Feat: ArrayOfNumber Tuple Sketch

[proost]

Summary

Add Go version ArrayOfDouble Tuple Sketch.

Design

ArrayOfDouble Tuple Sketch

// Number constraint for types that can be used in ArrayOfNumberSummary.
// NOTE: Only allowed fixed-size type.
type Number interface {
	~int8 | ~int16 | ~int32 | ~int64 |
		~uint8 | ~uint16 | ~uint32 | ~uint64 |
		~float32 | ~float64
}

type ArrayOfNumberSketch[V Number] interface {
	Sketch[*ArrayOfNumberSummary[V]]

	// NumValuesInSummary returns the number of values in ArrayOfNumberSummary.
	NumValuesInSummary() uint8
}

// ArrayOfNumberSummary is a fixed-size array of numeric values used as a summary in tuple sketches.
type ArrayOfNumberSummary[T Number] struct {
	values []T
	size   uint8
}

// Reset clears the content of the summary, restoring it to its initial state.
func (s *ArrayOfNumberSummary[T]) Reset()

// Clone creates and returns a deep copy of the current Summary instance.
func (s *ArrayOfNumberSummary[T]) Clone() Summary

// Update incorporates a new value into the summary, modifying its internal state based on the given input value.
func (s *ArrayOfNumberSummary[T]) Update(values []T)

// Values returns the values in the summary.
func (s *ArrayOfNumberSummary[T]) Values() []T

I’ve found that ArrayOfDouble sketch can be numeric array sketch. So i expand original one. Because of serialization, i decide to accept fixed size number type element array.

• ArrayOfNumberSketch inherit and extend Sketch to support numeric array summary sketch.
• ArrayOfNumberSummary will do same thing ArrayOfDouble Sketch Summary. So will add given values when Update method is called.

ArrayOfNumberUpdateSketch

// ArrayOfNumberUpdateSketch builds Tuple sketch from input data via update methods.
// This is a wrapper around a tuple sketch to match the functionality
// and serialization format of ArrayOfDoublesSketch in Java.
type ArrayOfNumberUpdateSketch[V Number] struct {
	table                   *hashtable[*ArrayOfNumberSummary[V]]
	numberOfValuesInSummary uint8
}

ArrayOfNumberUpdateSketch implements ArrayOfNumberSketch

ArrayOfNumberCompactSketch

// ArrayOfNumberCompactSketch is the immutable, serializable form of an array of number tuple sketch.
type ArrayOfNumberCompactSketch[V Number] struct {
	theta                   uint64
	entries                 []entry[*ArrayOfNumberSummary[V]]
	seedHash                uint16
	numberOfValuesInSummary uint8
	isEmpty                 bool
	isOrdered               bool
}

• ArrayOfNumberCompactSketch implements ArrayOfNumberSketch
• serialization and deserialization follows encoder & decoder pattern.

Release Schedule

I will upload 3 PRs.

1. A PR for ArrayOfNumberUpdateSketch and interfaces.
2. A PR for ArrayOfNumberCompactSketch and serialization / deserialization.
3. Set Operations.