adaptive precision linear floating-point addition/subtraction

[Ravenwater]

implement addition for arbitrary precision linear floating point

We have arbitrary precision and adaptive precision vocabulary.

arbitrary precision arithmetic is any arithmetic using a given fixed precision, whereas adaptive precision changes representations during computation to capture all relevant bits.

arbitrary precision arithmetic is, say float<nbits, es> where nbits = 32, and es = 8, which would represent single precision floating point.

adaptive precision arithmetic is afloat, without any template parameters as the size and representation will adapt to the computation.