int beetle(void)
{
const int replicates = 10;
const int n_samples = 10;
double s1[n_samples*replicates];
double s2[n_samples*replicates];
double s3[n_samples*replicates];
double s4[n_samples*replicates];
// 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
// Pars = {E, L, P, A, b, ue, ul, up, ua, ae, al, ap, cle, cap, cae, V}
double inits[NP] = {100, 0, 0, 0,
5.,
0, 0.001, 0, 0.003,
1/3.8, 1/(20.2-3.8), 1/(25.5-20.2),
0.1, 0.04, 0.1, 1000};
int n = n_samples;
double maxtime = 50;
int reps = replicates;
beetles(s1, s2, s3, s4, inits, &n, &reps, &maxtime);
printf("m1 = %g sd1 = %g nm2 = %g sd2 = %g m3 = %g sd3 = %g m4 = %g sd4 = %g\n",
gsl_stats_mean(s1, 1, n_samples),
sqrt(gsl_stats_variance(s1, 1, n_samples)),
gsl_stats_mean(s2, 1, n_samples),
sqrt(gsl_stats_variance(s2, 1, n_samples)),
gsl_stats_mean(s3, 1, n_samples),
sqrt(gsl_stats_variance(s3, 1, n_samples)),
gsl_stats_mean(s4, 1, n_samples),
sqrt(gsl_stats_variance(s4, 1, n_samples))
);
return 0;
}
/** \brief Euclidean distance between normalized vector x1 and x2.
Normalization is defined by
\f[
\hat{x}_i = \frac{x_i-\langle x_i \rangle_i}{\sqrt{\langle x_i^2\rangle_i}}
\f]
\param x1, x2 vectors of size p
\param optargs is ignored
*/
double vectordist_euclidean_normalized( const double *x1, const double *x2, int p,
OptArgList *optargs ){
int i;
double d;
double avg1, avg2,
rms1=0, rms2=0;
double norm1,norm2; /* normalized signal at time t */
avg1 = gsl_stats_mean(x1, 1, p);
avg2 = gsl_stats_mean(x2, 1, p);
for(i=0; i<p; i++){
rms1 += SQR( x1[i] );
rms2 += SQR( x2[i] );
}
rms1 = sqrt(rms1/(double)p);
rms2 = sqrt(rms2/(double)p);
d=0.0;
for( i=0; i<p; i++ ){
norm1 = (x1[i]-avg1)/rms1;
norm2 = (x2[i]-avg2)/rms2;
d += SQR( norm1-norm2 );
}
d = sqrt( d );
return d;
}
double compute_cov_mean(long nreps, double* runtimes_sec,
pred_method_info_t prediction_info) {
int my_rank, i,j;
double* mean_list = NULL;
int nmeans;
long current_nreps;
double* tmp_runtimes;
double cov_mean = COEF_ERROR_VALUE;
MPI_Comm_rank(MPI_COMM_WORLD, &my_rank);
if (my_rank == OUTPUT_ROOT_PROC) {
double mean_of_means = 0, q1,q3, sd;
long start_index, end_index;
double* runtimes;
nmeans = prediction_info.method_win;
if (nmeans > nreps) {
return COEF_ERROR_VALUE;
}
mean_list = (double*)malloc(prediction_info.method_win * sizeof(double));
tmp_runtimes = (double*)malloc(nreps * sizeof(double));
for (i = 0; i < nmeans; i++) {
current_nreps = nreps - i;
for (j=0; j < current_nreps; j++) {
tmp_runtimes[j] = runtimes_sec[j];
}
runtimes = tmp_runtimes;
gsl_sort(tmp_runtimes, 1, current_nreps);
if (current_nreps > OUTLIER_FILTER_MIN_MEAS) {
q1 = gsl_stats_quantile_from_sorted_data (tmp_runtimes, 1, current_nreps, 0.25);
q3 = gsl_stats_quantile_from_sorted_data (tmp_runtimes, 1, current_nreps, 0.75);
filter_outliers_from_sorted(tmp_runtimes, current_nreps, q1, q3, OUTLIER_FILTER_THRES, &start_index, &end_index);
runtimes = runtimes + start_index;
current_nreps = end_index - start_index + 1;
}
mean_list[i] = gsl_stats_mean(runtimes, 1, current_nreps);
}
mean_of_means = gsl_stats_mean(mean_list, 1, nmeans);
sd = gsl_stats_sd(mean_list, 1, nmeans);
cov_mean = sd/(mean_of_means);
//printf("cov_mean=%lf, nreps = %ld, thres=%lf (mean_of_means=%.10f)\n", cov_mean, nreps, prediction_info.method_thres, mean_of_means);
free(tmp_runtimes);
free(mean_list);
}
return cov_mean;
}
int main(void) {
double vol[NEXP]; /* calculated values of SE */
gsl_rng *r;
double mean, sd;
int j, n;
unsigned long seed = 1UL;
long p = pow(2, 19.);
/* allocate random number generator */
r = gsl_rng_alloc(gsl_rng_taus2);
/* seed the random number generator */
gsl_rng_set(r, seed);
for (n = 1; n < 10; n++)
{
for (j = 0; j < NEXP; j++) {
/* calculate pi using Monte Carlo algorithm */
vol[j] = mc_se(p, n, r);
}
mean = gsl_stats_mean(vol, 1, NEXP);
sd = gsl_stats_sd_m(vol, 1, NEXP, mean);
printf("%15ld %15d %14.8f %14.8f\n", p, n, mean, sd);
}
gsl_rng_free(r);
return(0);
}
static ERL_NIF_TERM arithmetic_mean(ErlNifEnv * env, int argc, const ERL_NIF_TERM argv[]) {
double_list dl = alloc_double_list(env, argv[0]);
ARG_ERROR_IF_DL_IS_NULL(dl);
ERL_NIF_TERM final = enif_make_double(env, gsl_stats_mean(dl.list, 1, dl.length));
free_double_list(dl);
return final;
}
