Let $J_{k,m}(N)$ be the space of Jacobi forms of weight $k$, index $m$, and congruence subgroup $\Gamma_{0}(N) \rtimes \mathbb{Z}^{2}$. I do not believe it is relevant here to specify what type of Jacobi form (weak, holomorphic, etc). I can't find hardly any references on how the index, and the congruence subgroup change when scaling the variables. So my question is simply:

Given $\varphi(\tau, z) \in J_{k,m}(1)$, what is the index and congruence subgroup of the Jacobi form $$\psi_{d_{1}, d_{2}}(\tau, z) = \varphi(d_{1} \tau, d_{2} z)$$ for general $d_{1}, d_{2} \in \mathbb{Z}_{>0}$?

There are a few relevant things I know. For an ordinary modular form $f(\tau)$ of weight $k$ on $SL_{2}(\mathbb{Z})$, it's straightforward to use the "slash operator" to show that $f(d \tau)$ is a weight $k$ form on $\Gamma_{0}(d)$. (See Exercise 1.2.11 of Diamond and Shurman)

I also know that in the case of $d_{1}=1$, $\varphi(\tau, dz) \in J_{k, d^{2}m}(1)$. This is easy to show from the transformations, and moreover it coincides with a well-known Hecke-like operator

$$U_{d} : J_{k,m}(1) \to J_{k, d^{2}m}(1).$$

I'm unsure what the right statement is for general $d_{1}, d_{2}$ specifically perhaps $\varphi(d \tau, z)$ and $\varphi(d \tau, d z)$.

Part of what's confusing me is that if $F(\tau, z, \sigma)$ is a weight $k$, degree 2 Siegel modular form, then it has a Fourier-Jacobi expansion in $Q = e^{2 \pi i \sigma}$:

$$F(\tau, z, \sigma) = \sum_{m=0}^{\infty} Q^{m} \varphi_{k,m}(\tau, z), \,\,\,\,\,\,\,\,\,\,\,\,\,\, \varphi_{k,m} \in J_{k,m}(1)$$

I believe it's true that $F(d\tau, dz, d\sigma)$ is a weight $k$ form for the degree 2 congruence subgroup $\Gamma^{(2)}_{0}(d) \subset Sp_{4}(\mathbb{Z})$. So we have a Fourier-Jacobi expansion:

$$F(d\tau, dz, d\sigma) = \sum_{m=0}^{\infty} Q^{dm} \varphi_{k,m}(d\tau, dz).$$

But I think we need the index of the Jacobi forms to match the exponent of $Q$. So somehow, $\varphi_{k,m}(d\tau, dz) \in J_{k, dm}(d)$? So the index is multiplied by $d$, not $d^{2}$ in this case.


1 Answer 1


$\varphi(d_1\tau,d_2 z)$ will generally be a Jacobi form for a congruence subgroup of this kind only when $d_1 | d_2$. Otherwise it will not transform correctly under $(\tau,z) \mapsto (\tau,z+\tau)$.

In this case its congruence group is $\Gamma_0(d_1) \rtimes \mathbb{Z}^2$ and its index is $m\frac{d_2^2}{d_1}$, as you can see using the Fourier-Jacobi expansions (the index indeed matches the exponent) or by checking the transformations directly.

  • $\begingroup$ This is great, thanks a lot! Maybe I'm just not looking in the right places, but it would be nice if this was remarked in the literature somewhere. By the way, you say a "congruence subgroup of this kind." Is something like, say, $\varphi(d\tau, z)$ possibly a Jacobi form for some more exotic subgroup? $\endgroup$
    – Benighted
    Sep 25, 2019 at 16:18
  • 1
    $\begingroup$ @Benighted Sure, you'll have to replace the $\mathbb{Z}^2$ by something else, like $d \mathbb{Z}^2$, and the index will no longer be an integer. $\endgroup$
    – user146384
    Sep 30, 2019 at 7:54

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.