int main(int argc, char * argv[]) {
int i = 0;
char *ptr = NULL;
char *string = NULL;
int fd = -1;
Header* p;
char *tab = NULL;
char size = sizeof(char)*7;
name_file(&i);
/*debug_println_ptr("dedans address ", &address);*/
fd = open_file("t");
if (fd != -1){
ptr = (char*)mmap_file();
string = (char*)mymalloc(size);
if (string == NULL) {
perror("Erreur d'allocation");
}
strcpy(string, "coucou ");
/*p = (void*)string - sizeof(Header);
printf("%s\n", string);
printf("string:%d size:%d address:%d\n", (int)string, p->size, address);*/
tab = (char*)mymalloc(size);
if (tab == NULL) {
perror("Erreur d'allocation");
}
strcpy(tab, "coucou ");
p = (void*)string - sizeof(Header);
/* printf("string -> %s | tab -> %s | poniteur address %d string address %d file %d\n", string, tab, (int)p, (int)string, (int)ptr); */
printf("\n p->size %d\n", p->size);
/*debug_println_ptr("dehors address ", &address);*/
myfree(string);
string = NULL;
/*printf("string -> %s | tab -> %s\n", string, tab);*/
}
else
printf("Erreur %d\n", fd);
exit(EXIT_SUCCESS);
}
// Runs experiments with one inputed data set and varies the memory
// Input: takes 5 command line arguments, lower and upper fraction of memory,
// names of files the data are in, and number of bins
// Output: creates 4 files, the log file holds all the data generated, the
// table file holds the pvalues deliminated by tabs, the extra file holds all
// the calculated statistics. and the pvalue holds the actual and estimated
// pvalues
int main(int argc, char* argv[])
{
if (argc < 5)
{
cout << "usage: VaryMemoryCategorical lower-memory upper-memory filename1 filename2" << endl;
exit(1);
}
double lower = atof(argv[1]), upper = atof(argv[2]);
char *filename1 = argv[3], *filename2 = argv[4];
double memory_percent, mem = lower;
int repeats = 0;
while (mem <= (upper + 0.00000001))
{
repeats++;
mem += 10;
}
double actual_values[repeats], estimated_values[repeats], percents[repeats];
long times[repeats];
long times2[repeats];
std::vector<double> data1, data2;
std::unordered_map<double,int> stream1, stream2;
// creates and initializes the log file
ofstream data_file;
char str[100];
name_file(str, argv, 0);
data_file.open(str);
ifstream input_file;
char output[100];
int stream_size1 = 0, stream_size2 = 0;
input_file.open(filename1);
while (!input_file.eof())
{
input_file >> output;
data1.push_back(atof(output));
stream_size1++;
if (stream1.find(atof(output)) == stream1.end())
stream1.insert(std::make_pair(atof(output),1));
else
stream1[atof(output)] += 1;
}
input_file.close();
input_file.open(filename2);
while (!input_file.eof())
{
input_file >> output;
data2.push_back(atof(output));
stream_size2++;
if (stream2.find(atof(output)) == stream2.end())
stream2.insert(std::make_pair(atof(output),1));
else
stream2[atof(output)] += 1;
}
input_file.close();
data_file << filename1 << endl << filename2 << endl;
data_file << "stream 1 size: " << stream_size1 << endl;
data_file << "stream 2 size: " << stream_size2 << endl;
data_file << "number of categories: " << stream1.size() << endl;
memory_percent = lower;
int i = 0, num_categories;
while (memory_percent <= (upper + 0.00000001)) //accounts for rounding
{
data_file << "memory percent: " << memory_percent << endl;
percents[i] = memory_percent;
// calculates the estimated statistic
ChiSquareCategorical sketch1(memory_percent);
ChiSquareCategorical sketch2(memory_percent);
timeval timeBefore, timeAfter; // initializes variables
long diffSeconds, diffUSeconds;
gettimeofday(&timeBefore, NULL);
for (std::vector<double>::iterator j = data1.begin(); j != data1.end();j++)
sketch1.insert(*j);
for (std::vector<double>::iterator j = data2.begin(); j != data2.end();j++)
sketch2.insert(*j);
gettimeofday(&timeAfter, NULL); // get time for insertion
