# Optimality Conditions Centre for Complex Dynamic Systems and Control

View again

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
PDF
58 pages
0 downs
6 views
Share
Description
Optimality Conditions Centre for Complex Dynamic Systems and Control
Tags

## Education

Transcript
Optimality Conditions Mar´ıa M. SeronSeptember 2004 Centre for Complex DynamicSystems and Control  Outline 1  Unconstrained OptimisationLocal and Global MinimaDescent DirectionNecessary Conditions for a MinimumNecessary and Sufﬁcient Conditions for a Minimum 2  Constrained OptimisationGeometric Necessary Optimality ConditionsProblems with Inequality and Equality ConstraintsThe Fritz John Necessary ConditionsKarush–Kuhn–Tucker Necessary ConditionsKarush–Kuhn–Tucker Sufﬁcient ConditionsQuadratic Programs Centre for Complex DynamicSystems and Control  Unconstrained Optimisation An unconstrained optimisation problem is a problem of the formminimise  f  ( x  ) ,  (1)without any constraint on the vector  x  .Deﬁnition (Local and Global Minima) Consider the problem of minimising f  ( x  )  over  R n  and let   ¯ x   ∈  R n  .If f  (¯ x  )  ≤  f  ( x  )  for all x   ∈  R n  , then   ¯ x is called a   global minimum  .If there exists an   ε -neighbourhood N  ε (¯ x  )  around   ¯ x such that f  (¯ x  )  ≤  f  ( x  )  for all x   ∈  N  ε (¯ x  ) , then   ¯ x is called a   local minimum  .If f  (¯ x  )  <  f  ( x  )  for all x   ∈  N  ε (¯ x  ) , x     ¯ x, for some   ε >  0 , then   ¯ x is called a   strict local minimum  . Centre for Complex DynamicSystems and Control  Local and Global Minima The ﬁgure illustrates local and global minima of a function  f   overthe reals. Strict local minimumGlobal minimaLocal minima f  Figure: Local and global minima Clearly, a global minimum is also a local minimum. Centre for Complex DynamicSystems and Control  Descent Direction Given a point  x   ∈  R n  , we wish to determine, if possible, whether ornot the point is a local or global minimum of a function  f  .For differentiable functions, there exist conditions that provide thischaracterisation, as we will see below.We start by characterising  descent directions  .Theorem (Descent Direction) Let f   :  R n  →  R be differentiable at   ¯ x. If there exists a vector d such that  ∇ f  (¯ x  ) d   <  0 , then there exists a   δ >  0  such that f  (¯ x   + λ d  )  <  f  (¯ x  )  for each   λ  ∈  ( 0 ,δ ) , so that d is a   descent direction   of f at   ¯ x. Centre for Complex DynamicSystems and Control
Related Search
Similar documents

View more...