Bourbaki

1. Open sets, neighbourhoods, closed sets

1. Open sets

Definition 1

A topological structure* (or, more briefly, a topology) on a set XX is a structure given by a set D\mathscr{D} of subsets of XX, having the following properties (called axioms of topological structures):

(O_I) Every union of sets of D\mathscr{D} is a set of D\mathscr{D}.

(O_{II}) Every finite intersection of sets of D\mathscr{D} is a set of D\mathscr{D}.

The sets of D\mathscr{D} are called open sets of the topological structure defined by D\mathscr{D} on XX.

Definition 2

A topological space is a set endowed with a topological structure*.

The elements of a topological space are often called points. When a topology has been defined on a set XX, this set is said to be the set underlying the topological space XX.

Axiom (O_I) implies in particular that the union of the empty subset of D\mathscr{D}, i.e. the empty set belongs to D\mathscr{D}. Axiom (O_{II}) implies that the intersection of the empty subset of D\mathscr{D}, i.e. the set XX, belongs to D\mathscr{D}.

To show that a set D\mathscr{D} of subsets of XX satisfies (O_{II}), it is often convenient to prove separately that it satisfies the following two axioms, whose conjunction is equivalent to (O_{II}): (O_{II\ a}) The intersection of two sets of D\mathscr{D} belongs to D\mathscr{D}. (O_{II\ b}) XX belongs to D\mathscr{D}.

Examples of topologies. Given any set XX, the set of subsets of XX consisting of XX and ∅\varnothing satisfies axioms (O1)(O_1) and (OII)(O_{II}) and therefore defines a topology on XX. So does the set P(X)\mathfrak{P}(X) of all subsets of XX: the topology it defines is the discrete topology on XX, and the set XX with this topology is called a discrete space.

A covering (Ui)i∈I(U_i)_{i \in I} of a subset AA of a topological space XX is said to be open if all the UiU_i are open in XX.

Definition 3

A homeomorphism of a topological space XX onto a topological space X′X' is an isomorphism of the topological structure of XX onto that of X′X';* that is to say, in accordance with the general definitions a bijection of XX onto X′X' which transforms the set of open subsets of XX into the set of open subsets of X′X'.

XX and X′X' are said to be homeomorphic if there is a homeomorphism of XX onto X′X'.

Example

If XX and X′X' are two discrete spaces, any bijection of XX onto X′X' is a homeomorphism.

The following criterion follows immediately from the definition of a homeomorphism: for a bijection ff of a topological space XX onto a topological space X′X' to be a homeomorphism, it is necessary and sufficient that the image under ff of each open set in XX is an open set in X′X', and that the inverse image under ff of each open set in X′X' is an open set in XX.

2. Neighbourhoods

Definition 4

Let XX be a topological space and AA any subset of XX. A neighbourhood of AA is any subset of XX which contains an open set containing AA. The neighbourhoods of a subset {x}\{ x \} consisting of a single point are also called neighbourhoods of the point xx*.

It is clear that every neighbourhood of a subset AA of XX is also a neighbourhood of each subset B⊂AB \subset A; in particular, it is a neighbourhood of each point of AA. Conversely, suppose AA is a neighbourhood of each of the points of a set BB, and let UU be the union of the open sets contained in AA; then U⊂AU \subset A, and since each point of BB belongs to an open set contained in AA, we have B⊂UB \subset U; but UU is open by virtue of (O1)(O_1), hence AA is a neighbourhood of BB. In particular:

Proposition 1

A set is a neighbourhood of each of its points if and only if it is open.

The everyday sense of the word "neighbourhood" is such that many of the properties which involve the mathematical idea of neighbourhood appear as the mathematical expression of intuitive properties; the choice of this term thus has the advantage of making the language more expressive. For this purpose it is also permissible to use the expressions "sufficiently near" and "as near as we please" in some statements. For example, Proposition 1 can be stated in the following form: a set AA is open if and only if, for each x∈Ax \in A, all the points sufficiently near xx belong to AA. More generally, we shall say that a property holds for all points sufficiently near a point xx, if it holds at all points of some neighbourhood of xx.

