einstein–Klein–Gordon system. In spherical symmetry, the equations reduce to the Poisson–Schrödinger system and admit spherically symmetric static states, which are used in the wave dark matter interpretation of the scalar field (Goetz, 2015). This is not a dispersive small-data theory in Minkowski space, but it shows that the same coupled wave–Klein–Gordon framework also supports static and approximately static structures in other asymptotic regimes.

These geometric and physical connections clarify why the subject is broader than an isolated dispersive model problem. The same coupling mechanisms arise in stability theory for Minkowski space, in asymptotic scattering theory, and in low-field static-state analyses.

6. Rigidity, damping, and other specialized phenomena

Beyond well-posedness and scattering, recent work studies channel-specific rigidity and sign-induced stabilization. One rigidity theorem considers

$u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v$7

in $u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v$8 for small compactly supported data and proves that if the Friedlander radiation field of the wave component vanishes at null infinity, then $u\,B^{\alpha\beta}\partial_\alpha\partial_\beta v$9, hence the wave component vanishes identically in spacetime (Li et al., 2024). The same paper stresses that this does not kill the Klein–Gordon field: if $$\left\{ \begin{aligned} &(\partial_t^2-\Delta_x)u=\mathbf N_1(v,\partial v)+\mathbf N_2(u,\partial v),\ &(\partial_t^2-\Delta_x+1)v=\mathbf N_1(v,\partial u)+\mathbf N_2(u,\partial u), \end{aligned} \right.$$0, the system reduces to a free Klein–Gordon equation (Li et al., 2024). A common misconception is therefore ruled out: vanishing wave radiation forces triviality only in the wave channel, not in the full coupled system.

Another recent direction isolates a nonlinear damping structure. For the $$\left\{ \begin{aligned} &(\partial_t^2-\Delta_x)u=\mathbf N_1(v,\partial v)+\mathbf N_2(u,\partial v),\ &(\partial_t^2-\Delta_x+1)v=\mathbf N_1(v,\partial u)+\mathbf N_2(u,\partial u), \end{aligned} \right.$$1-dimensional system

$$\left\{ \begin{aligned} &(\partial_t^2-\Delta_x)u=\mathbf N_1(v,\partial v)+\mathbf N_2(u,\partial v),\ &(\partial_t^2-\Delta_x+1)v=\mathbf N_1(v,\partial u)+\mathbf N_2(u,\partial u), \end{aligned} \right.$$2

global existence is proved under the sign condition

$$\left\{ \begin{aligned} &(\partial_t^2-\Delta_x)u=\mathbf N_1(v,\partial v)+\mathbf N_2(u,\partial v),\ &(\partial_t^2-\Delta_x+1)v=\mathbf N_1(v,\partial u)+\mathbf N_2(u,\partial u), \end{aligned} \right.$$3

where $$\left\{ \begin{aligned} &(\partial_t^2-\Delta_x)u=\mathbf N_1(v,\partial v)+\mathbf N_2(u,\partial v),\ &(\partial_t^2-\Delta_x+1)v=\mathbf N_1(v,\partial u)+\mathbf N_2(u,\partial u), \end{aligned} \right.$$4 is the semi-hyperboloidal $$\left\{ \begin{aligned} &(\partial_t^2-\Delta_x)u=\mathbf N_1(v,\partial v)+\mathbf N_2(u,\partial v),\ &(\partial_t^2-\Delta_x+1)v=\mathbf N_1(v,\partial u)+\mathbf N_2(u,\partial u), \end{aligned} \right.$$5-coefficient (Ma et al., 16 Jul 2025). The proof rewrites the wave variable through

$$\left\{ \begin{aligned} &(\partial_t^2-\Delta_x)u=\mathbf N_1(v,\partial v)+\mathbf N_2(u,\partial v),\ &(\partial_t^2-\Delta_x+1)v=\mathbf N_1(v,\partial u)+\mathbf N_2(u,\partial u), \end{aligned} \right.$$6

so that the $$\left\{ \begin{aligned} &(\partial_t^2-\Delta_x)u=\mathbf N_1(v,\partial v)+\mathbf N_2(u,\partial v),\ &(\partial_t^2-\Delta_x+1)v=\mathbf N_1(v,\partial u)+\mathbf N_2(u,\partial u), \end{aligned} \right.$$7-term becomes a barrier contribution and the first term in $$\left\{ \begin{aligned} &(\partial_t^2-\Delta_x)u=\mathbf N_1(v,\partial v)+\mathbf N_2(u,\partial v),\ &(\partial_t^2-\Delta_x+1)v=\mathbf N_1(v,\partial u)+\mathbf N_2(u,\partial u), \end{aligned} \right.$$8 is nonpositive (Ma et al., 16 Jul 2025). This gives a precise example in which the coefficients of the wave source induce a damping effect in the coupled Klein–Gordon dynamics.

Taken together, these results indicate that the modern theory of wave–Klein–Gordon systems is organized around a small set of recurrent themes: multiple propagation speeds, low-frequency wave effects, null or weak-null cancellations, quasilinear derivative loss, and long-range asymptotics. This suggests that the subject is best understood not as a single theorem family but as a collection of structurally distinct regimes unified by the interaction between a massless wave component and a massive Klein–Gordon component.