Вы находитесь на странице: 1из 5

Appendix A

Matrix Algebra Review

1. transpose A T
2. multiplication A B B A
(must be conformable)
3. addition A + B
4. x T x x T y xx T vector-matrix products that are scalars
T
x A x
-1
5. A A =I inverse

8 5 1 0
use row operations
3 2 0 1
6. det A (det = 0 singular matrix)

7. A x=b linear algebraic equations


-1
x=A b

8. Linear independence, row operations

9. eigenvalues
det (A - e I) = 0

solve for e

2 0 1 2
A , A
0 3 2 1

eigenvectors A v = e v
10. Hessian matrix fx x = H
2f
aij =
xix j

11. Jacobian matrix gx


gi
x j

12. Singular values and condition number


- find eigenvalues of A T A
- compute si = ei 1/2 (si > 0)
- condition number is ratio of largest si to smallest
non-zero si
- indicates ill-conditioning that can lead to numerical errors
in equation solving due to roundoff
Newton's method
h(x) = 0

scalar x k+1
x =
k
-h xk x k
dh
dx xk

h (x) = 0

vector
-1
k dh
x = h x
dx xk
k

(solve simultaneous nonlinear algebraic eqns)


Least Squares Data Fitting

model: Y=x+ (2.6)


min f = = (Y - x)T (Y - x)
T

1 x11 x12 x1p


Y1

2 Y2 .
= Y nx1 = .
px1
x

nxp


p Yn .
x x np
n1 x n2
T T
x x=x Y
T T
=(x x)-1 x Y

(Solve set of linear algebraic equations)

Вам также может понравиться