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 ...
126 The initial data problem Notice that Eqs. (8.1)-(8.2) involve a single hypersurface Σ0, not a foliation (Σt) t∈R . In particular, neither the lapse function nor the shift vector appear in these equations. Facing them, a naive way to proceed would be to choose freely the metric γ, thereby fixing the connection D and the scalar curvature R, and to solve Eqs. (8.1)-(8.2) for K. Indeed, for fixed γ, E, and p, Eqs. (8.1)-(8.2) form a quasi-linear system of first order for the components Kij. However, as discussed by Choquet-Bruhat [128], this approach is not satisfactory because we have only four equations for six unknowns Kij and there is no natural prescription for choosing arbitrarily two among the six components Kij. Lichnerowicz (1944) [177] has shown that a much more satisfactory split of the initial data (γ,K) between freely choosable parts and parts obtained by solving Eqs. (8.1)-(8.2) is provided by the conformal decomposition introduced in Chap. 6. Lichnerowicz method has been extended by Choquet-Bruhat (1956, 1971) [128, 86], by York and Ó Murchadha (1972, 1974, 1979) [273, 274, 196, 276] and more recently by York and Pfeiffer (1999, 2003) [278, 202]. Actually, conformal decompositions are by far the most widely spread techniques to get initial data for the 3+1 Cauchy problem. Alternative methods exist, such as the quasi-spherical ansatz introduced by Bartnik in 1993 [37] or a procedure developed by Corvino (2000) [98] and by Isenberg, Mazzeo and Pollack (2002) [162] for gluing together known solutions of the constraints, thereby producing new ones. Here we shall limit ourselves to the conformal methods. Standard reviews on this subject are the articles by York (1979) [276] and Choquet-Bruhat and York (1980) [88]. Recent reviews are the articles by Cook (2000) [94], Pfeiffer (2004) [201] and Bartnik and Isenberg (2004) [39]. 8.1.2 Conformal decomposition of the constraints The conformal form of the constraint equations has been derived in Chap. 6. We have introduced there the conformal metric ˜γ and the conformal factor Ψ such that the metric γ induced by the spacetime metric on some hypersurface Σ0 is [cf. Eq. (6.22)] γij = Ψ 4 ˜γij, (8.3) and have decomposed the traceless part A ij of the extrinsic curvature K ij according to [cf. Eq. (6.82)] A ij = Ψ −10 Â ij . (8.4) We consider here the decomposition involving Âij [α = −10 in Eq. (6.58)] and not the alternative one, which uses Ãij (α = −4), because we have seen in Sec. 6.4.2 that the former is well adapted to the momentum constraint. Using the decompositions (8.3) and (8.4), we have rewritten the Hamiltonian constraint (8.1) and the momentum constraint (8.2) as respectively the Lichnerowicz equation [Eq. (6.111)] and an equation involving the divergence of Âij with respect to the conformal metric [Eq. (6.112)] : ˜Di ˜ D i Ψ − 1 8 ˜ RΨ + 1 8 Âij Âij Ψ −7 + 2π ˜ EΨ −3 − 1 12 K2 Ψ 5 = 0 , (8.5) ˜Dj Âij − 2 3 Ψ6 ˜ D i K = 8π˜p i , (8.6)
8.2 Conformal transverse-traceless method 127 where we have introduce the rescaled matter quantities and ˜E := Ψ 8 E (8.7) ˜p i := Ψ 10 p i . (8.8) The definition of ˜p i is clearly motivated by Eq. (6.112). On the contrary the power 8 in the definition of ˜ E is not the only possible choice. As we shall see in § 8.2.4, it is chosen (i) to guarantee a negative power of Ψ in the ˜ E term in Eq. (8.5), resulting in some uniqueness property of the solution and (ii) to allow for an easy implementation of the dominant energy condition. 