void BoxCurve::drawSymbols(QPainter *painter, const QwtScaleMap &xMap,
const QwtScaleMap &yMap, double *dat,
int size) const {
const int px = xMap.transform(x(0));
QwtSymbol s = this->symbol();
if (min_style != QwtSymbol::NoSymbol) {
const int py_min = yMap.transform(y(0));
s.setStyle(min_style);
s.draw(painter, px, py_min);
}
if (max_style != QwtSymbol::NoSymbol) {
const int py_max = yMap.transform(y(size - 1));
s.setStyle(max_style);
s.draw(painter, px, py_max);
}
if (p1_style != QwtSymbol::NoSymbol) {
const int p1 =
yMap.transform(gsl_stats_quantile_from_sorted_data(dat, 1, size, 0.01));
s.setStyle(p1_style);
s.draw(painter, px, p1);
}
if (p99_style != QwtSymbol::NoSymbol) {
const int p99 =
yMap.transform(gsl_stats_quantile_from_sorted_data(dat, 1, size, 0.99));
s.setStyle(p99_style);
s.draw(painter, px, p99);
}
if (mean_style != QwtSymbol::NoSymbol) {
const int mean = yMap.transform(gsl_stats_mean(dat, 1, size));
s.setStyle(mean_style);
s.draw(painter, px, mean);
}
}
/**
* Get a summary statistic for the orbital elements; for instance,
* the median value calculated over all the elements of the list.
* @param kl List
* @param what Can be one of: STAT_MEAN, STAT_MEDIAN, STAT_STDDEV, STAT_MAD.
* Summary statistic is calculated correctly for angle parameters.
* @return A matrix whose entries are the summary statistic for the
* corresponding orbital element.
*/
gsl_matrix* KL_getElementsStats(const ok_list* kl, const int what) {
int npl = MROWS(kl->kernels[0]->elements);
if (npl == 0)
return NULL;
gsl_vector* v = gsl_vector_alloc(kl->size);
gsl_matrix* m = gsl_matrix_alloc(npl, ALL_ELEMENTS_SIZE);
gsl_matrix_set_all(m, 0.);
for (int i = 0; i < npl; i++)
for (int j = 0; j < ALL_ELEMENTS_SIZE; j++) {
for (int n = 0; n < kl->size; n++) {
VSET(v, n, MGET(kl->kernels[n]->elements, i, j));
}
switch (what) {
case STAT_MEAN:
if (j == MA || j == LOP || j == INC || j == NODE || j == TRUEANOMALY)
MSET(m, i, j, ok_average_angle(v->data, v->size, false));
else
MSET(m, i, j, gsl_stats_mean(v->data, 1, v->size));
break;
case STAT_STDDEV:
if (j == MA || j == LOP || j == INC || j == NODE || j == TRUEANOMALY) {
MSET(m, i, j, ok_stddev_angle(v->data, v->size, false));
}
else
MSET(m, i, j, gsl_stats_sd(v->data, 1, v->size));
break;
case STAT_MEDIAN:
if (j == MA || j == LOP || j == INC || j == NODE || j == TRUEANOMALY)
MSET(m, i, j, ok_median_angle(v->data, v->size, false));
else {
gsl_sort_vector(v);
MSET(m, i, j, gsl_stats_median_from_sorted_data(v->data, 1, v->size));
}
break;
case STAT_MAD:
if (j == MA || j == LOP || j == INC || j == NODE || j == TRUEANOMALY) {
double med = ok_median_angle(v->data, v->size, false);
MSET(m, i, j, 1.4826 * ok_mad_angle(v->data, v->size, med, false));
} else {
gsl_sort_vector(v);
double med = gsl_stats_median_from_sorted_data(v->data, 1, v->size);
MSET(m, i, j, 1.4826 * ok_mad(v->data, v->size, med));
}
break;
default:
// percentiles
gsl_sort_vector(v);
MSET(m, i, j, gsl_stats_quantile_from_sorted_data(v->data, 1, v->size, (double)(what)/100.));
};
}
gsl_vector_free(v);
return m;
}
void GbA::compute_likelihood(double *data, int nstats, int nbands, int nsim,
double mean[2], double cov[2][2]){
int i, j, k, l, idx;
gsl_vector *lsq = gsl_vector_alloc (_td->get_noftraces());
gsl_vector *input = gsl_vector_alloc (nbands);
gsl_vector *trace = gsl_vector_alloc (nbands);
double *mags, *dists;
double misfit_l2;
size_t *indices = new size_t[nsim];
mags = new double[nsim];
dists = new double[nsim];
for (i=0; i<nstats; i++){
for (j=0; j<nbands; j++){
gsl_vector_set(input, j, std::log10(data[i*nbands+j]));
}
for (k=0; k<_td->get_noftraces(); k++){
gsl_matrix_get_row(trace, _td->_tdata_gsl, k);
gsl_vector_sub(trace,input);
misfit_l2 = gsl_blas_dnrm2(trace);
gsl_vector_set(lsq, k, misfit_l2*misfit_l2);
}
gsl_sort_vector_smallest_index(indices, nsim,lsq);
mean[0] = 0;
mean[1] = 0;
for(l=0; l<nsim; l++){
dists[l] = std::log10(gsl_vector_get(_td->_r_gsl, indices[l]));
mags[l] = gsl_vector_get(_td->_m_gsl, indices[l]);
//std::cout << gsl_vector_get(_td->_m_gsl,indices[l]) << std::endl;
}
mean[0] = gsl_stats_mean(dists,1,nsim);
mean[1] = gsl_stats_mean(mags,1,nsim);
cov[0][0] = gsl_stats_variance(dists, 1, nsim);
cov[1][1] = gsl_stats_variance(mags, 1, nsim);