diffSeconds = timeAfter.tv_sec - timeBefore.tv_sec;
diffUSeconds = timeAfter.tv_usec - timeBefore.tv_usec;
times[i] = diffSeconds;
times2[i] = diffUSeconds;
double estimated_stat = sketch1.calculate_statistic(sketch2, 0);
estimated_values[i] = estimated_stat;
data_file << "estimate = " << estimated_stat << endl;
// calculates actual statistic
double constant1 = sqrt(double(stream_size2) / double(stream_size1));
double constant2 = sqrt(double(stream_size1) / double(stream_size2));
double actual_stat = 0;
for (std::unordered_map<double,int>::const_iterator j = stream1.begin(); j!= stream1.end(); j++)
{
double frequency1 = j->second;
double frequency2 = 0;
num_categories++;
if (stream2.find(j->first) != stream2.end())
frequency2 = stream2[j->first];
double value = frequency1 * constant1 - frequency2 * constant2;
actual_stat += (value * value) / (frequency1 + frequency2);
}
// have to loop through other stream to find when first one is 0
for (std::unordered_map<double,int>::const_iterator j = stream2.begin(); j!= stream2.end(); j++)
{
if (stream1.find(j->first) == stream1.end())
{
num_categories++;
int frequency1 = 0;
int frequency2 = j->second;
double value = frequency1 * constant1 - frequency2 * constant2;
actual_stat += (value * value) / (frequency1 + frequency2);
}
}
actual_values[i] = actual_stat;
data_file << "actual = " << actual_stat << endl;
i++;
memory_percent += 10;
}
data_file.close();
// creates pvalues file
name_file(str, argv, 3);
data_file.open(str);
// creates table file
ofstream data2_file;
name_file(str, argv, 1);
data2_file.open(str);
// creates time table file
ofstream time_file;
char str2[150];
name_file(str2, argv, 5);
time_file.open(str2);
// creates extra file
ofstream data3_file;
name_file(str, argv, 2);
data3_file.open(str);
int deg_freedom = num_categories - 1;
for (int i = 0; i < repeats; i++)
{
// adds values to the table
data2_file << percents[i] << "\t";
double error = abs(pochisq(estimated_values[i], deg_freedom) - pochisq(actual_values[i], deg_freedom));
data2_file << error << endl;
// adds values to the time table
time_file << percents[i] << "\t";
long double avg_time = times[i] + times2[i]/1000000.0;
time_file << avg_time << endl;
// adds values to pvalue file
data_file << pochisq(actual_values[i], deg_freedom) << " actual" << endl;
data_file << pochisq(estimated_values[i], deg_freedom) << " estimated" << endl;
// adds values to extra file
data3_file << percents[i] << "\t";
error = abs(estimated_values[i] - actual_values[i]) / actual_values[i];
data3_file << error << endl;
}
data3_file.close();
data2_file.close();
data_file.close();
return 0;
}
// Runs experiments with one inputed data set and varies the memory
// Input: takes 5 command line arguments, lower and upper fraction of memory,
// names of files the data are in, and number of bins
// Output: creates 4 files, the log file holds all the data generated. the tabl
// file holds the pvalues deliminated by tabs, the extra file holds all the
// calculated statistics. and the pvalue holds the actual and estimated pvalues
int main(int argc, char* argv[])
{
if (argc < 6)
{
cout << "usage: VaryMemoryReal lower-memory upper-memory filename1 filename2 num_bins\n";
exit(1);
}
double lower = atof(argv[1]), upper = atof(argv[2]);
char *filename1 = argv[3], *filename2 = argv[4];
int num_buckets = atoi(argv[5]);
if (lower <=0 || upper <= 0)
{
cout << "The memory must be greater than 0.\n";
exit(1);
}
if (num_buckets <= 0)
{
cout << "The number of buckets must be greater than 0.\n";
exit(1);
}
// finds the number of times the experiment will run
double memory_percent;
int repeats = 0;
double mem = lower;
while (mem <= (upper + 0.0000001)) // accounts for rounding error
{
repeats++;
mem *= sqrt(10);
}
double actual_values[repeats];
double GK_values[repeats];
