Skip to content

ADM Evolution Equations

Scrollable formula. Use Left and Right arrow keys to reveal hidden content.

Scrollable table. Use Left and Right arrow keys to reveal additional columns.

Scrollable figure. Use Left and Right arrow keys to inspect hidden content.

ADM evolution advances the spatial metric γij\gamma_{ij} and extrinsic curvature KijK_{ij} once lapse and shift have been chosen. The lapse and shift enter through normal acceleration and spatial Lie transport; they change the coordinate representative, not the covariant Einstein solution. Equivalence to that solution additionally requires the Hamiltonian and momentum constraints on the initial data and their preservation during evolution.

Required background. Hypersurface Normals and Tangential Projectors supplies the projected derivative and extrinsic geometry, while The Einstein Equation supplies the covariant equations being rewritten.

Helpful background. The Gibbons-Hawking-York Boundary Term explains how the same spatial metric and extrinsic curvature enter a boundary-aware variational problem.

Fix throughout

(∂t)μ=αnμ+βμ,γμν=gμν+nμnν,Kij=−12Lnγij.(\partial_t)^\mu=\alpha n^\mu+\beta^\mu, \qquad \gamma_{\mu\nu}=g_{\mu\nu}+n_\mu n_\nu, \qquad K_{ij}=-\frac12\mathcal L_n\gamma_{ij}.

It is convenient to write

∂⊥≡∂t−Lβ.\partial_\perp\equiv\partial_t-\mathcal L_\beta.

The pair (γij,Kij)(\gamma_{ij},K_{ij}) contains twelve component functions. The definition of KijK_{ij} supplies six metric-evolution equations; the ten components of Einstein’s equation split into six extrinsic-curvature-evolution equations and four constraints. On initial data, the constraints reduce twelve functions to eight, and four diffeomorphism directions leave four phase-space functions—the two gravitational polarizations and their conjugate momenta. The lapse and shift choose how that gauge freedom is represented during evolution.

Einstein’s equation does not by itself prescribe the lapse or shift. Algebraic, elliptic, or hyperbolic gauge conditions may do so, but the resulting enlarged system must be analyzed separately for well-posedness. The equations below are the continuum ADM identities before discretization.

The first evolution equation is kinematic:

∂⊥γij=−2αKij,\partial_\perp\gamma_{ij}=-2\alpha K_{ij},

or in coordinate components,

∂tγij=−2αKij+Diβj+Djβi.\partial_t\gamma_{ij} =-2\alpha K_{ij}+D_i\beta_j+D_j\beta_i.

Normal geometry appears only in the first term. The remaining terms are the change induced by tangential relabeling. If the shift is a Killing field of γij\gamma_{ij}, it can be nonzero while contributing nothing to ∂tγij\partial_t\gamma_{ij}.

This equation also evolves the inverse metric:

∂⊥γij=2αKij,\partial_\perp\gamma^{ij}=2\alpha K^{ij},

which follows by differentiating γikγkj=δij\gamma^{ik}\gamma_{kj}=\delta^i{}_j. That sign is needed when tracing the KijK_{ij} evolution equation.

The spatial projection of the four-dimensional Ricci tensor combines the contracted Gauss relation with the two-normal Ricci relation. Solving it for the normal derivative of KijK_{ij} and using

Rμν=8πG(Tμν−12Tgμν)+ΛgμνR_{\mu\nu} =8\pi G\left(T_{\mu\nu}-\frac12Tg_{\mu\nu}\right) +\Lambda g_{\mu\nu}

gives

∂⊥Kij=−DiDjα+α((3)Rij+KKij−2KikKkj)−8πGα[Sij−12γij(S−ρ)]−αΛγij.\begin{aligned} \partial_\perp K_{ij} ={}&-D_iD_j\alpha +\alpha\left( {}^{(3)}R_{ij} +K K_{ij} -2K_{ik}K^k{}_j \right)\\ &-8\pi G\alpha \left[ S_{ij}-\frac12\gamma_{ij}(S-\rho) \right] -\alpha\Lambda\gamma_{ij}. \end{aligned}

The matter variables use the fixed signs

ρ=Tμνnμnν,ji=−γiμnνTμν,Sij=γiμγjνTμν,S=γijSij.\rho=T_{\mu\nu}n^\mu n^\nu, \qquad j_i=-\gamma_i{}^\mu n^\nu T_{\mu\nu}, \qquad S_{ij}=\gamma_i{}^\mu\gamma_j{}^\nu T_{\mu\nu}, \qquad S=\gamma^{ij}S_{ij}.

The momentum density jij_i does not appear explicitly in this spatial projection; it sources the momentum constraint. The Hessian −DiDjα-D_iD_j\alpha is the geometric effect of accelerating the normal congruence. The shift enters only through ∂⊥\partial_\perp, again separating gauge transport from normal dynamics.

Gourgoulhon derives these projections and the adapted-coordinate system in Gourgoulhon 2007, §4.1.3, pp. 53–54 and §§4.3–4.4.3, pp. 59–64, PDF.

