Let $\Gamma = PSL(2,\mathbb{Z}) = \langle S,T \ | \ S^2=(ST)^3=1 \rangle$. Let $G$ be some mystery normal subgroup of $\Gamma$ that we happen to think may be congruence. Recall that a subgroup of $\Gamma$ is a congruence subgroup of level $N$ if it contains the principal congruence subgroup of level $N$.

Are there methods for establishing upper bounds on the possible levels of $G$, say given information about $\Gamma/G$? Establishing lower bounds is trivial: the minimal $N$ such that $T^N \in G$ (if such an $N$ exists) is the minimal possible level of $G$.

For example, suppose $\Gamma/G \cong (\mathbb{Z}_3 \times \mathbb{Z}_9) \rtimes S_3$ or more generally of the form (finite abelian group)-semidirect product-(symmetric group). I'm generally thinking about the case where $G$ is the kernel of some unitary representation with finite image.


1 Answer 1


If you only look at upper bounds, you can use the result by Lubotzky: For every algebraic group G, there is some $c$, such that the level of a congruence subgroup $U$ is bounded by the $c(G:U)$.

If you want to determine the level algorithmically, you can look at the image of the subgroup in $PSL(\mathbb{Z}/p^n\mathbb{Z})$. Determine the cusp split of the subgroup, if there are two cusp width $c_1, c_2$, and $p$ a prime with $p\nmid c_1c_2$, then $U$ contains different parabolic elements of order coprime to $p$, hence $U\Gamma(p)/\Gamma(p)$ contains two different parabolic subgroups, hence $U\Gamma(p)/\Gamma(p)=PSL(\mathbb{F}_p)$, thus $p$ does not divide the level.

  • $\begingroup$ Does "cusp split" mean the orbits of the rationals and infinity under the subgroup? $\endgroup$ Nov 22, 2015 at 18:53
  • $\begingroup$ $\mathbb{H}/\Gamma$ has one point at infinity. If $\Delta<\Gamma$, then $\mathbb{H}/\Delta$ has several points $p_1, \ldots, p_k$. Each $p_i$ has a neighbourhood intersecting $c_i$ copies of $\mathbb{H}/\Gamma$, where $\sum_i c_i=(\Gamma:\Delta)$. The partition $c_1, \ldots, c_k$ of $(\Gamma:\Delta)$ is the cusp split of $\Delta$ and is easy to compute, see Millington, Subgroups of the classical modular group, J. LMS 1 (1969) 351-357. $\endgroup$ Nov 23, 2015 at 13:08

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.