CHAPTER

19

Extrema of Function of
Several Variables

@Introduction
chapter we have alr eady derived the extrema (i.e., maxima or minima) of a
previous
variable. In the present chapter we will derive the extrema (both maxima
ctionofsingle
function of multi variable (mainly two or three variables). Further in this
minima)of a
find the extrema of a function of more than one variable subject to one
weintend to
chapter
functionalrestriction(s) by Lagrange's multiplier method.
more
or
ExtremePoint of a Function of Two Variables
19.1
for a function f!x, y) to have extreme point(s) at (a, b) is the first
Thenecessarycondition
order partialderivative should vanish at (a, b) [i.e.,rx (a, b) = 0 = fy (a, b)}. Now an extreme
point(a, b) of a function does not imply that the function is minimum (or maximum) at
(a,b). Below we state
the working rule to determine extrema as well as possible maxima or
of a function of two variables.
minima
Working Rule
Letttx,y) be a function of two variables and our objective is to find extrema of ftx, y).
Step-I : First solve the equations fx(x, y) = 0 and fy (x, y) = 0 to obtain the extreme points
andlet (a, b) be one such extreme point.
Step-2 : Find Hessian matrix H(x, y) = fxxfyy—(fxy)2.
Step-3: (i) If H (a, b) > 0, then f(x,y) is minimum at (a, b) if either fxx(a, b) > 0 or
G (a, b) > 0 and fix, y) is maximum at (a, b) if either (a, b) < 0 or f», (a, b) < 0.
(ii)IfH(a, b) < 0 then (a, b) is a saddle point of the function flx, y).
(iii)IfH (a, b)= 0, then we cannot draw any conclusion and further investigation is required.
Note

It shouldbe noted that instead of finding H(a, b) we may find d 2f (a, b), where
d2f= —dt+—dy /(x, y).
Ox DY
Inthiscase, in step-3, H(a, b) should be replaced by d2f (a, b) to check the nature of the
point(a, b).

[4031
 404 | Fundamental Engineering Mathematics
Working Rule for Function of Three Variables
Let flx, y, z) be a function of three variables x, y, z and our objective is to find the
possible
maxima and minimum of the function flx, y, z).
• Step-I : Solve the equations fr (x, y, z) = fy (x, y, z) = fz (x, y z) = O. Let (Q,b, c)
solution of the above systems so that (a, b, c) is an extreme point of the function Rx
• Step-2 : Find the value of d2f(a, b, c), where YR).

d2f= —dx+ —dy+—dz f

a2f dx2 + —idy 2 + —dz 2 +2 — dx dy + —dxdz +

Dy ö2z axaz ayöz
¯ ax2
+ Step-3 : (i) Ifd2f(a, b, c) is negative definite, then fix, y, z) is maximum at (a, b, c)
(ii) Ifd 2f(a, b, c) is positive definite then fix, y, z) is minimum at (a, b, c).
(iii) No conclusion can be done if d2f (a, b, c) is neither positive definite nor negative
definite.
Note

It should be noted that d2f (a, b, c) is a positive definite if fxx (a, b, c) 0

fxx(a, b, c) f (a, b, c) —{f (a, b, > 0 where as d 2f (a, b, c) is a negative definite.
(a, b, c) < 0, fxt (a, b, c) (a, b, c) —G (a, b, > 0.

Constrained Extremization : Lagrange's Multiplier

In this section we will introduce the Lagrange's multiplier technique to find the extreme
points of a function subject to an additional constraint.
Working rule to find extreme points of a function fix, y) subject to g(x, y) = c.
Step-I : To find extreme point(s) of a function subject to a constraint g(x, y) = c wefirst
form a Lagrangian L (x, y, A) = fix, y) + y) —c},
where A is known as Lagrange's undetermined coefficient.
Step-2 : Solve the equations La.= Ly = LA= O to find the extreme point(s) of the function
fix, y) [certainly LA = 0 gives nothing but the constraint g(x, y) = cl. Suppose (a, b) is one
of those extreme points.
• Step-3 : Find the Hessian matrix H(x, y) at (a, b) where H(x, y) = LEX
L,y—(Lxy) 2.
• Step-4 : (i) If H(a, b) > 0 then ftx, y) is minimum (maximum) at (a, b) if Lu (a,
b) > 0

(ii) If H (a, b) < 0, then (a, b) is a saddle point for the function fix, y).
(iii) If H (a, b) = 0, then no conclusion can be drawn regarding the maximum or
minimum
of the point (a, b).
 Extrema of Function of Several Variables 405

