Given an indexed family of sets $(X_i)_{i\in I}$, a canonical definition of Cartesian product is: $$ \prod_{i \in I} X_i = \left\{f\ \Big|\ f: I \longrightarrow \bigcup_{i \in I} X_i\ \wedge (\forall i\in I)\big(f(i) \in X_i\big)\right\} $$

I wonder if this definition is correct as well: $$ \prod_{i \in I} X_i = \left\{t\ \Big|\ t=(x_i)_{i\in I} \wedge (\forall i\in I)\big(x_i \in X_i\big)\right\} $$

Setting $\mathcal{X}=\bigcup_{i \in I} X_i$, the domain of each family $t$ (in the second notation) is not explicitly set, but values $t(i)$, that is $x_i$, should be in $X_i$ and so only those functions $t$ compatible with $\mathcal{X}$ are acceptable.
Also, the cardinality of $\mathsf{Rng} f$ (in the first notation) is usually smaller than the cardinality of $\mathcal{X}$ and the same as $I$. That is not possible for $t$, since families are surjections. So each $t$ should have a distinct domain, with the cardinality of $I$.


I reply here to a comment regarding indexed families.
I define an indexed family as an alias for surjective function.

Given a surjective function: \begin{align} x\colon I &\longrightarrow X \\ i &\mapsto x_i = x(i), \end{align} it can be denoted also with: $$(x_i)_{i\in I}$$ The range of the function $x$ might be denoted with: $$\{x_i\}_{i\in I}$$ or $\{ x_i \ \big|\ i \in I \} $. This is the standard definition I found in several books, e.g. Tourlakis, "Lectures in Logic and Set Theory" or Wikipedia.

  • $\begingroup$ What does $(x_i)_{i\in I}$ mean? You refer to it as a “family” and as a function, and you say “families are surjections.” I think it’s safer to define things in terms of more commonly-understood objects, like functions (in the top definition). $\endgroup$ – Steve Kass Jun 30 '18 at 21:48
  • $\begingroup$ Hi Antonio, the axiom of choice can be rewritten as "the Cartesian product of nonempty sets is a nonempty set itself." Hence, the first equality holds. The second equality is plainly odd. $\endgroup$ – Will M. Jun 30 '18 at 21:53
  • $\begingroup$ @SteveKass please, see the update on families $\endgroup$ – antonio Jun 30 '18 at 22:24
  • $\begingroup$ @WillM. The CP is a set whose elements are functions from $I$ to the bigunion. The indexed families in the second notation are functions from $I$ to a non-specified range, but the condition $x_i \in X_i$ should work as an implicit definition of the range. $\endgroup$ – antonio Jun 30 '18 at 22:30
  • $\begingroup$ @antonio a formula $\{x \mid p(x)\}$ is always assumed to be of the form $\{x \in S \mid p(x)\}$ where $S$ is a set. In other words, it is kind of a requirement of set theory that sets defined by "relations" are always subsets of a known set. $\endgroup$ – Will M. Jun 30 '18 at 22:39

$t = (x_i)_{i \in I}$ already means that $t$ is a function defined on $I$. A family of sets indexed by $I$ is a function on $I$. The right codomain can be deduced.

  • $\begingroup$ So, I understand this means the second definition is valid. $\endgroup$ – antonio Jul 1 '18 at 10:07
  • $\begingroup$ @antonio sort of. The first one is much better, being more explicit. $\endgroup$ – Henno Brandsma Jul 1 '18 at 10:29

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.