Multivariable Calculus

Dr. S. K. Gupta
Department of Mathematics
Indian Institute of Technology, Roorkee

Lecture - 08
Differentiability-II

Hello friends. So, welcome to lecture series on multivariable calculus. So, we were
discussing about differentiability, that what do we mean by differentiability of 2 variable
or more than 2 variable functions. We have seen that for differentiability existence of
first order partial derivatives at that point is necessary condition, and to prove that
whether a function is differentiable at a point or not we have to show either z = dz+1.
x+2. y and 1, 20 as (x, y) (0, 0)

(Refer Slide Time: 00:46)

So, these are first way out by which we can show a function is differentiable at a point
(x0, y0). And the second way out is simply take limit 0 (z-dz)/  =0 . where dz =
fx(x0, y0)x+fy(x0, y0)y and  = (x2+y2)1/2 . So, we can prove the differentiability of a
function f(x, y) by any one of the definition and try to show that 1, 20 as (x, y) (0, 0).

Now, we will discuss one more problem based on this. Show that this function is
continuous, process is partial derivative at (0, 0) but not differentiable at (0, 0).
(Refer Slide Time: 02:17)

So, how can we solve this problem let us see.

(Refer Slide Time: 04:38)

So, you know how to prove that this limit exists and equal to 0, we have to use delta
epsilon definition.

 x3  y 3
 , ( x , y )  (0,0)
f ( x, y )   x 2  y 2

0 ( x, y )  (0,0)

C ontinuity :
x3  y3
lim  f (0,0)
x2  y2
( x , y ) ( 0 , 0)

Let   0 be g iven
y3
x3  y 3
x3  y3 x3
 0    x 2
 y 2
x2  y 2
x2  y2 x 2
 y 2

| x | x2 | y | y 2 x2 | y 2
  | x | .1 | y | .1      2 sin ce  1 ,  1
x2  y2 x2  y 2 x 2
 y 2
x 2
 y 2

if w e choose, 2   , then
3 3
x  y
 0   , wh en ever 0 | x |  , o | y | 
x2  y 2
.So, in this way we have shown that this function is continuous at (0, 0).

Next, we have to show, that the first order partial derivative exist at (0, 0); so, f x at (0, 0)
(Refer Slide Time: 05:52)

h3  0
0
f (0  h,0)  f (0,0) h 2
 0 h3
f x |( 0,0)  lim  lim  lim 3  1
h 0 h h 0 h h 0 h

3
0h
0
f (0,0  h)  f (0,0) 2  h3
f y |( 0, 0)  lim  lim 0  h  lim 3  1
h0 h h0 h h0 h

So, we have shown that the first order partial derivative at (0, 0) exists. And the values of
fx is 1 and fy is -1 at (0, 0). Now the third part is function is not differentiable at (0, 0). It
is continuous at(0, 0), first order partial derivative exist at that point, but function is not
differentiable at that point. How a function is not differentiable? Now to show that a
function is not differentiable at (0, 0), we can use any one of the definition. So, try to
find out this limit.
(Refer Slide Time: 07:54)

z  (0  x,0  y )  f (0,0)
x 3  y 3

x 2  y 2
dz  f x (0,0)x  f y (0,0) y
 x  y
z  dz
lim
 0 
x 3  y 3
 x  y
x 2  y 2
lim
 0 
x 3  y 3  x 3  yx 2  xy 2  y 3
lim
x , y ( 0 , 0 ) ( x 2  y 2 ) 3 / 2
x 2 y  xy 2
lim
x , y ( 0 , 0 ) ( x 2  y 2 ) 3 / 2

along y  mx
x 2 mx  xm 2 x 2
lim
x 0 ( x 2  m 2 x 2 ) 3 / 2

x 3 ( m  m 2 )
lim
x 0 x 3 (1  m 2 ) 3 / 2

(m  m 2 )
(1  m 2 ) 3 / 2
 z  dz
lim
 0 
x 3  y 3
 x  y
x 2  y 2
lim
 0 
x3  y3  x3  yx 2  xy 2 
lim
x, y ( 0 , 0 )

So, this limits depends on m, and hence this limit does not exist. And this limit does not
exist means function is not differentiable at 0 comma 0.

So, that is how we can show that function is not differentiable. If it is not differentiable;
that means, we have to show that this limit does not exist ok. Now let us see some more
properties of differentiability.

(Refer Slide Time: 12:40)

Now, let a point x naught y naught be changed to x naught plus delta x and y naught plus
delta y. We want to predict the approximate value of change in the value of f ok, x, y
changing from x naught y naught to x naught plus delta x, y naught plus delta y. So, what
will be the approximate change the value of a function that you want to find out? How
can we find that? Suppose f is a differentiable function and its first order partial
derivative at x naught y naught are known. The value of a function is given by this
expression. We already know this thing. Now if we move from a point x naught y naught
to a point (x naught plus dx and y naught plus dy) nearby point the resulting differential
of f is given by df=fx(x0,y0)dx+fy(x0,y0)dy. (Refer Slide Time: 12:40)
 Now let us discuss few examples based on this. Basically, we are finding the
approximate change the value of f by this expression. This is the first order derivative at
x naught y naught. So, let us discuss few examples the first problem.

