5 PARTIAL DIFFERENTIATION

I N T R O D U C T I O N

Many applications require functions with more than one variable: the ideal gas

law, for example,

is pV=kT where p is the pressure, V is the volume, T is the

hsolute temperature of the gas, and k is

absolutetemperature a constant. Rearranging this equation as
Kshows that p is a function of T and V. If one of the variables, say 7, is
P
nt fixed and Vchanges, then the derivative ofp with respect to V measures the
is called the
rafe of change of pressure with respect to volume. In this case, it
nartial derivative of p
with respect to Vand written as

LEARNING OUTCOMES

At the end of this chapter you will be able to:

Manipulate partial derivative.

Perform second partial derivatives.

Use chain rule case 1 and case2.
multivariable.
Perform implicit differentiation for extrema, absolute
critical points, local
ldentify the extreme values using
extrema and Lagrange multipliers.

5.1 Partial Derivative

we treat y as a
two variables.
lf
f (x, v) of
Consider function called the partial
Deninition: a
The result is
with respect to
r.
Const differentiated
l d , may be let z
=f (x,y),
o r . I fwe
to x and is denoted by f,
Vative offwith respect

19
PartiadDflerentuatien

with respect to pis

The partial derivative lauly deli
Wewrile,
and differentiating /( y) with respect sto y.
respert
constant
as
treating r

Example 5. 1

derivatives of fr,y)=x'+*'y-2
Find the partial

Solution

.r.y)=3r+ 21
S.(.y)=2x'y-4y

5.1.1 Differentiability

A function is differentiable if it possesses the property of local linearity, For

functions of two variables this means we can accurately
approximate the surface
uSing a tangent plane.

Definition: A function f(x, y) is differentiable the

at
point (x,,y,)if
Ox
-(Vo) and ,y,) exist and Af can be written as
oy

Af=,:y,
Ox )Ar +ro.y,)Ay
Oy 6,Ar+6,Ay +

where &, >0 as

(Ar, Ay) -> (0,0).
First partials

ofX,y)= lim h.)-f(1,)

OX h->0

(T,) =lim
jx, y+h)- f(x.
h-0
h
y)

120
 Partial Differentiation

interpret partial derivatives as rates of Ifz= fx. y), then z/ôx

change.
an
He e rate of change of 2 with respect to x when y is fixed. Similariy
r e y r c s e n

re the rate
rate of
of change ofz with respect to y when x is fixed.
/yisthe

y) has continuous first

partial derivatives in a neighbourhood of
m:
T h e o r e m : Iff(x,
)then f(. y) isdifferentiable at the point (Xg+Yo).

Theorem: Differentiability=> continuity.

Example 5.2

Iff(ry)=Xy te cOsy, computef, and f.

i) fas in (), calculatef (1, #2).

For
Solution
we get
Treating y as a differentiating with respect to x,
constant and
a
y. Differentiating with respect y and considering x as
f(x. y) =_yte'cos
constant givesfx, y) =*-e'sin y.
e'cos(72) =
w2.
=
I and y =
m2, we getf(1,72) =w2 +
(i) Substituting
x

Increments

is the actual
is called the of fand inerement
Af= f(x+ Ar, y Ay)- f(t,y) +
is moved to (r+Ax,y +Aw).
change in the function fas (x,y)

Total differentials
differential offand is the tangent plane
df=Ar+ Ay is called the total
OX Oy The
is moved to (x+Ar, y+Av).
approximation to the change in / as (x,y)
for the
total differential is actually the equation
=/,(x. y) dr + f (x, y)dy
coordinates centered at
the point of tangency.
dngent plane in local
definced by the equation
) is
The
Ctotal differential of , y.
and are independent of each
d e + d +dt whether
or not . y z

y are
continuous,

ihprovided onlv that the partial derivatives involved

121
P a r t i a l
D i f e r e m i a t i o n

Example 5.3

=r +ity-y, find the differentia dz.

the differential
If:=fy)
from 2 to 2.05 andy changes from
from 3 to
2.96,
Ifr changes
)
value ofAz
and d. compare
Solution

( td+d=(2r+3y)dr+(3r-2))d
Ox

Puttingx = 2, dr = Ar = 0.05, y = 3, and dy = Ay = -00A

04. we ge
de=[2(2)+3(3)]0.05+[B(2)-2(3)-0.04) =0.65
The increment of z is

Az-f(2.05,2.96) -f(2,3)

[(2.05+32.05)(2.96)-(2.96)*]- [2 +3(2)(3)-3
= 0.6449

Notice that Az= dz but dz is easier to

compute.
5.2 Higher Derivative

5.2.1 Second
partial derivative
If z=
f(x, y), we use the
following notations.

aaff.
Ox Ox Ox
Cr Cy Cx
( -f
y x ycx
 Partial Diferentiation

Theorem: SupDose f is defined on a disk D that

contains the
and a r e both continuous on D,points
Clairut's
If the
she functions
functions
(4,b) then
I_8'

(a.b)=1,(a,
b) or

oyOx OxOy
or
= i fthe derivatives involved
.

are continuous.

