Notes on lecture 4

Jiten Chandra Kalita and Shyamashree Upadhyay

1 Parametrization of a curve with respect to

arc length
1.1 Re-parametrization of a curve
In the previous lecture, we have already seen the parametrization of a curve in
terms of the parameter t. But this parametrization is not unique. Instead of
t, we can use some other variable to parametrize the same curve. Eventually,
any parametrization (that we use) of the curve (if it is accurate), will give us
the same length of the curve.

Example 1.1.1. Suppose C is a curve which is parametrized by ~r(t) =<

t, t2 , t3 >, where 1 ≤ t ≤ 2. Suppose we define t = eu . Then we can re-
parametrize the same curve C as ~r(u) =< eu , e2u , e3u >, where 0 ≤ u ≤
ln 2. 

1.2 The arc length function

Let us consider a curve ~r(t) =< x(t), y(t), z(t) > in space, where t ∈ [a, b]. So
the initial point of the curve is ~r(a). Earlier, we were considering the length
of the whole curve. Suppose now, we would like to consider only a part of
this curve, that is, an arc (from the initial point to some other point on the
curve). Suppose we take a variable point t ∈ [a, b] and we simply want to
find out the length of the curve traced out from a to t. R
b
Earlier, the length of the whole curve was given by L = a |r~0 (t)|dt. But if
we are considering only the part of the curve traced out from a to t, then
we consider the integral only from a to t (NOT from a to b as considered for

1
getting the length of the whole curve), and we define
Z t
s(t) := |r~0 (u)|du. (∗ ∗ ∗)
a

This expression s(t) is going to give us the length of the curve traced out
from a to t. The expression on the right hand side of (∗ ∗ ∗) above is a
function of t, that’s why we have defined s to be a function of the variable t.
This function s(t) is known as the arc length function.
Remark 1.2.1. Observe that if we apply the fundamental theorem of integral
calculus on (∗ ∗ ∗), then we get
ds
= |r~0 (t)|.
dt
Remark 1.2.2. Whenever we talk about arc length or re-parametrization of a
curve (in books or in exercises), if nothing is stated, you should invariably assume
that it is “in the increasing direction of t”.

1.3 Re-parametrizing a curve with respect to the arc

length function
In subsection 1.2 above, we saw that the arc length function s is a function of
the variable t. Sometimes (not always), we may be able to express t explicitly
in terms of s. In that case, we are going to obtain a way of re-parametrizing
the same curve with respect to the arc length function s. Please see Example
1.3.1 below.
Example 1.3.1. Consider the curve ~r(t) =< cos t, sin t, t >, where t ∈
[0, 2π]. Suppose we want to re-parametrize this curve with respect to the arc
length function s. √
Observe that r~0 (t) =< − sin t, cos t, 1 >. This implies that |r~0 (t)| = 2. Now
(∗ ∗ ∗) tells us that Z t√ √
s(t) = 2du = 2t.
0
This implies that t = √s ,
where we denote s(t) by s. So, in this example, we
2
can express t explicitly in terms of s. Hence we get
s s s
~r(t) = ~r(t(s)) =< cos( √ ), sin( √ ), √ >,
2 2 2

2
 √
where s ∈ [0, 2 2π]. So we have re-parametrized the same curve with respect
to s. 
Homework: Re-parametrize the following curves with respect to arc length
measured from the point where t = 0 in the direction of increasing t:
• ~r(t) = 2tî + (1 − 3t)ĵ + (5 + 4t)k̂.

• ~r(t) = e2t cos 2tî + 2ĵ + e2t sin 2tk̂.

2 Curvature
Consider a curve C in the 3 dimensional space. Let the curve be such that
at each point lying on the curve, we can draw a unique tangent. Then the
curve is called smooth. See Figure 2.0.1 for an illustration.
The prerequisites for the smoothness of a curve C (where ~r(t) is a parametriza-

Figure 2.0.1: A smooth curve

tion of C) are:
• r~0 (t) exists and is continuous and

• r~0 (t) 6= ~0.

