10.03.2015 Views

v2009.01.01 - Convex Optimization

v2009.01.01 - Convex Optimization

v2009.01.01 - 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.

E.5. PROJECTION EXAMPLES 659<br />

E.5.0.0.5 Example. Projecting the origin on a hyperplane.<br />

(confer2.4.2.0.2) Given the hyperplane representation having b∈R and<br />

nonzero normal a∈ R m ∂H = {y | a T y = b} ⊂ R m (105)<br />

orthogonal projection of the origin P0 on that hyperplane is the unique<br />

optimal solution to a minimization problem: (1784)<br />

‖P0 − 0‖ 2 = inf<br />

y∈∂H ‖y − 0‖ 2<br />

= inf<br />

ξ∈R m−1 ‖Zξ + x‖ 2<br />

(1820)<br />

where x is any solution to a T y=b , and where the columns of Z ∈ R m×m−1<br />

constitute a basis for N(a T ) so that y = Zξ + x ∈ ∂H for all ξ ∈ R m−1 .<br />

The infimum can be found by setting the gradient (with respect to ξ) of<br />

the strictly convex norm-square to 0. We find the minimizing argument<br />

so<br />

and from (1786)<br />

P0 = y ⋆ = a(a T a) −1 a T x = a<br />

‖a‖ ‖a‖ x =<br />

ξ ⋆ = −(Z T Z) −1 Z T x (1821)<br />

y ⋆ = ( I − Z(Z T Z) −1 Z T) x (1822)<br />

a T<br />

a<br />

‖a‖ 2 aT x ∆ = A † Ax = a b<br />

‖a‖ 2 (1823)<br />

In words, any point x in the hyperplane ∂H projected on its normal a<br />

(confer (1848)) yields that point y ⋆ in the hyperplane closest to the origin.<br />

<br />

E.5.0.0.6 Example. Projection on affine subset.<br />

The technique of Example E.5.0.0.5 is extensible. Given an intersection of<br />

hyperplanes<br />

A = {y | Ay = b} ⊂ R m (1824)<br />

where each row of A ∈ R m×n is nonzero and b ∈ R(A) , then the orthogonal<br />

projection Px of any point x∈ R n on A is the solution to a minimization<br />

problem:<br />

‖Px − x‖ 2 = inf<br />

y∈A<br />

‖y − x‖ 2<br />

= inf<br />

ξ∈R n−rank A ‖Zξ + y p − x‖ 2<br />

(1825)

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

Saved successfully!

Ooh no, something went wrong!