Example 5.4

derivatives of f(r,y)=x+ry'-2y
ind the second partial
Find

Solution

f 2x+ 2.xy , =3x*y-6y

S, =6x'y-12y
f=2+2

J =6xy?

5.3 The Chain Rule

5.3.1 The chain rule (case 1)

is differentiable functions ofx and y, wherea
variable z =f{x, y) a
Then z is
One independent
both differentiable functions of t.
r =gt) and y
=
h{t) are

f dy
function off anddz
=J
dt oy dt
,

differentiable
dt Oxr

Example 5.5
du
find
x =e'cos 1, y=e sin 1, dt
If u=x -and

Solution

du u dr du dy
d x dt 0y di
(-2r)(e'sin/ +e'cos
t
sin1)+
2x(e cost-c'
1)-2esin 1(e sin14+ecOst)
2e cos(e'cost-e'
sin
123
Partial Differentiation

=e"2cos l-2costsint-2 sin"

/-2sin t cos
=2ecos -sin't-sin 21
2e(cos 2-sin 21).

5.3.2 The chain rule (case 2)

Two independent variables z = fr. y) is a differentiable functione
of x
where r =
g(s. 1) and y =h{s, 1) are both differentiable functions of s and
and y,
is a differentiable function of t and i. Then:

Cs Ox Os oy os

Example 5.6

Ifz= e siny, where x=st and y =st, find 0zl Os and Ôz/Ot.

Solution
Since the chain rule
gives
Oz oz Oy Oz z ox
OS ox s y s and Cz Oy
ô ax ot y Ot
we calculate the derivatives:

=
e sin Ox
y
Ox
OS =e cos y 2st
Oy
Oy
2st OX
OS 2s
yS
Ot
 Partial Differentiation

= ( e ' sin y\r)+(e' cos y)(2s/)

-resin(T) +2. cos(s°r)

Cx = (e' siny)(2st) + (e' cos y)(s*)

sin(s7)(2s1) +s'e cos(s't)

se" cos(s7).
2ste sin(sT)+
4 Implicit Differentiation

Example5.77

6ry.
Find if +"=

Solution

with
Finding means thaty Is a tunction ofr: that isy=fx). Differentiating y

TOspect to r gives

-3y =6 I.1-
dr
PartialDifterentiation

Example 5.8

Findandif r +y' +z' +6xyz =1

Solution

Findingand
Cx
means that z =f(x, y). erentiatingz with
we have
respect to

3 +0+3+612+x=0
Ox

x++2 I+=0

+2x=--2yz
Ox

0z +2yz
Ox
Similarly, +2.xy

_y+2xz
5.5 Extreme Values 2xy
Partial derivatives
just as derivatives can be used to
You could be used tofind the extrema of
should
variables. be able of functions of two
to find extrema
find extreme
and saddle functions of a
single variabi
points of
functions variad
of

126
 Partial Diferentiation

5.5.1 Critical point

Definition: A point (»Va) in the domain of the function fx, y) is called a

eritical point if

)
a )= oy )=0

(i) ) or -(x.) is undefined

(ii) (ro. Yo) is
a boundary point

Example 5.9

2r + 3y-4r +3y + 5.
Determine the critical pointfor flx, y)
=

Solution

First we need to fînd f(xy) and f,(x,y).

S,,y)=4r-4 S,(x,y)=6y+3

Now we find where , (x, y) = 0 and ,(x, y)=0.

0=6y+3
0 4x-4

is (1.-).
So the critical point for f(r.y)=2x
+3y-4x +3y +5

5.5.2 Local extrema

is open disk R
of/ if there
an
local maximum
We call/(a, b) a we
for all (r, 1) e R. Similarly,
Definition:

which f(a, b) 2f(r. y)

Centered at point (a, b) for /if thcre is an open
disk R centered al point
Cal f (a. h) a local minimum of is called
R. In either case, /(a. h)
for all (r. J) e

D 1or which f(a. h) sf(x, 1)

a local extremum.
 if they are local
t o sce
maximu
P R m t0 o r v i t i a t t o

ce
points deri
crifical
smooth
version
of the s e c o n d ve test.
to test
two-dimensiOnal
presihle
t 5s
a
m n i m u n s
using

D e r i v a t i v e
Test)
(Second
y) are
continuous in
Theorem

derivatives
o1 x,
secomd partial
Sugypoe

neighhourthood
the

of the point
(,X,)
and
(XYo)= ôy ,)=6
Define

(40,o) ycr
p L
=
o o ) /y,(a o)-|J, (p.-y)
(p-Jo)

i) If D> 0 and . Jo) <0 or

o»o)<0 then f(x,,V. a
is a

relative or local maximum.

) If D> 0and ,Y,)>0 or

Sao»yo) >0 then
relative or local minimum.
f(x,,y, )is a

ini) If D<0 then f(xg, Y,) is a saddle point.

