Ramanujan gave many fascinating formulas and approximations to pi. Using one of his examples, we can give its family. First, define the *fundamental units*,

These are involved in fundamental solutions to *Pell equations*. For example, for , it is {*x, y*} = {99, 13}, (see the values above), while for it is {*x, y*} = {1451, 110}. Using these solutions to Pell equations, then,

where,

Nice, isn’t it? The second to the last approximation is by Ramanujan which is accurate to 31 digits, while the last is by this author and is accurate to 42 digits. (*Can anyone find a nice expression for the next step?* ) The expression inside the log function is the exact value of,

where is the *Dedekind eta function*, and , , , , respectively.

The *fundamental discriminant* *d* = -4∙58 has *class number* *h*(d) = 2. Another one with the same class number is *d* = -4∙37. Hence, given,

then,

where the expression inside the log function is now the *absolute value* of the eta quotient at and .

Posted by Nikos Bagis on May 4, 2013 at 3:01 am

It is $\left(\frac{\eta(\tau)}{\eta(2\tau)}\right)^{24}=\left(\frac{k_r}{k_{4r}}\right)^2$, hence $\pi\approx 2\log\left(\frac{k_r}{k_4r}\right)$, where $k_r$ is the singular modulus. Have you got an algorithm of how we construct the inside of $\log$ ratio with Pell,s equations? This will be extremely helpful. Although constructions of how we can find the Weber invariant have been considered for certain values of $r=-\tau^2$. Is there any reason why using 37 and 58?

Posted by tpiezas on May 5, 2013 at 3:53 am

The algorithm generally applies only for even d = 4m with class number h(-d) = 2. Hence, m = 6, 10, 22, 58 and m = 5, 13, 37.