(Refer Slide Time: 14:21)

(Refer Slide Time: 14:21), Near the point 1 comma 2 is f x y, which is equal to x square
minus x y plus y square minus 3. More sensitive to changes in x or to change in y this
way to find out is more sensitive in changing x for changing in y. And what can we say
about this at 0.2 comma 1? So, what is the function? Function here is x square minus x y
plus y square minus 3 ok. First find f x, f x is 2 x minus y. And f x at point is 1 comma 2
add 1 comma 2. It is 2 minus 2 which is 0 now f y. F y is minus x plus 2 y. So, what is f
y at 1 comma 2? It is minus 1 plus 4, that is 3. So, what will be df? df will be fx at x
naught y naught, fx at 1 comma 2 which is 0 into dx plus 3 into dy which is equal to 3dy.
(Refer Slide Time: 16:13)

(Refer Slide Time: 16:13)

So, if you change the value of x, there will be no change in the value of f. Because there
is no term of dx here ok. If you change the value of x, there will be no change in the
value of f at a point (1,2). But if we change the value of y, the value of function will
change by 3 unit. Suppose we change y by 1 unit, that the change in the value of f will be
3 units. So, that means, it is more sensitive to change in the variable y not x. It remains
unaffected by the change in the value of x. Now if we can check the same sensitivity at a
point say (0.2,1).

So, at (0.2,1) it is equals to 4 minus 1 which is 3, and at (0.2,1) it is minus 2 plus 1

minus 2 plus 2 is 0. So, what will be df? df will be simply 3dx because this is 0. So,
what we can say now? We can say that the change in the value of a function will remains
unaffected, if you change in the y nearby. Nearby means near the (0.2, 1). Now if you
change the value of x by 1 unit the value of function will change by 3 units; that means,
at this point it is more sensitive to the change in the value of x and remains unaffected to
change in the value of y ok. Now the second problem using differentials, find the
approximate value of this term.

So, what is the term? Basically, it is under root of 298 whole square plus 401 whole
square.

(Refer Slide Time: 17:31)

x ( 298) 2  ( 40) 2 f ( x, y )  (x2  y2
x x0  300, y0  400
fx 
2
x  y 2 dx  2, dy  1
300 2 df  fx ( x0 , y0 ) dx  fy ( x0 , y0 ) dy
f x |( 200, 400 )  
500 5 df  3 / 5.  2  4 / 5.1  0.4
400 4 f ( 298,401)  df  f (300,400)
f y |( 300, 400 )  
500 5  0.4  500
f ( 298,401)  f (300,400) x  499.6

(Refer Slide Time: 20:34)

So, this will be 499.6. So, the approximate value of this x will be 499.6.
(Refer Slide Time: 21:18)

Now the third problem (Refer Slide Time: 21:18) , a certain function z is equals to f x y
has values f at 2, 3 is 5 f x is 3 and f y is 7 find the approximate value of f(1.98, 3.01).
So, again we can easily find out with the same concept. So, what are thinks given to us F
at 2 comma 3 f x at 2 comma 3 and f y at 2 comma 3. For solution (Refer Slide Time:
21:18)

So, the approximate value of a function at this point is 5.01. So, there is one type of
problem which can be solved using differential. Now we can also find absolute relative
and percentage change in the value of the function. How we can find out? So, if you
move from a point (x0,y0) to a point nearby, then we can describe the change in the value
of the function f x y in 3 ways.

(Refer Slide Time: 23:32)

First is absolute change, which is true is delta f, relative change with true is delta f
divided by f at (x0,y0) and estimate is d f upon f at (x 0,y0). The percentage change means
you will simply multiplied the relative change by 100 ok. So, how can we solve some
problems based on this let us see. Suppose a variables r and h, change from the initial
values r0, h0 which is 1, 5 by the amount dr = 0.03, and dh = - 0.1.

Then estimate the resulting absolute error relative and percentage change in the value of
function V=  2rh.

(Refer Slide Time: 27:25)

The initial point r0, h0 =1, 5. dr is the change in the value of r = 0.03. And dh is= - 0.1 ok.
Now we have to find the relative and absolute change in the value of V. How can you
find that? Now here V is a function of r and h ok. Instead of x and y we are having r and
h.

So, how can we define dV here dV will be dV=V r (r0, h0)dr +Vh(r0, h0)dh. Vr= 2rh and
Vr(1,5)= 2.1.5= 10. Again, Vh (1,5)= r2=.1=. dV=10dr+dh; dV=(10.0.03+1.-
0.1)=0.2
What will be f relative change? Relative will be dV/V= 0.2 /(25)=0.04. And the
percentage error will be dV/V.100 that is 0.04.100, = 4 percent. So, that is how we can
find out absolute relative and percentage change in the value of the function.

The next problem find the percentage error in the computed area of an ellipse, when an
error of 2 percent is made in increasing the major and minor axis. Area of the ellipse is a
b. For Solution see (Refer Slide Time: 28.49)

(Refer Slide Time: 28.49)

So, that is how we can find out the corresponding percentage error in the value of the
area of the ellipse. Now there are also some more problems based on this. Let us solve
one more problem.

(Refer Slide Time: 29:01)

The power consumed in the electric resistance given by P = E 2/R in watts. If E =80 volt,
and R = 5 ohms, how much power consumption will change if E is increased by 3 volt
and R is decrees by 0.1 ohm?. For Solution see (Refer Slide Time: 31:27)

(Refer Slide Time: 31:27)

So, we can simplify this and we can find out the corresponding change in the power
consumption. Similarly, we can solve the next problem also, which is given in electrical
circuit. Now the first problem of the next slide is say y= uv where u and v are
independent variables. If u is measured with 2 percent error and v is measured with 3
percent error, what is the percentage change in the value of y.

(Refer Slide Time: 31:44)

Refer Slide Time: 32:36)

So, here y = uv.

So, dy = yu .d u + yv . dv

vdu+udv

dy/y=du/u+dv/v

dy/y.100=2+3=5%

The corresponding change the value of y will be 5 percent. So, that is how we can solve
problems of this type.

So, thank you very much.