1 Eng Completare 05 2016 Differential Several Var

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Directional Derivative

In this section we shall consider real functions f defined on an


arbitrary open set G Rn .
Let s = (s1 , . . . , sn ) Rn , s 6= 0.

D 12 The directional derivative of a function f


at a point x G in the direction s (along s) is defined as the
limit
f (x + ts) f (x)
lim ,
t0 tksk
provided that it exists, and is denoted by
df f
(x) or (x).
ds s
w
21 / 33
T 13 If a function f : G R possesses continuous partial
derivatives at x G, then it has a derivative
along any unit vector s Rn given by
f
(x) = s f (x)
s

 Folosind formula (1), avem:

d (f (x + ts) f (x))
f f (x + ts) f (x)
(x) = lim = lim dt
s t0 t t0 1
   
= lim (s ) f (x + t s) = (s ) f (x)
t0

= s (f (x)) "
22 / 33
R 14 Using the notations = (1, 0) and = (0, 1), we have
f f f f
= , = .
x y

23 / 33
Lagranges Mean-value Theorem
Let D Rn be a convex set and a = (a1 , . . . , an ),
b = (b1 , . . . , bn ) be distinct points in D.

T 15 If f : D R possesses continuous partial


derivatives on D, then D such that

f (b) f (a) = (b a) f ()

 Define the auxiliary function F : [0, 1] R,


F (t) = f (a + t(b a)). By Lagranges mean-value theorem
for functions of one variable, c (0, 1) such that

F (1) F (0) = F 0 (c).

w
24 / 33
But
F (0) = f (a), F (1) = f (b)
and using Eq.(1), we get:
!
F 0 (t) = (b a) f a + t(b a) .

For = a + c (b a), we obtain

f (b) f (a) = ((b a) )f () = f () (b a) "

25 / 33
Differential of a Function. Definition
Let U and V be two normed linear spaces.
D 16 A function f : U V is said to be
Frechet differentiable at a point x0 U if there exists a continuous
linear mapping T : U V such that
kf (x) f (x0 ) T (x x0 )k
lim = 0.
xx0 kx x0 k

The functional T is called the Fr


echet differential of f at x0 ,
and is denoted by

df (x0).

w26 / 33
Differential of Functions of Several Variables
Let D Rn , a int(D) and f : D R be a continuous
function.
T 17 If f has continuous partial derivatives at a,
then it is differentiable at a and
df (a)(h) = (h ) f (a) = (f (a)) s, h Rn .

27 / 33
Using
df (a)(h) = (h ) f (a)
and taking
h = dx := (dx1 , . . . , dxn ),
we obtain
df (a)(dx) = (dx )f (a)
or n
X f
df (a)(dx) = (a) dxi ,
i=1
xi

28 / 33
which, with a slight abuse of notation, is written in the form
n
X f
df (a) = (a) dxi .
i=1
x

Similarly,

dk f (a) = (dx )k f (a)


 k

= dx1 + + dxn f (a). "
x1 xn

29 / 33
E 18 If f : R2 R has continuous partial derivatives of the second
order, then  2
2
d f = dx + dy f
x y
2f 2 2f 2f 2
= 2
dx +2 dx dy + 2
dy .
x xy y

R 19 The Lagrange mean-value formula can be written in the form

f (b) f (a) = df ()(b a)

w
30 / 33
Taylors Formula for Functions of Several Variables
Let f : Rn R be a function possessing m + 1 order
continuous partial derivatives in a convex neighborhood V of a
point a Rn .

T 20 (TAYLOR) For any x V V,


ka k < ka xk, s.t.
 m
(x a) f (a) (x a) f (a)
f (x) = f (a) + + +
1! m!
m+1
(x a) f ()
+ .
(m + 1)!

w
31 / 33
 Define the auxiliary function F : [0, 1] R,
F (t) = f (a + t(x a)), F (1) = f (x), F (0) = f (a),
0
 0

F (t) = (x a) f (a + t(x a)), F (0) = (x a) f (a),
................................................................
(k)
k 0
k
F (t) = (x a) f (a + t(x a)), F (0) = (x a) f (a),
Taylors formula for F gives
F 0 (0) F (n) (0) F (n+1) ()
F (1) = F (0) + 1!
+ + n!
+ (n+1)!
, 0 < < 1,
i.e.,
 m
(x a) f (a) (x a) f (a)
f (x) = f (a) + + +
1! m!
m+1
(x a) f ()
+ ,
(m + 1)!

where = a + (x a).

w
32 / 33
E 21 For n = 2, m = 1, Taylors formula becomes
f f
f (x, y) = f (a, b) + (x a) (a, b) + (y b) (a, b)
x y
 
1 2
2 f 2f 2
f
+ (x a) (, ) + 2(x a)(y b) (, ) + (y b)2
(, ) .
2 x2 xy y 2

James
GREGORY
(16381675)
a Scotish
mathematician
(he discovered
Taylor expansions
more than 40 years
" " before Taylor !!!);
33 / 33

You might also like