210 likes | 226 Views
This reading group session discusses the basics of convex optimization, including convex functions, unique minimum over convex domains, (sub)gradient methods, and decomposition methods. It also explores constrained convex optimization, optimality conditions, and network utility maximization. Implementation issues and the concept of primal-dual variables are also discussed.
E N D
DecomposableOptimisation Methods LCA Reading Group, 12/04/2011 Dan-Cristian Tomozei
Convexity • Convex function • Unique minimum over convex domain
Roadmap • (Sub)Gradient Method • Convex Optimisationcrash course • NUM • Basic Decomposition Methods • Implicit Signalling
Roadmap • (Sub)Gradient Method • Convex Optimisationcrash course • NUM • Basic Decomposition Methods • Implicit Signalling
(Sub)gradient method • Unconstrained convex optimisation problem • If objective is differentiable, • Else, • Gain sequence • Constant • Diminishing
Roadmap • (Sub)Gradient Method • Convex Optimisationcrash course • NUM • Basic Decomposition Methods • Implicit Signalling
Constrained Convex Optimisation • “Primal” formulation • Convex constraints unique solution • Lagrangian • “Dual” function • For all “feasible” points – lower bound • Slater’s condition zero duality gap
Optimality conditions • “Primal” and “dual” formulations • Karush-Kuhn-Tucker (KKT) Primal variables Dual variables (i.e., Lagrange multipliers) Optimum
Roadmap • (Sub)Gradient Method • Convex Optimisation crash course • NUM • Basic Decomposition Methods • Implicit Signalling
Network Utility Maximisation • Population of users • Concave utility functions (e.g., rates) • Typical formulation (e.g., [Kelly97]): • Network flows of rates • Physical links of max capacity • Routing matrix • Dual variables = congestion shadow prices
Roadmap • (Sub)Gradient Method • Convex Optimisation crash course • NUM • Basic Decomposition Methods • Implicit Signalling
Dual Decomposition • Coupling constraint • To decouple – simply write the dual objective • Iterative dual algorithm: • Each user computes • Use a gradient method to update dual variables, e.g.,
Primal Decomposition • Coupling variable • To decouple – consider fixed coupling variable • Iterative primal algorithm: • Solve individual problems and get partial optima • Update primal coupling variable using gradient method
Implementation issues • Certain problems can be decoupled • Dual decomposition dual algorithm • Primal vars (rates) depend directly on dual vars (prices) • Price adaptation relies on current rates • Always closed form? • Primal decomposition • The other way around… • Do we really need to keep track of both primal and dual variables? Can duals be “measured” instead?
Roadmap • (Sub)Gradient Method • Convex Optimisation crash course • NUM • Basic Decomposition Methods • Implicit Signalling
Multipath unicast min-cost live streaming • Graph • Supported rate region • Network cost function • Unsupported rate allocation • Marginal cost positive and strictly increasing • Source s wants to send data to receiver r at rate at minimum cost • Supported min-cut is at least
Optimisation formulation • Write Lagrangian • Primal-dual provably converges to optimum
Is it that hard? • Recall • Dual variables have queue-like evolution! • We already queue packets!
Implicit Primal-Dual • Rate control via • Rate on link (i,j) • Increase prop to backlog difference • Decrease prop to marginal cost (measurable – RTT, …) • Influence of parameter s • Small closer optimal allocation, huge queue sizes • Large manageable queue sizes, optimality trade-off
Conclusion • Finding a fit-all recipe is hard • We can handle some cases • Specific formulations may lead to nice protocols • See also • R. Srikant’s “Mathematics of Internet Congestion Control” • Kelly, Mauloo, Tan - *** • Palomar, Chiang - ***