added direct addition of barcode vectors to dense vectors, began experimentation with morphic RIVs

RIVLower.h

@@ -3,6 +3,7 @@
 #include <string.h>
 #include <string.h>
 #include <signal.h>
 #include <signal.h>
 #include <unistd.h>
 #include <unistd.h>
+#include <math.h>
 /* RIVSIZE macro defines the dimensionality off the RIVs we will use
 /* RIVSIZE macro defines the dimensionality off the RIVs we will use
  * 25000 is the standard, but can be redefined specifically
  * 25000 is the standard, but can be redefined specifically
  */
  */
@@ -64,6 +65,7 @@ typedef struct{
 struct RIVData{
 struct RIVData{
 	size_t RIVsize;
 	size_t RIVsize;
 	int nonZeros;
 	int nonZeros;
+	int I2SThreshold;
 	int *h_tempBlock;
 	int *h_tempBlock;
 	int tempSize;
 	int tempSize;
 	int thing;
 	int thing;
@@ -75,7 +77,6 @@ struct RIVData{
  * it sets global variables that practically all functions will reference,
  * it sets global variables that practically all functions will reference,
  * it checks that your base parameters are valid, and allocates memory for
  * it checks that your base parameters are valid, and allocates memory for
  * the functions to use, so that we can move fast with rare allocations.
  * the functions to use, so that we can move fast with rare allocations.
- * #TODO add signal redefinitions so that cache is saved even on crash
  */
  */
 void RIVInit();
 void RIVInit();
 
 
@@ -85,7 +86,7 @@ void RIVInit();
 void RIVCleanup();
 void RIVCleanup();
 
 
 /*consolidateD2S takes a denseRIV value-set input, and returns a sparse RIV with
 /*consolidateD2S takes a denseRIV value-set input, and returns a sparse RIV with
- * all 0s removed. it does nto automatically carry metadata, which must be assigned
+ * all 0s removed. it does not automatically carry metadata, which must be assigned
  * to a denseRIV after the fact.  often denseRIVs are only temporary, and don't
  * to a denseRIV after the fact.  often denseRIVs are only temporary, and don't
  * need to carry metadata
  * need to carry metadata
  */
  */
@@ -94,7 +95,7 @@ sparseRIV consolidateD2S(int *denseInput);  //#TODO fix int*/denseRIV confusion
 /* mapS2D expands a sparseRIV out to denseRIV values, filling array locations
 /* mapS2D expands a sparseRIV out to denseRIV values, filling array locations
  * based on location-value pairs 
  * based on location-value pairs 
  */
  */
-int* mapS2D(int* destination, sparseRIV input); //#TODO fix int*/denseRIV confusion
+
 
 
 /* makeSparseLocations must be called repeatedly in the processing of a 
 /* makeSparseLocations must be called repeatedly in the processing of a 
  * file to produce a series of locations from the words of the file
  * file to produce a series of locations from the words of the file
@@ -110,7 +111,7 @@ void makesparseLocations(unsigned char* word,  int *seeds, size_t seedCount);
  */
  */
 int fLexPush(denseRIV RIVout);
 int fLexPush(denseRIV RIVout);
 
 
-denseRIV fLexPull(FILE* lexWord, denseRIV output);
+denseRIV fLexPull(FILE* lexWord);
 
 
 /* creates a standard seed from the characters in a word, hopefully unique */
 /* creates a standard seed from the characters in a word, hopefully unique */
 int wordtoSeed(unsigned char* word);
 int wordtoSeed(unsigned char* word);
@@ -121,29 +122,16 @@ int wordtoSeed(unsigned char* word);
  */
  */
 int* mapI2D(int *locations, size_t seedCount);
 int* mapI2D(int *locations, size_t seedCount);
 
 
-int* addI2D(int* destination, int* locations, size_t seedCount);
-int cacheDump();
 
 
+sparseRIV consolidateI2SIndirect(int *implicit, size_t valueCount);
+sparseRIV consolidateI2SDirect(int *implicit, size_t valueCount);
+int cacheDump();
+int* addI2D(int* destination, int* locations, size_t seedCount);
+denseRIV denseAllocate();
 void signalSecure(int signum, siginfo_t *si, void* arg);
 void signalSecure(int signum, siginfo_t *si, void* arg);
 /* begin definitions */
 /* begin definitions */
 
 
-int* mapS2D(int* destination, sparseRIV input){// #TODO fix destination parameter vs calloc of destination
-	/* make sure our destination is a 0 vector */
-	memset(destination, 0, RIVKey.RIVsize*sizeof(int));
-	
-	int *locations_slider = input.locations;
-	int *values_slider = input.values;
-	int *locations_stop = locations_slider+input.count;
-	
-	/* apply values at an index based on locations */
-	while(locations_slider<locations_stop){
-		destination[*locations_slider] = *values_slider;
-		locations_slider++;
-		values_slider++;
-	}
-	
-	return destination;
-}
+
 
 
 int* addS2D(int* destination, sparseRIV input){// #TODO fix destination parameter vs calloc of destination
 int* addS2D(int* destination, sparseRIV input){// #TODO fix destination parameter vs calloc of destination
 	