So if ~r(t) is a parametrization of a curve C, and if, the conditions
• r~0 (t) exists and is continuous and

• r~0 (t) 6= ~0

3
are satisfied, then we say that the parametrization of the curve is
smooth. In order to be smooth, the curve should not have any sharp corners
or cusps.
We will now discuss curvature. Whenever we talk about curvature, we will
talk about it for a smooth curve only. The curvature is a measure of how
quickly a curve changes its direction.
Suppose we have a smooth curve in space. Let ~r(t) be a parametrization of
the curve. We know that the unit tangent to the curve is simply defined to
be
r~0 (t)
T~ = .
|r~0 (t)|
The curvature (of the curve) is denoted by the symbol κ and is defined as

dT~
κ= ,
ds

where s = s(t) denotes the arc length of the curve from the initial point upto
the point corresponding to t. That is, the curvature κ of the curve is the
magnitude of the rate of change of T~ with respect to the arc length s.
Observe that
dT~
dT~ dt
κ= = ds ,
ds dt
Rt
where s = a |r~0 (u)|du. We know from the previous lecture that ds dt
= |r~0 (t)|.
Hence
dT~
dt T~ 0 (t)
κ = ds = . (∗)
dt r~0 (t)
We are going to use the last expression (∗) (for κ) later on in order to derive
some formulae for the curvature.

Example 2.0.1. Find the curvature of a circle of radius a.

Whenever we want to define ~r(t) for a circle of radius a lying on a plane, we
are going to have:
~r(t) =< a cos t, a sin t > .
This implies that
r~0 (t) =< −a sin t, a cos t > .

4
So
r~0 (t)
T~ (t) =
|r~0 (t)|
< −a sin t, a cos t >
= =< − sin t, cos t > .
a
This implies that T~ 0 (t) =< − cos t, − sin t >. Clearly |T~ 0 (t)| = 1 and |r~0 (t)| =
a. So for a circle of radius a, we have:

T~ 0 (t) 1
κ= = .
r~ (t)
0 a


Remark 2.0.2. In Example 2.0.1 above, we saw that the curvature of a circle of
radius a equals a1 . That means larger the radius of the circle, lesser the curvature.
Suppose there are two circular roads of radii a1 and a2 , where a2 >> a1 . Now
if we consider the curvature at some point P1 (on the circle of radius a1 ) and
at some point P2 (on the circle of radius a2 ), then we can see that on the outer
circle, the curvature is less.
Theorem 2.0.3. The curvature of a curve given by the vector function ~r(t)
is
|r~0 (t) × r~00 (t)|
κ(t) = .
|r~0 (t)|3
~0
Proof: We know that T~ (t) = |rr~0 (t)|
(t)
. We also know that ds
dt
= |r~0 (t)|, where
s denotes the arc length function. This implies that
r~0 (t)
T~ (t) = .
ds/dt
In other words,
ds
r~0 (t) = T~ (t).
dt
2
ds ds
⇒ r~00 (t) = 2 T~ + T~ 0 .
dt dt
Now, from the properties of vector cross product, we know that if ~a.~b, ~c are
3 vectors and λ1 , λ2 , λ3 are scalars, then
λ1~a × (λ2~b + λ3~c) = (λ1 λ2 )~a × ~b + (λ1 λ3 )~a × ~c.

5
So
2
d2 s
   
ds ds
r~0 (t) × r~00 (t) = (T~ × T~ ) + (T~ × T~ 0 ) (∗∗)
dt dt2 dt
Now since the angle between T~ and itself is zero, it follows that T~ × T~ = ~0.
We also know from one of the earlier lectures that if ~u is a vector of constant
length, then ~u and u~0 are orthogonal. Since T~ is a unit vector, it is of constant
length 1. Hence T~ and T~ 0 are orthogonal. So we get from (∗∗) that:
 2
ds π
|r~0 (t) × r~00 (t)| = |T~ ||T~ 0 | sin
dt 2
 2
ds
= |T~ 0 |,
dt
as |T~ | = 1. Now since ds = |r~0 (t)|, we get from the above that
dt

|r~0 (t) × r~00 (t)|

|T~ 0 | = .
|r~0 (t)|2
So
|T~ 0 | |r~0 (t) × r~00 (t)|
κ(t) = = .
|r~0 (t)| |r~0 (t)|3


Theorem 2.0.4. For a curve y = f (x) in the cartesian coordinate plane,

|f 00 (x)|
κ(x) = .
[1 + {f 0 (x)}2 ]3/2
Proof: Let ~r(t) denote the position vector of any point (x(t), y(t)) on the
curve. Since we are in the cartesian coordinate system, we can write
~r =< x, y > .
Since the curve is given by y = f (x), we can further write:
~r =< x, f (x) > .
This implies that
r~0 =< 1, f 0 (x) > .
Now, if we simply follow the steps in the proof of Theorem 2.0.3 above, we
will get the proof of this theorem. This is left to you as an exercise. 

