Continued from this post

Ramanujan found this handy formula for $\pi$$$\frac 1\pi=\frac {\sqrt8}{99^2}\sum_{k=0}^{\infty}\binom{2k}k\binom{2k}k\binom{4k}{2k}\frac {26390k+1103}{396^{4k}}\tag1$$

Which is related to Heegner numbers.

Sometime after, the Chudnovsky brothers came up with another $\pi$ formula$$\frac 1\pi=\frac{12}{(640320)^{3/2}}\sum_{k=0}^\infty (-1)^k\frac {(6k)!}{(k!)^3(3k)!}\frac {545140134k+13591409}{640320^{3k}}\tag2$$ And according to Tito, $(2)$ has a total of $11$ other forms with integer denominators.

Question: What are all $11$ forms? And what $e$-transcendal number to they correspond to?


Chudnovsky-type pi formulas use the j-function $j(\tau)$ as the denominator. For $\tau=\frac{1+\sqrt{-d}}{2}$ and $\tau = \frac{1}{2}\sqrt{-d}$, there are exactly 9+4=13 discriminants $d$ such that $j(\tau)$ is an integer. (However, two of these do not yield series that converge.)

The j-function can easily be calculated in WA as, for example


For convenience, given the well-known sequence defined in three equivalent ways, $$h_k = \frac{(6k)!}{(3k)!\,k!^3}=\tbinom{2k}k\tbinom{3k}k\tbinom{6k}{3k}=\frac{12^{3k}\,\big(\tfrac{1}{2})_k\,\big(\tfrac{1}{6})_k\,\big(\tfrac{5}{6})_k}{k!^3}=1, 120, 83160, 81681600,\dots$$ with factorial $n!$, binomial coefficient $\binom nk$, and Pochhammer symbol $(a)_n$.

I. $\tau=\frac{1+\sqrt{-d}}{2}$

$$\begin{aligned} 1/π &= 3\sum_{k=0}^\infty (-1)^k\, h_k\, \frac{\color{blue}{7}\cdot 9k+8}{(15^3)^{k+1/2}}\\ 1/π &= 4 \sum_{k=0}^\infty (-1)^k\, h_k\, \frac{\color{blue}{11}\cdot 14k+15}{(32^3)^{k+1/2}}\\ 1/π &= 12 \sum_{k=0}^\infty (-1)^k\, h_k\, \frac{\color{blue}{19}\cdot 18k+25}{(96^3)^{k+1/2}}\\ 1/π &= \frac{4}{3} \sum_{k=0}^\infty (-1)^k\, h_k\, \frac{\color{red}{27}\cdot506k+837}{(3\cdot160^3)^{k+1/2}}\\ 1/π &= 12 \sum_{k=0}^\infty (-1)^k\, h_k\, \frac{\color{blue}{43}\cdot 378 k+789}{(960^3)^{k+1/2}}\\ 1/π &= 12 \sum_{k=0}^\infty (-1)^k\, h_k\, \frac{\color{blue}{67}\cdot3906 k+10177}{(5280^3)^{k+1/2}}\\ 1/π &= 12 \sum_{k=0}^\infty (-1)^k\, h_k\, \frac{\color{blue}{163}\cdot3344418 k+13591409}{(640320^3)^{k+1/2}}\end{aligned}$$

II. $\tau = \frac{1}{2}\sqrt{-d}$

$$\begin{aligned} 1/π &= 2\sum_{k=0}^\infty h_k\, \frac{\color{blue}{8}\cdot 14k+12}{(20^3)^{k+1/2}}\\ 1/π &= \frac{3}{2} \sum_{k=0}^\infty h_k\, \frac{\color{red}{12}\cdot 44k+48}{(2\cdot30^3)^{k+1/2}}\\ 1/π &= \frac{3}{\sqrt2} \sum_{k=0}^\infty h_k\, \frac{\color{red}{16}\cdot 63k+80}{(66^3)^{k+1/2}}\\ 1/π &= \frac{3}{2} \sum_{k=0}^\infty h_k\, \frac{\color{red}{28}\cdot513k+864}{(255^3)^{k+1/2}}\end{aligned}$$

with fundamental $d=3, 4, 7, 8, 11, 19, 43, 67, 163$ in blue and non-fundamental $d=12,16,27,28$ in red. However, the two smallest, namely $d=3,4,$ do not yield series that converges.

  • $\begingroup$ Just wondering, but is there a quick way to calculate $e^{\pi\sqrt{d}}$ for any number $d$ that involves the q-series and j-functions? And $d$ doesn't have to be a Heegner number, it can also be $58$ or $22$. (Both discovered by Ramanujan I think) $\endgroup$ – Frank Nov 23 '16 at 13:50
  • $\begingroup$ @Frank: Yes, the j-function input for WA and Mathematica is the same. $\endgroup$ – Tito Piezas III Nov 23 '16 at 13:53
  • $\begingroup$ @Frank: I don't understand your first question. Can you clarify? $\endgroup$ – Tito Piezas III Nov 23 '16 at 13:56
  • $\begingroup$ How would you calculate $e^{\pi\sqrt{67}}$ using the j-functions? How about $e^{\pi\sqrt{58}}$? Wikipedia describes a procedure for only Heegner numbers. So it'll work in calculating $e^{\pi\sqrt{67}}$. But not $e^{\pi\sqrt{58}}$ since $d=58$ is not a Heegner number. $\endgroup$ – Frank Nov 23 '16 at 14:08
  • $\begingroup$ @Frank: There is something wrong with your question. This post is about a pi formula using the jfunction. You want a formula for $e^{\pi\sqrt{67}}$ using the jfunction, is that it? What for? $\endgroup$ – Tito Piezas III Nov 23 '16 at 14:29

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.