Go to:

-Introduction
-Simulations
-Movies
-Presentation
-References

Relativistic Hydrodynamics
Gravitational Waveform Catalog

Laboratoire de l'Univers et de ses Théories

Observatiore de Paris

Departamento de Astronomía y Astrofísica

Universidad de Valencia

Departamento de Astronomía y Astrofísica

Universidad de Valencia

Numerical relativity is concerned with solving the Einstein equations, which are the field equations of general relativistic gravity, using computer simulations. In the 3+1 split of spacetime introduced by Arnowitt, Deser, and Misner (ADM) in 1960, the Einstein equations can be written in the form of the standard ADM metric equations. They consist of several coupled nonlinear partial differential equations, which divide into two classes: a set of time evolution equations, and a set of elliptic equations. In a dynamic situation, the evolution equations drive forward the time evolution of the spacetime.

The most common approach to numerically solve such equations is to use finite differences methods. These are particularly efficient in terms of computational speed for evolution equations. Therefore, many previous and current computer codes in numerical relativity solve only the evolution equations of the ADM metric system (free evolution method). However, a crucial tradeoff of this method is that after some evolution time the constraint equations, which are satisfied initially, are violated. These constraint violations can lead to very inaccurate results and will eventually lead to a crash of the computer code.

In the last years there have been many attempts find mathematically equivalent reformulations of the ADM metric equations which exhibit more robust numerical properties. Among these reformulations is the ADM-BSSN method (originally proposed by Nakamura, Oohara, and Kojima, and later refined by Baumgarte, Shapiro, and Shibata), which results in numerical evolutions which are stable for a longer time compared to the standard ADM free evolution method. An example for a 3D general relativistic hydrodynamics code in numerical relativity which utilizes the ADM-BSSN method is the Cactus code in combination with the Whisky thorn developed and used by the EU Research Training Network on Sources of Gravitational Waves. Still, experience with new simulations indicates that after some time constraint violations also occur and can grow without bound. This is a severe restriction for many astrophysical applications of numerical relativity like long-term simulations of merging black holes or black hole formation from stellar core collapse.

A completely different approach is to solve the constraint equations during the entire time evolution, which by definition prohibits constraint violations (constrained evolution). In that case however, the elliptic constraint equations have to be solved very often (i.e., at each time step). This is a severe obstacle in practical computations, as the numerical solution of elliptic equations is very time consuming even with fast modern supercomputers, if conventional finite difference methods are used.

But recently scientist have developed various alternative strategies for solving such equations efficiently. One such strategy is to use instead of finite difference methods a new class of methods, known as (pseudo-)spectral methods, where functions are not approximated by local low-order polynomials on a finite difference grid, but by global smooth functions in the form of a (small) number of coefficients. This approach is comparable to the Fourier expansion methods widely used in electrodynamics, and yields very accurate results if the desired solution is smooth (which is the case for the Einstein field equations). Spectral methods thus make it possible to use constrained evolution schemes in numerical relativity.

A very versatile and modern computational approach to using multi-domain spectral methods in numerical relativity is provided by the LORENE spectral methods toolbox developed by the numerical relativity group of the Laboratoire de l'Univers et de ses Théories at the Observatiore de Paris (for an introduction to spectral methods in the context of astrophysics, see [Bonazzola, 1999, Gourgoulhon, 2002]). It consist of a collection of C++ classes to solve various problems arising in numerical relativity, and more generally in computational astrophysics. Using spectral methods contained in this or other libraries, in the last years various research groups have performed numerical studies of astrophysical phenomena such as

- physical instabilities leading to gravitational collapse,
- gravitational collapse of a neutron star to a black hole,
- the infall phase of a triaxial stellar core in a core collapse supernova,
- the construction of equilibrium configurations of rapidly rotating neutron stars with magnetic fields,
- the tidal interaction of a star with a massive black hole,
- inertial modes of rotating stars,
- quasi-equilibrium configurations of co-rotating binary black holes, or
- the evolution of pure gravitational wave spacetimes.