 	
@@ -160,6 +148,8 @@ int* addS2D(int* destination, sparseRIV input){// #TODO fix destination paramete
 	
 	
 	return destination;
 	return destination;
 }
 }
+
+
 int* mapI2D(int *locations, size_t valueCount){// #TODO fix destination parameter vs calloc of destination
 int* mapI2D(int *locations, size_t valueCount){// #TODO fix destination parameter vs calloc of destination
 	int *destination = (int*)calloc(RIVKey.RIVsize,sizeof(int));
 	int *destination = (int*)calloc(RIVKey.RIVsize,sizeof(int));
 	int *locations_slider = locations;
 	int *locations_slider = locations;
@@ -193,12 +183,54 @@ int* addI2D(int* destination, int *locations, size_t valueCount){// #TODO fix de
 	
 	
 	return destination;
 	return destination;
 }
 }
+
+sparseRIV consolidateI2SIndirect(int *implicit, size_t valueCount){
+	int *denseTemp = mapI2D(implicit, valueCount);
+	
+	sparseRIV sparseOut = consolidateD2S(denseTemp);
+	
+	free(denseTemp);
+	
+	
+	return sparseOut;
+	
+	
+}
+sparseRIV consolidateI2SDirect(int *implicit, size_t valueCount){
+	sparseRIV sparseOut;
+	int *locationsTemp = RIVKey.h_tempBlock+RIVKey.RIVsize;
+	int *valuesTemp = RIVKey.h_tempBlock+2*RIVKey.RIVsize;
+	sparseOut.count = 0;
+	int add = 1;
+	int found;
+	for(int i=0; i<valueCount; i++){
+		found = 0;
+		for(int j=0; j<sparseOut.count; j++){
+			if(implicit[i] == locationsTemp[j]){
+				valuesTemp[i] += add;
+				add *= -1;
+				found = 1;
+			}
+		}
+		if(!found){
+			locationsTemp[sparseOut.count] = implicit[i];
+			
+			valuesTemp[sparseOut.count] = add;
+			sparseOut.count++;
+			add*= -1;
+		}
+	}
+	sparseOut.locations = malloc(2*sparseOut.count*sizeof(int));
+	sparseOut.values = sparseOut.locations+sparseOut.count;
+	memcpy(sparseOut.locations, locationsTemp, 2*sparseOut.count*sizeof(int));
+	return sparseOut;
+}
 sparseRIV consolidateD2S(int *denseInput){
 sparseRIV consolidateD2S(int *denseInput){
 	sparseRIV output;
 	sparseRIV output;
 	output.count = 0;
 	output.count = 0;
 	/* key/value pairs will be loaded to a worst-case sized temporary slot */
 	/* key/value pairs will be loaded to a worst-case sized temporary slot */
-	int* locations = RIVKey.h_tempBlock;
-	int* values = RIVKey.h_tempBlock+RIVKey.RIVsize;
+	int* locations = RIVKey.h_tempBlock+RIVKey.RIVsize;
+	int* values = locations+RIVKey.RIVsize;
 	int* locations_slider = locations;
 	int* locations_slider = locations;
 	int* values_slider = values;
 	int* values_slider = values;
 	for(int i=0; i<RIVKey.RIVsize; i++){
 	for(int i=0; i<RIVKey.RIVsize; i++){
@@ -236,7 +268,7 @@ sparseRIV consolidateD2S(int *denseInput){
 void RIVInit(){
 void RIVInit(){
 	RIVKey.RIVsize = RIVSIZE; //#TODO decide about macros vs global variables
 	RIVKey.RIVsize = RIVSIZE; //#TODO decide about macros vs global variables
 	RIVKey.nonZeros = NONZEROS;
 	RIVKey.nonZeros = NONZEROS;
-	
+	RIVKey.I2SThreshold = sqrt(RIVSIZE);
 	if(RIVKey.nonZeros%2){
 	if(RIVKey.nonZeros%2){
 		printf("your NONZEROS value must be an even number");
 		printf("your NONZEROS value must be an even number");
 		RIVKey.nonZeros++;
 		RIVKey.nonZeros++;
@@ -248,9 +280,9 @@ void RIVInit(){
 	struct sigaction action;
 	struct sigaction action;
 	action.sa_sigaction = signalSecure;
 	action.sa_sigaction = signalSecure;
 	action.sa_flags = SA_SIGINFO;
 	action.sa_flags = SA_SIGINFO;
-	for(int i=1; i<27; i++){
-		sigaction(i,&action,NULL);
-	}
+	//for(int i=1; i<27; i++){
+		sigaction(11,&action,NULL);
+	//}
 	 
 	 
 	RIVKey.h_tempBlock = (int*)malloc(3*RIVKey.RIVsize*sizeof(int));
 	RIVKey.h_tempBlock = (int*)malloc(3*RIVKey.RIVsize*sizeof(int));
 	RIVKey.tempSize = 3*RIVKey.RIVsize;
 	RIVKey.tempSize = 3*RIVKey.RIVsize;
@@ -286,11 +318,11 @@ int wordtoSeed(unsigned char* word){
 void makeSparseLocations(unsigned char* word,  int *locations, size_t count){
 void makeSparseLocations(unsigned char* word,  int *locations, size_t count){
 	locations+=count;
 	locations+=count;
 	srand(wordtoSeed(word));
 	srand(wordtoSeed(word));
-	for(int i=0; i<RIVKey.nonZeros; i++){
-		/* unrolled for speed, gauranteed to be an even number of steps */
+	int *locations_stop = locations+RIVKey.nonZeros;
+	while(locations<locations_stop){
+		/* unrolled for speed, guaranteed to be an even number of steps */
 		*locations = rand()%RIVKey.RIVsize;
 		*locations = rand()%RIVKey.RIVsize;
 		locations++;
 		locations++;
-		i++;
 		*locations = rand()%RIVKey.RIVsize;
 		*locations = rand()%RIVKey.RIVsize;
 		locations++;
 		locations++;
 	}
 	}
@@ -319,17 +351,19 @@ int fLexPush(denseRIV RIVout){
 	return 0;
 	return 0;
 }
 }
 
