Given a set $S$ we can define the filter consisting of all complements of finite sets, which is usually called Fréchet filter or cofinite filter.

For any $a\in S$ the set $\{A\subseteq S; a\in A\}$ is the principal ultrafilter defined by $a$.

What are "standard" (common, frequently used) notations for these two filters? What notation would you recommend?

I will be using this notation mostly in connection with $\mathcal F$-limits (see e.g. wikipedia or Hindman-Strauss, p.63) and Stone-Čech compactification.

For the Fréchet filter I was able to find

For principal ultrafilters I found:

  • $\mathcal F_a$, e.g. here

  • $\pi_a$, e.g. this paper

  • $e(a)$, e.g. Hindman-Strauss (I think the authors chose the notation $e$ to indicate that this gives the embedding of discrete space on $A$ into the Stone-Čech compactification.)

  • IIRC I have seen $a^*$ (in the context of Stone-Čech compactification), but I cannot find an example right now.

Since points of $A$ and principal ultrafilters are usually identified (i.e. $A$ is identified with the corresponding subspace of $\beta A$) maybe it would make sense in some situations to denote the principal ultrafilter given by $a$ again as $a$, but I think this would be too confusing.

  • $\begingroup$ In the context of $\beta\omega$ it’s not uncommon to denote by $n$ the principal ultrafilter over $n$, though I’m pretty sure that I recently saw $n^*$ in a proof that every fixed ultrafilter is in the centre of $\langle\beta\omega,+\rangle$. In general it seems common to use $\mathscr{U}_a$ when $\mathscr{U}$ is used for a generic ultrafilter, $p_a$ when $p$ is used for a generic ultrafilter, etc. $\endgroup$ Dec 4 '11 at 9:05
  • $\begingroup$ Since I tend to use uppercase script letters for filters and ultrafilters, I’ve always used $\mathscr{F}$ for the Fréchet filter. (I corrected your cofinal to cofinite.) $\endgroup$ Dec 4 '11 at 9:11
  • $\begingroup$ I think all the notations are fine, except $e(a)$ (why on earth $e(\cdot)$?) and I don't like the notation $a^\ast$ either: this is because it seems to indicate a contravariant dependence which it isn't. $\endgroup$
    – t.b.
    Dec 4 '11 at 9:53
  • $\begingroup$ @t.b.: I expect that $e(a)$ stems from the common use of $e$ to denote an embedding. My objection to $a^*$ is that I’m accustomed to $A^*$ denoting the set of free ultrafilters containing $A$, i.e., $\operatorname{cl}_{\beta D}(A)\setminus A$. $\endgroup$ Dec 4 '11 at 10:12
  • $\begingroup$ Thanks @Brian, that's an even better reason. Martin edited the explanation "e for embedding" in and this seems to be the right interpretation. The letter $e$ could also stand for "evaluation" if you think of the Stone-Čech compactification in terms of the Gel'fand spectrum of $C_b(X)$. $\endgroup$
    – t.b.
    Dec 4 '11 at 10:39

Summarizing the above comments by Brian M. Scott and t.b.: It seems that there is no generally accepted notation.

Notation for principal ultrafilters:

  • In general it seems common to use $\mathscr{U}_a$ when $\mathscr{U}$ is used for a generic ultrafilter, $p_a$ when $p$ is used for a generic ultrafilter, etc.

  • From the notations mentioned in the post, at least in some situations, $e(a)$ and $a^*$ are not advisable. (The letter $e$ is chosen rather arbitrarily. The notation $a^*$ might be confused $A^*$, which is often used for $\operatorname{cl}_{\beta D}(A)\setminus A$.)


A principal filter corresponding to a set $A$ is sometimes denoted as $\uparrow A$ (or $\uparrow^U A$ to mean a filter on the set $U$).


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.