_constrainedoptimization technique here also we can find d2L(x,y) at (a, b) in-

the Hessian matrix H(x, y) at (a, b). In this situation (12L(a, b) should be
bikd offinding
bYH(a, b).
Ola
rule to find extrema of a function flx, y, z) subject to g(x, y, z) = c.
working
1 : To find the extreme point offlx, y, z) subject to a constraint g(x, y, z) = c we first
Lagrangian
form
z) f (x, y, z) + A (g(x, y, z) —c), where A is known as Lagrangian undetermined
ultiplier.
equations lax= Ly = 1.4= LA = 0 to find the extreme point(s) of the
step-2: Solve the
[certainly LA = 0 gives nothing but the constraint g(x, y, z) = cl. Suppose
functionfix, y, z)
b, c) is one of those extreme points.
(a,
the value of d2L (a, b, c) where
Step-3: Find
y, Z) ax Oy Oz
d2L (X,
a 2L 2 + .dz 0214 021,
L dx 2 + —dy 2 +2 — dx dy + dy dz + — dz dx
Ox2 Oy2 OxOy öyOz özDx
2 is a negative definite, then ftx, y, z) is maximum at (a, b, c).
Step-4: (i) If d L (x, y, z)
(i) Ifd2L (x,y, z) is a positive definite, then f(x, y, z) is minimum at (a, b, c).
(iii)Noconclusion can be done if d2L (a, b, c) is neither a positive definite nor a negative
definite.
Notes

1. 1 is known as Lagrangian 'undetermined' coefficient since the explicit value of A does

notrequiredto determine the extrema of the objective function.
2. Insteadof a single constraint if there are more than one constraint, say gr (x, y, z) = Cl
andg2 (x,y z), then to find extrema of the function f(x, y, z), the Lagrangian should be
as follows : L(x, y, z, Al, 12) = /(x, y, z) + Al {gl (x, y, z) —Cl) + 12 {g2(x,y, z) —c2)
Nextthen we should proceed as earlier.

O Illustration-I : Find the extreme point of the function f(x, y) = x2 + 2t + 2y + 3.

Alsodetermine the nature of the function at the extreme point.
Herethe given function is
ftx, y) = x2 + —2x + 2y + 3
Wehave, —2, fy = 2y + 2, = 2 = fyyand fxy= 0
Nowat the extreme points fx = O = f . So we have,
21—2= 0 = 2y + 2 giving x —1, y
Thus(I, —1) is the extreme point of the function flx, y).
 406 Fundamental En ineering Mathematics
Now at (1, -- 1) the Hessian matrix is
H(l, -1) = [fxxfyy- =
and fxx(l, —1)= 2 (> 0).
Hence the extreme point (1, —1)is the minimum point for the function Rx,y)
corresponding minimum value is /tl, —1)= I + I -- 2 —e
2 + 3 = I. and
O Illustration-2 : Find the minimum value ofx2 + y2+ z2 subject to the

Here our targate is to minimize the function /(x, y, z) subject to xyz = c3.
We form the Lagrangian
L(x, y, z) = x2 + Y2 + z2 + A(xyz -- c3)
Taking first and second order partial derivatives, we have, lax= 2r + 2AY2,lay= O,
14 = 2z + 2Æty. +
LA c3, 2, 2, 2 and Lay= 2Az, lox = 2Ay,
At the extreme points we have, zye2
Lx = 0 giving 2(x + Ayz) = 0 — x = —Ayz
Ly = 0 giving 26' + Axz)= 0 * y = —Axz,
and = 0 giving 2(z + Axy)= 0 z = —Axy
Multiplying above three we have,
xzy = —13 x2y2 z2 i.e., A3xyz = —1
i.e., 13 c3 = —1 giving , since A is real.
c
Therefore at extreme point
— i.e., x 2 _ xyz = c giving x = ± c.
c c
Similarly, at extreme points we have y = ± c, z = ± c.
Since xyz = c2, therefore the extreme points are given by (c, c, c), (c, —c,—c), c,c)
and (—c,c, —c).
Now it can be shown that at each of above four extreme points d 2L(x,y, z) is positive
definite. Hence the function flx, y, z) = x2 + Y2 + z2 subject to xzy = c3 is minimum at four
points namely (c, c, c), (c, —c, —c),(—c, —c, c) and (—c, c, c) and minimum value offlx,yg)
isc + c + c2 = 3c2.

Illustrative Examples
Find the extrema of Ax, y) = x8 + —63(x + y) —12ty. [WBUT 20061

Solution Here the given function is

/(x, y) = x3 + y 3 —63(x + y) —12ty
Taking partial derivatives of first and second order, we have from (1)
 Extrema of Function of Several Variables 407
3y2- 63
-- 63,
12 and fyy = 6y

t the extreme points of/tx, y) we must have,

-63 3y2- - 63
-21=0

(3) we have,
(2)and

or X+ y +4=0
Ning s —Y
x = y and using it (2) and (3) we have, x2 4x -- 21 = 0
(4)we have,
7)(x + 3) = 0 i.e., x = 3
ving(x —
we have, (—3,—3)and (7, 7) are two stationary points 18incehere x = y)
Hencefrom (4)
(5) we have y = —x 4 and using it in (2) we have,
Againfrom

