### 4.8 The binary black hole inner boundary

It is clear that the three-dimensional inspiral and coalescence of black holes challenges the limits of
present computational know-how. CCM offers a new approach for excising an interior trapped region
which might provide the enhanced flexibility required to solve this problem. In a binary system,
there are major computational advantages in posing the Cauchy evolution in a frame which is
co-rotating with the orbiting black holes. Such a description seems necessary in order to keep the
numerical grid from being intrinsically twisted. In this co-orbiting description, the Cauchy evolution
requires an inner boundary condition inside the black holes and also an outer boundary condition
on a worldtube outside of which the grid rotation is likely to be superluminal. An outgoing
characteristic code can routinely handle such superluminal gauge flows in the exterior [31]. Thus,
successful implementation of CCM would solve the exterior boundary problem for this co-orbiting
description.
CCM also has the potential to handle the two black holes inside the Cauchy region. As described earlier
with respect to Figure 6, an ingoing characteristic code can evolve a moving black hole with long term
stability [105, 102]. This means that CCM might also be able to provide the inner boundary
condition for Cauchy evolution once stable matching has been accomplished. In this approach,
the interior boundary of the Cauchy evolution is located outside the apparent horizon and
matched to a characteristic evolution based upon ingoing null cones. The inner boundary for the
characteristic evolution is a trapped or marginally trapped surface, whose interior is excised from the
evolution.

In addition to restricting the Cauchy evolution to the region outside the black holes, this strategy offers
several other advantages. Although finding a marginally trapped surface on the ingoing null hypersurfaces
remains an elliptic problem, there is a natural radial coordinate system to facilitate its
solution. Motion of the black hole through the grid reduces to a one-dimensional radial problem,
leaving the angular grid intact and thus reducing the computational complexity of excising
the inner singular region. (The angular coordinates can even rotate relative to the Cauchy
coordinates in order to accommodate spinning black holes.) The chief danger in this approach
is that a caustic might be encountered on the ingoing null hypersurface before entering the
trapped region. This is a gauge problem whose solution lies in choosing the right location and
geometry of the surface across which the Cauchy and characteristic evolutions are matched. There
is a great deal of flexibility here because the characteristic initial data can be posed without
constraints.

This global strategy is tailor-made to treat two black holes in the co-orbiting gauge, as illustrated in
Figure 8. Two disjoint characteristic evolutions based upon ingoing null cones are matched
across worldtubes to a central Cauchy region. The interior boundaries of each of these interior
characteristic regions border a trapped surface. At the outer boundary of the Cauchy region, a matched
characteristic evolution based upon outgoing null hypersurfaces propagates the radiation to
infinity.

Present characteristic and Cauchy codes can handle the individual pieces of this problem. Their
unification appears to offer the best chance for simulating the inspiral and merger of two black holes. The
individual pieces of the fully nonlinear CCM module, as outlined in Section 4.2, have been implemented
and tested for accuracy. The one missing ingredient is long term stability in the nonlinear gravitational case,
which would make future applications very exciting.