APPROXIMATE_ORDERED_FIELD

const APPROXIMATE_ORDERED_FIELD: AlgebraicStructure

Defined in: structure.ts:175

What floating point actually is.

Every field operation is present and every one of them rounds, so associativity and distributivity fail: in binary64, (1e16 + 1) + -1e16 = 0 while 1e16 + (1 + -1e16) = 1. Overflow to infinity and underflow to zero break closure as well, and NaN is not equal to itself.

A BinaryFp* type therefore points here, not at ORDERED_FIELD. The distinction is invisible to a marker interface, which is the reason this information is data.