double QD_values[repeats];
double RS_values[repeats];
double percents[repeats];
std::vector<double> data1, data2;
// creates and initializes the log file
ofstream data_file;
char str[100];
name_file(str, argv, 0);
data_file.open(str);
ifstream input_file;
char output[100];
int stream_size1 = 0, stream_size2 = 0;
std::default_random_engine generator(1);
std::uniform_real_distribution<double> distribution(0.0, 1.0);
input_file.open(filename1);
while (!input_file.eof())
{
input_file >> output;
data1.push_back(atof(output) + (distribution(generator) * 0.000000001));
stream_size1++;
}
input_file.close();
input_file.open(filename2);
while (!input_file.eof())
{
input_file >> output;
data2.push_back(atof(output) + (distribution(generator) * 0.000000001));
stream_size2++;
}
input_file.close();
data_file << filename1 << endl << filename2 << endl;
data_file << "num_buckets: " << num_buckets << endl;
data_file << "stream 1 size: " << stream_size1 << endl;
data_file << "stream 2 size: " << stream_size2 << endl;
memory_percent = lower;
int i = 0;
while (memory_percent <= (upper + 0.00000001)) //accounts for rounding
{
data_file << "memory percent: " << memory_percent << endl;
percents[i] = memory_percent;
int sample_size1 = memory_percent * stream_size1;
int sample_size2 = memory_percent * stream_size2;
// calculates GK statistic
ChiSquareContinuous GK_sketch1(sample_size1,1);
for (std::vector<double>::iterator j = data1.begin(); j != data1.end();j++)
GK_sketch1.insert(*j);
ChiSquareContinuous GK_sketch2(sample_size2,1);
for (std::vector<double>::iterator j = data2.begin(); j != data2.end();j++)
GK_sketch2.insert(*j);
double GK_stat = GK_sketch1.two_sample_statistic(GK_sketch2, num_buckets);
GK_values[i] = GK_stat;
data_file << "GK = " << GK_stat << endl;
// calculates real statistic
double *upper_intervals = GK_sketch1.get_upper();
double *lower_intervals = GK_sketch1.get_lower();
double constant_1 = sqrt((double)stream_size2/stream_size1);
double constant_2 = sqrt((double)stream_size1/stream_size2);
double chi_squared = 0;
for (int i = 0; i < num_buckets; i++)
{
double frequency1 = 0, frequency2 = 0;
for (std::vector<double>::iterator j = data1.begin(); j!=data1.end();j++)
{
if (*j <= upper_intervals[i+1] && *j > lower_intervals[i+1])
frequency1++;
}
for (std::vector<double>::iterator j = data2.begin(); j!=data2.end();j++)
{
if (*j <= upper_intervals[i+1] && *j > lower_intervals[i+1])
frequency2++;
}
double lambda = frequency1 * constant_1 - frequency2 * constant_2;
chi_squared += (lambda * lambda) / (frequency1 + frequency2);
}
actual_values[i] = chi_squared;
data_file << "actual = " << chi_squared << endl;
memory_percent *= sqrt(10);
i++;
}
data_file.close();
// creates pvalue file
name_file(str, argv, 3);
data_file.open(str);
// creates table file
ofstream data2_file;
name_file(str, argv, 1);
data2_file.open(str);
// creates extra file
ofstream data3_file;
name_file(str, argv, 2);
data3_file.open(str);
int deg_freedom = num_buckets;
if (stream_size1 != stream_size2)
deg_freedom--;
for (int i = 0; i < repeats; i++)
{
// adds values to the table
data2_file << percents[i] * 100 << "\t";
double error = abs(pochisq(GK_values[i], deg_freedom) - pochisq(actual_values[i], deg_freedom));
data2_file << error << endl;
// adds values to pvalue file
data_file << pochisq(actual_values[i], deg_freedom) << " actual" << endl;
data_file << pochisq(GK_values[i], deg_freedom) << " GK" << endl;
// adds values to extra file
data3_file << percents[i] * 100 << "\t";
error = abs(GK_values[i] - actual_values[i]) / actual_values[i];
data3_file << error << endl;
}
data3_file.close();
data2_file.close();
data_file.close();
return 0;
}
string query_allow_login(void)
{
return name_file(TO) + ":query_stat_index|set_stat_index";
}