6
3 Normal and Binormal vectors
Suppose we are considering a curve C in the 3 dimensional space. Let C
be parametrized by ~r(t) =< x(t), y(t), z(t) >. Let P denote the point
(x(t), y(t), z(t)) on the curve C. Then r~0 (t) is the tangent to the curve C
at the point P . Now there are infinitely many vectors which are perpendicu-
lar to the tangent vector r~0 (t) in various directions. The unit tangent vector
~0 (t)
in the direction of r~0 (t) is given by T~ = |rr~0 (t)| . Throughout this section, we
will assume that r~0 (t) 6= ~0 (for the tangent vector at the point P to exist).
Now since T~ is a unit vector, it has constant length 1. So we know from one
of the previous lectures that T~ and T~ 0 will be orthogonal to each other. This
means that we will always be able to find out a unique direction (namely,
the direction of the vector T~ 0 ), which is perpendicular to the tangent vector
r~0 (t). So, we are going to choose the direction of T~ 0 as the direction of the
normal vector, which is perpendicular to the unit tangent vector T~ .

Remark 3.0.1. The vector T~ is a unit vector does not imply that the vector
T~ 0 is also a unit vector.

Take
~0
~ = T .
N
|T~ 0 |
Then N ~ is a unit vector. Also, T~ and N~ are perpendicular to each other.
This vector N~ is called the principal unit normal vector to the curve C
at the point P .
The vector T~ × N~ is perpendicular to both T~ and N
~ . Let

~ = T~ × N
B ~.

So the vectors T~ , N~ and B~ form a coordinate system at the point P (you

can consider a local coordinate system by considering 3 axes along the 3
directions T~ , N
~ , and B),
~ and this vector B~ is called the binormal vector
to the curve C at the point P . Figure 3.0.1 illustrates this.

Example 3.0.2. Let ~r(t) =< cos t, sin t, t >. Then ~r denotes a helix. And

r~0 (t) < − sin t, cos t, 1 >

T~ = = √ .
|r~ (t)|
0 2

7
 Figure 3.0.1: The normal and binormal vectors

Hence
1 1
T~ 0 = √ < − cos t, − sin t, 0 > and |T~ 0 | = √ .
2 2
~ is then given by
The unit normal vector N
~0
~ = T =< − cos t, − sin t, 0 > .
N
|T~ 0 |

î ĵ k̂
~ = T~ × N
Also B ~ = − sin
√t cos
√ t √1 . 
2 2 2
− cos t − sin t 0

4 Motion in space: Velocity and acceleration

Consider the space curve C in Figure 4.0.1. Let ~r(t) be a parametrization of
C. If we consider the curve C to be the trajectory of a particle moving in
space, and if the parameter t denotes time, then the derivative

d~r ~r(t + h) − ~r(t)

= lim
dt h→0 h
is nothing but the velocity vector ~v of the particle. In Figure 4.0.1, the point
P denotes the position of the particle at time t, and the point Q denotes the
position of the particle after time h. Let ~r(t) =< x(t), y(t), z(t) >. Then

8
 Figure 4.0.1: Motion of a particle in space

d~
r dx dy dz
from one of the earlier lectures, we know that dt
=< , ,
dt dt dt
>. So

dx dy dz
~v =< , , >.
dt dt dt
Again, the magnitude |~v | of ~v is nothing but the speed of the particle. Since,
we already know from one of the previous lectures that |r~0 (t)| = dsdt
(where s
denotes the arc length function), it follows that

ds
|~v | = |r~0 (t)| = .
dt
ds
This simply means that if we consider the arc length function s, then dt
is
nothing but the speed. Also, the acceleration is given by

d~v d2~r d2 x d2 y d2 z
= 2 =< 2 , 2 , 2 > .
dt dt dt dt dt

5 Functions of several variables

If we want to express some physical phenomenon mathematically, then it
always involves at least two variables, namely, space and time. We will now
study functions of more than one variables. Let us start with functions of 2
variables.

9
5.1 Functions of two variables
Let x and y be the two variables. Let R2 denote the set of all ordered pairs
(x, y), where x, y belong to R (Here R denotes the real line.). Let D be a
subset of R2 .

