2

The first statement: $\forall x \exists y (x\geq y^2 \Rightarrow x > 0)$
The Second statement: $\forall x (\exists y: x\geq y^2 \Rightarrow x > 0)$

Actually, my question is this: In the first statement, there exists $y$ for every $x$ such that what's in the parenthesis holds but in the second statement, "$\exists y$" is placed inside parenthesis. Does it make any difference to the first statement?

Hermis14
  • 2,597
Dahdo
  • 57
  • 3
    No, the 1st is the same as the 2nd, if in the 2nd you replace $\exists y$ by $\forall y$ – Physor Dec 26 '21 at 00:40
  • @Snaw The parentheses actually render the sentences have different meanings. – Hermis14 Dec 26 '21 at 02:49
  • @Physor You mean that we should make the replacement in the 1st. – Snaw Dec 26 '21 at 13:27
  • I restate my first comment: "No, the 1st is the same as the 2nd, if *in the 2nd* you replace $∃y$ by $∀y$" – Physor Dec 26 '21 at 13:29
  • @Physor Following the answer by Hermis14, the second statement is shorthand notation for $\forall x [(\exists y~x\leq y^2)\rightarrow x>0]$. If we turn this into $\forall x [(\forall y~x\leq y^2)\rightarrow x>0]$ the meaning is not the same as the first statement. I think you meant to write that if we change the first statement into $\forall x ~\forall y~(x\leq y^2 \rightarrow x>0)$ then it is equivalent to the second statement $\forall x [(\exists y~x\leq y^2)\rightarrow x>0]$. Am I right? – Snaw Dec 26 '21 at 13:50
  • If you take a outermost quantifier from the antecedent of an implication to apply it to the whole implication (with caution about not binding free variables in consequent), then that quantifier gets conjugated, i.e. $\forall \leftrightarrow \exists$ – Physor Dec 26 '21 at 13:54
  • @Physor Which is why I say the replacement has to occur in the first statement. I think we agree and it is just a matter of wording: we agree that $\forall x \forall y(x\leq y^2\rightarrow x>0)$ is equivalent to $\forall x [(\exists y~x\leq y^2)\rightarrow x>0]$. What I find confusing in your comment about replacement is that the first statement in the question is $\forall x \exists y(x\leq y^2\rightarrow x>0)$ and it is here that the $\exists y$ has to be changed to $\forall y$ in order for it to be equivalent to the second one. – Snaw Dec 26 '21 at 14:20
  • 1st $\iff$ (2nd: $\exists y \mapsto \forall y$) – Physor Dec 26 '21 at 14:26
  • (1st: $\exists y \mapsto \forall y$) $\iff$ 2nd – Physor Dec 26 '21 at 14:26
  • That still seems backwards to me.

    I made the comment just so that it helps the OP in case they are confused by this (and also so that it will not confuse me if I run into this again in the future). I think the situation is clarified by now. As long as we both understand what we mean that is all that matters.

    – Snaw Dec 26 '21 at 14:34

1 Answers1

2

To deal with a more general case, I modified the sentences to

$$ S_1)~~\forall x \exists y [ P(x,y) \to Q(x)] $$ $$ S_2)~~\forall x [ \exists y P(x,y) \to Q(x)] $$

$S_1$ is in what is called the prenex normal form. Converting quantified sentences containing material conditionals, $\to$ or $\Rightarrow$, to sentences in prenex form needs special attention.

You might know that $P \to Q$ is equivalent to $\neg P \lor Q$. Let us apply this to the sentence $S_2$.

$$ \begin{aligned} &\forall x [ \exists y P(x,y) \to Q(x)]\\ \Leftrightarrow~ &\forall x [ \neg \exists y P(x,y) \lor Q(x)]\\ \Leftrightarrow~ &\forall x [ \forall y \neg P(x,y) \lor Q(x)] \quad\text{by DeMorgan's law}\\ \Leftrightarrow~ &\forall x [ \forall y (\neg P(x,y) \lor Q(x))] \quad \text{ because $Q(x)$ is independent of $y$}\\ \Leftrightarrow~ &\forall x \forall y [P(x,y) \to Q(x)]\\ \end{aligned} $$

That is, $S_1$ and $S_2$ are not equivalent due to the quantification over $y$.

