2

I studied set theory in high school, but I'm seeing a period in my university course, that I can't recognize to mean something. Here is an example from my professor's slides:

The notation $\{ (x, y) \mid Φ \}$ abbreviates $\{ p \mid ∃x. ∃y. p = (x, y) ∧ Φ \}.$

ryang
  • 38,879
  • 14
  • 81
  • 179
nix
  • 23
  • 4
    It shouldn't really mean anything. It appears to be there in order to clarify which parts are their own grammatical object. Other authors might have done this with pure white space, others still might have used commas. Others still might have used words "There exists an $x$ and there exists a $y$ such that $p=(x,y)$ and $\Phi$" – JMoravitz Apr 21 '23 at 13:03
  • 4
    Without such periods, white space, commas, words, or however else is used... it could get confusing to see $\exists x \exists yp=(x,y)\wedge \Phi$. At a glance, you might have thought part of that was saying "There exists an object '$yp$'" or some other incorrect reading due to the lack of separation within what is written. – JMoravitz Apr 21 '23 at 13:06
  • 4
    I would just read those periods as "such that". – Gerry Myerson Apr 21 '23 at 13:33
  • 6
    Writing a dot after the quantifiers is the last remnant of the old "logician's dot notation", a historical alternative to parentheses. The basic idea was that you'd perform dotted operations in a formula after performing all non-dotted operations. So $\exists x P \wedge Q$ would mean the same thing as $(\exists x P) \wedge Q$, while $\exists x. P \wedge Q$ would mean $\exists x (P \wedge Q)$. Dotting is thoroughly outmoded, retained only in quantifiers and in the lambda calculus. – Z. A. K. Apr 21 '23 at 16:17
  • After quantifiers, I like to use a a colon ( : ) instead of a period. A period or dot ($\cdot$) is occasionally used as an AND-operator. – Dan Christensen Apr 22 '23 at 15:07
  • @Z.A.K. Using a dot or a bullet is common in the logical notation used in proof assistants such as HOL. Such systems typically take the quantifiers to have low precedence: so $\exists P \land Q$ means $\exists x (P \land Q)$. I think this is not outmoded, but is a good way to go: it fits better with the $\lambda$-calculs and with natural language. – Rob Arthan Apr 27 '23 at 21:47
  • @RobArthan: I think we actually agree here, so let me clarify what I wrote. There is nothing wrong with dotting variable binding constructs, such as $\forall, \exists, \lambda$ - indeed, I use the dotted versions exclusively in all my work, formal or informal. It's dotting connectives, such as writing $(A \land B) \rightarrow C$ as $A \land B .\rightarrow. C$, that's outmoded (but quantifier dotting evolved from it, and is a vestige of it). I wouldn't have let that past code review at CSIRO even in proof assistants that could in principle support it. I'm sure things were no different at ICL. – Z. A. K. Apr 28 '23 at 03:42
  • 1
    @Z.A.K. Thanks for the clarification. I think we are indeed happily and equally dotty $\ddot{\smile}$. – Rob Arthan Apr 29 '23 at 00:13

1 Answers1

4

The sentence $$∃x\; ∃y\; p = (x, y) ∧ Φ(x,y)\tag0$$ conventionally means $$\Big[∃x\; ∃y\; p = (x, y) \Big]∧ Φ(x,y)\tag1$$ $$\Big[∃a\; ∃b\; p = (a, b) \Big]∧ Φ(x,y).\tag1$$ If we add periods to sentence $(0)$ in the way that your lecturer does, then there is often ambiguity whether we are inserting them in a manner that it logically superfluous (for example, just to add visual space), OR whether we are using them as delimiters to specify sentence $(0)$'s intended meaning as $$∃x\; ∃y\; \Big[p = (x, y) ∧ Φ(x,y)\Big].\tag2$$ Note that sentences $(1)$ and $(2)$ are not equivalent to each other.

The notation $\{ (x, y) \mid Φ \}$ abbreviates $\{ p \mid ∃x. ∃y. p = (x, y) ∧ Φ \}.$

It turns out that your lecturer actually means sentence $(2)$ instead of sentence $(1),$ since the notation $$\left\{ (x, y) \mid Φ(x,y) \right\}$$ abbreviates $$\left\{ p \;\middle|\; ∃x\;∃y\; \Big[p = (x, y) ∧ Φ(x,y) \Big] \right\}.$$ (Notice that they had neglected to indicate that $Φ$ contains the free variables $x$ and $y.$) In other words, that slide is incorrect unless we read those periods as indicating that those existential quantifiers' scope extends beyond that symbol.

Please click on the second link for a fuller answer regarding the use of periods in formal sentences; in short, though: avoid them, or use them only with a prefacing usage note so that readers will be sure to read your logic formulae in your intended way.

ryang
  • 38,879
  • 14
  • 81
  • 179