-1

Suppose that $(R,\prec)$ is a totally ordered set, and let $S\subseteq R.$ Define: $$P:=\forall x\in S \ \forall y\in R(y\preceq x\Rightarrow y\in S),$$$$Q:=\forall x\in S (\forall y\in R(y\preceq x)\Rightarrow y\in S).$$

Are $P$ and $Q$ logically equivalent? I believe they are. However, their negations $$\neg P = \exists x\in S \ \exists y\in R(y\preceq x\land y\notin S),$$ $$\neg Q=\exists x\in S (\forall y\in R(y\preceq x)\land y\notin S)$$ do not appear to be logically equivalent. Where is my mistake?

ryang
  • 44,428
RFZ
  • 17,648
  • 11
  • 51
  • 144
  • Check parentheses: in the left one, the $\forall y$ quantifier applies also to the part $y \in S$... – Mauro ALLEGRANZA May 22 '25 at 10:58
  • I am not sure how to read your statement $\neg Q=\cdots$ – Henry May 22 '25 at 11:45
  • The criterion for logical equivalence is that the $2$ statements in question must always match in truth value. P is logically equivalent to Q IFF $P \leftrightarrow Q$, which would mean $\neg P \leftrightarrow \neg Q$. This condition is not met by your $2$ statements. – Hudjefa May 22 '25 at 22:48
  • 2
    Why so many downvotes? What’s the issue? The question is well-formulated and asks a specific question. I can't understand such rude behavior. – RFZ May 22 '25 at 23:09
  • @RFZ For posts looking for feedback or verification of a proposed solution. "Is this proof correct?" or "where is the mistake?" is too broad or missing context. Instead, the question must identify precisely which step in the proof is in doubt, and why so. This should not be the only tag for a question, and should not be used to circumvent site policies regarding duplication. – Sine of the Time May 25 '25 at 08:24

2 Answers2

1

$$P:=\forall x{\in} S \ \forall \color{cyan}y{\in} R\ (\color{cyan}y\preceq x\Rightarrow \color{cyan}y\in S)\tag1$$ $$Q:=\forall x{\in} S\ (\,\forall \color{green}y{\in} R\ (\color{green}y\preceq x)\Rightarrow \color\red y\in S\,)\tag2$$

$Q,$ but not $P,$ contains a free occurrence of $\color\red y$ (by convention, the scope of the green $\color{green}y$ is actually just ∀y∈R y⪯x). So, $P$ can't possibly be logically equivalent to $Q.$

$$\neg P = \exists x{\in} S \ \exists \color{cyan}y{\in} R\ (\color{cyan}y\preceq x\,\land\, \color{cyan}y\notin S)\tag3$$ $$\neg Q=\exists x{\in} S\ (\ \forall \color{green}y{\in} R\ (\color{green}y\preceq x)\,\land\, \color{red}y\notin S\ )\tag4$$

Yes, you've negated $P$ and $Q$ correctly.

Are $P$ and $Q$ logically equivalent?

Now, \begin{align}Q&\equiv\forall x{\in} S \,\big((\forall \color\green z{\in} R\;\color\green z\preceq x)\Rightarrow \color\red y\in S\big)\tag2\\&\equiv \forall x{\in} S \,\exists \color{green}z{\in} R\,\big(\color{green}z\preceq x\Rightarrow \color\red y\in S\big)\tag{2A}\end{align} \begin{align}\lnot Q&\equiv\exists x{\in} S \,\big((\forall \color\green z{\in} R\;\color\green z\preceq x)\,\land\, \color\red y\not\in S\big)\tag4\\&\equiv\exists x{\in} S \,\forall \color{green}z{\in} R\,\big(\color{green}z\preceq x\,\land\, \color\red y\not\in S\big)\tag{4A}\end{align} (the second and fourth logical equivalences () are explained in https://en.wikipedia.org/wiki/Prenex_normal_form#Conversion_to_prenex_form ).

Comparing (1) and (2A), and separately comparing (3) and (4A), confirms that $$P\not\equiv Q\\ \lnot P\not\equiv \lnot Q.$$

Where is my mistake?

Falsely believing that $(\forall y{\in} R\;y\preceq x)\Rightarrow y\in S\tag*{}$ is logically equivalent to $\forall y{\in} R\,\big(y\preceq x\Rightarrow y\in S\big)\tag*{}$ misled you, at the start, to think that $P$ is equivalent to $Q.$

Suppose that $(R,\prec)$ is a totally ordered set, and let $S\subseteq R.$

The logical equivalence of $P$ and $Q$ is actually indepdendent of this mathematics.


Reply to the OP's comment

Could you please explain why you're using a different color for y, and why you replaced y with z?

Sure. The green $\color{green}y,$ the cyan $\color{cyan}y$ and $x$ are distinct dummy/bound variables, so they can be renamed (though for each colour, any renaming must apply to every instance of the variable) to clarify this (instead of recycling the variables); in contrast, the red $\color{red}y$ is a free variable, so cannot be renamed. Analogously, $$\int_0^{\color{red}y} t\, \mathrm dt=\int_0^{\color{red}y} u\, \mathrm du\ne\int_0^x u\, \mathrm du.$$

ryang
  • 44,428
  • I’m a bit confused by your answer. First, you're using a different color for $y$, which is unclear to me. Also, you replaced $y$ with $z$, which I also don’t understand. Could you please explain these parts? Sorry, but I’m still not able to grasp your point. – RFZ May 22 '25 at 13:13
  • @RFZ I've added an addendum, and also added a straight reply to your "Where is my mistake?" query. – ryang May 22 '25 at 14:22
  • Sorry, but could we discuss this in chat, please? I'm afraid it might be too long for the comments. – RFZ May 22 '25 at 14:31
  • @RFZ Okay, I've re-expanded the Answer (including adding the grey parts and links to three pages) to further assuage your confusions. – ryang May 22 '25 at 17:05
0

In the statement $Q$, you cannot restrict the quantifier $\forall y\in R$ to the formula $y\preceq x$ only, since $y$ appears in the formula $y\in S$. So it doesn't make sense.

Some more explanations. You can have a free variable in a formula, or you can bind it with a quantifier. For instance,

  • $x\preceq y$ is a relation between two elements ;
  • $\forall x (x\preceq y)$ means that $y$ is a maximum ;
  • $\forall x,\forall y, x\preceq y$ means that your set cannot have two distinct element ;
  • $\exists x,\forall y:x\preceq y$ means that your set has a minimum element ;
  • $\exists x,\exists y:x\preceq y$ means only that your set is nonempty ;
  • $\forall x,\exists y:x\preceq y$ is always true for a reflexive relation.

But you cannot use the same variable as free and binded in the same formula, otherwise you would not know if it concerns the binded or the free variable.

  • My intuition tells me that this is the issue and the source of the mistake. Could you explain this more explicitly? It’s not clear to me whether the quantifier $\forall y \in \mathbb{R}$ should apply to both $y \preceq x$ and $y \in S$ – RFZ May 22 '25 at 12:31
  • I am still confused: in the formula $\forall y \in R,(y \preceq x) \Rightarrow y \in S$, why is the variable $y$ considered free? Isn’t it bound by the quantifier $\forall y$? – RFZ May 22 '25 at 14:35
  • If it is bound by $\forall y$, then it is the same proposition as $P$: $\forall x\in S, (\forall y\in R, (x\preceq y\Rightarrow y\in S))$. – Christophe Boilley May 22 '25 at 16:21