8.2 Conformal transverse-traceless method 8.2.1 Longitudinal/transverse decomposition of Âij In order to solve the system (8.5)-(8.6), York (1973,1979) [274, 275, 276] has decomposed Âij into a longitudinal part and a transverse one, by setting  ij = ( ˜ LX) ij + Âij TT , (8.9) where Âij TT is both traceless and transverse (i.e. divergence-free) with respect to the metric ˜γ: ˜γij Âij TT = 0 and Dj ˜ Âij TT = 0, (8.10) and ( ˜ LX) ij is the conformal Killing operator associated with the metric ˜γ and acting on the vector field X: ( ˜ LX) ij := ˜ D i X j + ˜ D j X i − 2 3 ˜ DkX k ˜γ ij . (8.11) The properties of this linear differential operator are detailed in Appendix B. Let us retain here that ( ˜ LX) ij is by construction traceless: ˜γij( ˜ LX) ij = 0 (8.12) (it must be so because in Eq. (8.9) both Âij and Âij TT are traceless) and the kernel of ˜L is made of the conformal Killing vectors of the metric ˜γ, i.e. the generators of the conformal isometries (cf. Sec. B.1.3). The symmetric tensor ( ˜ LX) ij is called the longitudinal part of Âij , whereas Âij TT is called the transverse part. Given Âij , the vector X is determined by taking the divergence of Eq. (8.9): taking into account property (8.10), we get ˜Dj( ˜ LX) ij = ˜ Dj Âij . (8.13) The second order operator ˜ Dj( ˜ LX) ij acting on the vector X is the conformal vector Laplacian ˜∆L: ˜∆L X i := ˜ Dj( ˜ LX) ij = ˜ Dj ˜ D j X i + 1 3 ˜ D i ˜ DjX j + ˜ R i j Xj , (8.14)
- Page 76 and 77: 76 3+1 equations for matter and ele
- Page 78 and 79: 78 3+1 equations for matter and ele
- Page 80 and 81: 80 3+1 equations for matter and ele
- Page 82 and 83: 82 3+1 equations for matter and ele
- Page 84 and 85: 84 Conformal decomposition equivale
- Page 86 and 87: 86 Conformal decomposition As an ex
- Page 88 and 89: 88 Conformal decomposition 6.2.4 Co
- Page 90 and 91: 90 Conformal decomposition 6.3.1 Ge
- Page 92 and 93: 92 Conformal decomposition where K
- Page 94 and 95: 94 Conformal decomposition to write
- Page 96 and 97: 96 Conformal decomposition hence Lm
- Page 98 and 99: 98 Conformal decomposition 6.5.2 Ha
- Page 100 and 101: 100 Conformal decomposition discuss
- Page 102 and 103: 102 Conformal decomposition Remark
- Page 104 and 105: 104 Asymptotic flatness and global
- Page 106 and 107: 106 Asymptotic flatness and global
- Page 108 and 109: 108 Asymptotic flatness and global
- Page 110 and 111: 110 Asymptotic flatness and global
- Page 112 and 113: 112 Asymptotic flatness and global
- Page 114 and 115: 114 Asymptotic flatness and global
- Page 116 and 117: 116 Asymptotic flatness and global
- Page 118 and 119: 118 Asymptotic flatness and global
- Page 120 and 121: 120 Asymptotic flatness and global
- Page 122 and 123: 122 Asymptotic flatness and global
- Page 124 and 125: 124 Asymptotic flatness and global
- Page 128 and 129: 128 The initial data problem where
- Page 130 and 131: 130 The initial data problem 8.2.3
- Page 132 and 133: 132 The initial data problem where
- Page 134 and 135: 134 The initial data problem Figure
- Page 136 and 137: 136 The initial data problem Figure
- Page 138 and 139: 138 The initial data problem In par
- Page 140 and 141: 140 The initial data problem Accord
- Page 142 and 143: 142 The initial data problem Remark
- Page 144 and 145: 144 The initial data problem Remark
- Page 146 and 147: 146 The initial data problem 8.4.1
- Page 148 and 149: 148 The initial data problem Since
- Page 150 and 151: 150 The initial data problem • fo
- Page 152 and 153: 152 Choice of foliation and spatial
- Page 154 and 155: 154 Choice of foliation and spatial
- Page 156 and 157: 156 Choice of foliation and spatial
- Page 158 and 159: 158 Choice of foliation and spatial
- Page 160 and 161: 160 Choice of foliation and spatial
- Page 162 and 163: 162 Choice of foliation and spatial
- Page 164 and 165: 164 Choice of foliation and spatial
- Page 166 and 167: 166 Choice of foliation and spatial
- Page 168 and 169: 168 Choice of foliation and spatial
- Page 170 and 171: 170 Choice of foliation and spatial
