I am working through problems in elementary relation theory in an Elias Zakon textbook and would like to ask for your feedback about robustness of my proof. Improvement suggestions are very much welcome.


$\text{Let } R \text{ be a relation with } D_R = D_R^{'} = A$ $R \text{ is trichotomic on } A \text{ if } R \cap R^{-1} = \emptyset = R \cap I_A \text{ and } A \times A \subseteq R \cup R^{-1} \cup I_A$


$D_R \text{ is } R$'s domain. $D_R^{'} \text{ is } R$'s codomain.

$I_A = \{ (x,x) \mid x \in A \}$

$R^{-1}$ is the inverse relation of $R$.

Trichotomy is defined as thus:

$R \text{ is trichotomic on } A \text{ iff for any } x \text{ and } y \in A \text{ we always have either } xRy, \text{ or } yRx, \text{ or } x = y, \text{ but never two of these together. }$


$A \times A \subseteq R \cup R^{-1} \cup I_A$ is given, so $R \cup R^{-1} \cup I_A$ contains all $(x,y), x \in A, y \in A$.

$R \cap I_A = \emptyset$ is given, so there are no $(x,x)$ $x \in A$ in $R$ and by definition not in $R^{-1}$ either.

Let $D = R \cup R^{-1} \cup I_A - I_A$. By definition of set difference, $D$ contains all $(x,y), x \in A, y \in A$ that are in $R \cup R^{-1} \cup I_A$ but not in $I_A$.

Since it is established that $R \cap I_A = \emptyset$ and $R^{-1} \cap I_A = \emptyset$ hence $D = R \cup R^{-1}$.

It is given that $R \cap R^{-1} = \emptyset$ hence for all $(x,y) \in R \cup R^{-1} \text{ and } x \neq y$ that are in $R$ $(y,x)$ does not also hold in $R$. For all $(x,y) \in R \cup R^{-1} \text{ and } x \neq y$ that are not in $R$ but are in $R^{-1}$ $(y,x)$ does hold in $R$.

Futhermore, it is established that $(x,x)$ does not hold in $R$ for each $x \in A$ hence only $x = x$ is satisfied.

So for all $(x,y)$ $x \in A, y \in A$ only one of $(x,y)$ or $(y,x)$ holds in $R$ or $x = y$.

$R$ is trichotomic on A. $\blacksquare$

  • $\begingroup$ It's ok but longer than needed. $R, R^{-1},$ and $I_A$ are subsets of $A\times A,$ so $ A\times A \subset R\cup R^{-1}\cup I_A$ implies $A\times A=R\cup R^{-1}\cup I_A.$.... Since $\phi=R\cap I_A,$ it follows by def'n of $R^{-1}$ (as you said) that $R^{-1}\cap I_A=\phi.$ We also have $R\cap R^{-1}=\phi $.... So $R, R^{-1},$ and $I_A$ are pairwise-disjoint sets and their union is $A\times A,$...so any $(x,y)\in A\times A$ belongs to exactly one of these sets. $\endgroup$ – DanielWainfleet May 12 '17 at 2:17
  • 1
    $\begingroup$ Hi, welcome to Math.SE. Please read the tag descriptions of the tags you want to use. For this question, the tag (elementary-set-theory) would be more appropriate than the tag (set-theory). $\endgroup$ – martin.koeberl May 12 '17 at 2:47
  • $\begingroup$ Thanks martin.koeberl for the note, will read more carefully before selecting tags in the future. DanielWainfleet, I enjoyed your proof and I see how you used the fact that $A = D_R = D_R^{'}$ to deduce $R$ and $R^{-1}$ must be subsets of $A \times A$. I had been somewhat concerned that my proof didn't really make explicit use of that given fact. $\endgroup$ – Alexander McLin May 12 '17 at 13:27
  • $\begingroup$ I did not actually deduce that $R$ and $R^{-1}$ are subsets of $A\times A.$ The set-theoretic definition is that a binary relation on $A$ is the same thing as a subset of $A\times A.$ $\endgroup$ – DanielWainfleet May 12 '17 at 16:56

I'm a bit hesitant to add this answer, since it doesn't directly answer your question, but instead shows a proof in a different style. However, it may help you get a different perspective on the problem and how to approach it.$% \newcommand{\calc}{\begin{align} \quad &} \newcommand{\op}[1]{\\ #1 \quad & \quad \unicode{x201c}} \newcommand{\hints}[1]{\mbox{#1} \\ \quad & \quad \phantom{\unicode{x201c}} } \newcommand{\hint}[1]{\mbox{#1} \unicode{x201d} \\ \quad & } \newcommand{\brk}{ \\ \quad & } \newcommand{\endcalc}{\end{align}} \newcommand{Ref}[1]{\text{(#1)}} \newcommand{\then}{\Rightarrow} \newcommand{\when}{\Leftarrow} \newcommand{\true}{\text{true}} \newcommand{\false}{\text{false}} \newcommand{\oneof}[1]{\text{exactly one of }#1\text{ holds}} %$

