jpayne@69: #ifndef HashHeapCounted_h
jpayne@69: #define HashHeapCounted_h
jpayne@69: 
jpayne@69: #include "HashList.h"
jpayne@69: #include "HashPriorityQueue.h"
jpayne@69: #include "HashSet.h"
jpayne@69: #include <math.h>
jpayne@69: #include "bloom_filter.hpp"
jpayne@69: 
jpayne@69: class MinHashHeap
jpayne@69: {
jpayne@69: public:
jpayne@69: 
jpayne@69: 	MinHashHeap(bool use64New, uint64_t cardinalityMaximumNew, uint64_t multiplicityMinimumNew = 1, uint64_t memoryBoundBytes = 0);
jpayne@69: 	~MinHashHeap();
jpayne@69: 	void computeStats();
jpayne@69: 	void clear();
jpayne@69: 	double estimateMultiplicity() const;
jpayne@69: 	double estimateSetSize() const;
jpayne@69: 	void toCounts(std::vector<uint32_t> & counts) const;
jpayne@69:     void toHashList(HashList & hashList) const;
jpayne@69: 	void tryInsert(hash_u hash);
jpayne@69: 
jpayne@69: private:
jpayne@69: 
jpayne@69: 	bool use64;
jpayne@69: 	
jpayne@69: 	HashSet hashes;
jpayne@69: 	HashPriorityQueue hashesQueue;
jpayne@69: 	
jpayne@69: 	HashSet hashesPending;
jpayne@69: 	HashPriorityQueue hashesQueuePending;
jpayne@69: 	
jpayne@69: 	uint64_t cardinalityMaximum;
jpayne@69: 	uint64_t multiplicityMinimum;
jpayne@69: 	
jpayne@69: 	uint64_t multiplicitySum;
jpayne@69: 	
jpayne@69:     bloom_filter * bloomFilter;
jpayne@69:     
jpayne@69:     uint64_t kmersTotal;
jpayne@69:     uint64_t kmersUsed;
jpayne@69: };
jpayne@69: 
jpayne@69: inline double MinHashHeap::estimateMultiplicity() const {return hashes.size() ? (double)multiplicitySum / hashes.size() : 0;}
jpayne@69: inline double MinHashHeap::estimateSetSize() const {return hashes.size() ? pow(2.0, use64 ? 64.0 : 32.0) * (double)hashes.size() / (use64 ? (double)hashesQueue.top().hash64 : (double)hashesQueue.top().hash32) : 0;}
jpayne@69: inline void MinHashHeap::toHashList(HashList & hashList) const {hashes.toHashList(hashList);}
jpayne@69: inline void MinHashHeap::toCounts(std::vector<uint32_t> & counts) const {hashes.toCounts(counts);}
jpayne@69: 
jpayne@69: #endif