 
-denseRIV fLexPull(FILE* lexWord, denseRIV output){
-	
+denseRIV fLexPull(FILE* lexWord){
+	denseRIV output;
+	output.values = malloc( (RIVKey.RIVsize+1) *sizeof(int));
+	output.frequency = output.values+RIVKey.RIVsize;
 	int diagnostic = 0;
 	int diagnostic = 0;
 	diagnostic += fread(output.frequency, 1, sizeof(int), lexWord);
 	diagnostic += fread(output.frequency, 1, sizeof(int), lexWord);
 	diagnostic += fread(&(output.magnitude), 1, sizeof(int), lexWord);
 	diagnostic += fread(&(output.magnitude), 1, sizeof(int), lexWord);
 	diagnostic += fread(output.values, RIVKey.RIVsize, sizeof(int), lexWord);
 	diagnostic += fread(output.values, RIVKey.RIVsize, sizeof(int), lexWord);
 			
 			
-	fclose(lexWord);
 	if(diagnostic != (RIVKey.RIVsize+2)){ 
 	if(diagnostic != (RIVKey.RIVsize+2)){ 
 		output.magnitude = -1;
 		output.magnitude = -1;
 	}
 	}
+	output.cached = 0;
 	return output;
 	return output;
 	
 	
 }
 }
@@ -350,9 +384,22 @@ int cacheDump(){
 	denseRIV* cache_stop = RIVKey.RIVCache+RIVKey.cacheSize;
 	denseRIV* cache_stop = RIVKey.RIVCache+RIVKey.cacheSize;
 	while(cache_slider<cache_stop){
 	while(cache_slider<cache_stop){
 		if((*cache_slider).cached){
 		if((*cache_slider).cached){
-			fLexPush(*cache_slider);
+			flag += fLexPush(*cache_slider);
 		}
 		}
 		cache_slider++;
 		cache_slider++;
 	}
 	}
 	return flag;
 	return flag;
 }
 }
+
+denseRIV denseAllocate(){
+	/* allocates a 0 vector */
+	denseRIV output;
+	output.values = calloc(RIVKey.RIVsize+1, sizeof(int));
+	/* for compact memory use, frequency is placed immediately after values */
+	output.frequency = output.values+RIVKey.RIVsize;
+	output.magnitude = 0;	
+	output.cached = 0;
+	return output;
+}
+
+/*TODO add a simplified free function*/

RIVaccessories.h

+/*isWordClean filters words that contain non-letter characters, and 
+ * upperCase letters, allowing only the '_' symbol through
+ */
+int isWordClean(char* word);
+/* used by wordClean */
+int isLetter(char c);
+
+
+
+int isLetter(char c){
+	
+	if((c>96 && c<123)||(c == 32) || (c == '_')) return 1;
+	else return 0;
+}
+int isWordClean(char* word){
+	char *letter = word;
+	char *word_stop = word+99;
+	while(letter<word_stop){
+		if(!(*letter)) break;
+		if(!(isLetter(*letter))){
+			
+			return 0;
+		}
+		letter++;
+	}
+	return 1;
+		
+}

RIVcullCPUlinux.c

@@ -5,7 +5,8 @@
 
 
 #define RIVSIZE 5000
 #define RIVSIZE 5000
 #define CACHESIZE 0
 #define CACHESIZE 0
-#define THRESHOLD 0.70
+#define NONZEROS 2
+#define THRESHOLD 0.7
 #define COSINEACTION do {\
 #define COSINEACTION do {\
 	if(cosine > THRESHOLD){ \
 	if(cosine > THRESHOLD){ \
 		printf("%s\t%s\n%f\n", baseRIV.name, (*multipliers).name, cosine);\
 		printf("%s\t%s\n%f\n", baseRIV.name, (*multipliers).name, cosine);\
@@ -33,14 +34,49 @@ int main(int argc, char *argv[]){
 
