Conformal method for initial data

In three spatial dimensions write the physical metric as psi to the fourth power times a conformal metric. Split extrinsic curvature into its trace and a trace-free part. In the conformal transverse-traceless construction, a particular power of psi rescales the contravariant trace-free tensor, simplifying its divergence. Indices on conformal quantities are raised and lowered with the conformal metric, while physical indices use the physical metric; silently mixing these operations changes powers of psi.

Typical free choices include a conformal metric, a mean-curvature field, and a transverse-traceless seed, together with boundary and matter information. A longitudinal vector contribution is solved from the momentum equation when needed. These inputs describe a construction, not a set of independently measurable observables. Different choices can change radiation content and physical charges. Conformal flatness, for example, is a restriction on the data and should not be presented as exact for every rotating black-hole configuration.

For vacuum data with flat conformal metric and zero mean curvature, a divergence-free conformal trace-free tensor solves the momentum condition. The remaining Hamiltonian equation is nonlinear elliptic: the flat Laplacian of psi balances a positive squared-tensor coefficient multiplied by a negative power of psi. Its sign follows from the earlier Hamiltonian convention and the scalar-curvature transformation. Setting the trace-free tensor to zero reduces the equation to a harmonic one away from singular points.

Boundary conditions are part of this equation. Asymptotically flat data commonly require psi to approach one at the outer end, while punctures or inner boundaries introduce additional conditions. A finite artificial outer boundary approximates infinity and can bias the resulting masses. Existence or uniqueness depends on the data class and formulation; an iterative solver converging to one discrete root does not prove either a continuum theorem or that all physically relevant branches have been found.

A tiny finite grid can show how the Laplacian couples a point to its neighbors and how nonlinear source terms affect an elliptic residual. It cannot establish that an arbitrarily assigned source is the norm of a globally divergence-free tensor. The one-dimensional reduction below is therefore explicitly an algebraic toy inspired by the nonlinear equation. The later seven-point example checks a genuine three-dimensional Laplacian stencil but still supplies only local sampled data.

For a smooth solution on a uniform Cartesian grid, a centered second derivative has second-order truncation error. An algebraic residual can be driven far below that truncation error by solving the discrete equations accurately; doing so does not erase discretization error. Conversely, an insufficiently converged elliptic iteration can dominate a nominal grid-refinement study. Record both the solver tolerance and how the physical observables change with resolution and outer boundary location.