'Lr—5= O giving (x l)(x + 5) = 0 i.e., x = 1, —5.

i.e.,
and (—5,1) are two stationary points, since here y = —x -- 'i.
Sofrom(5), (I, —5)
Thusfix, y) has four stationary points namely (—3,—3),(7, 7), (1,—5)and (—6,1).
we have, the Hessian matrix,
HO,y) = (G)2 = 6x•6y -- (-12)2 = 3(ity - 144 = 36(xy - 4).
—3)
NOW at
= 36 - 4) = 180 0) and fxx= 6 = 18 < O.
Sof (x,y) is maximum at (—3,—3)and
themaximumvalue is ft-3, -3) = (-3)3 + (-3) 3- 63(-3 - 3) = 405.
Againat (7, 7),
H(x,y) = 36 (7 x 7-4) = 36 = 1620 (>0) and At = 6 x 7 = 42 (>0).
Soflx,y) is minimum at (7, 7) and
the minimum value is 7) = (7)3 + (7)3 —63(7 + 7) 12 x 7 x 7 = 784.
Alsoat (1, —5) and (—5,1), we have respectively
H(x,y) I = 36 (1 x -4) = -324 (<0) and H(x, y) I 1)= 36(-5) x 1-4) = 324 (<0)
Thereforeboth the stationary points (1, —5)and (—5,1) are saddle points for the function
flx,y).

Find the extreme points of the function Ax, y) = x8 + ya — —12y+ 20 and

check the nature of those extreme points. (WBUT2001,2005,2010,2012)
Solution Here the given function is
flx,y) = x3 + y3 -3x- 12y + 20
 408 Fundamental En ineerin Mathematics
Taking partial derivatives of first and second order wc have from (1)
12,
fxx= (it, fyy= 6y and = =0
Now at the extreme points of /(x, y), we must have,
fx and fy = 0
i.e., 3x2 3 = 0 and 3y2 -- 12 = 0
i.e., x2 = 1 and = 4 giving x = ± 1, y = ± 2.
Therefore the extreme points of the function /(x, y) are (1, 2), (—1,2) Q).
Again the Hessian matrix is given by
H(x, y) = fxxfyy¯ (fry)2 = 36 xy.
Now at the four extreme points we have,
H(l, 2) = 36 xl x 2 = 72 0) and =
H(l, +2) = 36 x (-1) x 2 - 72 0)
H(l, -2) = 36 x 1 x = - 72 0)
and H(-l, -2) = 36 x(-l) x 72 (>0), _2) =
Hence we may conclude that the points (—1,2) and (1, —2)are the saddle points y)
—2)
while the function /(x, y) is maximum at (—1, and minimum at (1, 2). Further the maxi.
mum value of /tx, y) is —2)= (-1)3 + 12 x (-2) + 20 = 38 and minimum
value of/(x, y) is fli, 2) = (1)3 - 3 - 12 x (2) + 20 = 2.
Find the extreme values of the following function flx, y) = xa + - 3axy
[WBUT 2002, 2003,
20081

Solution Here the given function is

/(x, y) = x3 + y 3 —3axy
The first and second order partial derivatives of ftx, y) is given by
fx = 3x2 —gay, fy = 3y2 —3ax
= 6X, fyy = 6y and fxy = fyx = —3a•
Now at the extreme points of /(x, y) we have,
= 0 and fy = 0
i.e., 3x2 —3ay = 0 and 3)'2 —3ax = ()
i.e., x2 —ay = 0 and —ax = O
2
Combining we have, Y —ay = 0, since x =
a
i.e.. y(y 3 —a 3) = 0 giving — ay + a 2) = 0
Since y + ay + a 2 = 0 gives no real y, therefore from (2) we have, y = O ory = a
2
Again since at extrema, x — , so the extrme points are (0, 0) and (a, a).
 Extrema of Function of Several Variables 409
matrix is given by
Hessian
(fxy)2= (6x)(6y) (—3a)2= 9(dxy —aZ)
Il(x, y)
points (0, 0) we have,
extreme
= 9a2 (< 0)
9 (0 (12)
point (0, 0) is a saddle point of the function /(x, y).
the extreme
point (a, a) we have,
the extreme
= 27a2 (> 0) and
a) _ 9(4a a -- a2)
Il(a, a) ¯ [6Xl(a,a) = 6a 2 0 if a 2 0,
fix I (a,
is maximum at (a, a) if a < 0 and minimum at (a, a) if a > 0. However if az 0,
is a saddle point, which is established already.
a) 3= —03which is the maximum (or minimum) value according as
< O).
0 (or a
points of the following function
Find the extreme
g y) x3 + ary2 —3y2 —g.r2 + 4 [WBUT 2004, 2007, 2009)