Let us denote by B(x)\mathfrak{B}(x) the set of all neighbourhoods of xx. The sets B(x)\mathfrak{B}(x) have the following properties:

(V_I) Every subset of XX which contains a set belonging to B(x)\mathfrak{B}(x) itself belongs to B(x)\mathfrak{B}(x).

(V_{II}) Every finite intersection of sets of B(x)\mathfrak{B}(x) belongs to B(x)\mathfrak{B}(x).

(V_{III}) The element xx is in every set of B(x)\mathfrak{B}(x).

Indeed, these three properties are immediate consequences of Definition 4 and axiom (O_{II}).

(V_{IV}) If VV belongs to B(x)\mathfrak{B}(x), then there is a set WW belonging to B(x)\mathfrak{B}(x) such that, for each y∈Wy \in W, VV belongs to B(y)\mathfrak{B}(y).

By Proposition 1, we may take WW to be any open set which contains xx and is contained in VV.

This property may be expressed in the form that a neighbourhood of xx is also a neighbourhood of all points sufficiently near to xx.

These four properties of the sets B(x)\mathfrak{B}(x) are characteristic. To be precise, we have:

Proposition 2

If to each element xx of a set XX there corresponds a set B(x)\mathfrak{B}(x) of subsets of XX such that the properties (V_I), (V_{II}), (V_{III}) and (V_{IV}) are satisfied, then there is a unique topological structure on XX such that, for each x∈Xx \in X, B(x)\mathfrak{B}(x) is the set of neighbourhoods of xx in this topology.

By Proposition 1, if there is a topology on XX satisfying these conditions, the set of open sets for this topology is necessarily the set D\mathscr{D} of subsets AA of XX such that for each x∈Ax \in A we have A∈B(x)A \in \mathfrak{B}(x); hence the uniqueness of this topology if it exists.

The set D\mathscr{D} certainly satisfies axioms (O_I) and (O_{II}): for (O_I), this follows immediately from (V_I), and for (O_{II}), from (V_{II}). It

!Diagram showing sets V, U, W, and points x, y, z

Figure 1.

remains to show that, in the topology defined by D\mathscr{D}, B(x)\mathfrak{B}(x) is the set of neighbourhoods of xx for each x∈Xx \in X. It follows from (V_I) that every neighbourhood of xx belongs to B(x)\mathfrak{B}(x). Conversely, let VV be a set belonging to B(x)\mathfrak{B}(x), and let UU be the set of points y∈Xy \in X such that V∈B(y)V \in \mathfrak{B}(y); if we can show that x∈U,U⊂Vx \in U, U \subset V and U∈DU \in \mathscr{D}, then the proof will be complete. We have x∈Ux \in U since V∈B(x)V \in \mathfrak{B}(x); also U⊂VU \subset V, for every point y∈Uy \in U belongs to VV by reason of (V_III) and the hypothesis V∈B(y)V \in \mathfrak{B}(y). It remains to show that U∈DU \in \mathscr{D}, i.e. that U∈B(y)U \in \mathfrak{B}(y) for each y∈Uy \in U; now (Fig. 1) if y∈Uy \in U then by (V_IV) there is a set WW such that for each z∈Wz \in W we have V∈B(z)V \in \mathfrak{B}(z); since V∈B(z)V \in \mathfrak{B}(z) means that z∈Uz \in U, it follows that W⊂UW \subset U, and therefore, by (V_I), that U∈B(y)U \in \mathfrak{B}(y).

Q.E.D.

Proposition 2 shows that a topology on XX can be defined by means of the sets B(x)\mathfrak{B}(x) of neighbourhoods of points of XX, subject only to the axioms (V_I), (V_II), (V_III) and (V_IV).

