3+1 formalism and bases of numerical relativity - LUTh ...
3+1 formalism and bases of numerical relativity - LUTh ...
3+1 formalism and bases of numerical relativity - LUTh ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
9.3 Evolution <strong>of</strong> spatial coordinates 165<br />
<strong>and</strong> QTT ij is both traceless <strong>and</strong> transverse (i.e. divergence-free) with respect to the metric γ:<br />
DjQTT ij = 0. X is then related to the divergence <strong>of</strong> Q by Dj (LX)ij = DjQij. It is legitimate to<br />
relate the TT part to the dynamics <strong>of</strong> the gravitational field <strong>and</strong> to attribute the longitudinal<br />
part to the change in γij which arises because <strong>of</strong> the variation <strong>of</strong> coordinates from Σt to Σt+δt.<br />
This longitudinal part has 3 degrees <strong>of</strong> freedom (the 3 components <strong>of</strong> the vector X) <strong>and</strong> we<br />
might set it to zero by some judicious choice <strong>of</strong> the coordinates (xi ). The minimal distortion<br />
coordinates are thus defined by the requirement X = 0 or<br />
i.e.<br />
Qij = Q TT<br />
ij , (9.51)<br />
D j Qij = 0 . (9.52)<br />
Let us now express Q in terms <strong>of</strong> the shift vector to turn the above condition into an equation<br />
for the evolution <strong>of</strong> spatial coordinates. By means <strong>of</strong> Eqs. (4.63) <strong>and</strong> (9.9), Eq. (9.45) becomes<br />
Qij = −2NKijLβ γij + − 1<br />
<br />
−2NK + 2Dkβ<br />
3<br />
k<br />
γij, (9.53)<br />
i.e. (since Lβ γij = Diβj + Djβi)<br />
Qij = −2NAij + (Lβ)ij, (9.54)<br />
where we let appear the trace-free part A <strong>of</strong> the extrinsic curvature K [Eq. (6.53)]. If we insert<br />
this expression into the minimal distortion requirement (9.52), we get<br />
− 2NDjA ij − 2A ij DjN + Dj(Lβ) ij = 0. (9.55)<br />
Let then use the momentum constraint (4.66) to express the divergence <strong>of</strong> A as<br />
DjA ij = 8πp i + 2<br />
3 Di K. (9.56)<br />
Besides, we recognize in Dj(Lβ) ij the conformal vector Laplacian associated with the metric γ,<br />
so that we can write [cf. Eq. (B.11)]<br />
Dj(Lβ) ij = DjD j β i + 1<br />
3 Di Djβ j + R i j βj , (9.57)<br />
where R is the Ricci tensor associated with γ. Thus we arrive at<br />
DjD j β i + 1<br />
3 Di Djβ j + R i j βj = 16πNp i + 4<br />
3 NDi K + 2A ij DjN . (9.58)<br />
This is the elliptic equation on the shift vector that one has to solve in order to enforce the<br />
minimal distortion.<br />
Remark : For a constant mean curvature (CMC) slicing, <strong>and</strong> in particular for a maximal<br />
slicing, the term D i K vanishes <strong>and</strong> the above equation is slightly simplified. Incidentally,<br />
this is the form originally derived by Smarr <strong>and</strong> York (Eq. (3.27) in Ref. [246]).