However, one particular disadvantage of spectral methods is that they cannot accurately represent solutions which involve discontinuities such as shock fronts, which often occur in fluid dynamics. Particularly in the general multidimensional scenario of supernova core collapse, where a hydrodynamic shock drives the supernova explosion, spectral methods meet their limits. On the other hand, robust finite difference schemes to solve the fluid equations written in a hyperbolic form are known for a long time and have been employed successfully in computational fluid dynamics. In particular, the so-called high-resolution shock-capturing (HRSC) schemes have shown their advantages over other type of methods even when dealing with relativistic flows containing discontinuities. They both guarantee convergence to the physical solution of the investigated system, resolve discontinuities in the solution sharply and stably, and attain a high order of accuracy in smooth parts of the solution. Thus in the case of fluid equations, finite difference HRSC schemes are superior.

Therefore, in general relativistic numerical hydrodynamics, where both elliptic metric equations (with smooth solutions) and hyperbolic fluid equations (with possibly discontinous solutions) have to be solved simultaneously, it seems a promising strategy to use spectral methods for the former and HRSC schemes for the latter. In a computer code following this ansatz, this involves the usage of two different numerical grids (a spectral grid and a finite difference grid). Following early theoretical work on this at the Departamento de Astronomía y Astrofísica of the Universidad de Valencia in Spain and the Laboratoire de l'Univers et de ses Théories of the Observatiore de Paris in France, we call this approach "Mariage des Maillage", the French phrase for "grid wedding".

Based on previous computational studies of such a combination of the two methods in a work on spherically symmetric 1D core collapse in tensor-scalar gravity by [Novak and Ibáñez, 2000], or group was the first to successfully introduce the "Mariage des Maillage" concept in a 3D general relativistic hydrodynamics code [Dimmelmeier, et al., 2005]. To reduce the complexity of the fully general relativistic field equations for the spacetime metric, this code assumes conformal flatness for the three-metric (also known as conformal flatness condition (CFC) or Isenberg-Wilson-Mathews approximation [Wilson, et al., 1996]), as in our previous numerical simulations of general relativistic rotational core collapse [Dimmelmeier, et al., 2001, Dimmelmeier, et al., 2002a, Dimmelmeier, et al., 2002b].

Two recent approaches to approximate or rewrite the Einstein field equations as a system which prefers ellipticity of the equations by [Cerda, et al., 2005] (see also our previous work on general relativistic rotational core collapse in CFC+) and [Bonazzola, et al., 2004], respectively, open a new field of applications for implementing the "Mariage des Maillage" approach in numerical codes.

This work is dedicated to the memory of Jean-Alain Marck (1955-2000), who was one of the fathers and main contributors of the LORENE project.

In the following, we summarize some code tests and preliminary simulations, which demonstrate both the applicability of the "Mariage des Maillage" concept in a 3D general relativistic hydrodynamic code in general as well as the practical usability of that code for numerical simulations on current computer architectures. For more details, we refer to [Dimmelmeier, et al., 2005].

To ensure that the communication through the interface between the finite difference grid and the spectral grid works properly in both directions, we have performed a number of calculations which test the interpolation accuracy and efficiency. The two critical issues, on the one hand the accuracy and reduction of numerical noise in the interpolation from the finite difference grid to the spectral grid, and on the other hand the computational efficiency in the evaluation of a function represented on the spectral grid onto the finite difference grid, could be successfully resolved. We could also confirm the theoretical convergence orders of the spectral solver and the interpolation scheme in numerical calculations.

Next we have made a thorough comparison of the new spectral metric solver with the two old metric solvers, which are based on a finite difference grid, in the axisymmetric case. We could numerically reproduce the expected convergence properties. Additionally, we have tested the radial fall-off behavior of the metric components, where the compactified grid of the spectral solver yields superior results. Furthermore, in a test simulation of a dynamic core collapse of a stellar core to a neutron star, we have scrutinized the proper construction of radial domains of the spectral grid. At last we have compared the results from this core collapse simulation to results of the same model evolved using a fully general relativistic hydrodynamics code without the assumption of CFC [Shibata and Sekiguchi, 2003]. The excellent agreement between these two numerical approaches further proves the applicability of the "Mariage des Maillage" concept and also shows the very good quality of the CFC approximation in the context of supernova core collapse.

With the successful passing of the above tests, the code can be considered to be ready for applications which involve generic nonaxisymmetric configurations. We have first computed a nonaxisymmetric 3D test spacetime. Next we have checked the numerical stability of the code against perturbation when handling spherically or rotationally symmetric spacetimes in 3D.