 
 	directoryToL2s(rootString, &fileRIVs, &fileCount);
 	directoryToL2s(rootString, &fileRIVs, &fileCount);
 	printf("fileCount: %d\n", fileCount);
 	printf("fileCount: %d\n", fileCount);
-	getMagnitudes(fileRIVs, fileCount); 
-
+	
+	sparseRIV* fileRIVs_slider = fileRIVs;
+	sparseRIV* fileRIVs_stop = fileRIVs+fileCount;
+	while(fileRIVs_slider <fileRIVs_stop){
+		(*fileRIVs_slider).magnitude = getMagnitudeSparse(*fileRIVs_slider);
+		fileRIVs_slider++;
+		
+	}
+	
 	clock_t beginnsquared = clock();
 	clock_t beginnsquared = clock();
-	for(int i=1; i<fileCount; i++){
-		if(fileRIVs[i].boolean){
-			cosineCompare(fileRIVs[i], fileRIVs, i);
+	float cosine;
+	float minmag;
+	float maxmag;
+	denseRIV baseDense;
+	baseDense.values = malloc(RIVKey.RIVsize*sizeof(int));
+	fileRIVs_slider = fileRIVs;
+	sparseRIV* comparators_slider;
+	while(fileRIVs_slider<fileRIVs_stop){
+		comparators_slider = fileRIVs;
+			
+		memset(baseDense.values, 0, RIVKey.RIVsize*sizeof(int));
+		baseDense.values = addS2D(baseDense.values, *fileRIVs_slider);
+		baseDense.magnitude = (*fileRIVs_slider).magnitude;
+		minmag = baseDense.magnitude*.85;
+		maxmag  = baseDense.magnitude*1.15;
+		while(comparators_slider < fileRIVs_slider){
+			if((*comparators_slider).magnitude < maxmag && (*comparators_slider).magnitude > minmag && (*comparators_slider).boolean){
+				cosine = cosCompare(baseDense, *comparators_slider);
+			
+				
+				if(cosine>THRESHOLD){
+					printf("%s\t%s\n%f\n", (*fileRIVs_slider).name , (*comparators_slider).name, cosine);
+				(*comparators_slider).boolean = 0; 
+					RIVKey.thing++; 
+				}
+			}
+			comparators_slider++;
+		//cosineCompare(fileRIVs[i], fileRIVs, i);
 		}
 		}
 		
 		
+		
+		fileRIVs_slider++;
+		
 	}
 	}
 	clock_t endnsquared = clock();
 	clock_t endnsquared = clock();
 	double time = (double)(endnsquared - beginnsquared) / CLOCKS_PER_SEC;
 	double time = (double)(endnsquared - beginnsquared) / CLOCKS_PER_SEC;
@@ -85,7 +121,7 @@ void directoryToL2s(char *rootString, sparseRIV** fileRIVs, int *fileCount){
 		}else{
 		}else{
 			(*fileRIVs) = (sparseRIV*)realloc((*fileRIVs), ((*fileCount)+1)*sizeof(sparseRIV));
 			(*fileRIVs) = (sparseRIV*)realloc((*fileRIVs), ((*fileCount)+1)*sizeof(sparseRIV));
 			
 			
-			(*fileRIVs)[(*fileCount)] = fileToL2Clean(input);
+			(*fileRIVs)[(*fileCount)] = fileToL2(input);
 			strcpy((*fileRIVs)[(*fileCount)].name, pathString);
 			strcpy((*fileRIVs)[(*fileCount)].name, pathString);
 			
 			
 			fclose(input);
 			fclose(input);

RIVread.c

 #include <stdio.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <stdlib.h>
 #include <time.h>
 #include <time.h>
-#define CACHESIZE 1000
+#define CACHESIZE 10000
 #include "RIVtoolsCPUlinux.h"
 #include "RIVtoolsCPUlinux.h"
 #include <sys/stat.h>
 #include <sys/stat.h>
 #include <sys/types.h>
 #include <sys/types.h>

