The study of star cluster evolution has seen steady progress throughout the last several decades. Analytic estimates dating back from the late thirties, together with numerical simulations in the late sixties and seventies, have provided the stage for the break-through in our understanding in the early eighties, when the evolution through core collapse and beyond began to be understood.
In the ten years following this break-through, the various modeling
techniques have been pushed to the limit of their applicability. We
are now facing three barriers separating us from our goal of an
adequate treatment of star cluster evolution: a star-by-star -body
modeling, including stellar evolution as well as stellar dynamics.
Two of the three barriers are related to the stellar dynamics part of the problem: getting the raw speed to do the simulations, and developing the algorithms necessary to utilize that speed. The third barrier is related to the implementation of an utterly simple (but not too simple) treatment of the various stellar evolution effects mingling with stellar dynamics.
The identification and initial exploration of these barriers begun in earnest five years ago, with a detailed analysis of the requirements of a star-by-star simulation (Makino &Hut 1988; Hut, Makino &McMillan 1988). The resulting specification of the necessity of a computer speed of order a Teraflops at the time made our goal seem remote. And indeed, it is unreasonable to expect full-time access to a Teraflops computer before some time early in the next century.
Fortunately, we do not have to wait a decade. Thanks to the hardware
development of the Tokyo group (Makino et al. 1993), we now have a
definite time table for our first star-by-star globular cluster
simulations, with a number of stars in the range of -
: this is expected to take place some time in 1994.