cov[0][1] = gsl_stats_covariance_m(dists,1,mags,1,nsim,mean[0],mean[1]);
cov[1][0] = cov[0][1];
std::cout << "Distance: " << mean[0] << "; Magnitude: " << mean[1] << std::endl;
std::cout << "Covariance matrix:" << std::endl;
for(i=0; i<2; i++){
for(j=0;j<2;j++){
printf("%d, %d, %g\n",i,j,cov[i][j]);
}
}
}
}
void function_initial_estimate( const double* pdX, const double* pdY, int iLength, double* pdParameterEstimates ) {
double dMin;
double dMax;
gsl_stats_minmax( &dMin, &dMax, pdX, 1, iLength );
pdParameterEstimates[0] = gsl_stats_mean( pdX, 1, iLength );
pdParameterEstimates[1] = ( dMax - dMin ) / 2.0;
pdParameterEstimates[2] = gsl_stats_max( pdY, 1, iLength );
}
void shapeAlign::scaleMatrixZscore(gsl_matrix *M){
for (size_t i = 0; i < M->size1; i++){
gsl_vector_view row = gsl_matrix_row(M,i);
double mu = gsl_stats_mean(row.vector.data, row.vector.stride, row.vector.size);
double sigma = gsl_stats_sd_m(row.vector.data, row.vector.stride, row.vector.size, mu);
gsl_vector_add_constant(&row.vector,-mu);
gsl_vector_scale(&row.vector,1.0/sigma);
}
return;
}
/********************************************************************************
* clusterupdatebatch: Runs clusterupdate multiple times and gets physics as well
* as error estimates.
*******************************************************************************/
int
clusterupdatebatch(lattice_site * lattice, settings conf, double beta, datapoint * data )
{
int i,j;
double * e_block, * m_block, * e_block_avg, * m_block_avg, \
* e_block_error, * m_block_error, * c_block , * chi_block;
gsl_vector * mag_vector;
e_block = (double *) malloc(conf.block_size*sizeof(double));
m_block = (double *) malloc(conf.block_size*sizeof(double));
e_block_avg = (double *) malloc(conf.blocks*sizeof(double));
m_block_avg = (double *) malloc(conf.blocks*sizeof(double));
c_block = (double *) malloc(conf.blocks*sizeof(double));
chi_block = (double *) malloc(conf.blocks*sizeof(double));
e_block_error = (double *) malloc(conf.blocks*sizeof(double));
m_block_error = (double *) malloc(conf.blocks*sizeof(double));
mag_vector = gsl_vector_alloc(conf.spindims);
//Settle first
for(i = 0 ; i < conf.max_settle ; i++)
{
clusterupdate(lattice,conf,beta);
}
//Get averages and stdev for messurements
for(i = 0 ; i < conf.blocks ; i++)
{
for(j = 0 ; j < conf.block_size ; j++)
{
clusterupdate(lattice,conf,beta);
e_block[j] = total_energy(lattice,conf);
m_block[j] = magnetization(lattice,conf,mag_vector);
}
e_block_avg[i] = gsl_stats_mean(e_block,1,conf.block_size);
e_block_error[i] = gsl_stats_sd(e_block,1,conf.block_size);
m_block_avg[i] = gsl_stats_mean(m_block,1,conf.block_size);
m_block_error[i] = gsl_stats_sd(m_block,1,conf.block_size);
c_block[i] = gsl_pow_2(beta)*gsl_pow_2(e_block_error[i]);
chi_block[i] = beta*gsl_pow_2(m_block_error[i]);
}
(*data).beta = beta;
(*data).erg = gsl_stats_mean(e_block_avg,1,conf.blocks);
(*data).erg_error = gsl_stats_sd(e_block_avg,1,conf.blocks);
(*data).mag = gsl_stats_mean(m_block_avg,1,conf.blocks);
(*data).mag_error = gsl_stats_sd(m_block_avg,1,conf.blocks);
(*data).c = gsl_stats_mean(c_block,1,conf.blocks);
(*data).c_error = gsl_stats_sd(c_block,1,conf.blocks);
(*data).chi = gsl_stats_mean(chi_block,1,conf.blocks);
(*data).chi_error = gsl_stats_sd(chi_block,1,conf.blocks);
free(e_block);
free(m_block);
free(e_block_avg);
free(m_block_avg);
free(e_block_error);
free(m_block_error);
gsl_vector_free(mag_vector);
return(0);
}
void DistToPA(gsl_histogram * h, double mag, double sigma)
{
double m1 = 0.5*mag + 0.5;
double s1 = 0.25*sigma;
double n1 = m1*(1. - m1)/s1 - 1.;
double a1 = m1*n1;
double b1 = (1. - m1)*n1;
for (int pa = 1; pa < 8; pa++) {
char line[80];
double x;
double data[6000];
int size = 0;
int nbins = h->n;
gsl_histogram * hpa = gsl_histogram_alloc( nbins);
gsl_histogram_set_ranges_uniform( hpa, 0.0, 1.0);
char fname[50];
sprintf(fname,"/home/jonatas/mzanlz/mZ_pa%d.dat",pa);
FILE * fp;
fp = fopen(fname,"rt");
while(fgets(line,80,fp) != NULL){
x = atof(line);
size++;
data[size] = x;
gsl_histogram_increment( hpa, x);
}
double m2 = 0.5*gsl_stats_mean( data, 1, size ) + 0.5;
double s2 = 0.25*gsl_stats_variance( data, 1, size );
double n2 = m2*(1. - m2)/s2 - 1.;
double a2 = m2*n2;
