ARE211, Fall2013

ANALYSIS4: THU, SEP 12, 2013 PRINTED: AUGUST 18, 2013 (LEC# 4)

1. Analysis (cont) 1.8. 1.8.1. 1.8.2. 1.8.3. 1.8.4. Topology of Rn Open Sets Interior of a Set Closed Sets Accumulation Point

lv er sio r na fo ck

Contents

n
1 1 1 3 4 4

dr af

1.8. Topology of Rn

1.8.1. Open Sets.

Denition: Given a universe X , a metric d and x X , the set Bd (x, |X ) = {y X : d(x, y ) < } is called the -ball about x w.r.t. X and d. Important to note that the denition of openness depends on a metric and a universe. Technically, we should write Bd (x, |X ) Examples:
1

Pr el i

m in

ar y

to
1. Analysis (cont)

nl

y:

pl

ea

se ch e

ANALYSIS4: THU, SEP 12, 2013

PRINTED: AUGUST 18, 2013

(LEC# 4)

Suppose X = [0, 1] and consider the -ball about 1: it is (1 , 1]. Suppose d is the discrete metric and X = R. For 1, the -ball about x R is the point itself. For > 1, the -ball about x R is R. More examples: Suppose X = Rn and d is the max metric, i.e., d(x, y) = max{|xi yi | : i = 1, ..., n}. Whats the shape of the ball? Its an ndimensional cube. Suppose X = Rn and d is the absolute value metric, i.e., d(x, y) = the shape of the ball? Its the n-dimensional version of a diamond. From now on, we will usually take for granted that the universe is Rn and the metric is the Euclidean metric. Unless we specify that these are otherwise, we will simply write B (x, ). Question: Why do we care that some open sets with restricted universes look wierd, e.g., that (1 , 1] is in fact open in [0, 1]?
n i=1 |xi

yi |. Whats

Answer: Because in economics we are often (usually) talking about a restricted universe. For example, R+ is a restricted universe. Now, we say that a function f attains a strict local maximum at x0 if there exists an open neighborhood U of x0 such that f (x0 ) > f (x) for all x U . Consider the function f (x) = 1 x + x2 . If the universe is R, then theres nothing special about the point 0. But if the universe is R+ , then there does exist an open neighborhood U of 0 such that f (x0 ) > f (x) for all x U . Point of this illustration is that for what we do, we want to be able to say that for some universes, sets like [0, ) can be open. Denition: A set A X is said to be open in X w.r.t. a metric d if for every x A, there exists > 0 such that Bd (x, |X ) A. Example: the interval (a, b) is open in R but not open in R2 , e.g., when viewed as a horizontal or vertical line in R2 .

ARE211, Fall2013

the set {x Rn : ||x|| < 1} is an open set the half-open interval {x [0, 1] : 1/2 < x 1} is open in [0, 1] but not open in R. the singleton set {x} R is open in R w.r.t. the discrete metric. the empty set is always open, no matter what: look at the defn: every point in the set must be enclosed by an open ball thats in the set. This defn is certainly satised by the empty set!! (As well see later, the empty set is also closed.) Theorem: For every > 0 and x Rn , the -ball about x, B (x, ), is open in Rn . Proof: Pretty obvious. The following theorem contains the most important basic properties of open sets Theorem: The intersection of a nite number of open subsets of Rn is an open subset of Rn . The union of an any collection (i.e., even an innite number) of open subsets of Rn is an open subset of Rn . Note that the intersection of a innite number of open subsets need not be open: Example: X n = (1/n, 1/n) R. The intersection of X n s, for all n, is the singleton set {0}, which is not open in R in the usual metric. Suprising things happen with suprising metrics: every subset of R is an open set w.r.t. the discrete metric: reason is that each scalar in R is open in this metric, and any set X R is a union of scalars in R. Context is particularly important to us as economists because we are often dealing with functions dened on a restricted domain, and well be interested in whether or not a subset of that domain is open or not in that domain. For example, often the domain of our functions is R+ : subsets of the form [0, ) are open in R+ but not in R.