the given function is

solutionHere 3xy2 —3y2 —
/(x, y) = xa + 3x2 + 4
second order partial derivatives of the function /tx, y) is given by
Thefirst and
+ 3y2 6x, fy = 6xy 6y,
fit—6;fyy = 6x —6 and fxy = fyx = 6y
extreme points we have,
Nowat the
Oand fy = 0
i.e., 3(x2+ —
'E) = O and 6y(x -- 1) = O
y = Oor x = 1.
Nowfrom (3) we have
we have,
Wheny = 0, then from (2)
x2 —2x= 0 giving x = 0, 2
Againwhen x = 1, then from (2) we have,
1+ giving y = ± 1.
Hence(0, 0), (2, 0), (1, 1), (1,—1)are four extreme points of the function /(x, y).
•
Nowthe Hessian matrix is given by
H(x,y) - (fxy)2= (6x - - 6) 36y2= - - y21
Atthe extreme point (0, 0) we have,
H(O,0) = 36 - 1)2 -01 = 36 (>0) and (O,0) = 6 xo
Thereforeftx, y) is maximum at (0, 0) and the maximum value off(x, y) is /(O, 0) = 4.
At the extreme point (2, O)we have,
H(2,0) = 36 [(2-1)2 -01 = 36 0) and (2, 0) = 6 x 2-6 = 6 0).
Thereforeflx, y) is minimum at (2, 0) and
the minimum value offlx, y) is /(2, 0) = (2)3—3 x (2)2 + 4 = 0.
410 Fundamental EngineeringMathematics
Also at the extreme point (1, 1) and (1, —1)we have,
H(l,1)= 36 [(1 - 1) '2 36 ( < 0) and H (1, -1) = - 1)2-
Therefore the points (1, —1)and (1, 1) are saddle points to the function /tx, y)
and
function flx, y) does not attain any maximum or minimum at (1, —1)and (1, 1).

Find the extreme points of the function flx, y) = x2 + xy + Y2

and determine the nature of the function at extreme point(s).
Solution Here the given function is
ftx, y) = x2 + xy + Y2 + + 3y + 5
Differentiating flx, y) partially with respect to x and y respectively, we have,
fx=2t+Y+2, fy=x+2Y+3

Now at the extreme points of ftx, y) we have

0 = fy, i.e., we have,
2t + y + 2 = 0 and x + 2y + 3 = ()
1 4
Solving we have, x = ——y
Therefore, — is the only extreme point of the given function flx, y).
Also the Hessian matrix for flx, y) is H(x, y) = fÄfyy— (fxy)2 = 2 x 2 —(1)2 = 3 (>0)

Since at the extreme point both H(x, y) and fxx are positive so the
function

Åx,y) attain its minimum value at

Hence the minimum value of flx, y) is

1 4 16 2 12 8
9 9 9 {33
Prove that the function f(x, y) = x 2 —2xy + Y2 —x3 — + x5 has neither maxi.
mum nor minimum at the origin.
Solution Here the given function is
flx, y) = x2 —2xy + — —y3 + x5
Taking partial derivatives of flx, y) with respect to x and y respectively of first and second
order we have,
fx = —2y —3x2 + 4x4
fy=-2t+2y-3y2
fxx= 2 - 6X+16x3, G = -2 = fyx, 2 - 6y
 Extrema of Function of
Several Variables 411
is an extreme point of the function /(x, y)
clearly(0, O) since (0, 0) satisfies both = O and
at (0, 0
0)= 2 x 2 - O and fxx= 2
, 0).
er investigation are necessary whether a stationary point is an extreme point or
soforth it is maximum or minimum.
whether
or =
ve that fix, y) x4 + x2y + has a minimum at (O,
O).
Here the given function is /(x, y) = .t4 + x2y + Y2
solution derivative of first and second order, we
taking partial have
= 2y + 12r2
Ety + 4x3,

an extreme point of the given function ftx, y). Since (0, 0) satisfies both
clearly(O,0) is
0.
0 and
0), the Hessian matrix is
Nowat (0,
[Ox 2-0] -O.
are necessary to derive whether it is maximum or minimum at
Sofurtherinvestigations

0) = +2x4 > O for all values of x and y.

Also,fix,y) - fro,

Hencethe function f(x, y) has a minimum value at (0, 0).

Find the extreme point(s) of the function fix, y) = sinx + siny + cos(x + y),
0 <x, y < —. Also check the nature of the function at the extreme point(s).

