ECE210a_lecture9_small

ECE210a_lecture9_small - Gram-Schmidt Gram-Schmidt example...

Info iconThis preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Gram-Schmidt Gram-Schmidt example Consider finding an orthonormal basis from the vectors, x 1 = 1 1 × 10 − 3 1 × 10 − 3 , x 2 = 1 1 × 10 − 3 , x 3 = 1 1 × 10 − 3 , using 3-digit arithmetic. bardbl x 1 bardbl = radicalbig 1 + 1 × 10 − 6 + 1 × 10 − 6 = 1 , so q 1 = x 1 ; Now, q T 1 x 2 = bracketleftbig 1 1 × 10 − 3 1 × 10 − 3 bracketrightbig 1 1 × 10 − 3 = 1 . To find the direction of the second orthonormal vector, ˆ q 2 = x 2 − ( q 1 , x 2 ) q 1 = 1 1 × 10 − 3 − 1 × 1 1 × 10 − 3 1 × 10 − 3 = − 1 × 10 − 3 . Roy Smith: ECE 210a: 9 .2 Numerical aspects t-Digit Arithmetic. Examining numerical aspects. Given x ∈ R , define the t-digit floating point, representation, f t ( x ) = 0 . d 1 d 2 . . . d t × 10 e , where the digits, d i , and the exponent, e , minimize | x − f t ( x ) | . If f t ( x ) is not unique, round away from zero. Unfortunate properties 1. f t ( x + y ) negationslash = f t ( x ) + f t ( y ). 2. f t ( xy ) negationslash = f t ( x ) f t ( y ). Consider 2-digit arithmetic with x = 21 / 2 and y = 11 / 2. Roy Smith: ECE 210a: 9 .1 Gram-Schmidt Floating arithmetic example: Three digit arithmetic gives: q 1 = 1 1 × 10 − 3 1 × 10 − 3 , q 2 = − 1 , q 3 = − . 709 − . 709 bracehtipupleft bracehtipdownrightbracehtipdownleft bracehtipupright . These two vectors are at 45 degrees! Definitely NOT orthogonal. The problem arises from the fact that the first component in each vector dominates the differences in the second two components in the calculation. Modified Gram-Schmidt In the Gram-Schmidt procedure we have, q k = ( I − Q k − 1 Q ∗ k − 1 ) x k vextenddouble vextenddouble ( I − Q k − 1 Q ∗ k − 1 ) x k vextenddouble vextenddouble , Q k − 1 = bracketleftbig q 1 . . . q k − 1 bracketrightbig , Q = 0 . Consider expressing the calculation of q k as a product of matrix operations. Define the projections, P ⊥ = I, P i ⊥ = I − q i q ∗ i ←− removes the part aligned with q i Roy Smith: ECE 210a: 9 .4 Gram-Schmidt Normalizing ˆ q 2 gives, q 2 = − 1 ....
View Full Document

This note was uploaded on 04/06/2010 for the course ECE 210a taught by Professor Chandrasekara during the Spring '08 term at UCSB.

Page1 / 8

ECE210a_lecture9_small - Gram-Schmidt Gram-Schmidt example...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online