Due to the close dynamical interactions in dense stellar systems a variety of "strange" objects will form, which do not exist in the normal stellar population of the galactic disk. Among such objects will be: results of physical collisions: coalesced and stripped stars; X-ray binaries (neutron star or black hole plus ordinary star); Nova binaries (white dwarf plus ordinary star); double neutron stars, double black holes, double white dwarfs, and other combinations of neutron stars, black holes and white dwarfs. The predicted incidence and other properties of the X-ray binaries and binary pulsars (neutron star plus white dwarf), as a function of binary fraction, can then be compared with the observed incidence and properties of these objects, from which the initial binary fraction can be obtained.Using this fraction, one can then predict the incidence and formation rates of double neutron stars, double black holes and black hole plus neutron star systems in globular clusters. The coalescence events of these three types of double compact objects are expected to produce strong bursts of gravitational waves, detectable out to hundreds of Mega-parsecs distance in the Universe with future gravitational wave antennas like LIGO and TAMA300.
The first operational Laser Interferometer gravitational wave detector, TAMA300 [21], has demonstrated that on 19 September 1999 there has been no collisions between neutron stars within a distance of 5kpc from the Sun. Within the coming few years several other gravitational wave detectors will become available. These detectors are more sensitive than TAMA300 and they will open the possibility to qualitatively study the inner engines of globular clusters, which form ideal laboratories for gravitational wave antennas, because their distances are known and they are likely to host a wealth of sources [22]. The prediction of the rates of these events is therefore expected to be one the important scientific results from this project - in addition to all other results already mentioned.
A future most important result will be knowledge of the evolution of the dense stellar system through core collapse, as well as the result of this collapse, which so far is a completely uncharted territory. The first calculations to study the importance of the evolution of star clusters on the detection rates for gravitational wave detectors have been performed using the Starlab software environment (mentioned earlier) [20]. They found that in star clusters binaries are formed that contain two black holes. These black hole binaries have orbits so tight and eccentricities so high than the emission of gravitational waves causes them to coalsece in a few billion years. The predicted rate of these black-hole mergers is substantially greater than the corresponding rates from neutron star mergers [20].
The development of efficient N-body kernels, creating as much as possible data-locality to be exploited in efficient parallel code, and optimizing the cache behavior in heterogeneous architectures will be an important result of the research. Also the resulting algorithmic innovations are highly relevant, not only in the case of stellar dynamics but also in the field of molecular dynamics.Within High Performance Computing an important research item is the optimal use of heterogeneous computing resources, where e.g. a vector supercomputer and a massively parallel computer cooperate to solve a problem. The UvA N-body lab can also be considered as such a hybrid architecture. Optimal tuning of applications to use as efficiently as possible the computational power as offered by the UvA N-body lab therefore is a highly relevant research issue.