Definition 5.1.1. A function f : D → R is a rule which assigns to each

point (x, y) ∈ D, a unique point f (x, y) in R. The set D is called the domain
of the function f , and the subset of R in which the points f (x, y) belong for
all (x, y) ∈ D is called the range of f . We denote the range of f by the
symbol Rf . So Rf is given by

Rf = {z ∈ R|z = f (x, y) where (x, y) ∈ D ⊆ R2 }.

Remark 5.1.2. We often write z = f (x, y) to make explicit the value taken
on by f at the general point (x, y). In Definition 5.1.1 above, x and y are
independent variables, and z is the dependent variable. The domain D of f is
a subset of R2 and the range Rf of f is a subset of R. See Figure 5.1.1 for an
illustration.

Figure 5.1.1: Functions of two variables

10
Example 5.1.3. For each of the following functions, evaluate f (3, 2) and
sketch the domain
√ of f :
(a) f (x, y) = x+y+1
x−1
.
2
(b) f (x, y) = x ln(y − x).
To find out the domain of f , we have to find out all (x, y) ∈ R2 where the
function f is well defined.
For the function in (a), we can easily that f is well defined only when x + y +
1 ≥ 0 and x − 1 6= 0. So the domain of f in this case is the set of all points
in R2 lying towards the right hand side of the straight line x + y + 1 = 0
(including this line), and excluding the points on the straight line x = 1.
Figure 5.1.2 depicts this.
To find the domain of the function in part (b), observe that in this case, f

Figure 5.1.2: Domain of f in Example 5.1.3(a)

is well defined only when y 2 − x ≥ 0, that is, when x ≤ y 2 . So the domain of

f is shown by the shaded region in Figure 5.1.3.
One can easily calculate f (3, 2) for the functions given in (a) and (b) both.

p
Example 5.1.4. Find the domain and range of h where h(x, y) = 9 − x2 − y 2 .
For the function h to be well defined, we must have 9 − x2 − y 2 ≥ 0, that is,
we must have x2 + y 2 ≤ 9. So the domain D of h is given by:

D = {(x, y) ∈ R2 |x2 + y 2 ≤ 9}.

The range Rh of h is by definition the set {z ∈ R|z = h(x, y), (x, y) ∈ D}.
Observe that if (x, y) ∈ D, then 9 − x2 − y 2 ≤ 9. This implies that if (x, y) ∈

11
 Figure 5.1.3: Domain of f in Example 5.1.3(b)

p
D, then z = 9 − x2 − y 2 ≤ 3 (We are considering the positive square root
here. Throughout this course, whenever we consider square roots, they will
be positive square roots only, unless specifically mentioned). Clearly, since
we are considering the positive square root, we must also have z ≥ 0. So we
can now see that for any (x, y) ∈ D, 0 ≤ z = h(x, y) ≤ 3. Hence Rh = [0, 3].


5.2 Graph of a function of two variables

Consider a function f of a single variable x over an interval [a, b]. Then the
graph Gf of f is given by:

Gf = {(x, y) ∈ R2 |y = f (x), x ∈ [a, b]}.

Clearly, when f is a function of a single variable x, Gf ⊆ R2 .

Now consider a function f of two variables x and y. Let

f :D→R

be a function of two variables, where D ⊆ R2 is the domain of f . Then the

graph Gf of f is given by:

Gf = {(x, y, z) ∈ R3 |z = f (x, y), (x, y) ∈ D ⊆ R2 }.

Clearly in this case, we have Gf ⊆ R3 . See Figure 5.2.1 for an illustration.

In Figure 5.2.1, we have done the following:

12
 Figure 5.2.1: Graph of a function of two variables

For each point (x, y) in the domain D of f (that is, for each point (x, y, 0)
in the xy-plane, where (x, y) ∈ D), we have plotted a point in the space at
a height z = f (x, y) above it.

Now if we consider all the points of the form (x, y, z) where z = f (x, y) and
(x, y) ∈ D, we will obtain a surface S in the space (as shown in Figure 5.2.1).
This surface S is the graph of the function f of two variables.

Remark 5.2.1. In Figure 5.2.1, everything has been drawn in the upper half
of the xyz-space. But it is not always necessary that the figure has to lie in the
upper half.