double b2 = (1. - m2)*n2;
NormalizaGSLHistograma( hpa );
//char hname[100];
//sprintf(hname,"pa%d",pa);
//PrintGSLHistogram( hpa, hname );
gsl_histogram_sub( hpa, h);
double D = 0;
for (size_t i = 0; i < nbins; i++) {
D += gsl_histogram_get( hpa , i )*gsl_histogram_get( hpa , i );
}
// printf("%g %g ", sqrt(D), KLbetas(a1,b1,a2,b2));
fclose(fp);
gsl_histogram_free( hpa );
}
}
void couple::process () {
int nrand = 100;
double *dobs =new double[nind/2];
double *drandom = new double [nrand];
for (int i = 0 ; i < g->nind; i+=2) {
dobs[i/2] = distance (i,i+1);
}
for (int i = 0 ; i < nrand; i++){
int j = gsl_rng_uniform_int (rng_r, nind);
int k = j;
do {
k = gsl_rng_uniform_int (rng_r, nind);
} while (k==j);
drandom [i] = distance (j,k);
}
double mobs = gsl_stats_mean (dobs, 1 ,nind/2); double sobs = sqrt (gsl_stats_variance (dobs,1,nind/2));
double mrandom = gsl_stats_mean (drandom, 1 ,nrand); double srandom = sqrt (gsl_stats_variance (drandom,1,nrand));
cout << mobs << "\t" << sobs << endl;
cout << mrandom << "\t" << srandom << endl;
}
void matrix_mean(gsl_vector *mean, gsl_matrix *input){
// Function to extract the column mean of a gsl matrix
size_t col;
size_t NCOL = input->size2;
gsl_vector_view a_col;
#pragma omp parallel for private( a_col)
for (col = 0; col < NCOL; col++) {
a_col = gsl_matrix_column(input, col);
gsl_vector_set(mean, col, gsl_stats_mean(a_col.vector.data,
a_col.vector.stride, a_col.vector.size));
}
}
void ccl_mat_mean(const double *mat,const int i, const int j,const int axis,double*val){
gsl_matrix * mat_ = gsl_matrix_alloc(i,j);
memcpy(mat_->data,mat,i*j*sizeof(double));
int k;
if (axis == 0){// x axis
gsl_vector * vec = gsl_vector_alloc(j);
for (k=0;k<i;k++){
gsl_matrix_get_row(vec,mat_,k);
val[k] = gsl_stats_mean(vec->data,1,j);
}
gsl_vector_free(vec);
}
if (axis == 1){// y axis
gsl_vector * vec = gsl_vector_alloc(i);
for (k=0;k<i;k++){
gsl_matrix_get_col(vec,mat_,k);
val[k] = gsl_stats_mean(vec->data,1,i);
}
gsl_vector_free(vec);
}
gsl_matrix_free(mat_);
}
void
test_basic(const size_t n, const double data[], const double tol)
{
gsl_rstat_workspace *rstat_workspace_p = gsl_rstat_alloc();
const double expected_mean = gsl_stats_mean(data, 1, n);
const double expected_var = gsl_stats_variance(data, 1, n);
const double expected_sd = gsl_stats_sd(data, 1, n);
const double expected_sd_mean = expected_sd / sqrt((double) n);
const double expected_skew = gsl_stats_skew(data, 1, n);
const double expected_kurtosis = gsl_stats_kurtosis(data, 1, n);
double expected_rms = 0.0;
double mean, var, sd, sd_mean, rms, skew, kurtosis;
size_t i, num;
int status;
/* compute expected rms */
for (i = 0; i < n; ++i)
expected_rms += data[i] * data[i];
expected_rms = sqrt(expected_rms / n);
/* add data to rstat workspace */
for (i = 0; i < n; ++i)
gsl_rstat_add(data[i], rstat_workspace_p);
mean = gsl_rstat_mean(rstat_workspace_p);
var = gsl_rstat_variance(rstat_workspace_p);
sd = gsl_rstat_sd(rstat_workspace_p);
sd_mean = gsl_rstat_sd_mean(rstat_workspace_p);
rms = gsl_rstat_rms(rstat_workspace_p);
skew = gsl_rstat_skew(rstat_workspace_p);
kurtosis = gsl_rstat_kurtosis(rstat_workspace_p);
num = gsl_rstat_n(rstat_workspace_p);
gsl_test_int(num, n, "n n=%zu" , n);
gsl_test_rel(mean, expected_mean, tol, "mean n=%zu", n);
gsl_test_rel(var, expected_var, tol, "variance n=%zu", n);
gsl_test_rel(sd, expected_sd, tol, "stddev n=%zu", n);
gsl_test_rel(sd_mean, expected_sd_mean, tol, "stddev_mean n=%zu", n);
gsl_test_rel(rms, expected_rms, tol, "rms n=%zu", n);
gsl_test_rel(skew, expected_skew, tol, "skew n=%zu", n);
gsl_test_rel(kurtosis, expected_kurtosis, tol, "kurtosis n=%zu", n);
status = gsl_rstat_reset(rstat_workspace_p);
gsl_test_int(status, GSL_SUCCESS, "rstat returned success");
num = gsl_rstat_n(rstat_workspace_p);
gsl_test_int(num, 0, "n n=%zu" , n);
gsl_rstat_free(rstat_workspace_p);
}
CAMLprim value ml_gsl_stats_mean(value ow, value data)
{
size_t len = Double_array_length(data);
double result;
if(ow == Val_none)
result = gsl_stats_mean(Double_array_val(data), 1, len);
else {
value w = Unoption(ow);
check_array_size(data, w);
result = gsl_stats_wmean(Double_array_val(w), 1,
Double_array_val(data), 1, len);
}
return copy_double(result);
}