Example

We may define a topology on the set QQ of rational numbers by taking for open sets all unions of bounded open intervals; the set of these subsets certainly satisfies (O_I), and to see that it satisfies (O_II) it is enough to remark that if the intersection of two open intervals ]a,b[]a, b[ and ]c,d[]c, d[ is not empty, then it is the interval ]α,β[]\alpha, \beta[, where α=sup⁡(a,c)\alpha = \sup(a, c) and β=inf⁡(b,d)\beta = \inf(b, d). We get the same topology by defining for each x∈Qx \in Q the set B(x)\mathfrak{B}(x) of neighbourhoods of xx to be the set of subsets containing an open interval to which xx belongs. The topological space obtained by assigning this topology to QQ is called the rational line (cf. Chapter IV, § 1, no. 2). Notice that in this space every open interval is an open set. \* We can define a topology on the set RR of real numbers in the same way; RR with this topology is called the real line (cf. § 2, Exercise 5 and Chapter IV, § 1, no. 3). \

3. Fundamental systems of neighbourhoods; bases of a topology

Definition 5

In a topological space XX, a fundamental system of neighbourhoods of a point xx (resp. of a subset AA of XX) is any set G\mathfrak{G} of neighbourhoods of xx (resp. AA) such that for each neighbourhood VV of xx (resp. AA) there is a neighbourhood W∈GW \in \mathfrak{G} such that W⊂VW \subset V.

If G\mathfrak{G} is a fundamental system of neighbourhoods of a subset AA of XX, then every finite intersection of sets of G\mathfrak{G} contains a set of G\mathfrak{G}.

Examples. 1) In a discrete space (no. 1) the set {x}\{ x \} alone constitutes a fundamental system of neighbourhoods of the point xx. 2) On the rational line Q\mathbf{Q} the set of all open intervals containing a point xx is a fundamental system of neighbourhoods of this point. So is the set of open intervals ]x−1/n,x+1/n[]x - 1/n, x + 1/n[, and the set of closed intervals [x−1/n,x+1/n][x - 1/n, x + 1/n], where nn runs through all the integers >0> 0, or through any infinite strictly increasing sequence of integers >0> 0. \* There are analogous results for the real line. \*

Definition 6

A base of the topology of a topological space XX is any set B\mathfrak{B} of open subsets of XX such that every open subset of XX is the union of sets belonging to B\mathfrak{B}.

Proposition 3

If XX is a topological space, then for a set B\mathfrak{B} of open subsets of XX to be a base of the topology of XX it is necessary and sufficient that for each x∈Xx \in X the set of V∈BV \in \mathfrak{B} such that x∈Vx \in V is a fundamental system of neighbourhoods of xx.

It is clear that the condition is necessary. Conversely, if it is satisfied, then given any open set UU and any x∈Ux \in U there is an open set Vx∈BV_x \in \mathfrak{B} such that x∈Vx⊂Ux \in V_x \subset U. The union of the sets VxV_x for x∈Ux \in U is therefore equal to UU. This completes the proof.

Example 1

The discrete topology has as a base the set of subsets of XX which consist of a single point.

Example 2

The set of bounded open intervals is by definition a base of the topology of the rational line (no. 2). \* Likewise, the set of bounded open intervals is a base of the topology of the real line. \

4. Closed sets

Definition 7

In a topological space XX, the complements of the open sets of XX are called closed sets.

Taking complements, we find that the axioms (OI)(O_I) and (OII)(O_{II}) take the following form:

(OI′)(O'_I) Every intersection of closed sets is a closed set.

(OII′)(O'_{II}) Every finite union of closed sets is a closed set.

The empty set and the whole space XX are closed (and therefore both open and closed; cf. § 11).

On the rational line, every interval of the form [a,→[[a, \rightarrow[ is a closed set, for its complement ]\leftarrow, a[ is open; likewise, every interval of the form ]\leftarrow, a] is a closed set; hence so is every bounded closed interval [a,b][a, b], since it is the intersection of the intervals [a,→[[a, \rightarrow[ and ]\leftarrow, b].

The set ZZ of rational integers is closed in the rational line, since its complement ⋃n∈Z]n,n+1[\bigcup_{n \in Z} ]n, n+1[ is open.

A covering (Fi)i∈I(F_i)_{i \in I} of a subset AA of a topological space XX is said to be closed if each of the FiF_i is closed in XX.

