Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

In Algebra by Serge Lang, the author asserts, to prove the existence of a field extension where an irreducible polynom has a root, that if you take one set $A$ and a cardinal $\mathcal{C}$, that you can find a set $B$ such that $\text{card}(B)=\mathcal{C}$ and such that $A \cap B=\emptyset$.

How can one build $B$ ?

share|cite|improve this question
up vote 3 down vote accepted

Another approach, that avoids having to deal with ordinals or to invoke the axiom of foundation, is to note that there is a set $t$ not in $\bigcup\bigcup A$ (just on cardinality grounds: for any set $X$ there is a set $z$ not in $X$). Therefore if $B$ has size $\mathcal C$, letting $B'=\{t\}\times B$, we see that $B'$ is both of size $\mathcal C$, and disjoint from $A$. The point is that the usual set theoretic formalization of ordered pairs gives us that $(t,a)=\{\{t\},\{t,a\}\}$, so if such a pair $(t,a)$ is in $A$, then $t$ is in $\bigcup\bigcup A$.

share|cite|improve this answer
For any set $X$, there is a set $z$ not in $X$ because otherwise $X$ would be the set of all sets, which leads to the Russel paradox. – Florian Apr 9 '14 at 17:29
@Florian: Or Cantor's paradox (which is what I used to guarantee the existence of $t$, or $X$ in my answer). – Asaf Karagila Apr 9 '14 at 18:10

If you use the axiom of foundation, you can set $$ B := \left\{\{A,c\} \,\big|\, c \in \mathcal{C} \right\} \text{.} $$ $B$ obviously has cardinality $\mathcal{C}$, and if $x \in B \cap A$, then the axiom of foundation is violated because you then have an $\in$-circle $$ x \in A \in \{A,c\} = x \text{.} $$

Note that this assumes that cardinalities are sets - in particular, that $\mathcal{C}$ is a set. If your definition of cardinal doesn't make them sets, e.g. if your idea of a cardinal is the class of all sets with a given cardinality, then instead of $\mathcal{C}$, pick some set $C$ with $card(C) = \mathcal{C}$.

share|cite|improve this answer

There's an ordinal $\alpha$ such that all ordinals $\beta\ge\alpha$ are not in $A$. Take $\mathcal C$ ordinals starting at $\alpha$ as your set $B$.

Why does such ordinal $\alpha$ exist? Assume it doesn't, then the set $A$ contains arbitrarily big ordinals. By a transfinite recursion, we can construct a set of increasing ordinals from $A$ indexed by all ordinals. Just take an ordinal bigger than the supremum of all ordinals so far. This is a contradiction since ordinals are too numerous to form a set.

share|cite|improve this answer
Why is there an ordinal $\alpha$ such that all $\beta \ge \alpha$ are not in $A$ ? – Florian Apr 9 '14 at 16:23
@Florian I added details. I assume you know the ordinals aren't a set. Just take supremum of all ordinals you found so far to construct a new one. – user2345215 Apr 9 '14 at 16:27
Yes, one can build a sequence $(A_n)_{n\in\mathbb{N}}$, by recursion (why bother with limit ordinals ?) of strictly increasing ordinals. $\cup Im((A_n)_{n\in\mathbb{N}})$ is a set which is exactly the set of all ordinals (take an ordinal $\gamma$, then $\exists n\in\mathbb{N},\gamma<A_n$). And the set of all ordinals does not exist. We can then take the $\alpha$ described in the answer from user2345215, and have $B=(\mathcal{C}+A)\textbackslash A$. – Florian Apr 9 '14 at 16:57
@Florian Uhm no. Your sequence is too short. No matter how much a sequence of ordinals indexed by natural numbers (or any ordinal) increases, it's bounded. You need to do the full thing here. – user2345215 Apr 9 '14 at 17:06
@Florian Also you can add ordinals, so just set $B=\{\alpha+\gamma\mid\gamma\in\mathcal C\}$, that's what I meant in my answer. – user2345215 Apr 9 '14 at 17:09

Given any two sets, $A$ and $B$ we can find $A'$ and $B'$ such that $A'\cap B'=\varnothing$ and $|A|=|A'|$ and $|B|=|B'|$.

Simply take $A'=\{0\}\times A$ and $B'=\{1\}\times B$.

Since $A$ is fixed, using Cantor's theorem we know that that there exists some $X\subseteq A$ such that there are no ordered pairs of the form $\langle X,a\rangle\in A$ (for any $a$, that is). So now we can take $B'=\{X\}\times B$.

share|cite|improve this answer
But $A$ is fixed, you cannot replace it with an equipotent $A'$. – Andrés E. Caicedo Apr 9 '14 at 16:32
True. But now it works. – Asaf Karagila Apr 9 '14 at 16:35

Take the Cartesian product of $A$ with the set $\{1\}$.

share|cite|improve this answer
You miss-read, $B$ is supposed to have cardinality $\mathcal{C}$, which isn't necessarily the same as the cardinality of $A$. – fgp Apr 9 '14 at 15:16
There is also the requirement of a prescribed cardinality. – Thomas Apr 9 '14 at 15:16
Oops, you are right. What I ought to have typed was "of $\cal{C}$ with the set $\{a\}$ for some $a \in A$" which is what fgp got right first. – Lee Mosher Apr 9 '14 at 15:28

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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