/* Statistics code
* Make only one call to reading for a particular time range and compute all our stats
*/
void _compute_statistics(cdb_range_t *range, uint64_t *num_recs, cdb_record_t *records) {
uint64_t i = 0;
uint64_t valid = 0;
double sum = 0.0;
double *values = calloc(*num_recs, sizeof(double));
for (i = 0; i < *num_recs; i++) {
if (!isnan(records[i].value)) {
sum += values[valid] = records[i].value;
valid++;
}
}
range->num_recs = valid;
range->mean = gsl_stats_mean(values, 1, valid);
range->max = gsl_stats_max(values, 1, valid);
range->min = gsl_stats_min(values, 1, valid);
range->sum = sum;
range->stddev = gsl_stats_sd(values, 1, valid);
range->absdev = gsl_stats_absdev(values, 1, valid);
/* The rest need sorted data */
gsl_sort(values, 1, valid);
range->median = gsl_stats_median_from_sorted_data(values, 1, valid);
range->pct95th = gsl_stats_quantile_from_sorted_data(values, 1, valid, 0.95);
range->pct75th = gsl_stats_quantile_from_sorted_data(values, 1, valid, 0.75);
range->pct50th = gsl_stats_quantile_from_sorted_data(values, 1, valid, 0.50);
range->pct25th = gsl_stats_quantile_from_sorted_data(values, 1, valid, 0.25);
/* MAD must come last because it alters the values array
* http://en.wikipedia.org/wiki/Median_absolute_deviation */
for (i = 0; i < valid; i++) {
values[i] = fabs(values[i] - range->median);
if (values[i] < 0.0) {
values[i] *= -1.0;
}
}
/* Final sort is required MAD */
gsl_sort(values, 1, valid);
range->mad = gsl_stats_median_from_sorted_data(values, 1, valid);
free(values);
}
StatBox2D::BoxWhiskerData Layout2D::generateBoxWhiskerData(Column *colData,
int from, int to,
int key) {
size_t size = static_cast<size_t>((to - from) + 1);
double *data = new double[size];
for (int i = 0, j = from; j < to + 1; i++, j++) {
data[i] = colData->valueAt(i);
}
// sort the data
gsl_sort(data, 1, size - 1);
StatBox2D::BoxWhiskerData statBoxData;
statBoxData.key = key;
// basic stats
statBoxData.mean = gsl_stats_mean(data, 1, size);
statBoxData.median = gsl_stats_median_from_sorted_data(data, 1, size);
statBoxData.sd = gsl_stats_sd(data, 1, size);
statBoxData.se = statBoxData.sd / sqrt(static_cast<double>(size));
// data bounds
statBoxData.boxWhiskerDataBounds.sd_lower = statBoxData.mean - statBoxData.sd;
statBoxData.boxWhiskerDataBounds.sd_upper = statBoxData.mean + statBoxData.sd;
statBoxData.boxWhiskerDataBounds.se_lower = statBoxData.mean - statBoxData.se;
statBoxData.boxWhiskerDataBounds.se_upper = statBoxData.mean + statBoxData.se;
statBoxData.boxWhiskerDataBounds.perc_1 =
gsl_stats_quantile_from_sorted_data(data, 1, size, 0.01);
statBoxData.boxWhiskerDataBounds.perc_5 =
gsl_stats_quantile_from_sorted_data(data, 1, size, 0.05);
statBoxData.boxWhiskerDataBounds.perc_10 =
gsl_stats_quantile_from_sorted_data(data, 1, size, 0.10);
statBoxData.boxWhiskerDataBounds.perc_25 =
gsl_stats_quantile_from_sorted_data(data, 1, size, 0.25);
statBoxData.boxWhiskerDataBounds.perc_75 =
gsl_stats_quantile_from_sorted_data(data, 1, size, 0.75);
statBoxData.boxWhiskerDataBounds.perc_90 =
gsl_stats_quantile_from_sorted_data(data, 1, size, 0.90);
statBoxData.boxWhiskerDataBounds.perc_95 =
gsl_stats_quantile_from_sorted_data(data, 1, size, 0.95);
statBoxData.boxWhiskerDataBounds.perc_99 =
gsl_stats_quantile_from_sorted_data(data, 1, size, 0.99);
statBoxData.boxWhiskerDataBounds.max = data[size - 1];
statBoxData.boxWhiskerDataBounds.min = data[0];
// delete the double data pointer
delete[] data;
return statBoxData;
}
void single_sort(const char *dist_name, function_ptr dist_func, const char *sort_name, function_ptr sort_func, int runs, int side)
{
double cmean, csdev, tmean, tsdev;
struct timespec t0, t1;
int progress = (runs <= 10) ? 1 : runs / 10;
std::cout << std::setw(4) << side << " ";
std::cout << dist_name << " " << sort_name << " " << std::flush;
for (int i = 1; i <= runs; i++) {
random_seed(i);
tcomp = tread = twrit = 0;
(*dist_func)();
if ((i % progress) == 0) {
std::cout << '.' << std::flush;
}
clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &t0);
(*sort_func)();
clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &t1);
comp[i - 1] = tcomp;
msec[i - 1] = (t1.tv_sec - t0.tv_sec) * 1000.0 + (t1.tv_nsec - t0.tv_nsec) * 0.000001;