Given a set $\;A\;$ and a relation $\;R\;$ on $\;A\;$, and definition $$ \tag{0} R \text{ is trichotomic on } A \\\equiv\; \langle \forall x,y : x,y \in A : \oneof{xRy,\; yRx,\; x=y} \rangle $$ you are asked to prove $$ \tag{1} R \text{ is trichotomic on } A \\\when\; R \cap R^{-1} = \emptyset \;\land\; R \cap I_A = \emptyset \;\land\; A \times A \subseteq R \cup R^{-1} \cup I_A $$ Comparing the right hand sides of $\Ref{0}$ and $\Ref{1}$, it seems difficult to work from the latter to the former. Therefore we start at the former, and try to introduce $\;R^{-1}\;$, $\;I_A\;$, and $\;A \times A\;$.

Therefore we start to calculate as follows: $$\calc \langle \forall x,y : x,y \in A : \oneof{xRy,\; yRx,\; x=y} \rangle \op\equiv\hint{introduce $\;A \times A,\; R^{-1},\; I_A\;$; rewrite $\;xRy\;$ for symmetry} \langle \forall x,y : (x,y) \in A \times A : \brk \oneof{(x,y) \in R,\; (x,y) \in R^{-1},\; (x,y) \in I_A} \rangle \endcalc$$

Now we recognize that this statement really describes a partitioning of $\;A \times A\;$, and that the right hand side of $\Ref{1}$ also seems to be about that same concept.

Therefore we continue the above calculation:

$$\calc \dots \op\equiv\hints{element-level definition of $\;\text{partitions}\;$}\hint{-- to get rid of the individual elements $\;x,y\;$} \tag{*} \left\{R, R^{-1}, I_A\right\} \text{ partitions } A \times A \op\equiv\hint{set-level definition of $\;\text{partitions}\;$} R \cap R^{-1} = \emptyset \;\land\; R \cap I_A = \emptyset \;\land\; R^{-1} \cap I_A = \emptyset \brk\land\; R \cup R^{-1} \cup I_A = A \times A \endcalc$$

When now comparing this last expression with the right hand side of $\Ref{1}$, we see that there are two differences that we need to focus on: $\;R^{-1} \cap I_A = \emptyset\;$ and $\;R \cup R^{-1} \cup I_A \;\subseteq\; A \times A\;$.

For the first of these, the only thing I see is to try and use the definitions, in other words, calculate which pairs are in $\;R^{-1} \cap I_A\;$: for every $\;x,y\;$, $$\calc (x,y) \in R^{-1} \cap I_A \op\equiv\hint{definition of $\;\cap\;$} (x,y) \in R^{-1} \;\land\; (x,y) \in I_A \op\equiv\hint{definition of $\;I_{\dots}\;$} (y,x) \in R \;\land\; x=y \op\equiv\hint{definition of $\;I_{\dots}\;$} (y,x) \in R \;\land\; (y,x) \in I_A \op\equiv\hint{definition of $\;\cap\;$} (y,x) \in R \cap I_A \endcalc$$ and it immediately follows that $\;R^{-1} \cap I_A = \emptyset \;\equiv\; R \cap I_A = \emptyset\;$.

For the second difference, we calculate simply $$\calc R \cup R^{-1} \cup I_A \;\subseteq\; A \times A \op\equiv\hint{set theory} R \subseteq A \times A \;\land\; R^{-1} \subseteq A \times A \;\land\; I_A \subseteq A \times A \op\equiv\hint{use twice that $\;R\;$ is a relation on $\;A\;$; definition of $\;I_{\dots}\;$} \true \endcalc$$

With those we can conclude the main calculation:

$$\calc \dots \op\equiv\hint{logic and set theory, by the previous two calculations} R \cap R^{-1} = \emptyset \;\land\; R \cap I_A = \emptyset \;\land\; A \times A \;\subseteq\;R \cup R^{-1} \cup I_A \endcalc$$

which completes the proof that the right hand sides of $\Ref{0}$ and $\Ref{1}$ are equivalent.

Looking over the above proof, we now see that the key to this problem is that $\;A \times A\;$ is partitioned by $\;R\;$, $\;R^{-1}\;$, and $\;I_A\;$, and that that concept of partitioning can be expressed in two different ways: $\Ref{0}$ expresses it at the level of elements ($\;\oneof{\dots}\;$), and $\Ref{1}$ at the level of sets (empty intersections, union is all).

  • $\begingroup$ thank you for your answer, I enjoyed reading it and your approach was insightful. In the current pedagogical textbook I'm studying, partitions has yet to be introduced so your answer was a nice exposure to that concept. Your element-wise and set-wise approaches to partitioning was interesting and helped me get a better intuition of how those two are closely related. $\endgroup$ – Alexander McLin May 21 '17 at 13:23

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.