Given the j-function $j:=j(\tau)$, and $q=e^{2\pi i\tau} = \exp(2\pi i\tau)$ where, for convenience, we set $\tau=\sqrt{-n}$.

I. $\frac{A_2(q)}{A_1(q)} = \text{q-cfrac}:\;$ Icosahedral group

$$\begin{aligned} A_1(q) &= q^{-1/60} \prod_{n=1}^\infty \frac{1}{(1-q^{5n-1})(1-q^{5n-4})} = j\,^{1/60}\,_2F_1\left(\tfrac{-1}{60},\tfrac{19}{60},\tfrac{4}{5},\tfrac{1728}{j}\right)\\ A_2(q) &= q^{\color{blue}{11}/60} \prod_{n=1}^\infty \frac{1}{(1-q^{5n-2})(1-q^{5n-3})} = j\,^{-11/60}\,_2F_1\left(\tfrac{11}{60},\tfrac{31}{60},\tfrac{6}{5},\tfrac{1728}{j}\right) \end{aligned}$$



II. $\frac{B_2(q)}{B_1(q)} = \text{q-cfrac}:\;$ Octahedral group

$$\begin{aligned} B_1(q) &= q^{-1/24} \prod_{n=1}^\infty \frac{ (1-q^{4n-2})^2 }{ (1-q^{2n-1})^3}\quad =\quad j\,^{1/24}\,_2F_1\left(\tfrac{-1}{24},\tfrac{7}{24},\tfrac{3}{4},\tfrac{1728}{j}\right)\\ B_2(q) &= q^{\color{blue}{5}/24} \prod_{n=1}^\infty \frac{1}{(1-q^{2n-1})(1-q^{4n-2})^2} = j\,^{-5/24}\,_2F_1\left(\tfrac{5}{24},\tfrac{13}{24},\tfrac{5}{4},\tfrac{1728}{j}\right) \end{aligned}$$



III. $\frac{C_2(q)}{C_1(q)} = \text{q-cfrac}:\;$ Octahedral group?

$$\begin{aligned} C_1(q) &= q^{-1/16} \prod_{n=1}^\infty \frac{ 1 }{ (1-q^{8n-1})(1-q^{8n-4})(1-q^{8n-7})} =\,_2F_1\,\color{brown}{??}\\ C_2(q) &= q^{\color{blue}{7}/16} \prod_{n=1}^\infty \frac{1}{ (1-q^{8n-3})(1-q^{8n-4})(1-q^{8n-5})} =\,_2F_1\,\color{brown}{??} \end{aligned}$$



IV. $\frac{D_2(q)}{D_1(q)} = \text{not q-cfrac}:\;$ Tetrahedral group

$$\begin{aligned} D_1(q) &= q^{-1/12} \prod_{n=1}^\infty \frac{ (1-q^{n/3})^3 }{ (1-q^{n})^3}+3 D_2(q)= j\,^{1/12}\,_2F_1\left(\tfrac{-1}{12},\tfrac{1}{4},\tfrac{2}{3},\tfrac{1728}{j}\right)\\ D_2(q) &= q^{1/4} \prod_{n=1}^\infty \frac{ (1-q^{3n})^3 }{ (1-q^{n})^3}\quad = \quad j\,^{-1/4}\,_2F_1\left(\tfrac{1}{4},\tfrac{7}{12},\tfrac{4}{3},\tfrac{1728}{j}\right) \end{aligned}$$


$$D_1(q^{n})D_2(q)-D_1(q)D_2(q^{n})\overset{\color{brown}?}{=} 1\tag4$$

Note, however, that $\frac{D_1(q)}{D_2(q)} =4c^2+c^{-1}$ where $c$ is Ramanujan's cubic continued fraction.

V. Questions

  1. What is the hypergeometric formula for the two $C_i(q)$?
  2. It seems impossible to find integer $n$ for relation $(4)$. Why?
  3. Are there other infinite products with relations similar to $(1),(2),(3)$?

See also this and this.

  • $\begingroup$ Like $(1)$ which corresponding to Rogers-Ramanujan, do you have the continued fraction expansions for $B_{2}/B_{1}$ and $C_{2}/C_{1}$? If so please include that also in your post. Perhaps there are series expansions for each of these functions also. $\endgroup$ Commented Jul 7, 2015 at 5:33
  • $\begingroup$ @ParamanandSingh: They were also studied by Ramanujan. In Duke's paper they are given as eq. $(9.1)$ and $(9.2)$, $$u(\tau) = \frac{B_2(q)}{B_1(q)}= \cfrac{\sqrt{2}\,q^{1/8}}{1 + \cfrac{q} {1 +q+ \cfrac{q^2} {1+q^2+\ddots}}}$$ $$v(\tau) = \frac{C_2(q)}{C_1(q)}= \cfrac{q^{1/2}}{1 + q+\cfrac{q^2} {1 +q^3+ \cfrac{q^4}{1+q^5+\ddots}}}$$ $\endgroup$ Commented Jul 7, 2015 at 7:24
  • $\begingroup$ @ParamanandSingh: Just like $(1)$ has at least forty relatives, I wouldn't be surprised if $(2)$ and $(3)$ have relatives as well. $\endgroup$ Commented Jul 7, 2015 at 7:32

1 Answer 1


Still thinking about (2). As for (3), there are many such identities. For example, try the three level seven functions and the exponent 3. I can't at the moment find the reference that I was thinking of, but this seems to have some. As for (1), I think you have to look at a group for which $\Gamma_1(8)$ is a normal subgroup. Set $j_4=\frac{(q;q)_{\infty }^8}{q(q^4;q^4)_{\infty}^8} = \frac{\eta^8(\tau)}{\eta^8(4\tau)}$, where $\eta(\tau)$ is the Dedekind eta function. Then, \begin{equation*} C_1 = j_4^{1/16} \, _2F_1\left(-\frac{1}{8},\frac{3}{8};\frac{1}{2};\frac{-16} {j_4}\right)\\ C_2 = j_4^{-7/16} \, _2F_1\left(\frac{7}{8},\frac{3}{8};\frac{3}{2};\frac{-16}{ j_4}\right) \end{equation*} However, you will probably be disappointed because these 2F1's are just radicals. (And the symmetry is dihedral not octahedral.)

  • 1
    $\begingroup$ I made a small tweak expressing $j_4$ as eta quotients. This is just fine, at least I now know the $C_i$'s hypergeometric formula. And thanks for giving the link to Chadwick Gugg's thesis, "Modular Identities for the Rogers'-Ramanujan Functions and Analogues". It's a great collection. $\endgroup$ Commented Jul 8, 2015 at 5:00

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.