Let $\{ X_s \}_{s \in S}$ be a family of topological spaces, then the product topology defined on the cartesian product $X := \prod_{s\in S} X_s$ is the coarsest (i.e. smallest) topology such that every projection map $\tau_s: X \to X_s$ is continuous (see PlanetMath).

Now I am interested how the open sets look like in this product topology. In my notes and textbook's I find, that the sets of the form $$ \prod_{s \in S} W_s $$ with $W_s$ open in $X_s$ and $W_s \ne X_s$ only for finitely many $s \in S$ form a base of this topology. Now I know how does the base sets look, but how does the open sets look? I know every open set could be written as an union of base sets, but because in general $$ (A \times B) \cup (C \times D) \ne (A \cup B) \times (C \cup D) $$ (just $\subset$ holds) I can not say for example that the open sets are the sets $\prod_{s \in S} W_s$ with $W_s$ open and $W_s \ne X_s$ only for finitely many $s \in S$. So, could something be said about the form of the open sets?

  • $\begingroup$ Little more can be said than what is in the definition. For example one such open neighboorhodd of a point $x\in\mathbb R^{\mathbb N}$ says that $y\approx x$ if either $y_1\approx x_1$ or both $x_2\approx y_2$ and $x_3\approx y_3$ ... $\endgroup$ Jul 13, 2013 at 14:35
  • $\begingroup$ I think this question is a little too vague. $\endgroup$
    – tomasz
    Jul 13, 2013 at 14:40
  • 3
    $\begingroup$ This is indeed a great question to keep in your mind. But it does not take you anywhere just think about abstract definitions and set-theoretic operations. You need to see product topology in work in some specific examples. $\endgroup$
    – Hui Yu
    Jul 13, 2013 at 14:54
  • 1
    $\begingroup$ btw. if I just say a set in the product topology is open iff every component is an open set, in the infinite case I get the box topology, which differs from the product topology, so this straightforward characterisation is wrong. $\endgroup$
    – StefanH
    Jul 13, 2013 at 15:02
  • 2
    $\begingroup$ That a Cartesian product is closed iff all factors are closed in no way implies that all closed sets are Cartesian products. For example, in $\mathbb{R}^2$ every line is closed, but only the horizontal and vertical ones are products. $\endgroup$ Jul 13, 2013 at 16:18

2 Answers 2


As Hui Yu says, you are not going too far with just the definition and set-theoretic operations: think about specific examples.

For instance, before fighting against scary monsters like infinite arbitrary products, how about looking for examples in the humble $\mathbb{R}^2$? Are you sure you could find out a simple characterization of the open sets there (which happen to be the same for the product topology and for the Euclidian, usual one)?

E.g., what about a set like this one:

$$ U = \left\{ (x,y) \in \mathbb{R}^2 \ \vert \ xy > 1 \ , \ x > 0 \right\} \quad \text{?} $$

It's an open set. Do you think you could describe it easily (I mean, without just repeating the definition of open sets in $\mathbb{R}^2$) in terms of the open sets of the basis of the product topology?

  • $\begingroup$ Ok, got the point. Guess so it is simple not possible. $\endgroup$
    – StefanH
    Jul 13, 2013 at 17:44

Note that the following are true (in all spaces): (fix a base $\mathcal{B}$ for a space $X$)

  1. $f: X \to Y$ is open iff $f[B]$ is open in $Y$ for every $B \in \mathcal{B}$.

  2. $f: X \to Y$ is continuous at $x$ iff for every open set $O$ that contains $f(x)$, there exists some $B \in \mathcal{B}$ such that $x \in B$ and $f[B] \subset O$.

  3. $X$ is compact iff every cover of $X$ with elements from $\mathcal{B}$ has a finite subcover.

  4. $D \subset X$ is dense iff every non-empty $B \in \mathcal{B}$ intersects $D$.

  5. $f : Y \to X$ is continuous iff $f^{-1}[B]$ is open in $Y$ for all $B \in \mathcal{B}$.

Note that we can reason about continuity, openness, compactness, knowing only a base for the topology. So in most cases, all we really need is a good description for a base.

This is analogous to the situation of metric spaces $(X,d)$, where a base is specified (all sets of the form $B(x,r) = \{ y \in X: d(x,y) < r \}$, where $r>0$) and continuity between metric spaces is often expressed using the $\epsilon$-$\delta$ definition, which is just like 2., except using this base in both spaces. For product spaces as well, all proofs involving them essentially uses this base (or the subbase of all $\pi_s^{-1}[O]$ for open sets $O$ in $X_s$). We really do not need a description beyond the fact that they are unions of basic sets.

  • $\begingroup$ Sorry for necropost. Can you think of an open set in the product AxB that is not of the form $ U_a \times U_b $; $U_a, U_b$ open in $A,B$ respectively? $\endgroup$
    – MSIS
    Nov 19, 2019 at 2:19
  • 1
    $\begingroup$ @MSIS E.g. in $\Bbb R \times \Bbb R$ the open set $\{(x,y): x \neq y\}$ or $\{(x,y): x^2+y^2 > 1\}$ $\endgroup$ Nov 19, 2019 at 5:03
  • $\begingroup$ @HennoBrandsma Could you give an example of an open set in product topology where infinitely many factors are proper subsets? $\endgroup$
    – Akash Gaur
    Apr 16, 2020 at 15:50
  • $\begingroup$ @AkashGaur a product set with infinitely many factors are proper subsets has empty interior, so doesn’t contain a non-empty open set. $\endgroup$ Apr 16, 2020 at 15:57
  • $\begingroup$ +1 I liked the answer $\endgroup$ Feb 24 at 3:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.