SolutionHere the given function is

fix, y) = sinx + siny + cos(x + y)
Takingfirst and second order partial derivatives of fix, y) with respect to x and y respec-
tivelywe have,
cos x —sin (x + y), fy = cos y —sin(x + y)
—sint —cos(x + y), fry = —cos(x + y) = fyx, fyy= —siny —cos(x + y).
Nowat the extreme point(s) of f(x, y) we have,
cosx —sin(x + y) = 0, fy = cosy —sin(x + y) = 0
Solvingthese we have, x = y = since 0 < x, y < 2

Nowat the Hessian matrix,

412 Fundamental En ineerin Mathematics
111 1 1 3
H(x, y) = 2 4

and f (x y) sin—•cos

Therefore the extreme point is a maximum point and the maximum valueof
function is
= sin — + sin — + ms
6

Find the shortest distance from the point (O,b) to the parabolax2

Solution Let A be the distance from the point (0, b) to the parabola x-2= 4y.
Then we have 1 2 = x2 + (y — ,
where (x,y) is any P)int on the parabola.
To find the shortest distance, we form a Lagrangian as
L(x, y, A) = 12 + A(x2—4y) = + (y — + A(x2—4y)
Along the shortest path we have Lx = 0 = G. Now differentiating (2) partially with
to x and y we respectively have,
0 i.e., & —2Ax = 0 giving A = —1 or x = O.
Is. = O i.e., —b) —41 = 0 giving y = b + 21 = b —2, since A = —1
Therefore at A = —1, (O,b —2) is a stationary point.
Now from (2) we have,

d2L = —dx+—dy L — (dy)2 + 2 — dx dy

by ax2 by2 axoy
(03—2)

= ((2 + + 2(dy)2 1(0,b-2)= 2(dy)2 [since A = —1 at stationary)

Hence d2L is positive definite at (O,b 2)
Therefore the stationary point (0, b 2) gives minimum and at (0, b —2) the
shortest distance is = 2 unit.

Find the point on the plane x + 2y + 3z = 13 nearest to the point (I, 1,1)
using the method of Lagrange's multiplier. [WBIJT20011

Solution Lets be the distance from the point (1, 1, 1) to the plane x + 2y + 3z = 13,
we have,
s 2 = (x— 1)2 + (y —1)2 + (z—
where (x, y, z) is any point on the plane x + 2y + 3z = 13.
 Extrema of Function
of Several Variables
shortest distance, we form a Lagrangian 413
findthe + + as
fo A) ' + 13)
- (x- 1)2+(y- + (z- 1)2+
+2Y+3z- 13)
of L, we have,
extreme points
the

gives 2(x -- 1) + A = Ogiving x =

Lx
NOW
0 gives 2(y 1) + 21 = 0 giving y = 1
—A,
0 gives 2(z -- 1) + 31 = 0 giving z = 2-31
2
gives x + 2y + 3z 13 = 0
LA 0
2-1 + 2(1- A)+ 3 2-31 -13=0
I.e., 2 2
141 = 14 giving A = 1

A z:—1in the above we have, x = 3 5

Using 2
5 is a stationary point.
Therefore —,2, 72
have,
Nowwe
2

d2L
32.2,

= (dx) + —i (dy)2 + (dz)2 + 2 D2L +

a.x2 dy axöy Dxöz 2 Dyöz

= + 2(dy)2 + J = 2 [(dx)2 + (dy)2 + (dz)21 ( > 0)

Further 2
Thereforethe shortest distance from the point (1, 1, 1) to the plane x + 2y + 3z = 13 is
(3/2,2, 5/2). Hence the shortest distance is
along
2
1
2-1)2 4 4

Find the shortest distance from the origin to the hyperbola x2 + 8ry + 7y2
= 225. [CU 19961

SolutionHere the equation of the given hyperbola is

+ 7y2 - 225 = O
Lets be the distance from the origin to the hyperbola (1), then

where(x,y) is any point on the hyperbola (1)

414 Fundamental Engineering Mathematics
To find the shortest distance, we form a Lagrangian as
L(x, y, X) = s2 + I(x 2 + 8ty + 7y2 —225)
+ 7y2 - 225)
Alongthe shortest path we have both lux= 0 and Ly= 0.
Now differentiating (3) partially with respect to x and y respectively we have,
+ + 81y = '2(1 + 1) x + 81y,
'2y+ + 19y)= + + 71b,
Since at the point of shortest distance, lax= Ly = 0,
(1+1)x
so 2(1 + A)x + 81)' = 0 giving y = , 8Ax + 2(1 + 71b' = 0 giving y = — 4Rt
41 1+1 1+71
and comparing we have, 1+71 = 41 .(assuming x 0).

i.e., (1 + + =o
i.e., -1=0
i.e., (91 + —1) = 0 giving 1 = 1, —I9 .
1
when = 1, then y = — and when 1 = — , then y = ax.
2
Now when y = '2x, then we have from (1),
x2 + 16x2 + 2812 —225 = 0 giving x2 —9 = 0 i.e., x = ± 3, y = ± 6
Further when y = then from (1)
x2 —4x2 + —x2 —225 = 0, i.e., x2 + 180 = 0, which does not give any real value ofx.
Therefore the stationary points are (3, 6), (—3,—6). Also from (3) we have,
2 (1 + 1), + 71) and 81

so d2L = —dx+—dy L — Oxöy

Ox2 DY2
= + + 161dxdy+ +
Now at (± 3, ± 6),
d2L = 4(dx)2 + 16 dxdy + 16 (dy)2 [since A = Il

