19.07.2014 Views

Contents - Student subdomain for University of Bath

Contents - Student subdomain for University of Bath

Contents - Student subdomain for University of Bath

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.

218 APPENDIX A. ALGEBRAIC BACKGROUND<br />

Notation 30 In the context <strong>of</strong> V (k 1 , . . . , k n ), let P (z) =<br />

n∏<br />

(z − k i ).<br />

i=1<br />

i≠j<br />

n∏<br />

(z−k i ) and P j (z) =<br />

Proposition 76 The inverse <strong>of</strong> a Vandermonde matrix has a particularly simple<br />

<strong>for</strong>m: V (k 1 , . . . , k n ) −1 = (m i,j ) with<br />

m i,j = coeff(P j(z), z i )<br />

∏<br />

k≠j (k j − k k ) = coeff(P j(z), z i )<br />

. (A.9)<br />

P j (k j )<br />

i=1<br />

For example<br />

⎛<br />

V (k 1 , . . . , k 3 ) −1 = ⎜<br />

⎝<br />

k 2k 3<br />

(−k 2+k 1)(−k 3+k 1)<br />

−<br />

k 2+k 3<br />

(−k 2+k 1)(−k 3+k 1)<br />

−<br />

k 1k 3<br />

(−k 2+k 1)(k 2−k 3)<br />

k 1+k 3<br />

(−k 2+k 1)(k 2−k 3)<br />

1<br />

(−k 2+k 1)(−k 3+k 1)<br />

−<br />

1<br />

(−k 2+k 1)(k 2−k 3)<br />

k 1k 2<br />

(k 2−k 3)(−k 3+k 1)<br />

−<br />

k 1+k 2<br />

(k 2−k 3)(−k 3+k 1)<br />

1<br />

(k 2−k 3)(−k 3+k 1)<br />

⎞<br />

⎟<br />

⎠<br />

Corollary 25 If all the k i are distinct, then V (k 1 , . . . , k n ) is invertible.<br />

Equation (A.9) and the fact that the P j can be rapidly computed <strong>for</strong>m p suggest<br />

a rapid way <strong>of</strong> computing the elements <strong>of</strong> the inverse <strong>of</strong> a n × n Vandermonde<br />

matrix in O(n 2 ) time. In fact, we can solve a system <strong>of</strong> Vandermonde linear<br />

equations in O(n 2 ) time and O(n) space.<br />

Algorithm 43 (Vandermonde solver)<br />

Input: Vandermonde matrix V (k 1 , . . . , k n ), right-hand side w.<br />

Output: Solution x to V (k 1 , . . . , k n )x = w<br />

x := 0<br />

P := ∏ n<br />

i=1 (z − k i) #O(n 2 )<br />

<strong>for</strong> i := 1 to n<br />

Q := P/(z − k i ) #nO(n)<br />

D := P (k i ) #nO(n) by Horner’s rule<br />

<strong>for</strong> j := 1 to n<br />

coeff(Q,z<br />

x j := x j + w j−1 )<br />

i D<br />

In section 4.3.2 we will want to solve a slightly different system. Algorithm 43<br />

solves a system <strong>of</strong> the <strong>for</strong>m<br />

x 1 + k 1 x 2 + k1x 2 3 + · · · + k1 n−1 x n = w 1<br />

x 1 + k 2 x 2 + k2x 2 3 + · · · + k2 n−1 x n = w 1<br />

(A.10)<br />

.<br />

x 1 + k n x 2 + knx 2 3 + · · · + kn n−1 x n = w 1

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

Saved successfully!

Ooh no, something went wrong!