31.10.2014 Views

Chapter 3. Duality in convex optimization

Chapter 3. Duality in convex optimization

Chapter 3. Duality in convex optimization

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Th.<strong>3.</strong>17 (week duality) If x ∈ R n is primal feasible and<br />

Z ∈ R m×m is dual feasible, then<br />

c T x − A 0 • Z = F (x) • Z ≥ 0.<br />

Equality holds iff F (x) · Z = 0 (and <strong>in</strong> this case x, Z are<br />

optimal for (P), (D), respectively).<br />

Remark. [Without proof (see, e.g., [FKS, Ch.12])]<br />

Strong duality holds under the assumption that (P) is solvable<br />

and that there exists a strictly feasible x ∗ :<br />

F (x ∗ ) ≻ 0<br />

CO, <strong>Chapter</strong> 3 p 18/18

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!