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.

642 APPENDIX E. PROJECTION<br />

For any shape of matrix A of any rank, and given any vector b that may<br />

or may not be in R(A) , we wish to find a best Euclidean solution x ⋆ to<br />

Ax = b (1754)<br />

(more generally, Ax ≈ b given arbitrary matrices) by solving<br />

minimize<br />

x<br />

‖Ax − b‖ 2 (1755)<br />

Necessary and sufficient condition for optimal solution to this unconstrained<br />

optimization is the so-called normal equation that results from zeroing the<br />

convex objective’s gradient: (D.2.1)<br />

A T Ax = A T b (1756)<br />

normal because error vector b −Ax is perpendicular to R(A) ; id est,<br />

A T (b −Ax)=0. Given any matrix A and any vector b , the normal equation<br />

is solvable exactly; always so, because R(A T A)= R(A T ) and A T b∈ R(A T ).<br />

When A is skinny-or-square full-rank, normal equation (1756) can be<br />

solved exactly by inversion:<br />

x ⋆ = (A T A) −1 A T b ≡ A † b (1757)<br />

For matrix A of arbitrary rank and shape, on the other hand, A T A might<br />

not be invertible. Yet the normal equation can always be solved exactly by:<br />

(1745)<br />

x ⋆ = lim A + t I) −1 A T b = A † b (1758)<br />

t→0 +(AT<br />

invertible for any positive value of t by (1354). The exact inversion (1757)<br />

and this pseudoinverse solution (1758) each solve<br />

lim minimize<br />

t→0 + x<br />

‖Ax − b‖ 2 + t ‖x‖ 2 (1759)<br />

simultaneously providing least squares solution to (1755) and the classical<br />

least norm solution E.1 [287, App.A.4] (confer3.1.3.0.1)<br />

arg minimize ‖x‖ 2<br />

x<br />

subject to Ax = AA † b<br />

(1760)<br />

E.1 This means: optimal solutions of lesser norm than the so-called least norm solution<br />

(1760) can be obtained (at expense of approximation Ax ≈ b hence, of perpendicularity)<br />

by ignoring the limiting operation and introducing finite positive values of t into (1759).

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

Saved successfully!

Ooh no, something went wrong!