On: How to average floating point numbers (2015)

[amelius]

Does numpy.mean() do this?

[eesmith → amelius]

Simple testing shows it does not use the naive sum.

  >>> import numpy as np
  >>> arr = np.array( (1_000_000_000.1, 1.1) * 50_000)
  >>> arr.mean()
  500000000.6000003