Example 5.2.2. Sketch the graph of the function f (x, y) = 6 − 2x − 3y.

Let z = f (x, y) = 6 − 2x − 3y. Then we have:

2x + 3y + z = 6.

If we rewrite the above equation in the form xa + yb + zc = 1, then it will

represent a plane in space which intersects the x-axis at a = 3 units, the y-axis
at b = 2 units, and the z-axis at c = 6 units. This is because 2x + 3y + z = 6
is nothing but x3 + y2 + z6 = 1. Figure 5.2.2 depicts this. 
p
Example 5.2.3. p Sketch the graph of the function g(x, y) = 9 − x2 − y 2 .
Let z = g(x, y) = 9 − x2 − y 2 . Then we have:

x2 + y 2 + z 2 = 9.

13
 Figure 5.2.2: Graph of the function in Example 5.2.2

This is the equation of a sphere centered at the origin with radius 3.

Now, if we consider any point (x, y) in the xy-plane lying on the disc of
radius 3 centered at the origin, then corresponding to that point (x, y), we
will get two values of z on the sphere (Observe that if we draw a vertical
line at the point (x, y) which is parallel to the z-axis, then that line will
intersect the sphere at 2 points.). This violates the definition of a function,
because corresponding to each point in the domain of definition of a function,
we should get a unique point in the range. That’s why we have to consider
only the positive square root of 9 − x2 − y 2 while considering the function
g(x, y). This will give us the upper hemisphere as the graph of the function
g. Figure 5.2.3 depicts this. Note that if the function had been g(x, y) =

Figure 5.2.3: Graph of the function in Example 5.2.3

p
− 9 − x2 − y 2 , the lower hemisphere would have been the graph of g. 

14
 So we have now understood that the graph of a function of 2 variables can
be visualized as a surface. Figure 5.2.4 shows some functions of 2 variables
and their graphs.

Figure 5.2.4: Graphs of some functions

15
 2 2
Remark 5.2.4. Whenever the given function of 2 variables has e−x −y or
2 2
ex +y in it, we are going to get a “bell-shaped” surface as a graph. See (a) and
(b) of Figure 5.2.4 for an illustration.

5.3 Contour curves and Level curves

Motivation: we are considering the map of India in the xy-plane, and the
temperature in various places of India is denoted by z = T (x, y). Then we
will get a surface in the space as the graph of the function T (x, y). If we cut
this surface by the z = k plane (where k is a constant), then we will get a
curve on the surface, which represents all places in India having the same
temperature k. If we take the projection of such curves on the xy-plane, then
we will get some curves in the xy-plane (that is, on the map of India) called
isotherms. Isotherms are nothing but examples of level curves, which we
will define soon.
Example: Let z = x2 + y 2 . Then this equation represents a paraboloid. If
we cut the paraboloid by the z = k plane (where k is a constant), then the
plane cuts the paraboloid in the circle x2 + y 2 = k (Observe here that the
plane z = k is parallel to the xy-plane.). Take the projection of the circle
x2 + y 2 = k on the xy-plane. Then we will get the circle on the xy-plane. If
k1 > k and we do the same thing with the plane z = k1 , then we are going
to get a bigger circle as a projection on the xy-plane. Such circles (which are
the projections) on the xy-plane are examples of level curves.
Definition 5.3.1. Let f (x, y) be a function of 2 variables. Let k be a con-
stant in the range of f . Then the curve obtained by considering the intersec-
tion of the plane z = k with the graph of f (x, y) is called a contour curve.
And the projections of the contour curves on the xy-plane are called level
curves. 
Remark 5.3.2. Contour curves are curves along the surface formed by the
graph of the given function of 2 variables. Level curves are projections of the
contour curves on the xy-plane.
Remark 5.3.3. In Definition 5.3.1 above, it is important that the constant k
must lie in the range of f .
Figure 5.3.1 illustrates the concepts of a contour curve and a level curve.
Figure 5.3.2 depicts the contour curves and the level curves for a hill.

16
 Figure 5.3.1: Contour curve and level curve

Figure 5.3.2: Contour curves and level curves for a hill

Figure 5.3.3 gives an exercise on level curves.