if (! is_spatially_sorted()) {
std::cout << "result is not sorted!\n";
}
}
cmean = gsl_stats_mean(comp, 1, runs);
csdev = gsl_stats_sd(comp, 1, runs);
tmean = gsl_stats_mean(msec, 1, runs);
tsdev = gsl_stats_sd(msec, 1, runs);
std::cout << std::setprecision(0) << std::fixed;
std::cout << std::setw(12) << cmean << ' ' << std::setw(12) << csdev << ' ';
std::cout << std::setprecision(2) << std::fixed;
std::cout << std::setw(12) << tmean << ' ' << std::setw(12) << tsdev << '\n';
}
void update_chi2_t(chi2_t *chichi, hist_t *obsHist, double *dataMat)
{
//-- data should be n*d matrix
int N = chichi->N;
int d = chichi->d;
gsl_vector *X_model = chichi->X_model;
gsl_matrix *cov = chichi->cov;
gsl_matrix *cov2 = chichi->cov2;
double value;
//-- Update mean
int i;
for (i=0; i<d; i++) {
value = gsl_stats_mean(dataMat+i*N, 1, N);
gsl_vector_set(X_model, i, value);
gsl_vector_set(chichi->X_obs, i, (double)(obsHist->n[i]));
}
//-- Update covariance
int j;
for (j=0; j<d; j++) {
for (i=0; i<d; i++) {
if (i < j) {
value = gsl_matrix_get(cov, j, i);
gsl_matrix_set(cov, i, j, value);
continue;
}
value = gsl_stats_covariance_m(dataMat+i*N, 1, dataMat+j*N, 1, N, gsl_vector_get(X_model, i), gsl_vector_get(X_model, j));
gsl_matrix_set(cov, i, j, value);
}
}
//-- Make a copy, because the invertion will destroy cov
gsl_matrix_memcpy(cov2, cov); //-- Copy cov to cov2
//-- Make LU decomposition and invert
int s;
gsl_linalg_LU_decomp(cov2, chichi->perm, &s);
gsl_linalg_LU_invert(cov2, chichi->perm, chichi->invCov);
//-- Debias the C_{ij}^{-1}
//-- C^-1_nonbias = (N - d - 2) / (N - 1) * C^-1_bias
double factor = (N - d - 2) / (double)(N - 1);
gsl_matrix_scale(chichi->invCov, factor); //-- invCov *= factor
return;
}
int pxy_calc(hgt_stat_mean ***means, hgt_stat_variance ***vars, double *pxy, double *d1, double *d2,hgt_pop **ps, hgt_params *params, int rank, int numprocs, hgt_cov_sample_func sample_func, gsl_rng *r) {
char * func_name = "pxy_calc";
unsigned count, dest, tag, error_code;
count = params->maxl * 4;
dest = 0;
tag = 0;
unsigned i;
for (i = 0; i < params->replicates; i++) {
unsigned n;
for (n = 0; n < params->sample_size; n++) {
hgt_pop_calc_dist(ps[i], d1, d2, 1, sample_func, r);
hgt_pop_calc_pxy(pxy+(n*params->maxl*4), params->maxl, d1, d2, params->seq_len, 0);
}
unsigned l;
for (l = 0; l < params->maxl; l++) {
unsigned k;
for (k = 0; k < 4; k++) {
pxy[l*4+k] = gsl_stats_mean(pxy+l*4+k, params->maxl*4, params->sample_size);
}
}
if (rank != 0) {
error_code = MPI_Send(pxy, count, MPI_DOUBLE, dest, tag, MPI_COMM_WORLD);
check_mpi_error_code(error_code, "MPI_Send", func_name);
} else {
int j;
for (j = 0; j < numprocs; j++) {
if (j != 0) {
error_code = MPI_Recv(pxy, count, MPI_DOUBLE, j, tag, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
check_mpi_error_code(error_code, "MPI_Recv", func_name);
}
unsigned l;
for (l = 0; l < params->maxl; l++) {
unsigned k;
for (k = 0; k < 4; k++) {
hgt_stat_mean_increment(means[l][k], pxy[l*4+k]);
hgt_stat_variance_increment(vars[l][k], pxy[l*4+k]);
}
}
}
}
}
return EXIT_SUCCESS;
}
static double
robust_robsigma(const gsl_vector *r, const double s,
const double tune, gsl_multifit_robust_workspace *w)
{
double sigma;
size_t i;
const size_t n = w->n;
const size_t p = w->p;
const double st = s * tune;
double a, b, lambda;
/* compute u = r / sqrt(1 - h) / st */
gsl_vector_memcpy(w->workn, r);
gsl_vector_mul(w->workn, w->resfac);
gsl_vector_scale(w->workn, 1.0 / st);
/* compute w(u) and psi'(u) */
w->type->wfun(w->workn, w->psi);
w->type->psi_deriv(w->workn, w->dpsi);
/* compute psi(u) = u*w(u) */
gsl_vector_mul(w->psi, w->workn);
/* Street et al, Eq (3) */
a = gsl_stats_mean(w->dpsi->data, w->dpsi->stride, n);
/* Street et al, Eq (5) */
b = 0.0;
for (i = 0; i < n; ++i)
{
double psi_i = gsl_vector_get(w->psi, i);
double resfac = gsl_vector_get(w->resfac, i);
double fac = 1.0 / (resfac*resfac); /* 1 - h */
b += fac * psi_i * psi_i;
}
b /= (double) (n - p);
/* Street et al, Eq (5) */
lambda = 1.0 + ((double)p)/((double)n) * (1.0 - a) / a;
sigma = lambda * sqrt(b) * st / a;
return sigma;
} /* robust_robsigma() */
/**
* Get a summary statistic for the parameters; for instance,
* the median value calculated over all the elements of the list.