Therefore on the given hyperbola, from both (3, 6) and (—3,—6),we have the shortest
distance to the origin. Hence the required shortest distance is

12. If x, y, z denotes the three angles of a triangle, then find the maximum
value of cos x cos y cos z [CU19931
Solution Since x, y, z denotes the three angles of triangles then
 Extrema of Function of Several 415
Variables
coxz
cosXcosy
Ißt cosx cosy cos(lt —(x + y)) [From (I))
Theo cosx cosy cos(x + y)
cosx cosy(cosx cosy sinx siny)

sin2.t sin2y -- cos x cos2y

i.e.,
cos2x sin2y + 2sinx cosx cos y = —cos2x sin2y + sin2x cos2y

cos x sin2y
sin2x cos 2y +

sin '2xsin2y + 2cos2x cos2y,

+ 2 cos2x cos2y
-¯ sin2x sin2y
t/yy
tixy uyx= cos2x cos2y —sin2t sin2y = cos(2x + 2y)
and
at the extreme points we have,
Now
ux = 0 anduy=0

cos2x sin2y + sin2x cos2y = 0 and —sin2x cos2y + cos x sin2y = 0

i.e., 2

Sincefrom (1) we have 0 S x, y It, so solving above two equations we have, x 3

the point , the Hessian matrix of u is
at
Now

27t 411
sm2 —+2 cos— cos sm —+2cos — cos —cos2
3 3 3 3 3 3 3
2
47t
2 cos—cos — —sin —cos 2
3 3 3 3
411 47T 3
= 1—cos
2 sin2 (>0)
3

= 2cos — cos2 It sin 2

'27t
and u =
3 3 3

Thereforeat the point the function u (x,y) attains its maximum value,

and the corresponding maximum value is u = —cos —cos—cos—

3

Alsofrom(1), we may concludethat the extreme point is x and th

3 3 3
maximum value for the function cosx cosy coxz is —
corresponding 1
8
416 Fundamental En ineerin Mathematics
Note

The above problern can alB0be solved by Lagrange's multipliers method. In this
Lagrangian will be, the
cosy COBZ
L = COBX + y + z -- n)
= and the corresponding
U8ing lax= lay= L, = LA = 0, one may obtain x = y = z --
value of the function COBX cogy COBZ iB
8

13. Show that fXx,y) Isin xl + Isin yl has its minimum value at the origin
although rx and does not exist where — < x, y < n.
Solution Herc the given function is
f (x,y) = Isin xl+lsin yl
lim
f(O+h,0)- = lim
Now

f(O+h, 0)- f(0, 0) does not exist.

Therefore lim does not exist and hence
Similarly we may show that 0) also does not exist.
But we see that /(0, 0) = 0 and /(x, y) -- /(0, 0) = Isinxl+ Isin y O for any

Hence the function /(x, y) has its minimum value at the origin although and fydoesnot
exist at the origin [Proved].
14. Find the minimum value ofx 2 + + za subject to PX + qy + rz = a.
Solution Our objectiveis to minimize the function
2

subject to the condition PX + qy + rz —a = 0

We form the Lagrangian as follows :
L(x, y, z, A) = x2 + Y2 + z2 + A(px + qy + rz —a)
Now = + PA, 2y + (IA,Lz = 2z + rA
2= and O
At the minimum value of /(x, y, z),

Then + PA = 0 giving x =
2
2y + qA = 0 giving y = — ,
'2

2z + rl = 0 giving z = , andpx + qy + rz —a = 0
2
 Extrema of Function
of Several
we have, Variables 417
from (2)

2a

the co-ordinate of the extreme point of the

Hence function Rx, y, z) is
given by

pa qa
i.e.' p2+q +r p +q +r p +q +r
pa qa
Alsoat

d2L Ox Dy Oz

021. 2 0 210 2
—s(dy) + +2 02L
öxöy öxöz dydz
öyöz
+ (dy)2 + (dz)2) ( > 0) and Lu = 2 (> 0).
pa qa
Thereforethe extreme point is the minimum
pointforthe function ftx, y, z) and corresponding minimum value is
qa a 2(p2 +q + "2)
(p +q +r2)2 p +q +r

15. Prove that the volume of the greatest rectangular parallelopiped that can
abc
be inscribed in the ellipsoid is 8 [WBUT20021

SolutionLet 2t, 2y, 2z be the respective length, breadth and height of the rectangular
which is inscribed within the ellipsoid —
parallelopiped

Thenthe volume of the parallelopiped is given by V = 8xyz

z
Ourobjectiveis to minimize V subject to the ellipsoid —
Forthis purpose, let us formulate a Lagrangian as follows:

