A Basic Course in Real Analysis

Prof. P. D. Srivastava
Department of Mathematics
Indian Institute of Technology, Kharagpur

Lecture - 11
Ordered set, least upper bound, greatest lower bound of set

(Refer Slide Time: 00:34)

Hello. So, we will discuss the least upper bound; greatest lower bound of a set and compact sets
today. So, this requires the concept of ordered set; ordered set first. We define like this; let S be a
set, S be a set, an order on S an order on S is a relation, is a relation denoted by sign; this sign <
which we call later on, we will less than sign or a smaller sign, with the following, with the
following two properties. The first property is that if x  S and y  S , then one and only one,
only one of the following statement, of the statement of the statements, that is x  y, x  y, y  x
is true, is true. And second one is say, if x, y, z they are the elements of S, and if x  y and y  z
then then x  z .

So, if a set S together with this operation which we call it is together with this relation less than
sign, satisfy these two properties, then we say that this is an order on S. This is an order on S,
this sign less than sign we normally say; we say x related to the means x is smaller than y, x is
smaller than y. And the negation of this, the negation of this is x  y . This is the negation part of
it, means x  y ; the negation of this will be x is typically less than y. So, that will be the sign for.
Now, obviously when we say the set of rational number or set of the real numbers, then this
order is defined; one can identify 2 real number or 2 rational number, one can say which one is
low smaller than the other or whether they are equal or weather one is greater than the other, like
this.

(Refer Slide Time: 04:12)

Then the set, order set means, a set, an ordered set S, order set, is a set S, is a set S, in which an
order is defined, in which an order is defined; for example, set rational number, set of all rational
numbers is an ordered set, is an ordered set, because if we take r and s, suppose these are 2
rational number and define the relation r  s means, s-r means, s-r is a positive rational, is a
positive rational number, is a positive rational number. Then obviously, it was satisfy these two
properties, which you, which studied. So, set of rational number is an ordered set; set of real
number is an order set, set of real number is an ordered set, ordered set.

However set of complex number, set of complex number is not, complex number is not an
ordered set. We cannot introduce the order between the two elements of a C, complex number,
set of all complex number. Because for example, if suppose I take the complex number e and 1,
and when we say e  1 ; e is the idle element say 1 this is e or i. I will say i, if you i, i, this is
complex number.

So, i  1 , it means is the positive we are assuming. So, i.i  1.1, that i square is 1  1 , which is a
absurd. It means our ordering which we introduce is not correct; similarly, one can show weather
if i  1 , we can again lead a contradiction and like this. So, we are unable to introduce the order
in the over the set of all complex number, that is the set of a complex number is not an ordered
field, field we will discuss in algebra or some.

Then we are interested in particular in defining the least upper bound and the greatest lower
bound. So, let is see first what is an upper bound and lower. Suppose S is an ordered set, S is an
ordered set and E is a non-empty subset of S; now, if there exist some  , there exist a  in S.
Such that all the elements of E, that is x  E is less than or equal to  , and this is true for every
xE .

(Refer Slide Time: 08:06)

Then we say, then we say E is, then we say beta E is bounded above, E is bounded above,
bounded above and  is an upper bound and  is known as upper bound, upper bound for E.
Now; obviously,  is an upper bound were there are many infinity many real numbers will be
available; which will be an upper bound for E any number; which is greater than  will act as an
upper bound for E. So, we are interested in a get least upper bound for it.

(Refer Slide Time: 09:08)

The same case happens if it is lower bound, we define the lower bound in a similar way. Suppose
S is an ordered, suppose S is an ordered set, ordered set and E is a sub set of S, non-empty sub
set of S. If there exist, if there exist number say   S , such that   x, x  E . Then we say E is
bounded bellow and  is the lower bound, is a lower bound, is a lower bound of E. So, their
again, there will be many lower bound available as soon as we take any number less than  ,
which will also behave as a lower bound. So, we will be interested in knowing what will be the
greatest lower bound of the set E. So, we introduced the concept of the upper bound and lower
bound as follow: this is the concept least upper bound; suppose S is an ordered set, S is an
ordered set and E be a non-empty subset of S, and also assume E is bounded above, E is bounded
above. Now, suppose there exist, suppose there exist an   S with the following properties.

 is an upper bound, is an upper bound of E, of E; this is the first property and second one is if,
I take a number slightly lower than  , then it should not be a as an upper bound, and second is
if,  is any number less than  , then  is not an upper bound, a upper bound of E. So, E is a
bounded above and  is a such a number, which is an upper bound of this, but if we take a
number slightly lower than  , then that number will not be a when upper it means,  is the
least upper bound for E. Than  is called, then  is called the least upper bound, upper bound
of E and we denote this thing is and also or some we also say it is a supremum, supremum of E,
of E and we write is, we write that least upper bound of set E, each  or it is the same as when
we say supremum of E.

