resolvent and cayley operator - github pages€¦ · as a mnemonic (unverified), in a composite...

4
def: monotone operators definitions and related Lipschitz mapping, (e.g., nonexpansive and contraction mapping proof: (\lambda>=0, F monotone) = thm: for positive coefficient and monotone operator resolvent is a nonexpansive function Resolvent_a nd_Cayley... Resolvent and Cayley Operator 9:37 PM EE 364b Convex Optimization II Page 1

Upload: others

Post on 18-Feb-2021

0 views

Category:

Documents


0 download

TRANSCRIPT

  • def: monotone operators definitions and related

    Lipschitz mapping, (e.g., nonexpansive and contraction mapping

    proof: (\lambda>=0, F monotone) =

    thm: for positive coefficient and monotone operator resolvent is a nonexpansive function

    Resolvent_and_Cayley...

    Resolvent and Cayley Operator9:37 PM

    EE 364b Convex Optimization II Page 1

    onenote:#%7bNonexpansive,Contraction%7d%20mapping%20and%20Monotone%20Operators&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={94CA80D6-DFA1-4AED-A3D6-B2ABC1A8F4A7}&object-id={1C8E21C8-3E16-0E9F-30BF-126D2AD73D4C}&24&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.oneonenote:#%7bNonexpansive,Contraction%7d%20mapping%20and%20Monotone%20Operators&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={94CA80D6-DFA1-4AED-A3D6-B2ABC1A8F4A7}&object-id={BE94A8EC-0A0F-0F78-1D4C-2E0EDCD6168E}&E&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.oneonenote:#Resolvent%20and%20Cayley%20Operator&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={9EF9BC8C-D5A8-4637-8EF9-22E3CA7BFA42}&object-id={4F180EF6-8B0F-059D-3671-E6F8D440151C}&FE&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.one

  • statement: overloaded sum operator for relations has additivity

    As a mnemonic (unverified), in a composite relation, the constituent relation gulor majhe jara function, tader jonno ukto equ ation e functional operation chalano jabe

    Old proof , I suggest skip it.

    EE 364b Convex Optimization II Page 2

    onenote:#Introduction,%20Relations&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={7DCA72CC-6D53-A54E-8C6C-704EC43B7196}&object-id={A3991E73-C6A0-0641-366D-B93D6EFC72C3}&67&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.one

  • (\lambda>=0, F monotone) => C [*nonexpansive function*]

    thm: proximal operator is the resolvent of subdifferential operator

    normal cone operator

    thm: proximal operator is the resolvent of subdifferential operator

    def: multiplier to residual mapping

    def: multiplier to residual mapping

    thm: for positive coefficient and monotone operator resolvent is anonexpansive function

    [# Note, we want to find out z=R(y), but to find out z here we also need to know x*, so this is what weare going to do, solve for x* first, and then set the value in z=y-\lambda (b- Ax*) #]

    # resolvent for multiplier to residual mapping

    # Resolvent of normal cone operator

    EE 364b Convex Optimization II Page 3

    onenote:#1.2%20%5bBoyd%5d%20Interpretation%20of%20proximal%20mapping&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={0E80E461-59A9-4B89-A738-1DC23BAA0395}&object-id={BCB8BBB6-3602-07C2-2581-83845CD83707}&83&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.oneonenote:#%7bNonexpansive,Contraction%7d%20mapping%20and%20Monotone%20Operators&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={94CA80D6-DFA1-4AED-A3D6-B2ABC1A8F4A7}&object-id={CB71D46A-8DB3-0186-3778-1F06AABD753C}&F8&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.oneonenote:#1.2%20%5bBoyd%5d%20Interpretation%20of%20proximal%20mapping&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={0E80E461-59A9-4B89-A738-1DC23BAA0395}&object-id={BCB8BBB6-3602-07C2-2581-83845CD83707}&83&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.oneonenote:#%7bNonexpansive,Contraction%7d%20mapping%20and%20Monotone%20Operators&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={94CA80D6-DFA1-4AED-A3D6-B2ABC1A8F4A7}&object-id={09AA497C-9A86-086B-3C25-BF2A262CD688}&D&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.oneonenote:#%7bNonexpansive,Contraction%7d%20mapping%20and%20Monotone%20Operators&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={94CA80D6-DFA1-4AED-A3D6-B2ABC1A8F4A7}&object-id={09AA497C-9A86-086B-3C25-BF2A262CD688}&D&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.oneonenote:#%7bNonexpansive,Contraction%7d%20mapping%20and%20Monotone%20Operators&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={94CA80D6-DFA1-4AED-A3D6-B2ABC1A8F4A7}&object-id={09AA497C-9A86-086B-3C25-BF2A262CD688}&D&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.oneonenote:#Resolvent%20and%20Cayley%20Operator&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={9EF9BC8C-D5A8-4637-8EF9-22E3CA7BFA42}&object-id={93CB4476-9A25-0A30-1D06-06B0F725342E}&8F&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.one

  • [# Note, we want to find out z=R(y), but to find out z here we also need to know x*, so this is what weare going to do, solve for x* first, and then set the value in z=y-\lambda (b- Ax*) #]

    def: maximal monotone

    As a mnemonic (unverified), in a composite relation, the constituent relation gulor majhe jara function, tader jonno ukto equation e functional operation chalano jabe

    thm: for positive coefficient and monotone operator resolvent is anonexpansive function

    Why find fixed point of Cayley and resolvent of some operator?•

    why maximality is needed?

    Compact form: Resolvent of the multiplier to residual mapping

    EE 364b Convex Optimization II Page 4

    onenote:#%7bNonexpansive,Contraction%7d%20mapping%20and%20Monotone%20Operators&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={94CA80D6-DFA1-4AED-A3D6-B2ABC1A8F4A7}&object-id={6922464D-B07F-0D34-01E1-D5742EDFC42B}&5D&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.oneonenote:#Resolvent%20and%20Cayley%20Operator&section-id={9FDA29CD-B182-4008-B35A-143DC92D9F84}&page-id={9EF9BC8C-D5A8-4637-8EF9-22E3CA7BFA42}&object-id={93CB4476-9A25-0A30-1D06-06B0F725342E}&8F&base-path=https://d.docs.live.net/0ad95f7c63bcb36b/Documents/Convex_Optimization_II_Selfstudy/EE%20364b%20Convex%20Optimization%20II.one