Optimization and systems
Planning, logistics, recommendations and the performance of computer systems: optimization and probability applied to decisions at scale.
3 topics
Topics
Operations research and logistics
Optimal decisions under constraints: production plans, delivery routes, crew schedules, network flows. Continuous problems are solved with convex optimization and duality (Lagrange multipliers as shadow prices); discrete ones add integer programming on top.
Recommender systems
Predict how much a user will like an item as the dot product of learned vectors, , fitted by (stochastic) gradient descent on the known ratings with regularization.
Queueing theory and performance
Models of requests waiting for servers. With exponential (memoryless) arrivals and services, the M/M/1 queue gives closed formulas: as utilization , waiting time explodes like — why systems are run well below full capacity.