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 ...
24 Geometry of hypersurfaces The eigenvalues of the Weingarten map, which are all real numbers since χ is self-adjoint, are called the principal curvatures of the hypersurface Σ and the corresponding eigenvectors define the so-called principal directions of Σ. The mean curvature of the hypersurface Σ is the arithmetic mean of the principal curvature: where the κi are the three eigenvalues of χ. H := 1 3 (κ1 + κ2 + κ3) (2.41) Remark : The curvatures defined above are not to be confused with the Gaussian curvature introduced in Sec. 2.3.3. The latter is an intrinsic quantity, independent of the way the manifold (Σ,γ) is embedded in (M,g). On the contrary the principal curvatures and mean curvature depend on the embedding. For this reason, they are qualified of extrinsic. The self-adjointness of χ implies that the bilinear form defined on Σ’s tangent space by K : Tp(Σ) × Tp(Σ) −→ R (u,v) ↦−→ −u · χ(v) (2.42) is symmetric. It is called the second fundamental form of the hypersurface Σ. It is also called the extrinsic curvature tensor of Σ (cf. the remark above regarding the qualifier ’extrinsic’). K contains the same information as the Weingarten map. Remark : The minus sign in the definition (2.42) is chosen so that K agrees with the convention used in the numerical relativity community, as well as in the MTW book [189]. Some other authors (e.g. Carroll [79], Poisson [205], Wald [265]) choose the opposite convention. If we make explicit the value of χ in the definition (2.42), we get [see Eq. (2.7)] ∀(u,v) ∈ Tp(Σ) × Tp(Σ), K(u,v) = −u · ∇vn . (2.43) We shall denote by K the trace of the bilinear form K with respect to the metric γ; it is the opposite of the trace of the endomorphism χ and is equal to −3 times the mean curvature of Σ: K := γ ij Kij = −3H. (2.44) 2.3.5 Examples: surfaces embedded in the Euclidean space R 3 Let us illustrate the previous definitions with some hypersurfaces of a space which we are very familiar with, namely R 3 endowed with the standard Euclidean metric. In this case, the dimension is reduced by one unit with respect to the spacetime M and the ambient metric g is Riemannian (signature (+,+,+)) instead of Lorentzian. The hypersurfaces are 2-dimensional submanifolds of R 3 , namely they are surfaces by the ordinary meaning of this word. In this section, and in this section only, we change our index convention to take into account that the base manifold is of dimension 3 and not 4: until the next section, the Greek indices run in {1,2,3} and the Latin indices run in {1,2}.
2.3 Hypersurface embedded in spacetime 25 Figure 2.2: Plane Σ as a hypersurface of the Euclidean space R 3 . Notice that the unit normal vector n stays constant along Σ; this implies that the extrinsic curvature of Σ vanishes identically. Besides, the sum of angles of any triangle lying in Σ is α + β + γ = π, which shows that the intrinsic curvature of (Σ, γ) vanishes as well. Example 1 : a plane in R 3 Let us take for Σ the simplest surface one may think of: a plane (cf. Fig. 2.2). Let us consider Cartesian coordinates (X α ) = (x,y,z) on R 3 , such that Σ is the z = 0 plane. The scalar function t defining Σ according to Eq. (2.21) is then simply t = z. (x i ) = (x,y) constitutes a coordinate system on Σ and the metric γ induced by g on Σ has the components γij = diag(1,1) with respect to these coordinates. It is obvious that this metric is flat: Riem(γ) = 0. The unit normal n has components n α = (0,0,1) with respect to the coordinates (X α ). The components of the gradient ∇n being simply given by the partial derivatives ∇βn α = ∂n α /∂X β [the Christoffel symbols vanishes for the coordinates (X α )], we get immediately ∇n = 0. Consequently, the Weingarten map and the extrinsic curvature vanish identically: χ = 0 and K = 0. Example 2 : a cylinder in R 3 Let us at present consider for Σ the cylinder defined by the equation t := ρ − R = 0, where ρ := x 2 + y 2 and R is a positive constant — the radius of the cylinder (cf Fig. 2.3). Let us introduce the cylindrical coordinates (x α ) = (ρ,ϕ,z), such that ϕ ∈ [0,2π), x = r cos ϕ and y = r sinϕ. Then (x i ) = (ϕ,z) constitutes a coordinate system on Σ. The components of the induced metric in this coordinate system are given by γij dx i dx j = R 2 dϕ 2 + dz 2 . (2.45) It appears that this metric is flat, as for the plane considered above. Indeed, the change of coordinate η := R ϕ (remember R is a constant !) transforms the metric components into γi ′ j ′ dxi′ dx j′ = dη 2 + dz 2 , (2.46)
- Page 1: arXiv:gr-qc/0703035v1 6 Mar 2007 3+
- Page 4 and 5: 4 CONTENTS 3.3.6 Evolution of the o
- Page 6 and 7: 6 CONTENTS 8 The initial data probl
- Page 8 and 9: 8 CONTENTS
- Page 10 and 11: 10 CONTENTS
- Page 12 and 13: 12 Introduction 3D (no symmetry at
- Page 14 and 15: 14 Introduction
- Page 16 and 17: 16 Geometry of hypersurfaces in two
- Page 18 and 19: 18 Geometry of hypersurfaces 2.2.3
- Page 20 and 21: 20 Geometry of hypersurfaces Figure
- Page 22 and 23: 22 Geometry of hypersurfaces •
- Page 26 and 27: 26 Geometry of hypersurfaces Figure
- Page 28 and 29: 28 Geometry of hypersurfaces The no
- Page 30 and 31: 30 Geometry of hypersurfaces Since
- Page 32 and 33: 32 Geometry of hypersurfaces 2.4.3
- Page 34 and 35: 34 Geometry of hypersurfaces 2.5 Ga
- Page 36 and 37: 36 Geometry of hypersurfaces Exampl
- Page 38 and 39: 38 Geometry of hypersurfaces
- Page 40 and 41: 40 Geometry of foliations Figure 3.