The evolution system is accompanied by

H=(3)R+K2−KijKij−16πGρ−2Λ=0,\mathcal H ={}^{(3)}R+K^2-K_{ij}K^{ij} -16\pi G\rho-2\Lambda=0,

and

Mi=DjKji−DiK−8πGji=0.\mathcal M_i =D_jK^j{}_i-D_iK-8\pi G j_i=0.

The contracted Bianchi identity and ∇μTμν=0\nabla_\mu T^{\mu\nu}=0 imply a homogeneous subsidiary system for (H,Mi)(\mathcal H,\mathcal M_i) when the ADM evolution equations hold. Hence constraint-satisfying initial data remain on the constraint surface in the exact continuum problem, subject to suitable regularity and boundary compatibility.

This does not mean that every unconstrained solution of the twelve evolution equations is an Einstein spacetime. Off the constraint surface, one has an enlarged system. Nor does algebraic equivalence prove that a particular gauge choice gives a strongly hyperbolic Cauchy problem. Those principal-symbol questions belong to the next formulation layer.

In inertial Minkowski coordinates,

α=1,βi=0,γij=δij,Kij=0,Tμν=0,Λ=0.\alpha=1, \quad \beta^i=0, \quad \gamma_{ij}=\delta_{ij}, \quad K_{ij}=0, \quad T_{\mu\nu}=0, \quad \Lambda=0.

Every term in both evolution equations vanishes separately.

For Milne slices of the same spacetime,

γij=τ2qij,Kij=−1τγij,K=−3τ,(3)Rij=−2τ2γij.\gamma_{ij}=\tau^2q_{ij}, \qquad K_{ij}=-\frac1\tau\gamma_{ij}, \qquad K=-\frac3\tau, \qquad {}^{(3)}R_{ij}=-\frac2{\tau^2}\gamma_{ij}.

The metric equation gives

∂τγij=2τγij=−2Kij.\partial_\tau\gamma_{ij} =\frac2\tau\gamma_{ij} =-2K_{ij}.

For the second equation, α=1\alpha=1, βi=0\beta^i=0, and vacuum imply

(3)Rij+KKij−2KikKkj=(−2+3−2)1τ2γij=−1τ2γij.{}^{(3)}R_{ij}+K K_{ij}-2K_{ik}K^k{}_j =\left(-2+3-2\right)\frac1{\tau^2}\gamma_{ij} =-\frac1{\tau^2}\gamma_{ij}.

Direct differentiation gives the same result: ∂τKij=−γij/τ2\partial_\tau K_{ij}=-\gamma_{ij}/\tau^2. Thus nontrivial ADM evolution reconstructs a flat four-dimensional metric, confirming every quadratic sign.

Lapse and shift do not add physical solutions. They choose how the same Einstein geometry is sliced and labeled, provided the gauge construction stays regular.

Evolution equations alone are not Einstein’s equation. Initial and boundary data must satisfy the constraints, and the subsidiary system must not receive incoming constraint violations.

Algebraic equivalence is not well-posedness. Hyperbolicity depends on the chosen variables, reductions, constraint additions, and gauge equations.

Trace the KijK_{ij} evolution equation, taking account of the evolution of γij\gamma^{ij}, and use the Hamiltonian constraint to show that

∂⊥K=−DiDiα+α[KijKij+4πG(ρ+S)−Λ].\partial_\perp K =-D^iD_i\alpha +\alpha\left[ K_{ij}K^{ij}+4\pi G(\rho+S)-\Lambda \right].
Solution

Because K=γijKijK=\gamma^{ij}K_{ij},

∂⊥K=2αKijKij+γij∂⊥Kij.\partial_\perp K =2\alpha K^{ij}K_{ij} +\gamma^{ij}\partial_\perp K_{ij}.

Tracing the displayed evolution equation for KijK_{ij} gives

γij∂⊥Kij=−DiDiα+α((3)R+K2−2KijKij)+4πGα(S−3ρ)−3αΛ.\begin{aligned} \gamma^{ij}\partial_\perp K_{ij} ={}&-D^iD_i\alpha +\alpha({}^{(3)}R+K^2-2K_{ij}K^{ij})\\ &+4\pi G\alpha(S-3\rho)-3\alpha\Lambda. \end{aligned}

Adding the inverse-metric term cancels the explicit −2KijKij-2K_{ij}K^{ij}. The Hamiltonian constraint rearranges to

(3)R+K2=KijKij+16πGρ+2Λ.{}^{(3)}R+K^2 =K_{ij}K^{ij}+16\pi G\rho+2\Lambda.

Substitution combines the matter terms into 4πG(ρ+S)4\pi G(\rho+S) and the cosmological terms into −Λ-\Lambda, yielding the required trace equation.

  • Gourgoulhon, Éric. 3+1 Formalism and Bases of Numerical Relativity. Lecture notes, arXiv:gr-qc/0703035v1 (2007). arXiv.