RIVtoolsMorphic.h

+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <math.h>
+#include "RIVLower.h"
+#include "RIVaccessories.h"
+
+/* lexPush writes a denseRIV to a file for permanent storage */
+int lexPush(denseRIV RIVout);
+/* lexPull reads an existing lexicon entry (under directory "lexicon")
+ * and creates a denseRIV with those attributes.
+ * if the file does not exist, it creates a 0 vector with the name of word
+ */
+denseRIV lexPull(char* word);
+
+/* fileToL2 takes an input file, reads words (delimiting on " " and "\n") 
+ * and returns a sparse RIV which is the vector sum of the base RIVs of each 
+ * word contained
+ */
+sparseRIV fileToL2(FILE *input);
+/* fileToL2Clean operates the same as fileToL2 butkeeps only words 
+ * containing lowercase letters and the '_' symbol
+ * this is important if you will be lexPush-ing those words later
+ */
+sparseRIV fileToL2Clean(FILE *data);
+
+sparseRIV fileToL2direct(FILE *data);
+/*cosine determines the "similarity" between two RIVs. */
+float cosCompare(denseRIV baseRIV, sparseRIV comparator);
+
+sparseRIV wordtoL2(char* word);
+
+sparseRIV consolidateI2S(int *implicit, size_t valueCount);
+sparseRIV text2L2(char *text);
+sparseRIV text2L2(char *text){
+	unsigned int blockSize;
+	char word[100] = {0};
+	
+	/* locations (implicit RIV) are temp stored in temp block, and moved 
+	 * to permanent home in consolidation */
+	int *locations = RIVKey.h_tempBlock;
+	int locationCount = 0;
+	int displacement;
+	
+	while(sscanf(text, "%99s%n", word, &displacement)){
+		text += displacement+1;
+		if(!displacement){
+			break;
+		}
+		
+		if(!(*word)){
+			break;
+		}
+
+		blockSize = locationCount+RIVKey.nonZeros;
+		/* if this word would overflow the locations block, grow it */
+		if(blockSize>RIVKey.tempSize){
+			RIVKey.h_tempBlock = (int*) realloc(RIVKey.h_tempBlock, blockSize*sizeof(int));
+			locations = RIVKey.h_tempBlock;
+			RIVKey.tempSize+=RIVKey.nonZeros;
+		}
+		
+		/* add word's L1 RIV to the accumulating implicit RIV */
+		makeSparseLocations((unsigned char*)word, locations, locationCount);
+		locationCount+= RIVKey.nonZeros;
+		
+	}
+	sparseRIV output = consolidateI2S(locations, locationCount);
+	
+	/* frequency records the number of words in this file */
+	output.frequency = locationCount/RIVKey.nonZeros;
+	output.boolean = 1;
+	return output;
+}
+	
+sparseRIV fileToL2(FILE *data){
+	unsigned int blockSize;
+	unsigned char word[100] = {0};
+	
+	/* locations (implicit RIV) are temp stored in temp block, and moved 
+	 * to permanent home in consolidation */
+	int *locations = RIVKey.h_tempBlock;
+	int locationCount = 0;
+	
+	
+	while(fscanf(data, "%99s", word)){
+	
+		if(feof(data)){
+			break;
+		}
+		if(!(*word)){
+			break;
+		}
+		
+		blockSize = locationCount+RIVKey.nonZeros;
+		/* if this word would overflow the locations block, grow it */
+		if(blockSize>RIVKey.tempSize){
+			RIVKey.h_tempBlock = (int*) realloc(RIVKey.h_tempBlock, blockSize*sizeof(int));
+			locations = RIVKey.h_tempBlock;
+			RIVKey.tempSize+=RIVKey.nonZeros;
+		}
+		
+		/* add word's L1 RIV to the accumulating implicit RIV */
+		makeSparseLocations(word, locations, locationCount);
+		locationCount+= RIVKey.nonZeros;
+		
+	}
+	
+	sparseRIV output = consolidateI2S(locations, locationCount);
+	
+	/* frequency records the number of words in this file */
+	output.frequency = locationCount/RIVKey.nonZeros;
+	output.boolean = 1;
+	
+	return output;
+}
+
+sparseRIV fileToL2Clean(FILE *data){
+	
+		
+	unsigned char word[100] = {0};
+	int *locations = RIVKey.h_tempBlock;
+	unsigned int blockSize;
+	
+	int locationCount = 0;
+	
+	while(fscanf(data, "%99s", word)){
+		
+		if(feof(data)){
+			break;
+		}
+		
+		if(!(*word)){
+			break;
+		}
+		/* if the word is not clean, skip it */
+		if(!isWordClean((char*)word)){
+			continue;
+		}
+		blockSize = locationCount+RIVKey.nonZeros;
+		if(blockSize>RIVKey.tempSize){
+			RIVKey.h_tempBlock = (int*)realloc(RIVKey.h_tempBlock, blockSize*sizeof(int));
+			locations = RIVKey.h_tempBlock;
+			RIVKey.tempSize+=RIVKey.nonZeros;
+		}
+
+		makeSparseLocations(word, locations, locationCount);
+		locationCount+= RIVKey.nonZeros;
+		
+	}
+	
+	sparseRIV output = consolidateI2S(locations, locationCount);
+	
