HDA*[Z,Afeature]: Zobrist hashing of the abstract features (AZHDA*). More...

#include <PBeFS/astar/Distribution.h>

Public Member Functions
	AbstractZobristDistribution (const D &domain, std::uint64_t seed=1)
	constructor, drawing the table of the abstract features from the seed
std::uint64_t	hash (const state_t< D > &s) const
	the abstract Zobrist hash of s
std::uint32_t	key (const state_t< D > &s) const
	the key of s, the high 32 bits of its hash
unsigned	owner (const state_t< D > &s, unsigned p) const
	the owner of s among p workers, that of its key
std::uint32_t	key (std::uint32_t k, const state_t< D > &s, const state_t< D > &t) const
	the key of t, a successor of s whose key is k

Static Public Member Functions
static std::string	name (void)
	the name in the notation of Jinnai and Fukunaga

Detailed Description

template<AbstractFeatureDomain D>
class PBeFS_di_unipi_it::AbstractZobristDistribution< D >

HDA*[Z,Afeature]: Zobrist hashing of the abstract features (AZHDA*).

\( AZ( s ) = R[ A( x_0 ) ] \oplus \cdots \oplus R[ A( x_n ) ] \), where \( A \) is the feature projection of the Domain [see Jinnai and Fukunaga, JAIR 2017, equation (2)]. A successor that changes no abstract feature keeps the owner of its parent, which cuts the communication, while the Zobrist hashing of all the abstract features keeps the load balanced. The table is laid out by raw feature (Algorithm 3 of the paper), so that hashing costs exactly as much as with ZobristDistribution.