Tutorials › Topology (Undergraduate) › Why Finite Product and Box Topologies Agree

The Product Topology · Tutorial 229 of 799

Why Finite Product and Box Topologies Agree

See why finite products have identical product and box topologies by comparing their bases and checking examples with two and three factors.

Intermediate 9 min read

What You'll Learn

  • Define basic open sets for the product and box topologies
  • Prove the two topologies agree for finitely many factors
  • Describe product bases for two-factor and three-factor products
  • Verify the basis descriptions in real and finite discrete examples
  • Avoid confusing coordinate cylinders with restrictions on several coordinates

Two Descriptions of Open Sets in a Product

In a product space, an open set can restrict one coordinate, several coordinates, or—depending on the topology—all coordinates. The product topology and the box topology encode these possibilities differently when there are infinitely many factors. With only finitely many factors, however, every coordinate restriction is still a finite collection of restrictions. This is why the two topologies agree in that setting.

Let \((X_i,\mathcal{T}_i)\) be topological spaces indexed by a set \(J\), and let \(X=\prod_{i\in J}X_i\) be their Cartesian product. A point \(x\in X\) has coordinates \(x_i\in X_i\). In this tutorial, an open rectangle means a set of the form \(\prod_{i\in J}U_i\), where each \(U_i\) is open in \(X_i\). Some of these sets may be empty.

Definition: A basis for the box topology on \(\prod_{i\in J}X_i\) consists of all open rectangles \(\prod_{i\in J}U_i\), with every \(U_i\) open in \(X_i\). A basis for the product topology consists of rectangles in which \(U_i=X_i\) for all but finitely many indices \(i\). The topologies are the collections of all unions of their respective basis elements.

In the product topology, a basic open set can restrict only finitely many coordinates, although the restricted coordinates need not be specified in advance. In the box topology, a basic open set may impose an open condition in every coordinate. Each factor \(X_i\) itself is open, so a box that imposes conditions in only finitely many coordinates is also a box open in the box topology.

These collections do give bases. For the box topology, every point \(x\in X\) lies in the box \(\prod_{i\in J}X_i=X\). If two boxes \(\prod_i U_i\) and \(\prod_i V_i\) both contain \(x\), their intersection is \(\prod_i(U_i\cap V_i)\). Each \(U_i\cap V_i\) is open in \(X_i\), and this intersection box contains \(x\). For the product topology, the same intersection argument works, and the union of the two finite sets of restricted coordinates is finite. Thus the intersection again has only finitely many restrictions. These are the basis conditions: the basis sets cover the space, and the intersection of two basis sets is a union of basis sets (here, the intersection itself is a basis set).

The Finite-Product Theorem

Theorem: If \(J\) is finite, then the product topology and the box topology on \(\prod_{i\in J}X_i\) are equal.

Proof. If \(J\) is nonempty and finite, a box \(\prod_{i\in J}U_i\) restricts only coordinates in \(J\), which is a finite set. It is therefore a basic open set for the product topology. Every box basis element is product-open, so every union of box basis elements is product-open. This gives inclusion of the box topology in the product topology.

Conversely, a basic open set for the product topology has the form \(\prod_{i\in J}U_i\), with all \(U_i\) open: coordinates not restricted in its product-basis description are assigned \(U_i=X_i\), which is open. This is a box basis element. Thus every product-open set, being a union of product basis elements, is box-open. The two inclusions prove equality.

If some \(X_i\) is empty, the Cartesian product is empty, and both topologies on the empty set consist only of \(\varnothing\). If \(J\) is empty, the Cartesian product is the singleton consisting of the empty tuple. There is only one topology on a singleton, and both constructions give it. These cases also satisfy the conclusion. \(\square\)

The proof compares the basis elements, not just selected open sets. This distinction matters: equality of the two topologies means that every open set for either topology is open for the other. Since each open set is a union of basis elements, identifying the bases gives both required inclusions at once.

What the Bases Look Like in Two and Three Factors

For two factors, a product-basic set can be written \(U\times V\), where \(U\) is open in \(X\) and \(V\) is open in \(Y\). It can also be obtained by intersecting two coordinate cylinders: \(U\times Y\), which restricts the first coordinate, and \(X\times V\), which restricts the second. Their intersection is exactly \(U\times V\). Both factors can therefore be restricted without leaving the product topology.

Worked Example: A Basis in Two Real Factors

Give a product-basis description of \(A=(-1,2)\times(3,5)\) in \(\mathbb{R}\times\mathbb{R}\), where each copy of \(\mathbb{R}\) has its standard topology.

Both intervals are open in \(\mathbb{R}\), so \(A\) is an open rectangle. To express it using coordinate cylinders, take \(C_1=(-1,2)\times\mathbb{R}\) and \(C_2=\mathbb{R}\times(3,5)\). Each is a product-basic set because it restricts one coordinate and leaves the other unrestricted. Their intersection is

$$ C_1\cap C_2 =\bigl((-1,2)\cap\mathbb{R}\bigr)\times \bigl(\mathbb{R}\cap(3,5)\bigr) =(-1,2)\times(3,5)=A. $$

For a point check, \((0,4)\in A\), since \(-1<0<2\) and \(3<4<5\). It lies in both cylinders, so the intersection contains the point. More generally, every \((x,y)\in A\) satisfies \(-1<x<2\) and \(3<y<5\), hence belongs to both cylinders. A point outside \(A\) fails at least one of these two interval conditions and is excluded by the corresponding cylinder. Thus the intersection is exactly \(A\), as claimed.

