Steve McMillan - Gravitational dynamics of large stellar systems - Dynamics basics - Dynamical time (G M / R^3)^(-1/2) - Relaxation time ( ~ v^2); time to lose memory of initial state - Collisional/collisionless systems; relaxation time comparable/much larger than dynamical time - Young dense clusters and galactic nuclei: short relaxation times - Dynamical processes: mass segregation, core collapse (central density increases) - Mass segregation <- dynamical friction: massive objects slowed by collective response of stellar system (attraction from lagging "wake") - A rapid process compared to relaxation - Understandable as equipartition: = constant -> propto m^(-1/2) - Relaxation conducts energy from warm core to cooler halo; negative specific heat -> core heats up as it loses energy and runaway core collapse follows - Dynamical processes take interesting objects and puts them in a restricted region - Massive stars, giant stars, heavy white dwarfs, black holes... - Binary dynamics stops core collapse: binary scattering, energy equipartition, Heggie's law (hard binaries get harder, soft binaries get softer) - Heating rate independent of binary energy; eventually recoil of binaries will eject at 200-400 kT (kT ~ equipartition energy) - Binaries can support the core against collapse - Gravothermal oscillations: do not seem to exist in real (non equal-mass) systems - Compactification of binaries: even if the heavier NS do not form in binaries, they will tend to replace lighter component in interacting binaries - Number of X-ray sources seem to be correlated with collision rate (why?) - "Kitchen sink" cluster models - Multiphysics happening on comparable timescales - Leading models do star-by-star evolution, either N-body, Monte-Carlo (orbit averaged), or special-purpose hardware - GRAPE: massively parallel, deeply pipelined gravitational-force accelerator - since 10^18 flops for 10^6-star systems required - GRAPE-DR: SIMD, programmable, petaflop - Include special treatment of binaries and multiples (regularization, PN), stellar evolution (interpolation on precomputed grids), binary evolution, stellar interactions of collisions - Clusters: collisions imply creation of new stellar species not usually included in stellar-evolution models - MODEST (MOdeling DEnse STellar systems) - Intermediate-mass black holes in clusters - Interesting (if not compelling) evidence of existence: association between ultraluminous X-ray sources and young dense star clusters; in M82 X-1/MGG-11, if no beaming M_bh ~ 750 Msun - M15: 1000 Msun nonluminous matter in the cluster core; but NS, WD populations also possible - G1: M_bh ~ 20,000 Msun from velocity dispersion curves; claim fortified by X-ray and radio emission from center - Omega Centauri: M_bh ~ 40,000 Msun from dispersion - These BH fit at the bottom end of the M-sigma relation for galaxies and their BHs - Optimistic: good part of clusters may host IMBH - Dynamics of formation understood, stellar physics less so - Mass segregation, heavier stars form subcore (must do so before first supernova) - Runaway mergers: high densities lead to collisions, which naturally involve heavier stars [check Lombardi (2005)] - Stopped by mass loss from winds - Ejected black-hole binaries are LIGO targets - Black-hole encounters/mergers may also produce observable GW (O'Leary) - Summary - Dense stellar systems natural places for relativistic binaries (for LIGO) - Growing evidence for IMBHs in some clusters (100-1000 Msun, likely in binaries; EMRI of WD/NS/BH onto IMBH) - Clusters in galactic nuclei (EMRI of IMBH)