7

This one is from Velleman's "How to Prove It, 2nd Ed.", exercise 4.3.23.

Suppose $A$ is set, and $\mathcal{F}\subseteq\mathcal{P}(A)$. Let $$R=\{(a,b)\in A\times A : \text{for every } X\subseteq A\setminus\{a, b\}\text{, if } X\cup \{a\}\in\mathcal{F}\text{ then } X\cup\{b\}\in\mathcal{F}))\}$$ Show that $R$ is transitive.

  1. First of all, I'm not sure if I read this correctly and if my notation is correct: $$R=\{(a,b)\in A\times A : \forall X( X\subseteq A\setminus\{a, b\}\rightarrow(X\cup \{a\}\in\mathcal{F}\rightarrow X\cup\{b\}\in\mathcal{F}))\}$$
  2. To prove that $R$ is transitive, we need to prove that $$\forall a\in A\;\forall b\in A\;\forall c\in A\;((aRb\wedge bRc) \rightarrow aRc),$$ so for starters we suppose and let all the usual stuff:
    • let $a,b,c\in A$
    • suppose $aRb \wedge bRc$
    • expand $aRc$ to $\forall X( X\subseteq A\setminus\{a, c\}\rightarrow(X\cup \{a\}\in\mathcal{F}\rightarrow X\cup\{c\}\in\mathcal{F}))$
    • let $X$ be arbitrary and suppose $X\subseteq A\setminus\{a,c\}$
    • suppose $X\cup\{a\}\in\mathcal{F}$
    • show that $X\cup\{c\}\in\mathcal{F}$
  3. Now Velleman suggests splitting proof to cases: $b\not\in X$ and $b\in X$. Showing that $R$ is transitive for $b\not\in X$ is rather straightforward. All we need to do is to show from $b\not\in X \wedge X\subseteq A\setminus\{a,c\}$ that $X$ is also subset of both $A\setminus\{a,b\}$ and $A\setminus\{b,c\}$, and we just follow assumptions $aRb$ and $bRc$ to conclude $X\cup\{c\}\in\mathcal{F}$.
  4. Now we must show transitivity when $b\in X$. For this case Velleman suggests working with $X'=(X\cup\{a\})\setminus\{b\}$ and $X''=(X\cup\{c\})\setminus\{b\}$, and this is the part I totally don't get. Why would using $X'$ and $X''$ work for this proof, and how do actually connect them with all the givens/assumptions? I can see how all this connects after expanding $aRb$ and $bRc$, but I fail to see how this makes a correct proof.

So my questions are: is the 1. correct notation for given relation $R$ and how does 4. connect to givens.

If there is some other approach, I would be most thankful for any pointers.

LavaScornedOven
  • 401
  • 2
  • 8

1 Answers1

9

Yes, what you’ve written in (1) is equivalent to the definition in the problem, though I think that most readers will find the original version easier to absorb.

In the proof of transitivity, suppose that $b\in X\subseteq A\setminus\{a,c\}$; we want to show that if $X\cup\{a\}\in\mathscr{F}$, then $X\cup\{c\}\in\mathscr{F}$, so assume that $X\cup\{a\}\in\mathscr{F}$. Let $X_0=X\setminus\{b\}$; then Velleman’s sets $X'$ and $X''$ are given by $X'=X_0\cup\{a\}$, and $X''=X_0\cup\{c\}$.

Now $b,c\notin X'$, and $X'\cup\{b\}=X_0\cup\{a\}\cup\{b\}=X\cup\{a\}\in\mathscr{F}$, so $X'\cup\{c\}\in\mathscr{F}$, since $bRc$. Now $X'\cup\{c\}=X_0\cup\{a\}\cup\{c\}=X''\cup\{a\}$, so $X''\cup\{a\}\in\mathscr{F}$. Moreover, $X''\subseteq A\setminus\{a,b\}$, and $aRb$, so $X''\cup\{b\}\in\mathscr{F}$. Finally, $X''\cup\{b\}=X_0\cup\{c\}\cup\{b\}=X\cup\{c\}$, and we conclude that $X\cup\{c\}\in\mathscr{F}$, as desired.

Brian M. Scott
  • 616,228
  • 1
    To absorb this some time I need. :) (Wish I was as good at this as Yoda is in wielding the Force...) – LavaScornedOven Jan 23 '13 at 01:21
  • @LavaScornedOven: Just try to follow it one step at a time; don’t try to see the whole picture right away. – Brian M. Scott Jan 23 '13 at 01:29
  • Thanks for the help. I think I get the steps, but now for the big picture. BTW, is there some good book that could help me getting intuitive about relations? There's a load of books where I can learn about relations, but most textbooks I found are pretty dry/overly theoretical about it. – LavaScornedOven Jan 23 '13 at 01:42
  • @LavaScornedOven: I’m not sure that there really is any way to get an intuitive feel for relations in general. For the common properties of relations (e.g., transitivity) I think that your best bet is simply working with them and seeing lots of examples. It may also help to see more than one way of viewing a relation on $A$: a subset of $A\times A$, an $|A|\times|A|$ matrix of $0$s and $1$s, a directed bipartite graph. Have you seen either of the last two yet? – Brian M. Scott Jan 23 '13 at 01:48
  • I've seen the directed graph approach, but never worked with $|A|\times|A|$ matrices. I'm really new to all this, so I guess I just need to work it through. Once again, thanks a lot for all the help. – LavaScornedOven Jan 23 '13 at 01:54
  • @LavaScornedOven: You’re welcome. If you have access to a school library with a selection of discrete math texts, you might try digging through them to see different approaches and examples. – Brian M. Scott Jan 23 '13 at 02:01
  • 2
    Unfortunately, I'm not in school anymore (I'm 30. :D ), but I use Amazon. This is more of a hobby that started with me bumping on this list: http://www.amazon.com/gp/richpub/syltguides/fullview/R1GE1P236K3YSV – LavaScornedOven Jan 23 '13 at 02:25
  • @LavaScornedOven: Some freely available texts are noted in the answers to this question. Many of the books on this list are appropriate for a beginning discrete math course, though not all of those treat relations extensively. – Brian M. Scott Jan 23 '13 at 02:44
  • @Sibi: That seems an unnecessary step: we already know that $X_0\cup{b}=X$. – Brian M. Scott Mar 12 '15 at 01:18
  • Why did you define $X'$ and $X''$ so differently than in the hint? How are they even equivalent? – lightyourassonfire May 29 '23 at 20:04
  • 1
    They are equal because in this context, $b\ne a,c.$ – Anne Bauval May 29 '23 at 22:00
  • @lightyourassonfire: Anne Bauval has already explained why my definitions are equivalent to Velleman’s; I used mine because I think them a bit easier to work with and a bit more intuitively understandable than his. – Brian M. Scott May 30 '23 at 03:10