David M. Cooke wrote:
Basically, when doing numerical computations, there are two levels of
analysis: numerical analysis, where you're concerned about which algorithm to use, how stable it is, etc., and then the nitty-gritty things about floating point. For the last bit, you mainly worry about if rounding errors, etc. are big enough to worry about :) (In my experience, most people don't do that bit, but they're usually ok, except when they're not.)
Which, to bring the discussion back to SciPy, it is such an excellent idea that the basic algorithms in it are from the well tested NETLIB and related routine libraries. The people who wrote those did worry about rounding errors and the like, and the code is well crafted. Don't re-invent the wheel and all that.