pi_shaped.py

from functools import reduce as red; from math import \
(factorial as fact, comb);import sys;from decimal import \
      (getcontext as c,Decimal as dc);(a:=range,
       b:=int(sys.                 argv[1]));c\
       ().prec=b;                  ber=lambda\
       e,f=[dc(1)]:                [f.append(
       1-sum(comb(h                ,g)*f[g]/(h
       - g + 1) for                g in a(h)))for
       h in a(1, e                 + 1)] and abs(
      f[-1]);print                 ((2 * fact(b) /
      (ber(b)* 2**                  b * red(dc.__mul__
      , [1 - (1/dc                   (i) ** b) for i in
      [2, 3, 5, 7]                     ]))) ** (1 / dc(b)))
       #juliusgeo                         pi arb. precis.

Output:

(venv) ➜ python pi_shaped.py 200
3.1415926535897932384626433832795028841971693993751058209749445923078164062862089986280348253421170679821480865132823066470938446095505822317253594081284811174502841027019385211055596446229489549303819

It accepts one argument which is the number of digits of precision you want.

https://gist.github.com/juliusgeo/f4642c3a6835a2ea0676ef82a6fea1ac This is what I did before, it's basically the same implementation, but without all of the stuff to work with the whitespace changes.