mas01mc@292: // UNIT_TEST_LSH.cpp
mas01mc@292: 
mas01mc@292: #include <vector>
mas01mc@292: #include "lshlib.h"
mas01mc@292: #include "reporter.h"
mas01mc@292: 
mas01mc@292: #define LSH_IN_CORE
mas01mc@292: 
mas01mc@292: 
mas01mc@292: #define N_POINT_BITS 14
mas01mc@292: #define POINT_BIT_MASK 0x00003FFF
mas01mc@292: 
mas01mc@292: // Callback method for LSH point retrieval
mas01mc@292: void add_point(void* reporter, Uns32T pointID, Uns32T qpos, float dist)
mas01mc@292: {
mas01mc@292:   ReporterBase* pr = (ReporterBase*)reporter;
mas01mc@292:   pr->add_point(pointID>>N_POINT_BITS, qpos, pointID&POINT_BIT_MASK, dist);
mas01mc@292: }
mas01mc@292: 
mas01mc@292: int main(int argc, char* argv[]){
mas01mc@292: 
mas01mc@292:   int nT = 100; // num tracks 
mas01mc@292:   int nP = 1000;  // num points-per-track
mas01mc@292:   float w = 4.0;// LSH bucket width
mas01mc@292:   int k = 10;
mas01mc@292:   int m = 2;
mas01mc@292:   int d = 10;
mas01mc@292:   int N = 100000;
mas01mc@292:   int C = 200;
mas01mc@292: 
mas01mc@292:   float radius = 0.001;
mas01mc@292:   char FILENAME[] = "foo.lsh";
mas01mc@292: 
mas01mc@292:   assert(nP>=nT);
mas01mc@292: 
mas01mc@292:   int fid = open(FILENAME,O_RDONLY);
mas01mc@292:   LSH* lsh;
mas01mc@292:   bool serialized = false;
mas01mc@292:   Uns32T trackBase = 0;
mas01mc@292: 
mas01mc@292:   if(fid< 0){ // Make a new serial LSH file
mas01mc@292:     lsh = new LSH(w,k,m,d,N,C,radius);
mas01mc@292:     assert(lsh);
mas01mc@292:     cout << "NEW LSH:" << endl;
mas01mc@292:     }
mas01mc@292:   else{
mas01mc@292:     close(fid); // Load LSH structures from disk
mas01mc@292:     lsh = new LSH(FILENAME); 
mas01mc@292:     assert(lsh);
mas01mc@292:     cout << "MERGE WITH EXISTING LSH:" << FILENAME << endl;
mas01mc@292:     serialized=true;
mas01mc@292:     trackBase = (lsh->get_maxp()>>N_POINT_BITS)+1; // Our encoding of tracks and points
mas01mc@292:   }  
mas01mc@294:   cout << "k:" << lsh->get_numFuns() << " ";
mas01mc@294:   cout << "L:" << lsh->get_numTables() << " ";
mas01mc@294:   cout << "d:" << lsh->get_dataDim() << " ";
mas01mc@294:   cout << "N:" << lsh->get_numRows() << " ";
mas01mc@294:   cout << "C:" << lsh->get_numCols() << " ";
mas01mc@294:   cout << "R:" << lsh->get_radius() << " ";
mas01mc@294:   cout << "p:" << lsh->get_maxp() << endl;
mas01mc@292:   cout.flush();
mas01mc@292: 
mas01mc@292:   cout << endl << "Constructing " << nT << " tracks with " << nP << " vectors of dimension " << d << endl;
mas01mc@292:   cout.flush();
mas01mc@292:   // Construct sets of database vectors, use one point from each set for testing
mas01mc@292:   vector< vector<float> > vv = vector< vector<float> >(nP); // track vectors
mas01mc@292:   vector< vector<float> > qq = vector< vector<float> >(nP);// query vectors
mas01mc@292:   for(int i=0; i< nP ; i++){
mas01mc@292:     vv[i]=vector<float>(d);  // allocate vector
mas01mc@292:     qq[i]=vector<float>(d);  // allocate vector
mas01mc@292:   }
mas01mc@292:   
mas01mc@292:   for(int k = 0 ; k < nT ; k ++){
mas01mc@292:     cout << "[" << k << "]";
mas01mc@292:     cout.flush();
mas01mc@292:     for(int i = 0 ; i< nP ; i++)
mas01mc@292:       for(int j=0; j< d ; j++)
mas01mc@292: 	vv[i][j] =   genrand_real2() / radius; // MT_19937 random numbers