- Page 172 and 173: 172 Choice of foliation and spatial
- Page 174 and 175: 174 Choice of foliation and spatial
126 The initial data problem<br />
Notice that Eqs. (8.1)-(8.2) involve a single hypersurface Σ0, not a foliation (Σt) t∈R . In particular,<br />
neither the lapse function nor the shift vector appear in these equations. Facing them, a<br />
naive way to proceed would be to choose freely the metric γ, thereby fixing the connection D<br />
<strong>and</strong> the scalar curvature R, <strong>and</strong> to solve Eqs. (8.1)-(8.2) for K. Indeed, for fixed γ, E, <strong>and</strong> p,<br />
Eqs. (8.1)-(8.2) form a quasi-linear system <strong>of</strong> first order for the components Kij. However, as<br />
discussed by Choquet-Bruhat [128], this approach is not satisfactory because we have only four<br />
equations for six unknowns Kij <strong>and</strong> there is no natural prescription for choosing arbitrarily two<br />
among the six components Kij.<br />
Lichnerowicz (1944) [177] has shown that a much more satisfactory split <strong>of</strong> the initial data<br />
(γ,K) between freely choosable parts <strong>and</strong> parts obtained by solving Eqs. (8.1)-(8.2) is provided<br />
by the conformal decomposition introduced in Chap. 6. Lichnerowicz method has been extended<br />
by Choquet-Bruhat (1956, 1971) [128, 86], by York <strong>and</strong> Ó Murchadha (1972, 1974, 1979) [273,<br />
274, 196, 276] <strong>and</strong> more recently by York <strong>and</strong> Pfeiffer (1999, 2003) [278, 202]. Actually, conformal<br />
decompositions are by far the most widely spread techniques to get initial data for the <strong>3+1</strong><br />
Cauchy problem. Alternative methods exist, such as the quasi-spherical ansatz introduced by<br />
Bartnik in 1993 [37] or a procedure developed by Corvino (2000) [98] <strong>and</strong> by Isenberg, Mazzeo<br />
<strong>and</strong> Pollack (2002) [162] for gluing together known solutions <strong>of</strong> the constraints, thereby producing<br />
new ones. Here we shall limit ourselves to the conformal methods. St<strong>and</strong>ard reviews on this<br />
subject are the articles by York (1979) [276] <strong>and</strong> Choquet-Bruhat <strong>and</strong> York (1980) [88]. Recent<br />
reviews are the articles by Cook (2000) [94], Pfeiffer (2004) [201] <strong>and</strong> Bartnik <strong>and</strong> Isenberg<br />
(2004) [39].<br />
8.1.2 Conformal decomposition <strong>of</strong> the constraints<br />
The conformal form <strong>of</strong> the constraint equations has been derived in Chap. 6. We have introduced<br />
there the conformal metric ˜γ <strong>and</strong> the conformal factor Ψ such that the metric γ induced by the<br />
spacetime metric on some hypersurface Σ0 is [cf. Eq. (6.22)]<br />
γij = Ψ 4 ˜γij, (8.3)<br />
<strong>and</strong> have decomposed the traceless part A ij <strong>of</strong> the extrinsic curvature K ij according to [cf.<br />
Eq. (6.82)]<br />
A ij = Ψ −10 Â ij . (8.4)<br />
We consider here the decomposition involving Âij [α = −10 in Eq. (6.58)] <strong>and</strong> not the alternative<br />
one, which uses Ãij (α = −4), because we have seen in Sec. 6.4.2 that the former is<br />
well adapted to the momentum constraint. Using the decompositions (8.3) <strong>and</strong> (8.4), we have<br />
rewritten the Hamiltonian constraint (8.1) <strong>and</strong> the momentum constraint (8.2) as respectively<br />
the Lichnerowicz equation [Eq. (6.111)] <strong>and</strong> an equation involving the divergence <strong>of</strong> Âij with<br />
respect to the conformal metric [Eq. (6.112)] :<br />
˜Di ˜ D i Ψ − 1<br />
8 ˜ RΨ + 1<br />
8 Âij Âij Ψ −7 + 2π ˜ EΨ −3 − 1<br />
12 K2 Ψ 5 = 0 , (8.5)<br />
˜Dj Âij − 2<br />
3 Ψ6 ˜ D i K = 8π˜p i , (8.6)