A homeomorphism ff of a topological space XX onto a topological space X′X' (no. 1) can be characterized as a bijection of XX onto X′X' such that the image under ff of every closed subset of XX is a closed subset of X′X' and the inverse image under ff of every closed subset of X′X' is a closed subset of XX.

5. Locally finite families

Definition 8

A family (Ai)i∈I(A_i)_{i \in I} of subsets of a topological space XX is said to be locally finite if for each x∈Xx \in X there is a neighbourhood VV of xx such that V∩Ai=∅V \cap A_i = \varnothing for all but a finite number of indices i∈Ii \in I. A set S\mathscr{S} of subsets of XX is said to be locally finite if the family of subsets defined by the identity map of S\mathscr{S} onto itself is locally finite.

It is clear that if (Ai)i∈I(A_i)_{i \in I} is a locally finite family of subsets and if Bi⊂AiB_i \subset A_i for each i∈Ii \in I, then the family (Bi)i∈I(B_i)_{i \in I} is locally finite.

Every finite family of subsets of a topological space XX is obviously locally finite; the converse is not true in general.

\* For example, in R\mathbf{R}, the open covering formed by the interval ]\leftarrow, 1[ and the intervals ]n,→[n, \rightarrow[ for each integer n⩾0n \geqslant 0 is locally finite; and each interval ]n,→[n, \rightarrow[ meets an infinite number of sets of this covering. \*

Proposition 4

The union of a locally finite family of closed subsets of a topological space XX is closed in XX.

Let (Fi)i∈I(F_i)_{i \in I} be a locally finite family of closed subsets of XX, and suppose that x∈Xx \in X does not belong to F=⋃i∈IFiF = \bigcup_{i \in I} F_i; then xx has a neighbourhood VV which meets only those FiF_i whose indices belong to a finite subset JJ of II. For each i∈Ji \in J let UiU_i be the complement of FiF_i; then ∁F\complement F contains the set V∩⋂i∈JUiV \cap \bigcap_{i \in J} U_i, which is a neighbourhood of xx since each UiU_i is open and contains xx. Hence, by Proposition 1 of no. 2, ∁F\complement F is open and therefore FF is closed in XX.

We note that the union of an arbitrary family of closed subsets of XX is not necessarily closed; for example, on the rational line Q\mathbf{Q}, the set ]2,1[]2, 1[ is the union of the closed sets

[1n,1−1n]for n>2, \left[ \frac{1}{n}, 1 - \frac{1}{n} \right] \quad \text{for } n > 2,
but is not closed.

6. Interior, closure, frontier of a set; dense sets

Definition 9

In a topological space XX, a point xx is said to be an interior point of a subset AA of XX if AA is a neighbourhood of xx. The set of interior points of AA is called the interior of AA and is denoted by A˙\dot{A}.

According to Definitions 9 and 4, a point xx is an interior point of AA if there is an open set contained in AA which contains xx; it follows that A˙\dot{A} is the union of all the open sets contained in AA, and hence is the largest open set contained in AA: in other words, if BB is an open set contained in AA, then B⊂A˙B \subset \dot{A}. Consequently, if AA and BB are two subsets of XX such that B⊂AB \subset A, then B˙⊂A˙\dot{B} \subset \dot{A}; and AA is a neighbourhood of BB if and only if B⊂A˙B \subset \dot{A}.

Remark

The interior of a non-empty set can be empty; this is the case for a set consisting of a single point which is not open, for example on the rational line \* (or the real line) *.

Proposition 1 of no. 2 can be restated as follows:

A set is open if and only if it coincides with its interior.

The property (VII)(V_{II}) of no. 2 implies that every point which is an interior point of each of two subsets AA and BB is an interior point of A∩BA \cap B; consequently

(1)(1)
A˙∩B˙=A˙∩B˙. \dot{A} \cap \dot{B} = \dot{A} \cap \dot{B}.

Every point which is interior to the complement of a set AA is said to be an exterior point of AA, and the set of these points is called the exterior of AA in XX; a point x∈Xx \in X which is an exterior point of AA is therefore characterized by the property that xx has a neighbourhood which does not meet AA.