- Page 42 and 43: 42 Geometry of foliations 3.3.2 Nor
- Page 44 and 45: 44 Geometry of foliations means Eq.
- Page 46 and 47: 46 Geometry of foliations Remark :
- Page 48 and 49: 48 Geometry of foliations Note that
- Page 50 and 51: 50 Geometry of foliations
- Page 52 and 53: 52 3+1 decomposition of Einstein eq
- Page 54 and 55: 54 3+1 decomposition of Einstein eq
- Page 56 and 57: 56 3+1 decomposition of Einstein eq
- Page 58 and 59: 58 3+1 decomposition of Einstein eq
- Page 60 and 61: 60 3+1 decomposition of Einstein eq
- Page 62 and 63: 62 3+1 decomposition of Einstein eq
- Page 64 and 65: 64 3+1 decomposition of Einstein eq
- Page 66 and 67: 66 3+1 decomposition of Einstein eq
- Page 68 and 69: 68 3+1 decomposition of Einstein eq
- Page 70 and 71: 70 3+1 decomposition of Einstein eq
- Page 72 and 73: 72 3+1 equations for matter and ele
24 Geometry <strong>of</strong> hypersurfaces<br />
The eigenvalues <strong>of</strong> the Weingarten map, which are all real numbers since χ is self-adjoint,<br />
are called the principal curvatures <strong>of</strong> the hypersurface Σ <strong>and</strong> the corresponding eigenvectors<br />
define the so-called principal directions <strong>of</strong> Σ. The mean curvature <strong>of</strong> the hypersurface Σ<br />
is the arithmetic mean <strong>of</strong> the principal curvature:<br />
where the κi are the three eigenvalues <strong>of</strong> χ.<br />
H := 1<br />
3 (κ1 + κ2 + κ3) (2.41)<br />
Remark : The curvatures defined above are not to be confused with the Gaussian curvature<br />
introduced in Sec. 2.3.3. The latter is an intrinsic quantity, independent <strong>of</strong> the way the<br />
manifold (Σ,γ) is embedded in (M,g). On the contrary the principal curvatures <strong>and</strong> mean<br />
curvature depend on the embedding. For this reason, they are qualified <strong>of</strong> extrinsic.<br />
The self-adjointness <strong>of</strong> χ implies that the bilinear form defined on Σ’s tangent space by<br />
K : Tp(Σ) × Tp(Σ) −→ R<br />
(u,v) ↦−→ −u · χ(v)<br />
(2.42)<br />
is symmetric. It is called the second fundamental form <strong>of</strong> the hypersurface Σ. It is also called<br />
the extrinsic curvature tensor <strong>of</strong> Σ (cf. the remark above regarding the qualifier ’extrinsic’).<br />
K contains the same information as the Weingarten map.<br />
Remark : The minus sign in the definition (2.42) is chosen so that K agrees with the convention<br />
used in the <strong>numerical</strong> <strong>relativity</strong> community, as well as in the MTW book [189].<br />
Some other authors (e.g. Carroll [79], Poisson [205], Wald [265]) choose the opposite<br />
convention.<br />
If we make explicit the value <strong>of</strong> χ in the definition (2.42), we get [see Eq. (2.7)]<br />
∀(u,v) ∈ Tp(Σ) × Tp(Σ), K(u,v) = −u · ∇vn . (2.43)<br />
We shall denote by K the trace <strong>of</strong> the bilinear form K with respect to the metric γ; it is the<br />
opposite <strong>of</strong> the trace <strong>of</strong> the endomorphism χ <strong>and</strong> is equal to −3 times the mean curvature <strong>of</strong> Σ:<br />
K := γ ij Kij = −3H. (2.44)<br />
2.3.5 Examples: surfaces embedded in the Euclidean space R 3<br />
Let us illustrate the previous definitions with some hypersurfaces <strong>of</strong> a space which we are very<br />
familiar with, namely R 3 endowed with the st<strong>and</strong>ard Euclidean metric. In this case, the dimension<br />
is reduced by one unit with respect to the spacetime M <strong>and</strong> the ambient metric g is<br />
Riemannian (signature (+,+,+)) instead <strong>of</strong> Lorentzian. The hypersurfaces are 2-dimensional<br />
submanifolds <strong>of</strong> R 3 , namely they are surfaces by the ordinary meaning <strong>of</strong> this word.<br />
In this section, <strong>and</strong> in this section only, we change our index convention to take into account<br />
that the base manifold is <strong>of</strong> dimension 3 <strong>and</strong> not 4: until the next section, the Greek indices run<br />
in {1,2,3} <strong>and</strong> the Latin indices run in {1,2}.