The groups $U(N)$ and $SU(N)$ are the most important Lie groups in quantum field theory. The most popular are the $U(1),SU(2),SU(3)$ groups (these gauge groups form the Standard model). But is there mentioned a $SU(\infty)$ gauge theory in physics literature?

An example of such a theory could be the following: May be $g \in SU(\infty)$ smooth and for a function $f(x,y)$ with spacetime coordinate $x$ and the new $SU(\infty)$ degree of freedom $y$ it holds $gf(x,y) = \int d^4y (g(x,y,y')f(x,y'))$. Now it is straighforward to define a gauge connection and the gauge field strength.

In more non-theoretical words: Some quantum states have degeneracies and these degeneracies are based on a special symmetry (operator) that exists in a quantum system. If now the degeneracy symmetry operator is unitary and local symmetry one can define a gauge theory. Was this concept used in quantum mechanics or does such a concept makes sense?

Another interesting case is this: One can perform the following switch of coordinates $g(x,y,y') = g(x,x-y,x-y')$ and hence the generators $T_a(y,y')$ defined by $g(x,y,y') = \sum_a g_a(x)T_a(y,y')$ become dependent on the spacetime coordinate. Another question: Is it possible to define spacetime dependent generators of a Lie algebra?

  • 3
    $\begingroup$ How do you propose to define $SU(\infty)$? $\endgroup$
    – Danu
    Aug 28, 2015 at 21:02
  • $\begingroup$ Here, the internal variables $y \in \mathbb R^4$ represent an infinite-dimensional vector space; this vector space is a Hilbert space. $\endgroup$
    – kryomaxim
    Aug 28, 2015 at 21:12
  • $\begingroup$ $SU(N)$ is not even a vector space. $\endgroup$
    – Danu
    Aug 28, 2015 at 21:35
  • $\begingroup$ Also, your sentence doesn't make much sense to me in general. Could you try to rephrase it? $\endgroup$
    – Danu
    Aug 28, 2015 at 21:57
  • $\begingroup$ I am assuming that the $g$ is an unitary operator acting on the internal degree of freedom $y$. Because linear operators can be considered as infinite dimensional matrices while vectors are functions I talked about the matrix representations of $SU(\infty)$. $\endgroup$
    – kryomaxim
    Aug 28, 2015 at 22:08

2 Answers 2


Comments to the question (v2):

  1. The idea to consider the planar large $N_c\to \infty$ limit in $SU(N_c)$ QCD goes back to Ref. 1.

  2. In light-cone membrane theory, pioneered in Ref. 2, the group $SU(\infty)$ is naturally identified with area-preserving diffeomorphisms ${\rm SDiff}_0(T^2)$ on the torus $T^2$ connected to the identity.

  3. Concretely, OP's proposal resembles a Fourier series expansion of an extra (compact) spacetime dimension. Such exercises are customary in string theory.


  1. G. 't Hooft, A planar diagram theory for strong interactions, Nucl. Phys. B72 (1974) 461.

  2. J. Goldstone, unpublished; J. Hoppe, MIT Ph.D. Thesis, 1982.


There are apparently several thousand references to "SU(\infty)" on arxiv.org, and some of them are definitely talking about gauge fields or Yang-Mills.

I suspect that some of the time, this will just be a way of talking about the large N limit of SU(N), i.e., not referring to a literal SU(∞) field theory, but rather the N→∞ limit of some quantity in SU(N) field theory.

  • $\begingroup$ Thanks for your answer. I have asked whether it is possible that the generators of the $\lim_{N \mapsto \infty} SU(N)$ group are explicitely spacetime dependent. Generators of Yang-Mills theories like QCD are constant matrices (Gell-Mann matrices), but it is possible for a field theory that generators are explicitely dependent on the spacetime coordinate? $\endgroup$
    – kryomaxim
    Aug 29, 2015 at 9:43
  • $\begingroup$ @kryomaxim I think the fact that you choose a basis for a finite group is because it is finite. You would have have consider whatever theory you work with no consideration of basis. If everything works out fine, I suppose it works. I'm making this analogy with the way mathematicians deal infinite dimensional vector spaces. $\endgroup$
    – user28174
    Oct 27, 2015 at 4:38
  • $\begingroup$ Even for the Lorentz group, with which I gather you want your generalized SU(N) to commute, the matrices themselves do not depend on coordinates, even though, of course, they represent coordinate-dependent generators. I wonder if your question stems for the fact that infinite N normally demarcates infinite sets of integers which, when Fourier transformed, amount to continuous phase-space type surfaces cf. , as Qmechanic's answer addresses in 2 & 3; but such surfaces are ancillary world-sheet constructs, and not our spacetime. $\endgroup$ Apr 5, 2016 at 14:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.