* @param kl List
* @param what Can be one of: STAT_MEAN, STAT_MEDIAN, STAT_STDDEV, STAT_MAD.
* @return A vector whose entries are the summary statistic for the
* corresponding orbital parameter.
*/
gsl_vector* KL_getParsStats(const ok_list* kl, const int what) {
gsl_vector* v = gsl_vector_alloc(kl->size);
gsl_vector* ret = gsl_vector_calloc(PARAMS_SIZE + 1);
for (int j = 0; j < PARAMS_SIZE + 1; j++) {
if (j == PARAMS_SIZE)
for (int n = 0; n < kl->size; n++) {
VSET(v, n, kl->kernels[n]->merit);
}
else
for (int n = 0; n < kl->size; n++) {
VSET(v, n, VGET(kl->kernels[n]->params, j));
}
switch (what) {
case STAT_MEAN:
VSET(ret, j, gsl_stats_mean(v->data, 1, v->size));
break;
case STAT_STDDEV:
VSET(ret, j, gsl_stats_sd(v->data, 1, v->size));
break;
case STAT_MEDIAN:
gsl_sort_vector(v);
VSET(ret, j, gsl_stats_median_from_sorted_data(v->data, 1, v->size));
break;
case STAT_MAD:
gsl_sort_vector(v);
double med = gsl_stats_median_from_sorted_data(v->data, 1, v->size);
VSET(ret, j, 1.4826 * ok_mad(v->data, v->size, med));
break;
default:
// percentiles
gsl_sort_vector(v);
VSET(v , j, gsl_stats_quantile_from_sorted_data(v->data, 1, v->size, (double)(what)/100.));
};
};
gsl_vector_free(v);
return ret;
}
int
main(void)
{
double data[5] = {17.2, 18.1, 16.5, 18.3, 12.6};
double mean, variance, largest, smallest;
mean = gsl_stats_mean(data, 1, 5);
variance = gsl_stats_variance(data, 1, 5);
largest = gsl_stats_max(data, 1, 5);
smallest = gsl_stats_min(data, 1, 5);
printf ("The dataset is %g, %g, %g, %g, %g\n",
data[0], data[1], data[2], data[3], data[4]);
printf ("The sample mean is %g\n", mean);
printf ("The estimated variance is %g\n", variance);
printf ("The largest value is %g\n", largest);
printf ("The smallest value is %g\n", smallest);
return 0;
}
int ZeroOutliersOne (double *buf, int len, double alpha) {
/*
Replace outliers in the input buffer with zeros using the Grubbs
test (at the significance level alpha). The buffer is
overwritten. Return the number of outliers found.