For three factors, the same description gives basic sets \(U\times V\times W\). They can be formed by intersecting the three cylinders \(U\times Y\times Z\), \(X\times V\times Z\), and \(X\times Y\times W\), where \((X,\mathcal{T}_X)\), \((Y,\mathcal{T}_Y)\), and \((Z,\mathcal{T}_Z)\) are the factors. Each cylinder is product-basic, and the intersection imposes all three coordinate conditions.

Worked Example: A Basis in Three Real Factors

In \(\mathbb{R}^3\), with the product of the standard topologies on the three copies of \(\mathbb{R}\), consider \(B=(-2,1)\times(0,4)\times(5,7)\). Write \(B\) as an intersection of product-basic cylinders and check membership of \(p=(0,2,6)\).

The three cylinders are \(C_1=(-2,1)\times\mathbb{R}\times\mathbb{R}\), \(C_2=\mathbb{R}\times(0,4)\times\mathbb{R}\), and \(C_3=\mathbb{R}\times\mathbb{R}\times(5,7)\). Each has open coordinate factors and restricts only one coordinate, so each is product-basic. Coordinate by coordinate,

$$ C_1\cap C_2\cap C_3 =\bigl((-2,1)\cap\mathbb{R}\cap\mathbb{R}\bigr) \times\bigl(\mathbb{R}\cap(0,4)\cap\mathbb{R}\bigr) \times\bigl(\mathbb{R}\cap\mathbb{R}\cap(5,7)\bigr) =B. $$

The point \(p=(0,2,6)\) belongs to \(B\): its coordinates satisfy \(-2<0<1\), \(0<2<4\), and \(5<6<7\). Thus \(p\) lies in each of \(C_1,C_2,C_3\), and hence in their intersection. Conversely, any point in the intersection must satisfy all three displayed interval conditions, so it belongs to \(B\). The intersection check verifies the basis description in both directions.

Checking the Agreement with Discrete Factors

The same basis argument does not depend on intervals or on any special property of the real line. It applies to arbitrary topological factors. A finite example makes the basis elements especially visible: in a discrete space every subset is open, so the product rectangles can be checked one tuple at a time.

Worked Example: Three Finite Discrete Factors

Let \(X_1=\{a,b\}\), \(X_2=\{0,1\}\), and \(X_3=\{u,v\}\), each with the discrete topology (every subset is open). Consider \(D=\{(a,0,u),(b,1,v)\}\) in \(X_1\times X_2\times X_3\).

For each tuple, the product of its three singleton coordinate sets is a basic rectangle. In particular, \(\{a\}\times\{0\}\times\{u\}=\{(a,0,u)\}\) and \(\{b\}\times\{1\}\times\{v\}=\{(b,1,v)\}\). Therefore

$$ D= \bigl(\{a\}\times\{0\}\times\{u\}\bigr) \cup \bigl(\{b\}\times\{1\}\times\{v\}\bigr). $$

Each singleton coordinate set is open, so both displayed rectangles are box-basic. They are also product-basic, since there are only three factors and hence only three coordinate restrictions. Their union is open in both topologies. The equality is exact: the first rectangle contains only \((a,0,u)\), and the second contains only \((b,1,v)\), so their union contains precisely the two points listed in \(D\).

In fact, every subset of this finite product is open in both topologies: it is the union of the singleton rectangles belonging to its tuples. The empty subset is the empty union. This verifies the agreement directly for this example as well as illustrating the general basis argument.

A Common Mistake and Why the Result Matters

Common pitfall. One might think that a basic open set in the product topology can restrict only one coordinate. That confuses a coordinate cylinder with the full collection of product-basic sets. In \(\mathbb{R}^2\), for example, the incorrect conclusion would be that \((0,1)\times(0,1)\) is not product-basic because it restricts both coordinates. The correction is to write

$$ (0,1)\times(0,1) =\bigl((0,1)\times\mathbb{R}\bigr) \cap \bigl(\mathbb{R}\times(0,1)\bigr). $$

Each cylinder on the right is product-basic, and a finite intersection of open sets is open. More precisely, the rectangle itself is a product-basis element: it has only two restricted coordinates, and two is finite. The product topology permits any finite number of coordinate restrictions, not just one.

The finite-product theorem is useful whenever a product is grouped or described by its coordinates. For two or three factors, one may work with open rectangles without first deciding whether the topology was described by the product basis or the box basis. The reason is not that the two definitions impose the same rule for every index set; it is that a finite index set makes “restrict every coordinate” a finite-support condition. The number of factors is the decisive hypothesis in the proof.

ProductBox basis elementProduct-basis description
Two factors \(X\times Y\)\(U\times V\)\((U\times Y)\cap(X\times V)\)
Three factors \(X\times Y\times Z\)\(U\times V\times W\)\((U\times Y\times Z)\cap(X\times V\times Z)\cap(X\times Y\times W)\)
Finitely many factorsOne open set in each factorA finite collection of coordinate restrictions

Check Your Understanding

Use the basis descriptions and the finite-product theorem to answer the following.

  1. Write \(U\times V\subseteq X\times Y\) as an intersection of two coordinate cylinders.
  2. In \(X\times Y\times Z\), write \(U\times V\times W\) as an intersection of three product-basic cylinders.
  3. Explain why a box basis element in a product with five factors is product-basic.
  4. In two copies of \(\mathbb{R}\), express \((2,4)\times(-1,1)\) as an intersection of coordinate cylinders.
  5. What is the product when the index set is empty, and why do its product and box topologies agree?