13dualmethods

# 13dualmethods - EE236C(Spring 2008-09 13 Dual methods dual...

This preview shows pages 1–4. Sign up to view the full content.

EE236C (Spring 2008-09) 13. Dual methods dual of convex problem with linear constraints differentiability of dual function dual decomposition rate control 13–1 Convex problem with linear constraints minimize f ( x ) subject to Gx precedesequal h Ax = b ( G R m × n , A R p × n ) dual function g ( λ, ν ) = inf x dom f L ( x, λ, ν ) = inf x dom f ( f ( x ) + ( G T λ + A T ν ) T x h T λ b T ν ) = h T λ b T ν f * ( G T λ A T ν ) f * ( y ) = sup x dom f ( x T y f ( x )) is the conjugate of f this lecture: methods that solve dual by a first-order method Dual methods 13–2

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Dual feasibility ( λ, ν ) is dual feasible if λ followsequal 0 and L ( x, λ, ν ) is bounded over x dom f : G T λ A T ν dom f * , λ followsequal 0 to solve dual problem when feasible set is not R m + × R p make dual domain explicit ( e.g. , when using gradient projection) minimize L with a method that detects unboundedness, and use unbounded direction to generate cutting-plane for dual domain ( e.g. , when using ACCPM) dual methods simplify if feasible set is R m + × R p some sufficient conditions: dom f is bounded ( e.g. , after addition of variable bounds) f strongly convex ( e.g. , after addition of a regularization term) Dual methods 13–3 Subgradients of conjugate function f * ( y ) = sup x dom f ( x T y f ( x )) from page 2–7, f * is subdifferentiable at all points in intdom f * weak subgradient rule if ˆ x maximizes x T ˆ y f ( x ) over x dom f , then ˆ x ∂f * y ) f * ( y ) = sup x dom f ( x T y f ( x )) ˆ x T y f x ) = ˆ x T ˆ y f x ) + ˆ x T ( y ˆ y ) = f * y ) + ˆ x T ( y ˆ y ) Dual methods 13–4
Conjugate of closed functions a convex function is closed if epi f is a closed set (see BV § A.3.3) properties: for f closed and convex f ** = f : conjugate of conjugate of f is f (BV exercise 3.39) y ∂f ( x ) if and only if x ∂f * ( y ) proof: if ˆ y ∂f x ) , then x T ˆ y f ( x ) reaches its maximum over x at ˆ x ; therefore ˆ x ∂f * y ) ; this shows ˆ y ∂f x ) = ˆ x ∂f * y ) reverse implication follows from f ** = f Dual methods 13–5 Differentiability of conjugate if f is closed and strictly convex, then f *

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

### What students are saying

• As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

Kiran Temple University Fox School of Business ‘17, Course Hero Intern

• I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

Dana University of Pennsylvania ‘17, Course Hero Intern

• The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

Jill Tulane University ‘16, Course Hero Intern