17
 Figure 5.3.3: An exercise

5.4 Functions of three or more variables

A function f of 3 variables, is a rule that assigns to each ordered triple
(x, y, z) in a domain D ⊆ R3 , a unique real number denoted by f (x, y, z).
For instance, the temperature T at any point on the surface of the earth
depends on the longitude x, the latitude y of the point, and the time t. So
we can write T = f (x, y, t).
Suppose we have the function f (x, y, z) (where (x, y, z) ∈ D ⊆ R3 ) and we
want to find the graph of this function f . Observe that in this case, the

18
graph Gf of f is given by:

Gf = {(x, y, z, u) ∈ R4 |u = f (x, y, z), (x, y, z) ∈ D ⊆ R3 }.

But it is not possible to visualize a graph in R4 . However, if we take

f (x, y, z) = k, where k is a constant belonging to the range of f , then it
will represent a surface in R3 . For example, if we take x2 + y 2 + z 2 = 9, then
it is going to represent a sphere of radius 3.
Just like in case of level curves, if for a function f of 3 variables x, y, z, we
equate f (x, y, z) = k (where k belongs to the range of f ), then we are going to
obtain a surface in R3 . Such surfaces are called level surfaces. Figure 5.4.1
gives certain level surfaces for the function f (x, y, z) = x2 + y 2 + z 2 .
It is not possible to visualize the graph of a function f of 3 variables, since

Figure 5.4.1: Level surfaces

that will lie in a 4 dimensional space. However, we do gain some insight

about the function f by examining its level surfaces, which are the surfaces
with equation f (x, y, z) = k, where k is a constant lying in the range of f .
If the point (x, y, z) moves along a level surface, then the value of f (x, y, z)
remains fixed.

19
6 Limit of a function of two variables
Definition 6.0.1. Let f be a function of 2 variables whose domain D ⊆ R2
includes points arbitrarily close to the point (a, b). Then we say that the
limit of f (x, y) as (x, y) approaches (a, b) is L and we write
lim f (x, y) = L,
(x,y)→(a,b)

if for every number  > 0, there exists a number δ > 0 such that
p
|f (x, y) − L| <  whenever (x, y) ∈ D and 0 < (x − a)2 + (y − b)2 < δ.


6.1 Geometrical interpretation of limit

Let Dδ (a, b) denote the discpin the xy-plane having center at (a, b) and radius
δ. Then the condition 0 < (x − a)2 + (y − b)2 < δ simply means all those
points in the domain D which are not equal to (a, b) and which lie in the
interior of the disc Dδ (a, b). The definition of limit then says that for all
(x, y) ∈ D which are not equal to (a, b) and which lie in the interior of the
disc Dδ (a, b), we must have that f (x, y) lies in the interval (L − , L + ).
Figure 6.1.1 explains this.
Figure 6.1.2 depicts the phenomenon in the 3 dimensional space. Observe

Figure 6.1.1: Geometrical interpretation of limit

that the disc on the surface S in Figure 6.1.2 is the image of Dδ (a, b) un-
der the map f . The highest point and the lowest point on the surface Sδ

20
corresponding to the disc Dδ (a, b) (See Figure 6.1.2) must lie in the interval
(L − , L + ).

Figure 6.1.2: Limit of a function of 2 variables

6.2 Independence of path

In this subsection, we will see that the limit lim(x,y)→(a,b) f (x, y) exists if and
only if whatever path we take in order to make the variable point (x, y) to
approach the point (a, b), we obtain the same value of the limit. Let us begin
with an example first.

Example 6.2.1. Let us compare the behavior of the functions f (x, y) and
g(x, y) as (x, y) approaches the origin (0, 0), where

sin(x2 + y 2 )
f (x, y) =
x2 + y 2
and
x2 − y 2
g(x, y) = .
x2 + y 2
In Figure 6.2.1, we can see a table for some values of f (x, y) as (x, y) ap-
proaches the origin (0, 0). Observe that in the table given in Figure 6.2.1, if
one goes from −1 to 0 along the x and y axes both, then the value of f (x, y)
is approaching 1. The same is true if we go from 1 to 0 along the x and the
y axes.
However, if we look at a similar table (Table 2 in Figure 6.2.2) for g(x, y),

21
 Figure 6.2.1: Table 1 for Example 6.2.1

we can see that as (x, y) approaches the origin, the values of g(x, y) are not
converging anywhere. So we can’t conclude anything about the behavior of
the function g(x, y) near the origin.
So we can say intutively that

lim f (x, y) = 1
(x,y)→(0,0)

and
lim g(x, y) does not exist.
(x,y)→(0,0)