(Refer Slide Time: 13:33)

The similar way, we define the greatest lower bound, in the similar way we can introduced the
concept of the greatest lower bound of E. So, what we assume is suppose S is an ordered set, S is
an ordered set; E is a sub set non-empty sub set of S and E is bounded below, bounded below.
Suppose they are exist and say   S ,   S with the following property, with the following
properties; the first is  is a lower bound, is a lower bound of E, is a lower bound of E. Second
one is if, a number if I choose, if a number say    . Then  is not an is not a lower bound of
E, is not a lower bound of E. If we take any number  slightly I have done it will not be lower
bound; then this  , then  is known as, as known as greatest lower bound of E or we can also
say it is the infimum, infimum of the set E, and we denote this as, denote it as,   infE  g.l.bE

So, this is the way we define the greatest lower bound and upper and the least upper bound,
greatest and least. Now let us take an example, suppose I take the set A, let A be the, A be the set
of all positive rational number p, p is positive rational number. So, it is positive rational numbers
such that, p 2  2 and let B is the set of all positive rational numbers; such that p 2  2 , p square
is greater than 2.

Now if we look that A and B; A is the set of all positive rational number who’s square is less
than 2 and B is the set of all positive rational number is square is greater than 2; obviously, A is
bounded above. In fact, all the elements of B will be the upper bound, will be act will act as an
upper bound for A, but A does not act the least of upper bound because we cannot get the
rational number which is for, which we can say is a really a number, which is an least upper
bound for A.

Similarly, for the B; if we look the, all the elements of B, satisfy this condition, then it has a
lower bound, and all the elements of A behave as a lower bound of this plus all rational number,
which are negative or 0 will behave as a lower bound for this, but neither A nor B has an upper
bound and the greatest lower bound. So, this will show that A contains no largest number and B
is clearly, A contains no largest number; while the B and B contains no smallest number. So, in
this case the greatest lower bound A has no greatest, B has no greatest lower bound and A has no
largest number.

(Refer Slide Time: 18:48)

So this, so therefore, we can say A has no least upper bound, least upper bound; where the B has
no greatest lower bound that is obviously true. let us take an another example, suppose I take the
set of rational number, let E1  {r  Q : r  0} , and E2 be the set of all rational number Q
rationales, such that r is greater than equal to zero, r is sorry less than equal to zero; same such
that r is less than equal to 0. Suppose I take this thing, than the set of rational number which is
less than 0; so obviously it has an upper bound 0, here also is upper bound 0. It means the
supremum value of E, so therefore supremum of E1 will be 0; supremum of E2 will also be 0, but
you see that supremum value of E1 does not belongs to E1 by the supremum value of the E2
belongs to it.

So, it is not necessary when we say their least upper bound or greatest lower bound, then it is not
require; it is not necessary that the that supremum value will be a point of the set, it may or may
not be the point of set, that we have observed here. Similarly, but both are having the same;
similarly, now another interesting property which is a connection with the order set is this is
known as the least upper bound property, least upper bound property.

What is the least upper bound property? An ordered set, an ordered set S is said to, is said to
have, is said to have the least upper bound, least upper bound property. If the following is true, is
true, following is true, but therefore if, if, E is a non-empty sub set of S, is not empty, is a non
empty sub set of S and E is bounded above, bounded above; then the supremum of E, that is least
upper bound of E will exist, an exist in S, exist in S. So, this is the least upper bound property of
a set; an order set as a said to have a least upper bound property if, the following is true. That is
if we take any sub set non-empty sub set of which is bounded above; then supremum we exist in
S, then we say this set S is a least upper bound property. If, for any set this supremum does not
exist, then the set will not have a least of upper bound property; for example, set of rational
numbers, set of rational numbers that is which is denoted by Q, does not have, does not have
least upper bound property, bound property and (Refer Slide Time: 13:33) that we have seen
already with this example.

Because, basically A and B these are the two sub sets of the rational numbers, and neither the A
nor B has an upper bound. It is not neither A has does not have a upper bound, B does not have
the lower bound for it. So, basically the set of rational number you can say does not have a least
upper bound property. Now, there is a relation between the greatest lower bound, least upper
bound and the least upper bound property. In fact, it is shown that, if the set is having the least
upper bound property, then it must have a greatest lower bound property also and that can be
judged in the next theorem.

