Today I Learned

20250413

python floating point sucks (?): https://docs.python.org/3/tutorial/floatingpoint.html

• commonly known as python floats, according to this article:

Since at least 2000, almost all machines use IEEE 754 binary floating-point arithmetic, and almost all platforms map Python floats to IEEE 754 binary64 “double precision” values.

• IEEE 754 binary64 values have 53 bits of precision, which is represented as J/2**N
  • base/radix: 2 or 10
  • precision
  • exponent range from emin to emax where emin = 1-emax
• memomry layout: https://en.wikipedia.org/wiki/IEEE_754
  • 53 digits, exponent is in range [-1022, 1023], which takes 10 bits, in total 63 bits
  • with one additional sign bit