69

We get cos B Isin A cos (A + B)+ cos A sin (A +B)

A
= cOs B sin (2A + B)
and cos A sin A + 28) 1h
OB
dA 0, =0
only when A=B = n/3.
Also = 2 cos B cos (2A + B), t = = 2 cos A cos ( i + 2B)
dB
-
sin B sin (24 + B)+cos Bcos (2A+ B) = cos (2A + 28)
When A B = w/3, r = 1, s
-
=
1/2,t -
1 so that rt -s" 3/4.
- =
These show that fA, B) is maximum for A B nd3.
Then C n-(A +B)= /3.
Hence cos A cos B cos C is maximum when each of the
maximum value = 1/8. angles is n3 i.e., triangle is equilateral and
12 LAGRANGE'S METHOD OF UNDERTERMINED
MULTIPLIERS
Sometimes it is required to find the
independent but are connected by some stationary values of a function of several variables which are nat
having least number of given relations. Ordinarily, we try to convert the
one,
independent variables with the given
help of given relations. Then solve itfünction to
method. When such a by the ab
explain this method.
procedure becomes impracticable, Lagrange's method* proves very convenient. Now
Let u f(x, y, 2)
be a function of three variables x, z which
y, are connected by the relation.
For u to have
x, y, a)0
stationary values, it is necessary that
duldr =
0,
duldy 0, duldz 0. = =
ddyddu=0
Also differentiating (2), we getd+dy+
dr
d dz =do =0
Multiply (4) by a parameter A and add to (3). Then
This equation will be satisfied i -0,+-0,+ =
0.
These three equations together with (2) will
determine the values ofx, y, z and A for which
Working rule: 1. Write F f(%, z)+ is
z)
=
y, A{a, y,
u
station
2. Obtain the equations 0,0 0
3. Solve the above
The values equations together with o (x, y, 2) 0. =
ofx, y, z so obtained will give the stationary value of
flx, y, z).
Obu. Although the
deterine the nature of theLagrange's
method is often very useful in
stationary point. This can sometimes, be application
determined
yet the drawback is that oe cns
problem from physical eonsideratios
 70
A reetangular bos open at the top is to huwe polume of 32 pubic
eSaumiple
material
deast
g least materi for its construction. , Find the daensioria of
urukshetru, 2006;P.TU, 2006 UPTU, 2005)
y and z A. be the edges of the box and S
ot T.
be its surface.
S o l u t i o n .

S = xy + 2y2 + 22x
Then ry2 = 32
.)
. ..(ii)
. from
Eliminating 2 frn ) with the help of (i), we get S =
xy +2y +
) =xy 64}
+

aS/dr =
y 64/x 0 and dS/dy 64/y2 =0.
-
=
= x -

get x y 4.
we
= =
these,
Solving r = RS/02 1 2 8 / ,s = 0*S/dxdy = 1,t = ?s/ay2 = 128/
Now = y w 4, r t -s*= 2x 2-1 = + ve and r is also + ve.
At from (ii),
S is minimum for x =
y =4. en 2 = 2.
Hence
method)
Otherwise (by Lagrange's
F=xy + 2yz + 22x + Alxyz - 32)
Write

= y +2z + Ayz =0o

Then

=x+ 22 + zx =0 ..iv)

2 y + 2x + Axy =0
we get 2zx -2ay 0
Multiplying (iii) by x and (iv) byy and subtracting,
= or x =y.

a s it will not satisfy (ii)}

The value 2 0 is neglected,
=

z and subtracting, we get y 2.

Again multiplying (iv) by y and (v) by
=

ui)
Hence the dimensions of the box are x
=
y = 22 = 4
a small value to a large one.
When z is small, the box is
Now let us see what happens as z increases from andS also
with a large base showing that S is large.
As z increases, the base of the box decreases rapidly
t a s z increases. Thus S must
be a minimun at some
After a certain stage, S again starts increasing
reases. when x y 4ft and z 2 t. =

is given by (vi). Hence S is minimum

= =

ermediate stage which

Anpe,2009
Example 46. Given x a find themaximum valoe ofy
+z-a.
Solution. Let flx, y,z) =x"yz" and ¢lx,y, z) = * +y
Then Flx, y, 2) flx, y,z) + 0lx, y, z) =

*"yzP + x +y +2-a).