Definition 10

The closure of a subset* AA of a topological space XX is the set of all points x∈Xx \in X such that every neighbourhood of xx meets AA, and is denoted by A‾\overline{A}.

This definition can be reformulated by saying that a point xx lies in the closure of a set AA if there are points of AA as near xx as we please to.

Every point which is not in the closure of AA is exterior to AA, and conversely; thus we have the formulae (which are duals of each other)

(2)(2)
CA‾=C^A,CA˙=C‾A. \mathscr{C}\overline{A} = \widehat{\mathscr{C}}A,\qquad \mathscr{C}\dot{A} = \overline{\mathscr{C}}A.

Hence, to any proposition on interiors of sets, there corresponds by duality a proposition on closures, and vice versa. In particular, the closure of a set AA is the smallest closed set which contains AA; in other words, if BB is a closed set such that A⊂BA \subset B, then A‾⊂B\overline{A} \subset B. If AA and BB are two subsets of XX such that A⊂BA \subset B, then A‾⊂B‾\overline{A} \subset \overline{B}.

A set is closed if and only if it coincides with its closure.

The dual of formula (1) is

(3)(3)
A∪B‾=A‾∪B‾. \overline{A \cup B} = \overline{A} \cup \overline{B}.

Proposition 5

Let* AA be an open set in XX; then for every subset BB of XX *we have

(4)(4)
A∩B‾⊂A∩B‾. A \cap \overline{B} \subset \overline{A \cap B}.

For suppose x∈A∩B‾x \in A \cap \overline{B}; then if VV is any neighbourhood of xx, V∩AV \cap A is a neighbourhood of xx, since AA is open; hence V∩A∩BV \cap A \cap B is not empty and therefore xx lies in the closure of A∩BA \cap B.

If xx lies in the closure of AA but not in AA, then every neighbourhood of xx contains a point of AA other than xx; but if x∈Ax \in A it can happen that xx has a neighbourhood which contains no point of AA except xx. We say then that xx is an isolated point of AA. In particular, xx is isolated in the whole space XX if and only if {x}\{x\} is an open set.

A closed set which has no isolated points is called a perfect set.

Definition 11

In a topological space XX, a point xx is said to be a frontier point of a set AA if xx lies in the closure of AA and in the closure of CA‾\overline{CA}. The set of frontier points of AA is called the frontier of AA.

The frontier of AA is therefore the set A‾∩CA‾‾\overline{A} \cap \overline{\overline{CA}}, which is closed. A frontier point xx of AA is characterized by the property that every neighbourhood of xx contains at least one point of AA and at least one point of CA‾\overline{CA}; xx may or may not belong to AA. The frontier of AA is the same as the frontier of CA‾\overline{CA}. The interior of AA, the exterior of AA and the frontier of AA are mutually disjoint and their union is the whole space XX.

Definition 12

A subset AA of a topological space XX is said to be dense in XX (or simply dense, if there is no ambiguity about XX) if A‾=X\overline{A} = X, i.e. if every non-empty open set UU of XX meets AA.

Example

\* We shall see in Chapter IV, § 1 that the set of rational numbers and its complement are dense on the real line. \* In a discrete space XX the only dense subset of XX is XX itself. On the other hand, every non-empty subset of XX is dense in the topology on XX for which the only open sets are ∅\varnothing and XX.

Proposition 6

If B\mathfrak{B} is a base of the topology of a topological space XX, there is a dense set DD in XX such that Card(D)≤Card(B)\mathrm{Card}(D) \leq \mathrm{Card}(\mathfrak{B}).

We may restrict ourselves to the case in which none of the sets of B\mathfrak{B} is empty (the non-empty sets of B\mathfrak{B} already form a base of the topology of XX). For each U∈BU \in \mathfrak{B}, let xVx_V be a point of UU; it follows from Proposition 3 of no. 3 that the set DD of the points xUx_U is dense in XX, and we have Card(D)≤Card(B)\mathrm{Card}(D) \leq \mathrm{Card}(\mathfrak{B}) (Set Theory, Chapter III, § 3, no. 2, Proposition 3).

Exercises

See the exercises for § 1.

Transcribed from the printed volume. Not translated.