mas01mc@292:     lsh->insert_point_set(vv, (trackBase+k)<<N_POINT_BITS);
mas01mc@292:     qq[k] = vv[k]; // One identity query per set of database vectors
mas01mc@292:   }
mas01mc@292:   cout << endl;
mas01mc@292:   cout.flush();
mas01mc@292: 
mas01mc@292:   cout << "Writing serialized LSH tables..." << endl;
mas01mc@292:   // TEST SERIALIZED LSH RETRIEVAL
mas01mc@292:   lsh->serialize(FILENAME);
mas01mc@292: 
mas01mc@292:   // TEST LSH RETRIEVAL IN CORE
mas01mc@292:   printf("\n********** In-core LSH retrieval from %d track%c **********\n", 
mas01mc@292: 	 (lsh->get_maxp()>>N_POINT_BITS)+1,(lsh->get_maxp()>>N_POINT_BITS)>0?'s':' ');
mas01mc@292:   fflush(stdout);  
mas01mc@292:   for(int i = 0; i < nT ; i++ ){
mas01mc@292:     trackSequenceQueryRadNNReporter* pr = new trackSequenceQueryRadNNReporter(nP,nT,(lsh->get_maxp()>>N_POINT_BITS)+1); 
mas01mc@292:     lsh->retrieve_point(qq[i], i, &add_point, (void*)pr); // LSH point retrieval from core
mas01mc@292:     printf("query vector %d] t1:%u t2:%0X\n", i, lsh->get_t1(), lsh->get_t2());
mas01mc@292:     fflush(stdout);
mas01mc@292:     pr->report(0,0);
mas01mc@292:     delete pr;
mas01mc@292:   }
mas01mc@292:   delete lsh;
mas01mc@292: 
mas01mc@292:   cout << "Loading Serialized LSH functions from disk ..." << endl;
mas01mc@292:   cout.flush();
mas01mc@292:   lsh = new LSH(FILENAME);
mas01mc@292:   assert(lsh);  
mas01mc@292:   //  lsh->serial_dump_tables(FILENAME);
mas01mc@292:   printf("\n********** Serialized LSH retrieval from %d track%c **********\n", (lsh->get_maxp()>>N_POINT_BITS)+1,(lsh->get_maxp()>>N_POINT_BITS)>1?'s':' ');
mas01mc@292:   fflush(stdout);  
mas01mc@292:   for(int i= 0; i < nT ; i++ ){
mas01mc@292:     trackSequenceQueryRadNNReporter* pr = new trackSequenceQueryRadNNReporter(nP,nT,(lsh->get_maxp()>>N_POINT_BITS)+1); 
mas01mc@292:     lsh->serial_retrieve_point(FILENAME, qq[i], i, &add_point, (void*) pr); // LSH serialized point retrieval method  
mas01mc@292:     printf("query vector %d] t1:%u t2:%0X\n", i, lsh->get_t1(), lsh->get_t2());
mas01mc@292:     fflush(stdout);
mas01mc@292:     pr->report(0,0);
mas01mc@292:     delete pr;
mas01mc@292:   }  
mas01mc@292:   delete lsh;
mas01mc@292: 
mas01mc@292: #ifdef LSH_IN_CORE  
mas01mc@292:   cout << "Loading Serialized LSH functions and tables from disk ..." << endl;
mas01mc@292:   cout.flush();
mas01mc@292:   // Unserialize entire lsh tree to core
mas01mc@292:   lsh = new LSH(FILENAME,1);
mas01mc@292:   
mas01mc@292:   // TEST UNSERIALIZED LSH RETRIEVAL IN CORE
mas01mc@292:   printf("\n********** Unserialized LSH in-core retrieval from %d track%c **********\n", (lsh->get_maxp()>>N_POINT_BITS)+1,(lsh->get_maxp()>>N_POINT_BITS)>1?'s':' ');
mas01mc@292:   fflush(stdout);
mas01mc@292:   for(int i = 0; i < nT ; i++ ){
mas01mc@292:     trackSequenceQueryRadNNReporter* pr = new trackSequenceQueryRadNNReporter(nP,nT,(lsh->get_maxp()>>N_POINT_BITS)+1); 
mas01mc@292:     lsh->retrieve_point(qq[i], i, &add_point, (void*) pr); // LSH point retrieval from core
mas01mc@292:     printf("query vector %d] t1:%u t2:%0X\n", i, lsh->get_t1(), lsh->get_t2());
mas01mc@292:     fflush(stdout);
mas01mc@292:     pr->report(0,0);
mas01mc@292:     delete pr;
mas01mc@292:   }  
mas01mc@292:   delete lsh;
mas01mc@292: #endif
mas01mc@292: 
mas01mc@292: }