“JUST THE MATHS”

UNIT NUMBER

14.9

PARTIAL DIFFERENTIATION 9
(Taylor’s series)
for
(Functions of several variables)

by

A.J.Hobson

14.9.1 The theory and formula

14.9.2 Exercises
UNIT 14.9 - PARTIAL DIFFERENTIATION 9

TAYLOR’S SERIES FOR FUNCTIONS OF SEVERAL VARIABLES

14.9.1 THE THEORY AND FORMULA

Initially, we shall consider a function, f (x, y), of two independent variables, x, y, and obtain
a formula for f (x + h, y + k) in terms of f (x, y) and its partial derivatives.

Suppose that P,Q and R denote the points with cartesian co-ordinates, (x, y), (x + h, y) and
(x + h, y + k), respectively.

y
6 R

P Q -x
O

(a) As we move in a straight line from P to Q, y remains constant so that f (x, y) behaves
as a function of x only.

Hence, by Taylor’s theorem for one independent variable,

h2
f (x + h, y) = f (x, y) + fx (x, y) + fxx (x, y) + . . .,
2!

∂f ∂2f
where fx (x, y) and fxx (x, y) mean ∂x
and ∂x2
respectively, with similar notations encountered
in what follows.

In abbreviated notation,

h2
f (Q) = f (P) + hfx (P) + fxx (P) + . .
2!

1
(b) As we move in a straight line from Q to R, x remains constant so that f (x, y) behaves
as a function of y only.

Hence,

k2
f (x + h, y + k) = f (x + h, y) + kfx (x + h, y) + fxx (x + h, y) + . . .;
2!

or, in abbreviated notation,

k2
f (R) = f (Q) + kfy (Q) + fyy (Q) + . .
2!

(c) From the result in (a)

h2
fy (Q) = fy (P) + hfyx (P) + fyxx (P) + . . .
2!

and

h2
fyy (Q) = fyy (P) + hfyyx (P) + fyyxx (Q) + . . .
2!

(d) Substituting the results into (b) gives

1 h 2 i
f (R) = f (P) + hfx (P) + kfy (P) + h fxx (P) + 2hkfyx (P) + k 2 fyy (P) + . .
2!

It may be shown that the complete result can be written as

!
∂ ∂
f (x + h, y + k) = f (x, y) + h +k f (x, y)+
∂x ∂y

!2 !3
1 ∂ ∂ 1 ∂ ∂
h +k f (x, y) + h +k f (x, y) + . . .
2! ∂x ∂y 3! ∂x ∂y

2
Notes:

(i) The equivalent of this result for a function of three variables would be

!
∂ ∂ ∂
f (x + h, y + k, z + l) = f (x, y, z) + h +k +l f (x, y, z)+
∂x ∂y ∂z

!2 !3
1 ∂ ∂ ∂ 1 ∂ ∂ ∂
h +k +l f (x, y, z) + h +k +l f (x, y, z) + . . .
2! ∂x ∂y ∂z 3! ∂x ∂y ∂z

(ii) Alternative versions of Taylor’s theorem may be obtained by interchanging x, y, z... with
h, k, l....

For example,

!
∂ ∂
f (x + h, y + k) = f (h, k) + x +y f (h, k)+
∂x ∂y

!2 !3
1 ∂ ∂ 1 ∂ ∂
x +y f (h, k) + x +y f (h, k) + . . .
2! ∂x ∂y 3! ∂x ∂y

(iii) Replacing x with x − h and y with y − k in (ii) gives the formula,

!
∂ ∂
f (x, y) = f (h, k) + (x − h) + (y − k) f (h, k)+
∂x ∂y

!2 !3
1 ∂ ∂ 1 ∂ ∂
(x − h) + (y − k) f (h, k) + (x − h) + (y − k) f (h, k) + . . .
2! ∂x ∂y 3! ∂x ∂y

This is called the “Taylor expansion of f (x, y) about the point (a, b)”

3
(iv) A special case of Taylor’s series (for two independent variables) is obtained by putting
h = 0 and k = 0 in (ii) to give

! !2
∂ ∂ 1 ∂ ∂
f (x, y) = f (0, 0) + x +y f (0, 0) + x +y f (0, 0) + . . .
∂x ∂y 2! ∂x ∂y

This is called a “MacLaurin’s series” but is also the Taylor expansion of f (x, y) about
the point (0, 0).

EXAMPLE
 
π
Determine the Taylor series expansion of the function f x + 1, y + 3
in ascending powers
of x and y when

f (x, y) ≡ sin xy,

neglecting terms of degree higher than two.

Solution

We use the result that

! !2 
π π ∂ ∂ π 1 ∂ ∂ π
      
f x + 1, y + = f 1, + x +y f 1, + x +y f 1, + . . .,
3 3 ∂x ∂y 3 2! ∂x ∂y 3

√
in which the first term on the right has value 3/2.

The partial derivatives required are as follows:

∂f π π
≡ y cos xy giving − at x = 1, y = ;
∂x 6 3

∂f 1 π
≡ x cos xy giving at x = 1, y = ;
∂y 2 3

√
∂2f 2 π2 3 π
≡ −y sin xy giving − at x = 1, y = ;
∂x2 18 3

4
 √
∂2f 1 π 3 π
≡ cos xy − xy sin xy giving − at x = 1, y = ;
∂x∂y 2 6 3

√
∂2f 3 π
2
≡ −x2 sin xy giving − at x = 1, y = .
∂y 2 3

Neglecting terms of degree higher than two, we have

√ √ 2 √ ! √
3 π 1 3π 2 1 π 3 3 2
sin xy = + x+ y− x + − xy − y +. . .
2 6 2 36 2 6 4

14.9.2 EXERCISES

1. If f (x, y) ≡ x3 − 3xy 2 , show that

 
f (2 + h, 1 + k) = 2 + 9h − 12k + 6 h2 − hk − k 2 + h3 − 3hk 2 .

2. If f (x, y) ≡ sin x cosh y, evaluate all the partial derivatives of f (x, y) up to order five at
the point, (x, y) = (0, 0), and, hence, show that

1 3  1  5 
sin x cosh y = x − x − 3xy 2 + x − 10x3 y 2 + 5xy 4 + . . .
6 120
3. If z is a function of two independent variables, x and y, where y ≡ z − x sin z, evaluate
all the partial derivatives of z(x, y) up to order three at the point, (x, y) = (0, 0), and,
hence, show that

z(x, y) = y + xy + x2 y + . . .