Por stationary values of F, = 0 , 0 =

=0
mr 0, nx yn-1 2+A=0, pxy'zP+
zP +2 =

A= mn-lyz = n"" y-l P =pr"y"z

( a+y +z =a
- " t n * p mtntpP

The maximum value off occurs when +n +p)

apl{m
+p), 2
=
an/m + n
am/ln + n +p),y =

Hence the maximum value of flx, y, z)= PM m"-np

mn m+n+ p"***P

sphe
distaneesofthe poine , 1)fhothe
UTU, 20101
d the noximum.and minimunt
so that
12) the given point
ation. Let Plx, y,2 the sphere and A(3,4,
2 ) be any point
on

- 3 P + y - 4 ¥ + ( a -12-fla,
y,2), say
AP'=
 59
vu,eompute dfP, a,HYeu,b,w
7. IfF-+r-y,Geu?
uy +, i zu-u tht . B, el, =uo.
y,zVdlu,
lfu=r+y+2,u y
+D, uvu =2,show (Kurukhetra, 2009 P.7U, s000 a n ea
&
atu,v1
show that 2 wu-p)
B. Ifue =r+y
andu+t=r*y'., .
e. du,.Are u andu functionally related,
and vfunetionally t s0.f
related. If
10. Ifu 1-ry and ve an'x+ tan 'y,find da, y) Areu
te if
o
=xir + 2y-),ahow that they
nally relate
arefanetionally related, anid indNuapar
11. f a r + a r e r - y + z n o d the n
INTERPRETATION
8 (1) GEOMETRICAL OX, OY,
point referred to
axes
IfPx, y, 2) be the of a
coordinates
5.1)
represents a surface. (Fig.
O2 then the equation z =ft,y)
through P
Let a plane y b parallel to the XZ-plane pass
=
the curve APB given by
cutting the surface along
2 =fla, b).
moves along the
x varies then P
As y remains equal to b and
derivative off (r, b) w.r.t. x.
curve APB and dz/ot is the ordinary
angle which the tangent at
Hence de/ar at Pis the tangent ofthe
Pto the section ofthesurfuce z =flx, y) by a plane through P purallel to
to the x-axis.
the plane XOZ, mukes uwith a line parallel
which the tan
Similarly, dz/dy at P is the tangent of the angle
the surfuce z flx, y) and the
gent at P to the curve of intersection of
=
plane x a, makes with a line parullel to the y-axis.
=
.1
(2) Tangent plane and Normal to a surface. Let Plr, y,
2)
and Qr +r,y + 8y, 2 + &) be two neighbouring points on the surface Flx, y, z)= 0. (Fig, 5.2)
Let the arc PQ be õs and the chord PQ be oc, so that (as for plane curves)
Lt (s/8c) = 1.
P
or os oy 8s & s
The
The direction cosines of PQ
direction cosines oy
of PQ are are
& 'as &c
When &s 0,Q-P and PQ tends to tangent line PT. Then noting that the coordinatesofanypointn
PQ are functions ofs only, the direction cosines of PT are
dr dy de
ds' ds' ds
aF d
Differentiating (i) with respect to s, we obtaind dF dy,
d a ds
d0
This shows that the tangent line whnse direction cosines are given
by (ii), is perpendicular to the line having direction ratios
..ii
Since we can take different curves joining Q toP, we get a number of
tangent lines at P and the line having direction ratios (iüi) will be perpen-
dicular to all these tangent lines at P. Thus all the tangent lines at P lie in
a plane through P perpendicular to line (iü).
Hence the equution of the tungent plune to (i) at the
point Pis
F
- dy
-y)z-)=0
dz
Fig
where X, Y. Z) are the current coordimates of any point on this tangent plane.
 60
Lation of the normal to the surfacr at P (i.e., the line through P. perpendicular to the tangent
A l s ot h e

is
P)
ane
at
X -x Y - y Z - z
aFx Fy aFdz
S1. Pind the eguationa of thuetangent flan and the nomnal w the eurface 41+
Example

Solution. We have Flx,y, z) = 4r2+ 4y2- 2+4.