Additional comment: As I showed for a more general case that $S_2$ and $\forall x \forall y [P(x,y) \to Q(x)]$, which is different from $S_1$, are logically equivalent, the only thing I can do now is to prove each proposition.

Let us restrict our domain of discourse to real numbers.

  1. Consider $\forall x \exists y [ x \ge y^2 \to x > 0]$. Let $x$ be arbitrary. Now, we choose $ y = 1$. Then, if $x \ge y^2 = 1$, we have $x \ge 1 > 0$. Therefore, the claim is true!

  2. Now, consider $\forall x [\exists y: x \ge y^2 \to x > 0]$. Let $x = 0$. We have to show that if $\exists y: x \ge y^2$ is true, so is $x > 0$. However, we know that $x = 0 \ge 0^2$; hence, there is $y = 0$ such that $x \ge y^2$ holds, but $x > 0$ is false. We found $x = 0$ such that $\exists y: x \ge y^2$ but not $x > 0$. Therefore, the claim is false.

There are different.

Hermis14
  • 2,597
  • It seems by your derivation that you interpreted $P_2$ as $\forall x:((\exists y:x\leq y^2)\rightarrow x>0)$. Whether that was the intention of OP or not remains to be seen, but simply by the logical formula itself it does not follow. You would need the extra parenthesis I added to change the meaning of the formula. The two formulas as written in the question are equivalent. – Snaw Dec 26 '21 at 03:18
  • @Snaw But it's a standard notation in first-order logic. Extra parentheses are not required when the quantified formula is atomic. – Hermis14 Dec 26 '21 at 03:22
  • I'm not sure what you mean. Could you quote an example where $\exists y~P(y)\to Q$ is used in the meaning of $(\exists y~P(y))\to Q$ rather than the standard meaning of $\exists y~(P(y)\to Q)$? – Snaw Dec 26 '21 at 03:27
  • @Snaw I found a site https://nokyotsu.com/qscripts/2014/07/distribution-of-quantifiers-over-logic-connectives.html. And almost every logic book conforms to the standard notation. – Hermis14 Dec 26 '21 at 03:30
  • Which line in that table? – Snaw Dec 26 '21 at 03:32
  • See the 5th and 6th row from the bottom in the first table. – Hermis14 Dec 26 '21 at 03:34
  • It seems other answers on MSE agree with you, e.g. https://math.stackexchange.com/a/1150782/998310 and https://math.stackexchange.com/a/495816/998310 - good to know this. I'm pretty sure my logics course did not use these conventions, but I could be forgetting.. – Snaw Dec 26 '21 at 03:54
  • 1
    @Snaw I am glad that I helped you :) – Hermis14 Dec 26 '21 at 03:57
  • Thank you @Hermis14 and Snaw for your help, but I still have doubts to clear. Yes, I understood that those two statements are different but I don't really understand well what ∀x[∃yP(x,y)→Q(x)] which is equivalent to ∀x∀y[P(x,y)→Q(x)] really means. It would really help if you explain to me its exact difference from this one: ∀x∃y[P(x,y)→Q(x)] – Dahdo Dec 26 '21 at 12:35
  • Asking, I was actually on a question to say if the statement is true or false and give reason for your answer. I got stuck on these two questions: Q1) ∀x (∃y x ≥ y²) ⇒ x >0 Q2) ∀x ∃y (x ≥ y² ⇒ x > 0). That's where my confusion arose from. – Dahdo Dec 26 '21 at 12:45
  • @Daniel Would you check it out? I added some comments. – Hermis14 Dec 26 '21 at 16:17