20.07.2013 Views

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 ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

4.5 ADM Hamiltonian formulation 65<br />

field on Σ0 <strong>and</strong> p a vector field on Σ0, which obeys the constraint equations (4.65)-(4.66):<br />

R + K 2 − KijK ij = 16πE (4.93)<br />

DjK j<br />

i − DiK = 8πpi, (4.94)<br />

does there exist a spacetime (M,g,T) such that (g,T) fulfills the Einstein equation <strong>and</strong> Σ0 can<br />

be embedded as an hypersurface <strong>of</strong> M with induced metric γ <strong>and</strong> extrinsic curvature K ?<br />

Darmois (1927) [105] <strong>and</strong> Lichnerowicz (1939) [176] have shown that the answer is yes for<br />

the vacuum case (E = 0 <strong>and</strong> pi = 0), when the initial data (γ,K) are analytical functions<br />

<strong>of</strong> the coordinates (x i ) on Σ0. Their analysis is based on the Cauchy-Kovalevskaya theorem<br />

mentioned in Sec. 4.4.2 (cf. Chap. 10 <strong>of</strong> Wald’s textbook [265] for details). However, on physical<br />

grounds, the analytical case is too restricted. One would like to deal instead with smooth (i.e.<br />

differentiable) initial data. There are at least two reasons for this:<br />

• The smooth manifold structure <strong>of</strong> M imposes only that the change <strong>of</strong> coordinates are<br />

differentiable, not necessarily analytical. Consequently if (γ,K) are analytical functions<br />

<strong>of</strong> the coordinates, they might not be analytical functions <strong>of</strong> another coordinate system<br />

(x ′i ).<br />

• An analytical function is fully determined by its value <strong>and</strong> those <strong>of</strong> all its derivatives at a<br />

single point. Equivalently an analytical function is fully determined by its value in some<br />

small open domain D. This fits badly with causality requirements, because a small change<br />

to the initial data, localized in a small region, should not change the whole solution at all<br />

points <strong>of</strong> M. The change should take place only in the so-called domain <strong>of</strong> dependence <strong>of</strong><br />

D.<br />

This is why the major breakthrough in the Cauchy problem <strong>of</strong> general <strong>relativity</strong> has been<br />

achieved by Choquet-Bruhat in 1952 [127] when she showed existence <strong>and</strong> uniqueness <strong>of</strong> the<br />

solution in a small neighbourhood <strong>of</strong> Σ0 for smooth (at least C 5 ) initial data (γ,K). We shall<br />

not give any sketch on the pro<strong>of</strong> (beside the original publication [127], see the review articles<br />

[39] <strong>and</strong> [88]) but simply mentioned that it is based on harmonic coordinates.<br />

A major improvement has been then the global existence <strong>and</strong> uniqueness theorem by Choquet-<br />

Bruhat <strong>and</strong> Geroch (1969) [87]. The latter tells that among all the spacetimes (M,g) solution<br />

<strong>of</strong> the Einstein equation <strong>and</strong> such that (Σ0,γ,K) is an embedded Cauchy surface, there exists<br />

a maximal spacetime (M ∗ ,g ∗ ) <strong>and</strong> it is unique. Maximal means that any spacetime (M,g) solution<br />

<strong>of</strong> the Cauchy problem is isometric to a subpart <strong>of</strong> (M ∗ ,g ∗ ). For more details about the<br />

existence <strong>and</strong> uniqueness <strong>of</strong> solutions to the Cauchy problem, see the reviews by Choquet-Bruhat<br />

<strong>and</strong> York [88], Klainerman <strong>and</strong> Nicolò [169], Andersson [15] <strong>and</strong> Rendall [212].<br />

4.5 ADM Hamiltonian formulation<br />

Further insight in the <strong>3+1</strong> Einstein equations is provided by the Hamiltonian formulation <strong>of</strong><br />

general <strong>relativity</strong>. Indeed the latter makes use <strong>of</strong> the <strong>3+1</strong> <strong>formalism</strong>, since any Hamiltonian<br />

approach involves the concept <strong>of</strong> a physical state “at a certain time”, which is translated in<br />

general <strong>relativity</strong> by the state on a spacelike hypersurface Σt. The Hamiltonian formulation

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!