Finally, as the most stringent and demanding test application of a strongly relativistic 3D matter distribution, we have evolved a rapidly rotating neutron star in equilibrium, to which a strong bar-like density perturbation is added. As noticeable in the figures below, which encode the density distribution in the equatorial plane, the matter from the initial bar perturbation is partially accreted onto the neutron star, while some of it is shed and evolves into spiral arms. With the completion of this test suite we have demonstrated that in the framework of the "Mariage des Maillage" approach, spectral method can be successful implemented in a finite difference HRSC hydrodynamic code in 3D.

(JavaScript must be enabled to display these figures.)

The dynamics of the uniformly rotating neutron star model with a strong bar perturbation can be viewed in a movie. Color coded is the density on an equatorial slice, overplotted by vectors encoding the velocity field.

The bar perturbation with a strength of 10% is initially corotating with the neutron star. As it is not in equilibrium, the central part of it is accreted onto the neutron star, while matter from the outer parts is shed into spiral arms.

Note that some matter leaves the computational grid during the evolution, as the outer grid boundary is located close to the spiral arms. However, a special treatment of the boundary conditions prevents that numerical artifacts arise from this.

Regular movie in AVI format (1.5 MB).

Slow motion movie in AVI format (4.3 MB).

A more detailed overview about the "Mariage des Maillage" project, the test calculations, and the background of the envisaged applications in astrophysical simulations can be found in a color slide presentation:

Presentation in PDF format (1.2 MB).

• Bonazzola, S., Gourgoulhon, E., and Marck, J.-A.,
  "Spectral methods in general relativistic astrophysics",
  J. Comput. Appl. Math., 109, 433, (1999),
  [Article in gr-qc].


• Bonazzola, S., Gourgoulhon, E., Grandclement, P., and Novak, J.,
  "A constrained scheme for Einstein equations based on Dirac gauge and spherical coordinates",
  Phys. Rev. D, 70, 104007, (2004),
  [Article in gr-qc].


• Cerdá-Durán, P., Faye, G., Dimmelmeier, H., Font, J.A., Ibáñez, J.M., Müller, E.,
  and Schäfer, G.,
  "CFC+: Improved dynamics and gravitational waveforms from relativistic core collapse simulations",
  Astron. Astrophys., 439, 1033-1055, (2005),
  [Article in astro-ph].


• Dimmelmeier, H., Font, J.A., and Müller, E.,
  "Gravitational waves from relativistic rotational core collapse",
  Astrophys. J. Lett., 560, L163-L166, (2001),
  [Article in astro-ph].


• Dimmelmeier, H., Font, J.A., and Müller, E.,
  "Relativistic simulations of rotational core collapse. I. Methods, initial models, and code tests",
  Astron. Astrophys., 388, 917-935, (2002),
  [Article in astro-ph].


• Dimmelmeier, H., Font, J.A., and Müller, E.,
  "Relativistic simulations of rotational core collapse. II. Collapse dynamics and gravitational radiation",
  Astron. Astrophys., 393, 523-542, (2002),
  [Article in astro-ph].


• Dimmelmeier, H., Novak, J., Font, J.A., Ibáñez, J.M., and Müller, E.,
  "`Mariage des Maillages': A new numerical approach for 3D relativistic core collapse simulations",
  Phys. Rev. D, 71, 064023-1-30, (2005),
  [Article in astro-ph].


• Gourgoulhon, E.,
  "Introduction to spectral methods",
  Talk at the 4th EU Network Meeting, Palma de Mallorca, (2002),
  [Article in PDF format].


• Novak, J. and Ibáñez, J.M.,
  "Gravitational waves from the collapse and bounce of a stellar core in tensor-scalar gravity",
  Astrophys. J., 533, 392-405, (2000),
  [Article in astro-ph].


• Shibata, M. and Sekiguchi, Y.,
  "Gravitational waves from axisymmetrically oscillating neutron stars in general relativistic simulations",
  Phys. Rev. D, 68, 104020, (2003),
  [Article in gr-qc].


• Wilson, J.R., Mathews, G.J., and Marronetti, P.,
  "Relativistic numerical model for close neutron-star binaries",
  Phys. Rev. D, 54, 1317-1331, (1996),
  [Article in gr-qc].