Abstract. We describe Cactus, a framework for building a variety of computing applications in science and engineering, including astrophysics, relativity and chemical engineering. We first motivate by example the need for such frameworks to support multi-platform, high performance applications across diverse communities. We then describe the design of the latest release of Cactus (Version 4.0) a complete rewrite of earlier versions, which enables highly modular, multi-language, parallel applications to be developed by single researchers and large collaborations alike. Making extensive use of abstractions, we detail how we are able to provide the latest advances in computational science, such as interchangeable parallel data distribution and high performance IO layers, while hiding most details of the underlying computational libraries from the application developer. We survey how Cactus 4.0 is being used by various application communities, and describe how it will also enable these applications to run on the computational Grids of the near future. Application Frameworks in Scientific ComputingVirtually all areas of science and engineering, as well as an increasing number of other fields, are turning to computational science to provide crucial tools to further their disciplines. The increasing power of computers offers unprecedented ability to solve complex equations, simulate natural and man-made complex processes, and visualise data, as well as providing novel possibilities such as new forms of art and entertainment. As computational power advances rapidly, computational tools, libraries, and computing paradigms themselves also advance. In such an environment, even experienced computational scientists and engineers can easily find themselves falling behind the pace of change, while they redesign and rework their codes to support the next computer architecture. This

We demonstrate that evolutions of three-dimensional, strongly non-linear gravitational waves can be followed in numerical relativity, hence allowing many interesting studies of both fundamental and observational consequences. We study the evolution of time-symmetric, axisymmetric and non-axisymmetric Brill waves, including waves so strong that they collapse to form black holes under their own self-gravity. An estimate for the critical amplitude for black hole formation in a particular interpolating family of initial data is obtained. The gravitational waves emitted in the black hole formation process are compared to those emitted in the head-on collision of two Misner black holes.PACS number͑s͒: 04.25. Dm, 04.30.Db, 95.30.Sf, 97.60.Lf Gravitational waves have been an important area of research in Einstein's theory of gravity for years. Einstein's equations are nonlinear, and therefore can cause waves, which normally would disperse if weak enough, to be held together by their own gravity. This property characterizes Wheeler's geon ͓1,2͔ proposed more than 40 years ago, and is responsible for many interesting phenomena. Even in planar symmetric spacetimes, there are many interesting results, such as the formation of singularities from colliding plane waves ͑see ͓3͔ and references therein͒. In axisymmetry, Ref.͓4͔ studied the formation of black holes ͑BHs͒ by imploding gravitational waves, finding critical behavior ͓5͔.These discoveries are all in spacetimes with special symmetries, but they raise important questions about general three-dimensional ͑3D͒ spacetimes, e.g., the nature of critical phenomena in the absence of symmetries has only recently been studied through a perturbative approach ͓6͔. A few studies of gravitational wave evolutions have been performed in the linear and near linear regimes ͓7-9͔, in preparation for the study of nonlinear, strong field 3D wave dynamics. However, until now no such studies have been successfully carried out.In this paper we present the first successful simulations of highly nonlinear gravitational waves in 3D; i.e., we study the process of strong waves collapsing to form BHs under their own self-gravity. We obtain an estimate for the critical amplitude for the formation of BHs based on a particular family of interpolating initial data. We show that one can now carry out these evolutions for long times. For waves that are not strong enough to form BHs, we follow their implosion, bounce and dispersal. For waves strong enough to collapse to a BH under their own self-gravity, we find the dynamically formed apparent horizons ͑AHs͒, and extract the gravitational radiation generated in the collapse process. These wave forms can be compared in axisymmetry to head-on BH collisions ͑performed earlier and reported in ͓10͔͒. The wave forms are similar at late times, dominated by the quasinormal modes of the resulting BHs as expected. The difference in the wave forms at early times for these two very different collapse scenarios shows to what extent one can extract information about...

We present results for two colliding black holes (BHs), with angular momentum, spin, and unequal mass. For the first time, gravitational waveforms are computed for a grazing collision from a full 3D numerical evolution. The collision can be followed through the merger to form a single BH, and through part of the ringdown period of the final BH. The apparent horizon is tracked and studied, and physical parameters, such as the mass of the final BH, are computed. The total energy radiated in gravitational waves is shown to be consistent with the total initial mass of the spacetime and the apparent horizon mass of the final BH. DOI: 10.1103/PhysRevLett.87.271103 PACS numbers: 04.25.Dm, 04.30.Db, 95.30.Sf, 97.60.Lf The collision of two black holes (BHs) is considered by many researchers to be a primary candidate for generating detectable gravitational waves. As the first generation of gravitational wave detectors [1], with enough sensitivity to potentially detect waves, is coming online for the first time next year, the urgency of providing theoretical information needed not only to interpret, but also to detect the waves, is very high. However, even in axisymmetry, the problem has proven to be extremely difficult, requiring nearly 20 years to solve in even limited cases (e.g., [2 -5]). In full, 3D progress has been rather slow due to many factors, including (but not limited to) unexpected numerical instabilities, limited computer power, and the difficulties of dealing with spacetime singularities inside BHs. The first true 3D simulation of spinning and moving BHs was performed in [6]. In [6], the two BHs start out close to each other, much closer than the separation for the last stable orbit of a particle in the Schwarzschild spacetime, and the evolution proceeds through parts of the plunge and ring-down phase of a "grazing collision" within a very short time interval. The spacetime singularities are dealt with by a particular choice of coordinates, singularity avoiding slicing and vanishing shift. BH excision [7,8] has allowed improvements in the treatment of the spacetime singularities to the extent that highly accurate simulations of single BHs can be carried out [9][10][11][12] and recent applications to the grazing collision of BHs show promise [13]. One of the key limiting factors in the existing two approaches to the grazing collision is the achievable evolution time for which useful numerical data can be obtained, which due to numerical problems has been limited to 7M in [6], and to about 9M-15M in [13]. Here time is measured in units of the total Arnowitt-Deser-Misner (ADM) mass M of the system as opposed to using the bare mass m of one of the BHs.In this paper we consider singularity avoiding slicing as in [6]. We combine the application of a series of recently developed physics analysis tools and techniques with significant progress made in overcoming the problems mentioned above. Early, preliminary results from this series of simulations have been presented in [14,15], but we now provide the first de...

The ability to harness heterogeneous, dynamically available "Grid" resources is attractive to typically resource-starved computational scientists and engineers, as in principle it can increase, by significant factors, the number of cycles that can be delivered to applications. However, new adaptive application structures and dynamic runtime system mechanisms are required if we are to operate effectively in Grid environments. In order to explore some of these issues in a practical setting, we are developing an experimental framework, called Cactus, that incorporates both adaptive application structures for dealing with changing resource characteristics and adaptive resource selection mechanisms that allow applications to change their resource allocations (e.g., via migration) when performance falls outside specified limits. We describe here the adaptive resource selection mechanisms and describe how they are used to achieve automatic application migration to "better" resources following performance degradation. Our results provide insights into the architectural structures required to support adaptive resource selection. In addition, we suggest that this "Cactus Worm" is an interesting challenge problem for Grid computing.

No abstract