The Variable-Processor Cup Game
summary
Shows that allowing P to vary changes guarantees drastically: the greedy emptier remains optimal versus deterministic adversaries but its backlog jumps from O(log N) to Ω(N). Provides a recursive filling strategy that forces backlog ~N within time about 2^N, and a refined construction that still forces backlog N^{1-ε} in quasi-polynomial time against randomized greedy-like emptiers. Highlights dependence on adaptive versus oblivious fillers and leaves open whether the N^{1-ε} gap is avoidable.
AI-written summary · about
chapters (16)
- 0:00Intro
- 0:14THE SINGLE-PROCESSOR CUP GAME
- 1:39THE P-PROCESSOR CUP GAME
- 2:33THE GREEDY ALGORITHM
- 3:14LOTS OF OTHER DIRECTIONS OF RESEARCH
- 4:09ON TIME-SHARED SYSTEMS, RESOURCES VARY
- 4:59THIS PAPER: THE VARIABLE-PROCESSOR CUP G.
- 6:27How DOES GREEDY PERFORM?
- 7:52WHAT ABOUT FOR SHORTER GAMES?
- 8:50WHAT IF THE EMPTIER IS RANDOMIZED? Important Note: The filler is an oblivious adversary
- 10:50PUTTING THE RESULTS TOGETHER
- 12:03THE GOAL
- 12:15INDUCTIVE HYPOTHESIS: n-1 CUPS
- 14:06FREEZE CUP 1 DURING THE RECURSION
- 15:25SET p = 1 AND PLACE 1/2 IN EACH CUP
- 16:09DOWNSIDES OF THIS CONSTRUCTION
