11 - Multiplying a matrix by its inverse should give the...

This preview shows pages 1–4. Sign up to view the full content.

2/12/2010 1 Multiplying a matrix by its inverse should give the identity matrix: A = 150 -100 0 -100 150 -50 0 -50 50 >> Ainv = inv(A) Ainv = 0.0200 0.0200 0.0200 0.0200 0.0300 0.0300 0.0200 0.0300 0.0500 >> format long >> test = A * Ainv test = 1.000000000000000 0 0 -0.000000000000000 1.000000000000000 0 0 0 1.000000000000000 >> test == eye(3) ans = 0 1 1 0 1 1 1 1 1 The answer is close, but not quite right… Some calculator exercises: Evaluate (1 + 1e 13) 1 The answer should be 1e 13 but is zero…. Evaluate (1e13 + 1)–1e13 The answer should be 1 but is zero …. Evaluate (1 + 4e 11 + 8e 11)–1 The answer should be 1.2e 10 but is zero (Casio fx 991) …. . Evaluate (4e 11 + 8e 11 + 1)–1 The answer should be 1.2e 10 and is (Casio fx 991) …. . All these issues are a result of the way in which values are represented within computers (and within calculators).

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

View Full Document
2/12/2010 2 +/ 1 0 0 0 0 . 1 1 Mantissa Exponent Sign Th t i td i 49” ft A Pseudo IEEE Standard Floating Point Format The exponent is stored in “excess 49” format. Stored exponent = actual exponent + 49 Actual exponent = stored exponent–49 A stored exponent value of 99 is reserved for special cases (infinity, not a number). Possible actual exponent values range from 49 to +49. Except when the actual exponent is 49, values are always normalized (leftmost mantissa digit is non zero). Very realistic apart from base 10 rather than base 2. +/ 9 9 9 9 9 . 9 8 Largest possible magnitude = 9.9999 x 10 49 +/ 0 0 0 0 1 . 0 0 Smallest possible magnitude = 0.0001 x 10 49 = 1 x 10 53 (note–only 1 significant digit) +/ 1 0 0 0 0 . 0 0 Smallest possible magnitude with five significant digits = 1.0000 x 10 49
2/12/2010 3 Values 0.0001 x 10 49 = 1x10 53 apart Values 0.0001 x 10 48 = 1x10 52 apart Values 0.0001 x 10 47 = 1x10 51 apart “Subnormals” (loss of precision) zero 1.0000 x 10 48 1.0000 x 10 49 1.0000 x 10 47 Rounding A number of strategies are possible: Round to nearest, ties to even Round to nearest, ties away from zero (ties round up) Round towards zero (truncation) Round towards + Round towards ‐∞ First possibility is the default behaviour for computers.

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

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

This document was uploaded on 04/14/2010.

Page1 / 11

11 - Multiplying a matrix by its inverse should give the...

This preview shows document pages 1 - 4. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online