The relation between the, relation between the least upper bound, greatest lower bound and this;
so theorem says or you can before this, you can write the remark; I would write the remark. Here
is every ordered set, there is a relation between the greater lower bound, that every order set,
order set with the least upper bound property, upper bound properties with the least also has
there is their remark is there is a close relation, close relation between, between greatest lower
bound and the least upper bound and that, and that, this that every order set with the least upper
bound property; also, has the has the greatest lower bound property.

(Refer Slide Time: 25:42)

This can be seen with the help of this result the theorem says that if suppose, suppose S is an
order set, ordered set with the least upper bound property, with the least upper bound property,
upper bound property and suppose B is non empty sub set of S, B is a non empty sub set of s
having and B is bounded below, bounded below. So, S is an order, order set which has a least
upper bound property and a set B has a property which is bounded below. Now; this together will
implies the least relation between the greatest lower bound property. So, what he say is if B is
bounded below and let L be the set of all all lower bounds of B, lower bounds of B; then the
supremum of B, that is the least upper bound of B; that is  will exist, exist in S and and this 
will be the infimum value of B; that is it will the greatest lower bound for B and in particular
infimum of this B exists in S.

If you want let us see the proof of this. What is given is S ordered set which has a upper bound
property; upper bound property means if B if any set is there, which is a sub set non empty sub
set of this and if has in upper bound; then the supremum of this will exist in this. Now here we
are assuming that S has a upper bound property and a non-empty sub set B is bounded below.
Then; because of this upper bound property and this condition we will saw the B will have the
greatest lower bound and infimum of B will exists in S; that is what it is. So, since B is given B
is bounded below; this is given and what is our L? L is the set of all lower bound of B. B is
bounded below it is already given is means there is a bound available. So, L is non-empty.

So, so this implies L is non-empty, now L is the set of all lower bound of B. So, what is L? So,
clearly since L is the collection of, L is the set of all lower bound, bounds, lower bounds of B.
So, basically L consist of those y, it means L  { y  S : y  x, x  b} , because L is the collection
of the lower bound. So, the y  x and y will be the lower bound for B and or such why we
satisfy this condition, will come in the class L and this will be a non-empty set; this one thing is
clear, now every x. So, if we look the L. L is the collection of those points which are low less
than equal to x for a B. It means every point of B behave as an upper bound for x.

So, clearly clearly every x  B is an upper bound, is an upper bound of L. So, it means L is
bounded above, thus L is bounded above, bounded above. So, L is a non-empty set which is
bounded above, it is a sub set of S, L  S . So, bounded above an is a sub set of S, is it not? Is it
not? So, we can apply the property because S is an ordered set, having the least of upper bound
property. So, by the property since S has a, since S, since S has a least upper bound property. So,
so by this L will have a supremum value.
(Refer Slide Time: 31:31)

So, L has a supremum value, supremum in S, in S exists, supremum value will exists and let it be
let  is that supremum value of L is  . Suppose supremum value of this  . Now if we choose
 ; if  is any number less than  , then  is not an upper bound of L, upper bound of L,
because  is the least upper bound. So, if we take any number lower than  , lower than  , than
that cannot behave even upper bound for it. Otherwise  will be the least upper bound. So,  if it
is less than  , it cannot be upper bounded L and what is our B? B is the set of those points which
are for such that every point of this, is an upper bound for it and here  is not coming an upper
bound for L. So; obviously,  cannot be a point in B. Because all the points B must be an upper
bound, is an upper bound which we have shown, but  is not an upper bound of L therefore,
gamma cannot be a point of B. So, what the this follows it implies, that it implies that
  x, x  B , because any number less than  cannot be a point of B. So,  will be the least
number and then   x . So, this shows that   L , that is would   L ; that is the supremum will
exists and it is in L.

Now if any number which is less than  , any number    , then  cannot be in L, because 
is the least upper bound and all the  is greater than. So, again it will not be in L. So, once it is
not in L,  is not in L,  in L, then what happened; that this  which is greater than  , in
other words that  will be the infumum value of B, because then  will be the, in which  will
be in B. So, it is a alpha beta. So, this shows  is a lower bound, lower bound of B, is a lower
bound for B; but  is not, but  is not as if    it will not the lower bound for. So, it is the
least. Therefore,   inf B and that proofs the result. So, this shows the. Now have having proof
this thing, we will come back again to the sets we are, we are discussing the open sets and close
sets etcetera, we are the supremum value, infimum value will be required. So, we need basically
we wanted to saw that result, we are the supremum concept and infimum concept is required.