*/
int noutliers = 0, idx, n, outlier = 1;
double meanval, maxval, sdval, tn, critval, v, *tmp;
tmp = (double *) malloc (len*sizeof (double));
while (outlier) {
n = 0;
for (idx=0; idx<len; idx++)
if ((v=*(buf+idx)))
*(tmp+n++) = v;
if (n>2) {
meanval = gsl_stats_mean (tmp, 1, n);
maxval = meanval;
for (idx=0; idx<n; idx++) {
v = *(tmp+idx);
if (fabs(v-meanval) > fabs(maxval-meanval))
maxval = v;
}
sdval = gsl_stats_sd_m (tmp, 1, n, meanval);
tn = fabs ((maxval-meanval)/sdval);
critval = zcritical (alpha, n);
if (tn>critval) {
for (idx=0; idx<len; idx++)
if (*(buf+idx) == maxval) {
*(buf+idx) = 0;
noutliers++;
}
} else
outlier = 0;
} else
outlier = 0;
}
free (tmp);
return noutliers;
}
void print_measurement_results_prediction(const job_t* job_p, const reprompib_common_options_t* opts_p,
double* maxRuntimes_sec, const nrep_pred_params_t* pred_params_p, const pred_conditions_t* conds_p) {
int j;
int my_rank, np;
double mean_runtime_sec, median_runtime_sec;
FILE* f;
MPI_Comm_rank(MPI_COMM_WORLD, &my_rank);
MPI_Comm_size(MPI_COMM_WORLD, &np);
f = stdout;
if (my_rank == OUTPUT_ROOT_PROC) {
if (opts_p->output_file != NULL) {
f = fopen(opts_p->output_file, "a");
}
if (job_p->n_rep == 0) { // no correct results
// runtime_sec = 0;
mean_runtime_sec = 0;
median_runtime_sec = 0;
} else { // print the last measured runtime
//runtime_sec = maxRuntimes_sec[job_p->n_rep - 1];
mean_runtime_sec = gsl_stats_mean(maxRuntimes_sec, 1, job_p->n_rep);
qsort(maxRuntimes_sec, job_p->n_rep, sizeof(double), cmpfunc);
median_runtime_sec = gsl_stats_median_from_sorted_data(maxRuntimes_sec, 1, job_p->n_rep);
}
for (j = 0; j < conds_p->n_methods; j++) {
fprintf(f, "%s %ld %ld %.10f %.10f %s %.10f\n", get_call_from_index(job_p->call_index), job_p->n_rep, job_p->count,
mean_runtime_sec, median_runtime_sec, get_prediction_methods_list()[pred_params_p->info[j].method],
conds_p->conditions[j]);
}
if (opts_p->output_file != NULL) {
fclose(f);
}
}
}
void streamRtsFile(xmlTextReaderPtr reader)
{
if (reader == NULL) {
fprintf(stderr, "Unable to access the file.\n");
exit(EXIT_FAILURE);
}
int ret;
if (verbose) {
printf("=== Analyzing ===\n");
}
ret = xmlTextReaderRead(reader);
while (ret == 1 && cont > 0) {
processNode(reader);
ret = xmlTextReaderRead(reader);
}
if (ret != 0 && cont > 0) {
printf("Failed to parse.\n");
return;
}
printf("=== FU result ===\n");
double mean = gsl_stats_mean(fuArray, 1, limit);
printf("Mean:\t\t%.5f (%.3f)\n", mean, mean * 100.0);
double variance = gsl_stats_variance_m(fuArray, 1, limit, mean);
printf("Variance:\t%.5f\n", variance);
double stddev = gsl_stats_sd_m(fuArray, 1, limit, mean);
printf("Std. Dev:\t%.5f\n", stddev);
double max = gsl_stats_max(fuArray, 1, limit);
printf("Max:\t\t%.5f (%.3f)\n", max, max * 100.0);
double min = gsl_stats_min(fuArray, 1, limit);
printf("Min:\t\t%.5f (%.3f)\n", min, min * 100.0);
printf("Total RTS:\t%d\n", rtsNr);
printf("Valid RTS:\t%d\n", validFuCnt);
printf("Invalid RTS:\t%d\n", invalidFuCnt);
}
void
test_basic(const size_t n, const double data[], const double tol)
{
gsl_rstat_workspace *rstat_workspace_p = gsl_rstat_alloc();
const double expected_mean = gsl_stats_mean(data, 1, n);
const double expected_var = gsl_stats_variance(data, 1, n);
const double expected_sd = gsl_stats_sd(data, 1, n);
const double expected_skew = gsl_stats_skew(data, 1, n);
const double expected_kurtosis = gsl_stats_kurtosis(data, 1, n);
double expected_rms = 0.0;
double mean, var, sd, rms, skew, kurtosis;
size_t i;
/* compute expected rms */
for (i = 0; i < n; ++i)
expected_rms += data[i] * data[i];
expected_rms = sqrt(expected_rms / n);
/* add data to rstat workspace */
for (i = 0; i < n; ++i)
gsl_rstat_add(data[i], rstat_workspace_p);
mean = gsl_rstat_mean(rstat_workspace_p);
var = gsl_rstat_variance(rstat_workspace_p);
sd = gsl_rstat_sd(rstat_workspace_p);
rms = gsl_rstat_rms(rstat_workspace_p);
skew = gsl_rstat_skew(rstat_workspace_p);
kurtosis = gsl_rstat_kurtosis(rstat_workspace_p);
gsl_test_rel(mean, expected_mean, tol, "mean n=%zu", n);
gsl_test_rel(var, expected_var, tol, "variance n=%zu", n);
gsl_test_rel(sd, expected_sd, tol, "stddev n=%zu", n);
gsl_test_rel(rms, expected_rms, tol, "rms n=%zu", n);
gsl_test_rel(skew, expected_skew, tol, "skew n=%zu", n);
gsl_test_rel(kurtosis, expected_kurtosis, tol, "kurtosis n=%zu", n);
gsl_rstat_free(rstat_workspace_p);
}
int main(void)
{
double v[NEXP]; /* calculated volumes */
gsl_rng *r;
double mean, sd = 1;
int n, j;
unsigned long seed = 1UL;
long p = POINTS;
/* allocate random number generator */
r = gsl_rng_alloc(gsl_rng_taus2);
/* seed the random number generator */
gsl_rng_set(r, seed);
for (n = 1; n <= 9; n++)
{
while (sd >= .03)
{
for (j = 0; j < NEXP; j++)
{
/* calculate volume using Monte Carlo algorithm */
v[j] = mc_se(p, n, r);
}
mean = gsl_stats_mean(v, 1, NEXP);
sd = gsl_stats_sd_m(v, 1, NEXP, mean);
p *= 2;
}
printf("%d %10.8f %10.8f\n", n, mean, sd);
/* reset the standard deviation and the number of points at each iteration */
sd = 1;
p = POINTS;
}
gsl_rng_free(r);
return(0);
}