1.8.2. Interior of a Set. Denition: For any set A X a point x A is called an interior point of A if there is an open set

ANALYSIS4: THU, SEP 12, 2013

PRINTED: AUGUST 18, 2013

(LEC# 4)

U X such that x U A. The interior of A is the collection of all interior points of A and is denoted int(A). (Note that int(A) may be empty.) As always whether a point is an interior point or not depends on universe and metric. Let A = [0, 1] R. In this case, int(A) = (0, 1). Let A = [0, 1] [0, 1]. In this case, int(A) = [0, 1]. Generally, given any universe X , the set X X is equal to int(X ). Theorem: For any set A X , the interior of A w.r.t. X is the union of all subsets of A that are open in X . Proof: Consider a point x contained in some open subset of A. Then its an interior point and hence, immediately, belongs to interior A. Now consider x A such that x is not contained in any open subset of A. Then it fails the defn of an interior point. Theorem: A set A is open in X i A = int(A).

1.8.3. Closed Sets. Informally, a set is closed if it contains its edges (boundary points). Denition: A set B X is closed in X if its complement in X (i.e., the set X \B ) is open. Note that X is closed in X , since its complement in X , i.e., the empty set is open. Sets can be neither open nor closed. An example is the set [0, 1) R.

1.8.4. Accumulation Point. Denition: A point x X is called an accumulation point of a set A X if for every > 0, the ball B (x, |X ) contains a point y A, y = x. Important note: At this point, some of my defns depart from the defns that appear in important texts, including Simon and Blume. Dont be too concerned about this. Unless youre super obsessive compulsive, just go with my defns. (As in when you are in Hong Kong you drive on the right side of the road.) If you are super obsessive compulsive and are worried about defns in books that dier from mine, come see me and Ill discuss.

ARE211, Fall2013

That is, a point is an accumulation point of a set if there are points in the set that are distinct from x, but arbitrarily near to x. But an accumulation point of a set neednt belong to the set. People are always confused about the relationship between an accumulation point and a limit. Limits are things that sequences have; accumulation points are things that sets have. (Some books dont necessarily make this distinction, however.) Example: The point 1 R is an accumulation point of the set (0, 1) R in the usual metric: > 0, 1 /2 B (1, ) and 1 /2 (0, 1). Example: Note that a set consisting of isolated points (e.g., the integers) contains no accumulation points. (Have to be careful here: the set S = {1/n : n N} consists entirely of isolated points. Moreover, 0 is an accumulation point of S . However, 0 is not contained in S . Example: More interesting example is the rational numbers Q. Turns out that every point in R is an accumulation point of Q in the usual metric, i.e., there is a rational number arbitrarily close to any real number. Once again, the notion of accumulation depends on the metric you use. Think about the discrete metric: take the set (0, 1); what are the accumulation points of this set; there arent any, since any two points that arent the same are distance 1 away from each other. As always, the notion of accumulation depends on the context or universe. Suppose that X = (0, 1). Consider the set [1/2, 1). What are its accumulation points in X ? Note that 1 isnt in X , so isnt an accumulation point in this context. This example provides another instance of how the universe matters. In X = (0, 1), all of the points in [1/2, 1) are accumulation points, so this set is closed in (0, 1). There is a very close relationship between the notions of accumulation points and closed sets. Theorem: A set A X is closed in X i A contains all of its accumulation points.

ANALYSIS4: THU, SEP 12, 2013

PRINTED: AUGUST 18, 2013

(LEC# 4)

That is, in order for A X to be closed, if x X and there are points arbitrarily close to x that are in A, then x must be in A also. Note that the integers are closed in R under the usual metric, since there are no accumulation points to contain. Think about the discrete metric. What sets are closed in the discrete metric? Theorem: Every set A X is closed in the discrete metric. Note that every set X X is closed with respect to itself. Cant be any accumulation points of X that arent included in X . Also, every point in X X is an interior point, hence X is open. Note that this implies that the empty set is both closed and open.