(Refer Slide Time: 35:19)

So, that is why all these things were taking. So, this result we wanted to saw the result is let E be
a non-empty, non-empty set of real numbers, non-empty set of real numbers which is bounded
above, which is bounded above and let y is the supremum value of E, that is least upper bound
for E. Then the result says that y will be a point of closer of E, closer of E, means set E together
which is limit point is the closer set. Hence y  E , if E is closed. So; obviously, when E is close

set E  E . So, this is a second part follows immediately, they nothing. The first part we wanted
to saw first. So, that E is a non-empty set of the real number which is bounded above and
supremum of E is B suppose y, then y will be a point in E . Now since E is a basically,
E  E  E  , where E  , is the set of all limits points, set of limits points of E. Collection of all the
limits points of E denoted by E  , now if y  E , if y  E , y  E then; obviously, y will be the

element of E because it is union of this and y bar, so nothing to prove.

So, let us suppose y is not in E, but y is a limit point of E, we will saw then y is a limit point. So,
assume y  E , but we wanted to saw y is in E closer it means, y must be a limit point of E so
that we wanted to prove. So, let us see for every h  0 their exists, their exists, their exists, then
a point there exists, then a point say x  E , there exists a point x  E . Such that, such that
y  h  x  y holds. Why? Here this is our set say y. E, this is the set E; the point y is not in E is
outside of it. Then we can find for each h  0 , we can find at least some point which lies in the
interval say this is y, in between y-h to y, this point x will always be available, otherwise if it is
not so. Then y-h will behave as a, y-h will behave as a upper bound for E, if it is not so.

(Refer Slide Time: 39:10)

And this is true because otherwise y-h will act as an upper bound, least upper bound for E, as a
least upper bound of E; which is not true. Because y is given to be the upper bound. So, as soon
as you take a number slightly lower than this, then this number must be available. It means in
between y minus y n and y one can always get a at least one number of x, E which is available,
but what is the y minus y n. But this interval by y-h, is it not in real x lies. So, is it not a
neighborhood of y, with radius says h it is the left hand neighborhood left side, left hand left side
neighborhood of y in h. So, they are exists of neighborhood of y, which includes the point of x. It
means every neighborhood of this y will inputs at least some point of E.

So, this shows that, but this is the neighborhood, is a neighborhood of y with radius h; in which
the point x  E , in which they will exist a point x  E . So, this shows that y is the limit point of
E. Because the definition of the limit point of the set in means every neighborhood around the
point y, every neighborhood to y how small the radius may be must includes the points of E and
this is true here that if we take any neighborhood of y, at least one point x is available. The
otherwise, if it is not available then it will contradict to the fact that y is the supremum value of
E. So, if it is a limit point then y must be the point in the closer, E  hence y is in closer of this
set. So, this proves the result.

Now remark we can see; we know if E  Y  X , suppose we are X is a metric space. Then we
have seen that a set may be open in X, may be open in Y and may not remain open in X. This we
have seen just like a open interval (a, b) which we have seen it is open in R1 , but it is not open in
R2 . So, in case of the open set or close set the space which incluses the set is important. Here the
set is E is open. So, that is why we introduced the concept of an open set relative to the space,
relative to Y or relative to X.

So, we introduce here two definition; that one is, a set E is open, a set E is open, E is open sub
set of X. Means means to each p  E , belongs to E there is associated associated a positive
number, a positive number r such that such that the condition, such that the condition
d ( p, q)  r , q  X  q  E . Than we say E is open with respect to X, open sub set of X means
then E is said to be open open relative to X, relative to X. Similarly we can say Y, see Y is also
may take a place under the same may take topology d. So, E may also be open to respect to Y.
(Refer Slide Time: 44:24)

Then we say define, we define that a the set E is open, the set E is open open relative to Y,
relative to (Y, d) or metric a place (Y, d) if to each, if to each here also (X, d) you write (X, d) is
a matric a place. Relative to X Y, relative to Y if to each p  E there is, there is, there is
associated, they are each an associated an r  0 ; such that, such that the condition condition
d ( p, q )  r , q  Y  q  E .
Then we say it is a, it means what, suppose we have the set (X, d) which is a metric space a set E
this, is a set E we say it is open in X means, that if we take any point p; In E then one can always
find out neighborhood around the point p or a ball centered at p with a suitable radius say r, such
that all the points inside this is a point of E, is a point of E. Then we say that E is open in all the
point q, q which are of X, all the point q which are in X if they are there are the point E then we
say it is a open. It means that every point is interior point with respect to (X, d) but when we say
this is our Y, Y is a sub set of X. So, it is also a metric space with respect to d, then we say that E
is open with respect to Y.

