I have a document in which I have a named proposition, which I wish to reference by its acronym. Since it is named, it ought not be numbered. Still, I'd like to reference it using a cross-reference, rather than hard-coding the acronym, in case I change it (so it gets updated everywhere) or remove it (so I get a warning).

The best I've been able to come up with uses a combination of thmtools, cleveref and manual definitions of the cleveref reference format and of the representation of the relevant LaTeX counter.

All of this seems rather circuitous and - at least in the case of the counter, which I don't even need - unnecessary. However, no simpler approach seems to work.

Dropping the following are NOT really viable options for me at this point as lots now depends on them: mathtools, amsthm and cleveref.

However, thmtools can be dropped if, for some reason, a nicer solution is possible without it. (I only started using it to try to deal with this case - I can easily revert to plain amsthm.)

Loading hyperref is NOT an option, as I can't submit a paper to a journal if it's got hyperlinks. So, even if this would help, I can't use such an approach.


\declaretheorem[numbered=yes, name=Big Friendly Giant, refname={BFG,BFG}]{bfg}

\begin{bfg} Big Friendly Giant theorem \label{thm:bfg}

Here's a reference: \cref{thm:bfg}.

output (target and desired)

Note that this output is both what I get and what I want. What is the most elegant way to get it?

If I use numbered=no or numbered=unless unique, I get an unnumbered proposition (good), but no cross-reference (bad). If I don't redefined \thebfg, I get a cross-reference (good), but a numbered proposition (bad). If I set \thebfg to \relax, I get an unwanted space, which \unskip gets rid of. And so on. That is, hackish though this code is, it is the most elegant - or the least inelegant - I've been able to come up with.

  • You approach will produce any reference to a bfg theorem/proposition as BFG then; I do not think that this is what you have in mind? As another solution I suggest to use a normal environment that only 'fakes' the look of a theorem.
    – user31729
    Sep 24, 2017 at 20:48
  • I mean 'your approach' of course... sorry about that typo
    – user31729
    Sep 24, 2017 at 21:39
  • @ChristianHupfer No, that really is what I have in mind. That is, the code I posted does what I want. It just seems a very circuitous way of doing it.
    – cfr
    Sep 24, 2017 at 22:43
  • @ChristianHupfer That is, it might be more convenient to have a more generic environment, of course, such that the tag was specified on each use. My approach requires a distinct theorem type for each named result. However, since I only have one such result in my actual document, this aspect isn't really a big deal.
    – cfr
    Sep 24, 2017 at 22:44

1 Answer 1


I'm not quite sure that I did understand the question. But I would probably do something like this:

\declaretheorem[numbered=no,name=Big Friendly Giant, postheadhook=\def\@currentlabel{BFG}]{bfg}

\begin{bfg}[label=thm:bfg] Big Friendly Giant theorem 

Here's a reference: \ref{thm:bfg}.

enter image description here

  • Thanks. So use \ref rather than \cref? I guess I could do that, provided this still works with cleveref loaded. It isn't quite as seamless as using \cref, but certainly doable.
    – cfr
    Dec 8, 2017 at 17:20
  • No, \ref is not the main point. The main point is to define \@currentlabel. I only removed the \cref definitions as I didn't really understand their purpose. Dec 8, 2017 at 17:56

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .