Anda di halaman 1dari 2

APPENDIX C

DIFFERENTIATION WITH
RESPECT TO A VECTOR

The first derivative of a scalar-valued function f (x) with respect to a vector


x = [x1 x2 ]T is called the gradient of f (x) and defined as
 
d f/x1
f (x) = f (x) = (C.1)
dx f/x2

Based on this definition, we can write the following equation.


 
T T y
x y= y x= (x1 y1 + x2 y2 ) = 1 = y (C.2)
x x x y2
 
T 2 x
x x= (x1 + x22 ) = 2 1 = 2x (C.3)
x x x2

Also with an M N matrix A, we have

T T T
x Ay = y A x = Ay (C.4a)
x x
T T T
y Ax = x A y = AT y (C.4b)
x x

where

M 
N
xT Ay = amn xm yn (C.5)
m=1 n=1

Applied Numerical Methods Using MATLAB, by Yang, Cao, Chung, and Morris
Copyright 2005 John Wiley & Sons, Inc., ISBN 0-471-69833-4
471
472 DIFFERENTIATION WITH RESPECT TO A VECTOR

Especially for a square, symmetric matrix A with M = N , we have

T if A is symmetric
x Ax = (A + AT )x 2Ax (C.6)
x
The second derivative of a scalar function f (x) with respect to a vector x =
[x1 x2 ]T is called the Hessian of f (x) and is defined as
 2 
d2 f/x12 2 f/x1 x2
H (x) = 2 f (x) = f (x) = (C.7)
dx2 2
f/x2 x1 2 f/x22

Based on this definition, we can write the following equation:

d2 T if A is symmetric
2
x Ax = A + AT 2A (C.8)
dx
On the other hand, the first derivative of a vector-valued function f(x) with
respect to a vector x = [x1 x2 ]T is called the Jacobian of f (x) and is defined as
 
d f1 /x1 f1 /x2
J (x) = f(x) = (C.9)
dx f2 /x1 f2 /x2

Anda mungkin juga menyukai