+	/* frequency records the number of words in this file */
+	output.frequency = locationCount/RIVKey.nonZeros;
+	output.boolean = 1;
+	return output;
+}
+
+sparseRIV consolidateI2S(int *implicit, size_t valueCount){
+	if(valueCount>RIVKey.I2SThreshold){
+		return consolidateI2SIndirect(implicit, valueCount);
+	}else{
+		return consolidateI2SDirect(implicit, valueCount);
+	}	
+	
+}
+void aggregateWord2D(denseRIV destination, char* word){
+	
+	//makeSparseLocations((unsigned char*)word, locationSlot, 0);
+	srand(wordtoSeed((unsigned char*)word));
+	for(int i=0; i<RIVKey.nonZeros; i++){
+
+		destination.values[(rand()%RIVKey.RIVsize)] +=1;
+		destination.values[(rand()%RIVKey.RIVsize)] -= 1;
+	}		
+}
+
+float cosCompare(denseRIV baseRIV, sparseRIV comparator){
+		
+	int dot = 0;
+	
+	int *values = comparator.values;
+	int *locations = comparator.locations;
+	int *locations_Stop = locations+comparator.count;
+		
+	while(locations<locations_Stop){
+		/* we calculate the dot-product to derive the cosine */
+		dot += (*values)*(*(baseRIV.values+(*locations)));
+		locations++;
+		values++;
+	}
+	float cosine = dot/(baseRIV.magnitude*comparator.magnitude);
+
+	return cosine;
+}
+
+float getMagnitudeSparse(sparseRIV input){
+	unsigned long long int temp = 0;
+	int *values = input.values;
+	int *values_stop = values+input.count;
+	while(values<values_stop){	
+		temp += (*values)*(*values);
+		values++;
+			
+	}
+	float magnitude = sqrt(temp);
+	input.magnitude = magnitude;
+	return magnitude;
+}
+
+denseRIV lexPull(char* word){
+	#if CACHESIZE > 0
+	
+	/* if there is a cache, first check if the word is cached */
+	srand(wordtoSeed((unsigned char*)word));
+	int hash = rand()%RIVKey.cacheSize;
+	if(!strcmp(word, RIVKey.RIVCache[hash].name)){
+		
+		/* if word is cached, pull from cache and exit */
+		return RIVKey.RIVCache[hash];
+	}
+	#endif /* CACHESIZE > 0 */
+	denseRIV output;
+	
+	
+	
+	char pathString[200];
+	
+	sprintf(pathString, "lexicon/%s", word);
+	FILE *lexWord = fopen(pathString, "rb");
+	
+	/* if this lexicon file already exists */
+	if(lexWord){
+		/* pull data from file */
+		output = fLexPull(lexWord);
+		fclose(lexWord);
+	}else{
+		/*if file does not exist, return a 0 vector */
+		output = denseAllocate();
+	}
+	
+	strcpy(output.name, word);
+	return output;
+}
+int lexPush(denseRIV RIVout){
+	//printf("%s\n", (*RIVout).name);
+	#if CACHESIZE == 0
+		fLexPush(RIVout);
+		return 0;
+	#else /* CACHESIZE != 0 */
+	
+		/* if our RIV was cached, there are two options (hopefully)
+		 * either the RIV is still cached, and the data has been updated to the cache
+		 * or the RIV was pushed out from under it, in which case it has already been pushed*/
+		 
+		if(RIVout.cached){
+			return 0;
+		}
+		
+		srand(wordtoSeed((unsigned char*)RIVout.name));
+		int hash = rand()%RIVKey.cacheSize;
+		
+		if(!RIVKey.RIVCache[hash].cached){
+			RIVKey.RIVCache[hash] = RIVout;
+			RIVKey.RIVCache[hash].cached = 1;
+			return 0;
+			
+		/*if the current RIV is more frequent than the RIV holding it's slot */
+		}else if(*(RIVout.frequency) > *(RIVKey.RIVCache[hash].frequency) ){
+			//scanf("%f", &(*RIVout).magnitude);
+			//printf("%s replacing %s\n", (*RIVout).name, RIVKey.RIVCache[hash].name);
+			/* push the current cache entry to a file */
+			int diag = fLexPush(RIVKey.RIVCache[hash]);
+			/* replace the cache entry with the currrent RIV */
+			
+			RIVKey.RIVCache[hash] = RIVout;
+			RIVKey.RIVCache[hash].cached = 1;
+			return diag;
+		}else{
+			/* push current RIV to file */
+			fLexPush(RIVout);
+		}
+		return 0;
+	#endif /* CACHESIZE == 0 */
+}
+
+sparseRIV fileToL2direct(FILE *data){;
+	unsigned char word[100] = {0};
+	denseRIV denseTemp;
+	// a temporary dense RIV is stored in the tempBlock 
+	denseTemp.values = RIVKey.h_tempBlock;
+	memset(RIVKey.h_tempBlock, 0, RIVSIZE*sizeof(int));
+	int count = 0;
+	while(fscanf(data, "%99s", word)){
+		count++;
+		if(feof(data)){
+			break;
+		}
+		if(!(*word)){
+			break;
+		}
+		
+		
+		// add word's L1 RIV to the accumulating implicit RIV 
+		aggregateWord2D(denseTemp, (char*)word);
+		
+	}
+	sparseRIV output = consolidateD2S(denseTemp.values);
+	
+	// frequency records the number of words in this file 
+	output.frequency = count;
+	output.boolean = 1;
+	return output;
+}
+

