Approximate dynamic programming using fluid and diffusion approximations with applications to power management.
Wei ChenDayu HuangAnkur A. KulkarniJayakrishnan UnnikrishnanQuanyan ZhuPrashant G. MehtaSean P. MeynAdam WiermanPublished in: CoRR (2013)
Keyphrases
- approximate dynamic programming
- power management
- diffusion approximations
- heavy traffic
- queueing networks
- power consumption
- linear program
- long run
- dynamic programming
- energy saving
- stochastic dynamic programming
- average cost
- energy consumption
- reinforcement learning
- data center
- steady state
- control policy
- energy efficiency
- step size
- asymptotically optimal
- policy iteration
- single server
- call center
- arrival rate
- state space
- design considerations
- convergence rate
- fixed point
- evolutionary algorithm
- multistage
- neural network