OF/ar= 8x, OFldy 8y, dFld2
=
22, and at the point (2, 2, 6)
= -

F = 16, dFly =16,dFTet - 12

nce the equation ofthe tangent plane at (2, 2,6) is 16(X-2)+ 16(Y-2)-12(Z-6)= 0
Hence t
4X+4Y-3Z + 2 = 0 ..i)
t h e equation of the normal at (2, 2, 6) [being perpendicular to (i] is

2 -2Y-2 9
PROBLEMS 8
t h e equations of the tangent plane and normal to each ofthe following surfaces at the given points
L2y3-2at (2,1,-3) (Asam. 1998) 2. +Bye=3 at(1,2.-1) Osmania, 2003 S)
& =a at ( i) 4. 2-3ty -4x = 7 at (1,1, 2.
Show theplanex+12y6r-17-0 touches theconicoid ar-6y917 0Find alsothe peintof contae
&Show that the plane ax + by +c2 + d=0 touches thesurfaçepr+ 2=0,f 2od0
Find the equation af the normal to the surface +y+z2= PTU. 2009 S

9TAYLOR'S THEOREM FORFUNCTIONS OF TWoVARIABLES

Consideringf(x + h,y + k) as a function of a single variable z, we have by Taylor's theorem
y+,*afa,y +k)
fa +h,y+ k)=f(x,y + k) + h "z,dx ...0)

Now expanding f(x, y + k) as a function ofy only,

, , E f a ,),
int fa,y+ k)=fa,y) +k
,TE,y)
akes the
form/(r + h,y +k)=fa,y)+h

dy
..(1)
Hence, fr + h.y + k) =fa,y)+ h-
ambols we write it as f(x + h,y + t) =f,y)+
Taking x=a andy =b, (1) becomes
a, b)+ 2hMs, a,b) *1,a,bN+
+
ah,b+ ) =f(a,b) +
Ihf, la, b) +af,(la,bl+
 61

Putting a +h =x andb+k =y so that h * - a , k =y -6, we get =

b) f, (a, b)]
fia, b) + [(x a) f, (a, b) + (y
-

fix, y) =
-

lx-a), (a, b)+ 2/x-aMy - ,

b) (a, b) + (y- b" ,a, b)]
a) and (y b). t is used to expand fs, vi
of (x
-

Taylor's expansion of fx, y) in powers -

This is
neighbourhood of (a, b).
Cor. Putting a = 0,b = 0, in (2), we get
1
0)+ 2xy , (0, 0)+ y, (0,0)).
fx,y) =
fo, 0) lxf,0, 0)+yf, (0, 0)]+x(0,
+

This is Maclaurin's expansion of f(x, y.

Example 32. Erpand e log (+ y) in powery oft andy upto terms of third degre,
VTU 2010; P.TU, 2009 J.N.TU,
fa, y) = e log (1 + y) f0,0) = 0
Solution. Here
f a ,y) = e log (1 + y) 0,0) 0

f,0, 0) 1
,,y)=e 1+y 0,0)
=

=0
a , y)= e log (1 +y)

, y) = e (0, 0) 1 =

1+y
lyt,y) = - e (1 +y2 , 0,0)= 1 -

0 , 0)=0
, )=ëlog(1 +y)
1
,)= e ,(0,0)=1
1+
-1
y , y ) =- e (1+y (0, 0) =

yny 7, y) yN = 2e" (1 + 0 , 0)=2

Now Maclaurin's expansion of flr, y) gives

fa, y)=f(0, 0)+f,0, 0)+yf,0, 0)+ 0,0)+2xyf0,0)+f0,0

f0, 0)+3 0,0) +ay,,0, 0)+ 0,0))+

1
e log (1+)0+f0)+ (1) + ()+21)+*(-1}

3 r0)+a1)+3ay(-1)+(2)1 +

Example 33. Epunid y+ 2 i h poneryof ) and (y +uing TuoloYsthirorim

(P.T.U, 2010: V.TU, 2008: U.P.TU., 2006 Anna,
Solution. Taylor's expansion off lx, y) in powers of tr - a) and (x - b) is given by

f,y)=fu,b)+I-a)f,(a, b)+(y -b)/ta, b)+ x-aI.a,b)

+2r-uXy-b)f,a, b)+(y-bP,,la,b))+ r -a,a, b)

+3 (r aP (y- 6)1a, b) + 3 (r -

a) (y -6Pfla, b)
+(y-b ,m (a, b))+
Hence a= 1,b = -
2 and f(«, y) =xy +3y-2