machine epsilon

In math code, you might see values for epsilon, or ε. This stands for the smallest representable increment, often of floating point numbers. Due to the nature of floating point numbers, this value is different at different points along the number-line.

Some libraries do this pretty well: in particular the Julia Language provides a function epsilon(value) that gives a value-specific epsilon. Doing so is harder in JavaScript because number types are less simple: the singular Number object is always represented as Float64 behind the scenes, and though we can use Typed Arrays to reinterpret and modify the underlying bits, we can't shuttle a float into a single int, since there are no Int64 arrays.

This library explores this in detail. It isn't ready for prime-time, since it uses Float32 rather than Float64 units.

api

reintepret(value, fromtype, totype)

Reintepret the bits in a number from one type to another:

// read a floating point number as an int
reinterpret(value, Float32Array, Int32Array);

floatlexorder(value, delta)

Given a floating point number, move it by delta increments.

epsilon(value)

Return the difference between this floating point value and the next value on the number line.