L (x,y, z, A) = 8xyz + A

IdamentalEngineering Mathematics (SEM-1)—27

418 | Fundamental EngineeringMathematics
Now at the extreme point of (1) we have,

Then from (2) we have,

0 i.e., 8yz +
a
8rz +
0 i.e.,
b2

0 i.e., 8ty +
c
2

c
Now multiplying (3), (4), (5) respectively by x, y and z and then adding we have,
2
24xyz + 21
c
i.e., 24 xyz + 21 = 0 [From (6)1
giving 1 = -- 12xyz.
Hence from (3) we have, at extreme point,

8 yz—3x 2yz giving X = , since yz 0 and x > 0.

a
Similarly from (4) and (5), we have at the extreme point, y —
b
and z =

Therefore, is an extreme point of the Lagrangian (2).

Now to check the nature of that extreme point, we have from (2),
21 L and
2 YY¯ 2 zz¯-PA2
a

Therefore when I = 3xyz , then at the extreme point

we have,
2 (å'å'å)
2
d2L = Ox Dy Oz

2 0214
—(dy)2 2 +2
x Y öXÖZdd
021,
dz
Dy2 DyOz Y

= —(dx) 2 +4 (dy)2 + % (dz)2 + 16(zdxdy + ydxd2+xdydz)

8 abc 8 abc 2 - 8 abc (dz)2 +

(dx)2 ——(dy) (c dx dy + bdx dz + a dy dz)
Väa2 Cib 2
 Extrema of Function of Several Variables 419
2 2
dy dz dxdy dxdz dydz
abc 8 b c ac
a bc
3

16 abc dx dy dz
3
2
we have dz
Since
from
I

giving (lady dxdz dydz 2

-0 2 dx dy
bc a c

gives the maximum value and the corresponding maxim um

the parallelopiped is cube units.