Now here when you draw the neighborhood around the p, then the point q which you are
choosing, will must be a point of Y. Because you are not getting the one off course the point of Y
is also the point of X, but there may be some point which are in X, but not in Y. So, this relation
when d ( p, q )  r , q  Y  q  E . Then we say E is open in Y. So, as if there is no X, only E is a
sub set of Y and E will be the open set in Y every point of p is an interior point with respect to Y;
that is all. Then we say E is open relative to Y similarly E is open, now this has been shown that
a set E may be open with respect to Y, sub set may not be open with respect to the large set and
that example we have seen; however, in case of the compact set, we will see this result, this
restriction is not there. So, that is more fluently than our open set or close set. So, that is cover.

So, we will see that, before going for the compact set definition we have one more results, that
result shows, what will be the form of the open sets in the relative case. Suppose Y is a non
empty sub set of X, (X, d) be a metric space where, (X, d) is a metric space. Let (X, d) be a
metric and by be a non empty sub set of X. A sub set E of X, a sub set E of (X, d), of (X, d) is
open sub set E of (X, d) a sub set E of Y I am sorry. So, I was sub set E of (Y d), (Y, d) let it be
(Y, d), a sub set E of Y, a sub set E of Y is open is open relative to Y, relative to Y if an only if,
if an only if, if an only if E can be expressed as E  Y  G for some open sub set G of X.

So, what this results says is, let (X, d) be a metric space and Y is a non empty sub set of X, So, Y
under the same metric d will also be metric space and suppose E, is a sub set of Y. Then we say a
sub set E of Y will be open with respect to Y or related to Y if E can be expressed into this form,
for some open set G of X. If an only that is if E is of this form then E will be a open set, sub set
of Y and if E is open, then it can be express into this form. So, vice versa; let us see the proof of
this.

(Refer Slide Time: 50:45)

Suppose E is open relative to Y, E is open relative to Y, Y. We wanted to show E will be of this

once. So, why the definition of the relative to Y means to each p  E , there is a, there is a
positive number, positive number say our rp such that, such that the condition condition

d ( p, q)  rp , q Y  q  E . This is by definition when E is open relative to Y. Now let us

consider Vp  {q  X : d ( p, q)  rp } where q is in the elements of Y; let us q is in X, q is in X. So,

this is already there, Let us see now; obviously, this is a neighborhood. So, once in
neighborhood, it is an open set, it is an open set and G if I take the union of all these V p where
the p  E then this collection of the open set will also be able. So, it is an open set is an open
sub set of X. Clear? Nothing to...

Now, since p is in the neighborhood Vp with centered p and radius say rp , p is in the for all

p  E . This is while construction V, because p is the center of this neighborhood. So, it is clear

that, than it is clear, that E will be contend in G, which is G Y . G Y . Because this Vp, p is a
set in E and all the points in E belongs to Vp and G is the union of Vp. So, E every point of E is in
Y as well as G. as well as in G. So, it is intersection of this thing is obvious. By our choice, but
by our choice of Vp we can say, we can say that we have that Vp  Y  E . By our choice means,

because we have Vp constructed like this way, set of all that. This is There is a positive such that
this one is, so when restrict q to Y. Then all these points basically they are the points common to
E; intersection with this and containing E. So, by our choice, because this will be the set to each
p there is a positive because E is open, because E is an open set. So, this entire thing is available
in E, because E giving to be an open set related this. So, his is our choice therefore, for our for
Vp. So, that. So, the G Y , take the union of this G Y  E . And hence E will be equal to
G Y . So, one result is complete.

(Refer Slide Time: 54:54)

Conversely just one more, conversely if G is open, G is open in X and E is of the form
E  G Y , then every p  E , every p  E has a neighborhood Vp this is totally contend in G.
Because G is open and E is of this form. So, E will also be for any p  E means, it will be in G
and G is open. So, neighborhood must be available in G. Then the G, Vp  Y , neighborhood

inter section Y will be contend in E; because p is in there and G is in this form. So, intersection
will be available in E. So, that E is open. So, that E is open relative to Y. Y, relative to Y and that
is proves the results.

Thank you very much.

Thanks.