seqan::hibf::sketch::minhashes Struct Reference

MinHash sketches design to be used for Locality sensitive hashing. More...

#include <hibf/sketch/minhashes.hpp>

Public Member Functions
void	fill_incomplete_sketches (std::span< uint64_t > const &more_smallest_values)
	Adds more minhash values to an existing but incomplete table.
bool	is_valid () const
	Checks whether the minHash table is completely filled.
Constructors, destructor and assignment
	minhashes ()=default
	Defaulted.
	minhashes (minhashes const &)=default
	Defaulted.
minhashes &	operator= (minhashes const &)=default
	Defaulted.
	minhashes (minhashes &&)=default
	Defaulted.
minhashes &	operator= (minhashes &&)=default
	Defaulted.
	~minhashes ()=default
	minhashes (std::vector< uint64_t > const &smallest_values)
	construct from a vector of the smallest values in a set (sorted ascending).

Static Public Member Functions
static void	push_to_heap_if_smaller (uint64_t const value, std::vector< uint64_t > &heap)
	Pushes value to the heap if it is smaller than the current largest element.

Public Attributes
std::vector< std::vector< uint64_t > >	table {}
	A table of sketches. For LSH we need multiple sketches, stored in a table.

Static Public Attributes
static constexpr size_t	num_sketches {16}
	...00001111
static constexpr uint64_t	register_id_mask {15}
static constexpr size_t	sketch_size {40}

Friends
class	cereal::access

Detailed Description

MinHash sketches design to be used for Locality sensitive hashing.

The partitioned HIBF as well as the Fast Layout need minHash sketches for a locality sentive hashing algorithm that improves the user bin distribution.

The sketches are specifically designed for this purpose:

The minhashes struct keeps a table of several original MinHash sketches to be used for repeated iterations of LSH.

Constructor & Destructor Documentation

~minhashes()

seqan::hibf::sketch::minhashes::~minhashes ( )

default

Defaulted.

---

The documentation for this struct was generated from the following file:

• minhashes.hpp