olurneof

Cfxetcise
Answer Type Questions 9. Prove that the function/(x, y) = 4xy
ID Short
point of the function has neither maximum value nor mini-
1. Findthe critical mum value at (0, O).
ftx, y) = x2 + Y2—xy + 2x + 2y —4.
10. Find the extreme point of the function
2. Find the stationary point of the func- /tx, y) = .r2 + Y2 + (x + y + 1)2.
tion f(x, y) = x2 + Y2 — + 2y + 3.
ID Long Answer Type Questions
3. Find the extreme points of the func-
tion /tx, y) = x3+ —3x —12y + 20.
1. Examine the extreme values of the
4. Find the critical points of the function function /tx, y) = x2 + y + x + y + 1.
flx, y) = 2(x — —x 4 —y 4. [CU 20051
5. Provethat if a < 0, then the function 2. Find the extreme values of the func-
fix, y) = x3 + y 3 —3axy is maximum at tion /tx, y) = 2x2 —xy + 2y2 —20x.
[CU 20001
2 3. Examine the extreme values of the
6. Prove that ftx, y) = x3 + 3xy2 —3y2 —3x
+ 4 has a maximum at (0, O). function Rx, y) = x3 + 3x2+ 4xy + y 2
7. Prove that the function ftx, y) = x2—2ry 4. Find the maximum and minimum
+ Y2 + 1 3 —y3 + x 5 has neither maximum value of the function
nor minimum value at ((),0). /tx, y) = 4x2 —xy + 4y2 + x3y + xy 3—4.
[CU 20071
8. Prove that is a saddle point of
5. Examine for extreme values of the
the function function
f(x, y) = x3y + xy3 + 4x2 —xy + 4y2 —4. /tx, y) = x8 + 3xy2 —15x2 —15y2 + 72x.
420 Fundamental Engineering Mathematics
6. Determine the saddle points of the tox2+Y2+z2=c2at ±c
function flx, y) = x3 + y 3—3x s—12y + 20.
3'h).
7. Examine for the existance of maxima Hence derive (x2y 2z2)i s x2+
or minima of the function
19. Prove that the area of a triangle J.
ßx, y) = xy . perimeter is constant wk
is
when it is an maxirR
equilateraltriangle
8. Prove that (—1,2) and (1, —2)are the
saddle points for the function 20. Show that the maximum value
of
f(x, y) = x3 + y3 3x —12y + 20. subject to 4x + 3y = 5 is 25
48
9. Find the extreme point(s) of the func- 21. Prove that among all the
tion flx, y) = x2 + xy + Y2 + x + y + 1 parallelopiped of same volume
and examine the nature of the func- cube has the least surface area. the
tion at extreme point(s).
Multiple Choice Questions (MCQ)
10. Show that origin is a minimum point
+2 I. For a function f(x, y) the solutions
offlx,y) = 2x4—x2Y+ equations fx(x, y)
ofthe
= fy (x, y) = 0
weget
Il. Show that (1, 1) is a minimum point of a. extreme points b. saddlepoints
c. natural points d. none of these
the function ftx, y) = xy +
2. Iffx = fy = 0 at (a, b) then the
critical
12. Show that f!x, y) = 284 + 2r 2y + Y2 has point of f(x, y) is
a minimum at (0, 0).
13. Show that the minimum value of d. none of these
ftx, y) = Ixl+lyl occurs at the origin 3. If a function f(x, y) has a maximum
or
although neither fx(0, 0) nor fy(O,0) minimum at (2, 2), then fy (2, 2) =
exist.
14. Find the maximum value of f(x, y) = d. none of these
sinx siny sin(x + y), where 0 x, y It 4. The sufficient condition for the
and 0 S x + y lt. function flx, y) to be maximumat its
15. Minimize x2 + Y2 + z2 subjct to xyz = 8. extreme point (a, b) is
16. Find the minimum value ofxy + yz + zx
subject to xyz = c3 b.f (a b) f (a b) —{fxy(a,
17. Find the maximum and minimum c.f (a b) f (a b) —{fxy(a,
value of x2 + Y2 + z2 + 2 subject to
2
d. none of these
and x + y = z. 5. For the function ftx, y), if fx(a,b)=
4 5 25
18. Show that the arithmatic mean of fy(a, b) = 0, then (a, b) will bea
saddle point if
three positive numbers is never less
than their geometric mean. (a b) —{fxy(a,
a. fxx(a b) fYY
b.f (a b) —{fxy(a,
(a b) fYY >O
[Hint. Show that the function
/(x, y, z) = x2y2z2 is maximum subject c.f (a b) —{fxy(a,
(a b) fYY
d. none of these
 Extrema of Function of Several Variables 421
y), if dz = 0 at 15. For what value of a, origin is an ex-
function z = fix,
b) is a maximum point treme point for nx, y) = x2 + ay2 + 2
6. forb),then (a,
(a,
when b. d 2z < O
d. none of these
c. any real value of a
for the function
extremepoint d. none of the above
1 2 is 16. The function ftx, y) = 8t4— + Y2
has a minimum at
a. d. does not exist
value of the function
Themaximumis
cosz
coss cosy 1 17. If L(x,y, z) be a Lagrangian such that
b.
8 at (a, p, y). Then
36 the point (a, p, y) is
minimum value of the function a. maximum point b. minimum point
9. The d. extreme point
sin z is c. saddle point
sinx sin y
18. For a function flx, y) iffx = 2t + y and
d. none of these fy = x + y then
value of the function a. origin is a saddle point
10.Themaximum R exists at
sinx cosy; x, y e b. origin is a maximum point
b. two points
a. one point c. (1, 0) is a maximum point
c. infinitepoints d. no where d. origin is a minimum point
(a, b, c), fx = fy =
ll. Supposeat a point 19. If the Lagrangian of a functionflx, y, z)
fz = O and d2f is a positive definite subject to the constraint g(x, y, z) = c is
then for the function ffx, y, z) the point L(x, y, z) = flx, y, z) + y, z) —c},
(a, b, c) is then at the extreme points
a. maximum point b. minimum point
c. saddle point d. none of these
2
12.The minimum value of x2 + Y2 + z b. = Obut
subjectto x + y + z = 6 is c. O but O, *O
a. 12 b. 20
2
13. The maximum value of x-2 + Y2 + 22
subjectto x + y + z = 10 is 20. If at a point (1, 2, 3), the function
f!x, y, z) is maximum and at (2, 3, 1),
400
a. 100 b. fix, y, z) is minimum then
9
d. none of these a. fx(l, 2, 3) = O but 3, 1) = O
14. The maximum value of xyz subject to b. fx(l, 2, 3) = O,but fy(2, 3, 1) * O
x + y + z = 15 is
a. 15 b. 125 c. fx(l, 2, 3) = O = fy(2, 3, 1)
c.1000 d. none of these d. fy(l, 2,
422 Fundamental Engineering Mathematics

Answers
3. (1, 2) (-1, 2), (1, -2) and (-1, -2) 7. Minimum at (2, 2).
4. (a, Vä) 10 9. One extreme point 1
3' i '-j 'a
nd
B. 1. Minimum at 1 this is minimumpoint.
2
16 and fmax =
2. Minimumat 15.12
3 l) 8
2 16. Minimum point (c, c, c) and
3. Minimumat 3'
mum value 3c 2.
4. Minimum at (0, 0), neither a maxi-
17. Maximum value = 12 and
mum nor a minimum at
mum value = 109
17

5. Minimum at (6, 0) and minimum

value = 108; Maximum at (4, O)and 10.
maximum value = 112; Neither a 11. (b) 12. (a) 13. (d) 14. (b) 15. c
maximum nor a minimum at (5, 1)
16 (a) 17. (c) 18. (d) 19. (a) 20. c