(iv) If D=
0 then the test is inconclusive or
fails.
Example 5. 10
Find all
relative extrema
4x+3y+5. and/or saddle
points for f(.
v)=X -
Solution
First weneed to
That is we find the critical
need to find point(s) for
where f (r,
1) =
0
/(x. v)= 3 4r +31+
and f.(x, T) =0.
 Partial Differentiation

Ar-4
I,a.y)= -6y+3
117=,
,
0=4r-4 and 0=-6y +3

TPhs,t n ec
ical point for the given function is at |-|.
' rlical

Todelcmune i t .
:f any relative extrema or saddle points exist we need to find the
artial derivatives.

s m dp r a r t i

(x.y)=4. J, ,y) -6, 1, (T,y) = f,(x,y)=0

evaluate
need to
We
O

D=f.(a,b) S,(a,5)-[S,(a,b)
= (-4)-(-6)-0

=24.

=-4 <0 we conclude that a relative maximum exists

Cinre D>0and s(a,b)
s--L)=(-1,7,7.75).

Example 5.11

Findallrelative extrema
for f(r.y) =y' -x'.

Solution
maximum
<0 but it is not a
saddle point because D
* -r'at (0,0) is
,)= a

OI minimum.

S5.3 Absolute or global extrema

O:We call f(a, b) the absolute

maximum of fiff(a, b) 2f(r, y) forb)all
E
offiff(a,
the absolute minimum
E domain. Similarly. we call f(a, b)
orall ()edomain.

120
 Partial Differentiation

Theorem
Theorem: Extreme
Value

bounded region
closed and
D.
a
continuous on
Suppose that fr, v)
has both an
is
absolute maximum
and a b s o l u t e m i n i m u m on R.
.Then
values of a
function can only ocur at
ical po
The extreme
Theorem:
anywhere clse.
they cannot o c c u r

Extreme value critical point

cxtreme value
Critical point

Example 5.12

function /(x,y)
= ar +2jy" -4y on therei
regon
Find the absolute extrema ofthe
R:r.)y2x' and y s4

Solution

The first thing to do is find any critical points.

6x >* = 0
f(x,y) =

=1
f(x,y) 4y -4>4y =4~y
=

So we find a critical point at (0, 1). Is this in the region?

su
region. the
Yes, sokeep it. Incidentally, by the D-test, this is a minimum on
what we are looking for on the boundary will be a maximum. If we had n

than one point or a saddle point, we could not conclude this.

Now we need to test the boundary conditions. One boundary is at the line
Put this value into thefcquation and reduce to just a function of N.

(x,4) 3x+2(4)-4(4) 3r +32-16=3r +16

r=0

Taking the first derivative, we get/ ' r . 4) - 6r, which has a eritical pom
So have the
now we
point (0, 4) to test later.

The sccond
boundary is y=x, so makc this replacement.
f(x.x)=3r +2(r )-4()=31 +2r - 4r2
 Partial Differentiation

aking the first derivative, we get 'x,r)=8x -2x =2x(4x-1). This

Gunction has critical points at x = 0: (0,0), and at (1- 1 The
hsintersect at (2, 4), and (-2, 4). This gives us, finally, seven points to test
function.
in the original

f(0,1) =3(0) +2(1-4(1)=2-4= -2

0,4) =3(0) +2(4) -4(4) =32-16=16
f(0,0)=3(0) +2(0) -4(0) =0

S(2.4)=3(2) +2(4)* -4(4) =12+32-16=28

f-2.4) 3(-2) +2(4) -4(4) 28
= =

the minimum on the region at is the absolute minimum.

(0, 1, -2)
As expected,
maximum on the region occurs at the intersections of the
And the absolute
houndary conditions, (2, 4, 28) and (-2, 4, 28). Since the two values are the
ame, both are absolute maximums.

5.6 Lagrange Multipliers

subject to the constraint
To find the maximum and minimum values of fx, y)
=c (assuming these extreme
values exist).
gx,y)
. Find all values ofx,y and 7 where is a scalar such that

f. =ig, (1) F(x,y)=S(x,y)-2g(x.y) V .y) = AVg(x, )

and gr, y) =c
,=ig, (2) F =0 (1)
(2)
glr,y)=c (3)F =0
Notice that
gx, y) = c (3)

dx dy
or

Vg dei dg

131
 Partial Differentiation
2 Evaluatef(x. F) at all of the points found
is the maximum value
in
(). Thc
of f ( . )and the small largest of thes
vaiue off (a. y). est value 1s
the r
Example 5.13
Find the smallest value of r v subject to the constraint
Solution
33
Using Lagrange multipliers, we havc
f =g (1)
f.is (2)
glx. v} c(3)
This gives us the three cquatsons
2x-3 (1)
2y- (2)
3x3(3)
Solve for x and y in (I) and (2).
respetIveiy, w
Plugging these into (3).
we have
Plugging in . 3
nto
into our
ou
cquations aaboye