Recall from single variable calculus that

lim f (x)
x→a

exists and equals L if and only if

lim f (x) = lim− f (x) = L.

x→a+ x→a

22
 Figure 6.2.2: Table 2 for Example 6.2.1

That is, it does not matter whether x approaches the point a from the left
or the right hand side. In both the cases, the value of the function f (x)
approaches L.
In the case of functions of a single variable, the variable point x can approach
the point a from the left and the right only. But for a function of 2 variables,
this is not the case. There are infinitely many directions by which the variable
point (x, y) can approach the point (a, b). Moreover, the variable point (x, y)
can approach (a, b) not just through straight lines, but also via curves (A
straight line can also be considered as a curve.).
Suppose
lim f (x, y) = L1 along a curve C1
(x,y)→(a,b)

and
lim f (x, y) = L2 along a curve C2 .
(x,y)→(a,b)

If L1 6= L2 , then definitely we can say that the limit of the function f (x, y)
DOES NOT EXIST at (a, b).
So whatever path we take in order to make the variable point (x, y) to ap-
proach the point (a, b), we must obtain the same value of the limit of the

23
function. Then only we can say that the limit of f (x, y) EXISTS at the point
(a, b). In other words, the independence of path must be maintained. Fig-
ure 6.2.3 illustrates this.

Figure 6.2.3: Independence of path

Example 6.2.2. Recall the function g(x, y) from Example 6.2.1. We had

x2 − y 2
g(x, y) = .
x2 + y 2

In Example 6.2.1, we had intutively concluded that the limit of g(x, y) does
not exist as (x, y) approaches the origin. Let us now see a formal proof of
this fact. Observe that
 2
x − y2 −y 2

lim = lim = −1,
(x,y)→(0,0) along y−axis x2 + y 2 y→0 y 2

as we have x = 0 along the y-axis. And

 2
x − y2 x2

lim = lim = 1,
(x,y)→(0,0) along x−axis x2 + y 2 x→0 x2

as we have y = 0 along the x-axis.

Since −1 6= 1, we can conclude that the limit does not exist.

24
Another way to prove that the limit does not exist is the following: Let (x, y)
approach the origin along the straight line y = mx. Then we have:
 2
x − y2 x2 − m 2 x2 1 − m2

lim = lim = .
(x,y)→(0,0) along y=mx x2 + y 2 x→0 x2 + m2 x2 1 + m2

For different values of m, that is, if (x, y) approaches (a, b) through different
straight lines having slope m, we will get different limits. So we can conclude
that the limit does not exist. 

Remark 6.2.3. It is always easy to prove that a limit does not exist by con-
sidering two or more different paths. But it is NOT EASY to prove that a limit
exists by considering the path method. That’s because, there are infinitely many
paths via which the variable point (x, y) can approach the point (a, b). So for
proving that a limit exists, we should always go by the  − δ definition (given
earlier in this section).
2
Example 6.2.4. Let f (x, y) = x3x y
2 +y 2 . Find the limit if it exists at the point

(0, 0).
Now if we consider the limit of f along the x-axis or the y-axis or along the
y = mx line, it is easy to see that all these 3 approaches will give 0 as the
value of the limit. But there are infinitely many directions in which (x, y)
can approach the origin. Since these 3 approaches were giving us the same
value 0 as the limit, therefore we may assume (for the time being) that L = 0
and verify it via the  − δ definition. That is, we have to verify/prove the
following:

Given any  > 0, there exists a δ > 0 such that

3x2 y p
| − 0| <  whenever 0 < (x − 0)2 + (y − 0)2 < δ.
x2 + y 2
Observe now that
3x2 y 3x2 |y|
| − 0| = .
x2 + y 2 x2 + y 2
x2
Since x2 ≤ x2 + y 2 , we have x2 +y 2
≤ 1. This implies that

3x2 y 3x2 |y|

| − 0| = ≤ 3|y|.
x2 + y 2 x2 + y 2

25
 p p p
But 3|y| = 3 y 2 <  whenever 0 < x2 + y 2 < 3 . This is because y 2 ≤
p
x2 + y 2 . So, if we choose δ = 3 , then we can see that we can readily
prove/verify the required statement. Hence we conclude that

lim f (x, y) exists and equals 0.

(x,y)→(0,0)

—————-END——————

26