RIVunused.h

+int* mapS2D(int* destination, sparseRIV input); //#TODO fix int*/denseRIV confusion
+int* addI2D(int* destination, int* locations, size_t seedCount);
+/* cosine determines the "similarity" between two RIVs.  */
+float* cosineCompare(sparseRIV baseRIV, sparseRIV *multipliers, size_t multiplierCount);
+/* magnitudes will be used later in cosine comparison */
+void getMagnitudes(sparseRIV *inputs, size_t RIVCount);
+unsigned char *sscanAdvance(unsigned char **string, unsigned char *word);//unused except in text2l2
+sparseRIV text2L2(unsigned char *text);//unused
+float* cosineCompareUnbound(sparseRIV baseRIV, sparseRIV *multipliers, size_t multiplierCount);
+/*lexPush writes a denseRIV to a file of the same name, under the directory "lexicon"
+ * it is up to the programmer to ensure that the name of the RIV is a valid filename
+ * although it will of course attempt to create the file if it does not exist
+ */
+int* mapS2D(denseRIV destination, sparseRIV input){// #TODO fix destination parameter vs calloc of destination
+	// make sure our destination is a 0 vector 
+	memset(destination.values, 0, RIVKey.RIVsize*sizeof(int));
+	
+	int *locations_slider = input.locations;
+	int *values_slider = input.values;
+	int *locations_stop = locations_slider+input.count;
+	
+	// apply values at an index based on locations 
+	while(locations_slider<locations_stop){
+		destination[*locations_slider] = *values_slider;
+		locations_slider++;
+		values_slider++;
+	}
+	strcpy(destination.name, input.name);
+	*(destination.frequency) = input.frequency;
+	destination.magnitude = input.magnitude;
+	
+	return destination;
+}
+int* addI2D(int* destination, int *locations, size_t valueCount){// #TODO fix destination parameter vs calloc of destination
+	int *locations_slider = locations;
+	int *locations_stop = locations_slider+valueCount;
+
+	/*apply values +1 or -1 at an index based on locations */
+	while(locations_slider<locations_stop){
+	
+		destination[*locations_slider] +=1;
+		locations_slider++;
+		destination[*locations_slider] -= 1;
+		locations_slider++;
+	}
+	
+	
+	return destination;
+}
+float* cosineCompare(sparseRIV baseRIV, sparseRIV *multipliers, size_t multiplierCount){
+	float *results = calloc(multiplierCount, sizeof(float));
+	float* results_slider = results;
+	int *baseDenseRIV = RIVKey.h_tempBlock;
+	memset(RIVKey.h_tempBlock, 0, RIVKey.RIVsize*sizeof(int));
+	addS2D(baseDenseRIV, baseRIV);
+	float cosine;
+	sparseRIV *multipliersStop = multipliers+multiplierCount;
+	
+	/* if two vectors are too different in size, we can ignore the risk of similarity */
+	float minsize = baseRIV.magnitude * .85;
+	float maxsize = baseRIV.magnitude * 1.15;
+	int dot = 0;
+	
+	int *values;
+	int *locations;
+	int *locations_Stop;
+	/* check the baseRIV against each multiplier */
+	while(multipliers<multipliersStop){
+		/* skip a pair if the multiplier has already been culled, or if
+		 * the size difference is too great */
+		if(((*multipliers).boolean) 
+		&& (((*multipliers).magnitude < maxsize) 
+		&& ((*multipliers).magnitude > minsize))){
+			
+			dot = 0;
+			values = (*multipliers).values;
+			locations = (*multipliers).locations;
+			locations_Stop = locations+(*multipliers).count;
+			
+			while(locations<locations_Stop){
+				/* we calculate the dot-product to derive the cosine */
+				dot += (*values)*(*(baseDenseRIV+(*locations)));
+				locations++;
+				values++;
+			}
+			/* magnitudes had better already be calculated at this point*/
+			cosine = dot/((baseRIV.magnitude)*((*multipliers).magnitude));
+			*results_slider = cosine;
+			results_slider++;
+			/* perform the action defined by the COSINEACTION macro */
+			COSINEACTION;
+		}
+		multipliers++;
+		
+	}
+	return results;
+}
+float* cosineCompareUnbound(sparseRIV baseRIV, sparseRIV *multipliers, size_t multiplierCount){
+	float *results = calloc(multiplierCount, sizeof(float));
+	float* results_slider = results;
+	int *baseDenseRIV = RIVKey.h_tempBlock;
+	memset(RIVKey.h_tempBlock, 0, RIVKey.RIVsize*sizeof(int));
+	addS2D(baseDenseRIV, baseRIV);
+	float cosine;
+	sparseRIV *multipliersStop = multipliers+multiplierCount;
+	
+	/* if two vectors are too different in size, we can ignore the risk of similarity */
+	int dot = 0;
+	
+	int *values;
+	int *locations;
+	int *locations_Stop;
+	/* check the baseRIV against each multiplier */
+	while(multipliers<multipliersStop){
+			
+		dot = 0;
+		values = (*multipliers).values;
+		locations = (*multipliers).locations;
+		locations_Stop = locations+(*multipliers).count;
+		
+		while(locations<locations_Stop){
+			/* we calculate the dot-product to derive the cosine */
+			dot += (*values)*(*(baseDenseRIV+(*locations)));
+			locations++;
+			values++;
+		}
+		/* magnitudes had better already be calculated at this point*/
+		cosine = dot/((baseRIV.magnitude)*((*multipliers).magnitude));
+		*results_slider = cosine;
+		results_slider++;
+		/* perform the action defined by the COSINEACTION macro */
+		COSINEACTION;
+	}
+	multipliers++;
+	
+
+	return results;
+}
+
+void getMagnitudes(sparseRIV *inputs, size_t RIVCount){
+	for(int i=0; i<RIVCount; i++){
+		unsigned long long int temp = 0;
+		int *values = inputs[i].values;
+		int *values_stop = values+inputs[i].count;
+		while(values<values_stop){	
+			temp += (*values)*(*values);
+			values++;
+				
+		}
+		float magnitude = sqrt(temp);
+		inputs[i].magnitude = magnitude;
+	}
+}
+
+

books/pg15193clean/1

-project gutenberg scientific american supplement no 
\ No newline at end of file

books/pg15193clean/10

-title scientific american supplement no february 
\ No newline at end of file

books/pg15193clean/100

-rate feet work 
\ No newline at end of file

books/pg15193clean/1000

-cities herculaneum pompeii several smaller towns on slope 
\ No newline at end of file

books/pg15193clean/1001

-mountain destroyed lava buried under mass 
\ No newline at end of file

books/pg15193clean/1002

-pumice stones ashes second 
\ No newline at end of file

books/pg15193clean/1003

-contiguous mountains in iceland in two enormous lava 
\ No newline at end of file

books/pg15193clean/1004

-streams one miles wide over ft deep other 
\ No newline at end of file

books/pg15193clean/1005

-scarcely inferior flowed first miles other till 
\ No newline at end of file

books/pg15193clean/1006

-reached sea pouring flood white hot lava 
\ No newline at end of file

books/pg15193clean/1007

-ocean destroying in paths killing in water 
\ No newline at end of file

books/pg15193clean/1008

-ocean fish mainstay inhabitants who 
\ No newline at end of file

books/pg15193clean/1009

-reduced disaster directly indirectly less 
\ No newline at end of file

books/pg15193clean/1010

-former strength third 
\ No newline at end of file

books/pg15193clean/1011

-in devastated such an immense area in java but all 
\ No newline at end of file

books/pg15193clean/1012

-eruptions known besides as mere childs play terrible one 
\ No newline at end of file

books/pg15193clean/1013

-krakatoa in 
\ No newline at end of file

books/pg15193clean/1015

-reader will examine map east indies will find 
\ No newline at end of file

books/pg15193clean/1016

-represented in straits lie between sumatra 
\ No newline at end of file

books/pg15193clean/1017

-java little island krakatoa in maps make before will 
\ No newline at end of file

books/pg15193clean/1018

-hunt in vain name like bull run before then 
\ No newline at end of file

books/pg15193clean/1019

-unknown fame though navigators who passed through straits knew 
\ No newline at end of file

books/pg15193clean/102

-vi educational competitive examinations 
\ No newline at end of file

books/pg15193clean/1020

-as beautiful tropical isle an extinct volcanic cone in 
\ No newline at end of file

books/pg15193clean/1021

-center in beginning however little well behaved 
\ No newline at end of file

books/pg15193clean/1022

-island showed symptoms wrath boded no good larger 
\ No newline at end of file

books/pg15193clean/1023

-islands in vicinity noted fine fruits 
\ No newline at end of file

books/pg15193clean/1024

-abounded famous picnic ground towns cities even 
\ No newline at end of file

books/pg15193clean/1025

-miles away subterranean rumblings mutterings 
\ No newline at end of file

books/pg15193clean/1026

-wrath became conspicuous people capital java 
\ No newline at end of file

books/pg15193clean/1027

-put steamboat requisition visited island in large 
\ No newline at end of file

books/pg15193clean/1028

-number time island constantly in slight tremor 
\ No newline at end of file

books/pg15193clean/1029

-subterranean roar like continue but distant mutterings 
\ No newline at end of file

books/pg15193clean/103

-interesting details famous examinations 
\ No newline at end of file

books/pg15193clean/1030

-thunder but crisis reached august am 
\ No newline at end of file

books/pg15193clean/1031

-beautiful sunday morning water straits 
\ No newline at end of file

books/pg15193clean/1032

-like sea glass as clear as crystal john in 
\ No newline at end of file

books/pg15193clean/1033

-apocalyptic vision speaks beauty morning enhanced 
\ No newline at end of file

books/pg15193clean/1034

-extraordinary transparency tropical air distant mountain 
\ No newline at end of file

books/pg15193clean/1035

-ranges seem so near seem possible strike 
\ No newline at end of file

books/pg15193clean/1036

-stone cast hand only mysterious rumblings mutterings 
\ No newline at end of file

books/pg15193clean/1037

-pent up forces beneath island disturbed breathless calm 
\ No newline at end of file

books/pg15193clean/1038

-silence lay on calm before terrible 
\ No newline at end of file

books/pg15193clean/1039

-mightiest most awful on record burst forth sudden 
\ No newline at end of file

books/pg15193clean/104

-consequences overworked competitors 
\ No newline at end of file

books/pg15193clean/1040

-night snatched away day eyes terrified beholders on 
\ No newline at end of file

books/pg15193clean/1041

-mainland but vivid play lightnings around ascending 
\ No newline at end of file

books/pg15193clean/1042

-column dust penetrated even deep obscurity distance 
\ No newline at end of file

books/pg15193clean/1043

-miles awful darkness stretched within circle diameter 
\ No newline at end of file

books/pg15193clean/1044

-miles while more less darkness reigned within circle 
\ No newline at end of file

books/pg15193clean/1045

-diameter three times as great within latter area dust 
\ No newline at end of file

books/pg15193clean/1046

-fall like snow sky breaking off limbs trees its weight 
\ No newline at end of file

books/pg15193clean/1047

-miles distant while in miles away scene 
\ No newline at end of file

books/pg15193clean/1048

-disaster fall depth several inches explosions 
\ No newline at end of file

books/pg15193clean/1049

-so loud as distinctly heard in miles away 
\ No newline at end of file

books/pg15193clean/1050

-sound like constant roar cannon in field 
\ No newline at end of file

books/pg15193clean/1051

-battle finally whole island blown pieces now came 
\ No newline at end of file

books/pg15193clean/1052

-most awful contest battle death between neptune 
\ No newline at end of file

books/pg15193clean/1053

-vulcan sea poured down chasm millions tons only 
\ No newline at end of file

books/pg15193clean/1054

-first converted vapor millions tons 
\ No newline at end of file

books/pg15193clean/1055

-seething white hot lava beneath over shores miles away waves 
\ No newline at end of file

books/pg15193clean/1056

-over ft high rolled such fury even 
\ No newline at end of file

books/pg15193clean/1057

-part bedrock swept away blocks stone tons 
\ No newline at end of file

books/pg15193clean/1058

-weight carry two miles inland on sumatra side 
\ No newline at end of file

books/pg15193clean/1059

-straits large vessel carry three miles inland wave 
\ No newline at end of file

books/pg15193clean/106

-vii electrical speed engine 
\ No newline at end of file

books/pg15193clean/1060

-course growing less in intensity traveled across whole indian 
\ No newline at end of file

books/pg15193clean/1061

-ocean miles cape good hope around 
\ No newline at end of file

books/pg15193clean/1062

-atlantic waves in atmosphere traveled around globe three 
\ No newline at end of file

books/pg15193clean/1063

-times rate miles hour dust volcano 
\ No newline at end of file

books/pg15193clean/1064

-carry up atmosphere fully twenty miles finest 
\ No newline at end of file

books/pg15193clean/1065

-distribute through whole body air reader doubtless 
\ No newline at end of file

books/pg15193clean/1066

-remembers beautiful reddish purple glow sunrise sunset 
\ No newline at end of file

books/pg15193clean/1067

-fully six months after august glow caused 
\ No newline at end of file

books/pg15193clean/1068

-volcanic dust in atmosphere interfering passage 
\ No newline at end of file

books/pg15193clean/1069

-suns rays upper part solar spectrum more manifest 
\ No newline at end of file