void GeomData::calculate_CDL() {
int i,dom, row;
const int* indices = NULL;
// elements' coordinates
const double* coords[4] = {NULL,NULL,NULL,NULL};
// vector created over elements' edges
int num_edges = 3*(dim-1);
double* vectors[6];
for (i=0; i<num_edges; i++) {
vectors[i] = new double[dim];
}
// indices to automate edge vectors creation
int idx[6][2] = {{0,1},{1,2},{2,0},{0,3},{1,3},{2,3}};
// Auxiliary variable
int pos = dim+1;
int k = 0;
for (dom=0; dom<_ndom; dom++) {
int nelements = numDomElem[dom];
// loop over domains' elements
for (row=0; row<nelements; row++) {
getElement(dom,row,indices);
// get vertices coordinates
for (i=0; i<pos; i++) {
getCoordinates(indices[i+pos],coords[i]);
}
// create edge vectors
for (i=0; i<num_edges; i++) {
const double* vtx1 = coords[ idx[i][0] ];
const double* vtx2 = coords[ idx[i][1] ];
for (int j=0; j<dim; j++) {
(vectors[i])[j] = vtx1[j] - vtx2[j];
}
}
// vectors lengths summation
for (i=0; i<num_edges; i++) {
elem_CDL[k] = sqrt( inner_product(vectors[i],vectors[i],dim) );
}
// take average lenght
elem_CDL[k] /= pos;
k++;
}
}
for (i=0; i<num_edges; i++) {
delete[] vectors[i];
vectors[i] = 0;
}
}
int main(void)
{
double a[N]={1, 2, 3}, b[N]={1, 2, 3}, *p = a, *q = b;
printf("%f\n", inner_product(p, q, N));
return 0;
}
void innerproductTest() {
std::cout << "inner product ... ";
std::vector<std::complex<double>> x = {1,2,3,4};
std::vector<std::complex<double>> y = {1,std::complex<double>(0,1),1,std::complex<double>(0,1)};
std::complex<double> expected = std::complex<double>(4,-6);
bool pass = std::abs(expected - inner_product(x,y)) < 1e-8;
if(pass) std::cout << "PASS" << std::endl;
else std::cout << "FAIL" << std::endl;
}
/*solution saved in b */
void conjugate_gradient(sparse_matrix A, double *b, int size) {
double *x, *r, *p, *Ap, *aux, rnew, rold, alfa;
int i;
x = (double*) malloc(size*sizeof(double));
r = (double*) malloc(size*sizeof(double));
p = (double*) malloc(size*sizeof(double));
Ap = (double*) malloc(size*sizeof(double));
aux = (double*) malloc(size*sizeof(double));
for (i = 0; i < size; i ++) {
x[i] = 0;
r[i] = b[i];
p[i] = b[i];
}
rold = inner_product(r, r, size);
/* result of operations from all void functions used here are stored in the last argument */
while (1) {
matrix_vector_product(A, p, Ap);
alfa = rold / inner_product(p, Ap, size); /*step length*/
vector_scalar_product(p, alfa, size, aux);
vector_sum(x, aux, size, x);
vector_scalar_product(Ap, alfa, size, aux);
vector_subtraction(r, aux, size, r);
rnew = inner_product(r, r, size);
if (sqrt(rnew) < E)
break;
vector_scalar_product(p, rnew / rold, size, p);
vector_sum(p, r, size, p);
rold = rnew;
}
for (i = 0; i < size; i ++) {
b[i] = x[i];
}
free(x);
free(r);
free(p);
free(Ap);
free(aux);
}
double sPredModule::solveCoef(double wrss) {
Rcpp::NumericMatrix
cc = d_fac.solve(CHOLMOD_L, d_fac.solve(CHOLMOD_P, d_Vtr));
double ans = sqrt(inner_product(cc.begin(), cc.end(),
cc.begin(), double())/wrss);
Rcpp::NumericMatrix
bb = d_fac.solve(CHOLMOD_Pt, d_fac.solve(CHOLMOD_Lt, cc));
copy(bb.begin(), bb.end(), d_coef.begin());
return ans;
}
/**
* Solve (V'V)coef = Vtr for coef.
*
* @param wrss weighted residual sum of squares (defaults to 1.)
* @return convergence criterion
*/
double dPredModule::solveCoef(double wrss) {
if (d_coef.size() == 0) return 0.;
copy(d_Vtr.begin(), d_Vtr.end(), d_coef.begin());
d_fac.update(d_V);
d_fac.dtrtrs('T', d_coef.begin()); // solve R'c = Vtr
double ans = sqrt(inner_product(d_coef.begin(), d_coef.end(),
d_coef.begin(), double())/wrss);
d_fac.dtrtrs('N', d_coef.begin()); // solve R beta = c;
return ans;
}
double blok::deviation(){
double total = std::accumulate(content.begin(), content.end(), 0);
double mean = total / content.size();
vector<unsigned int> zero_mean( content );
transform( zero_mean.begin(), zero_mean.end(), zero_mean.begin(),bind2nd( minus<double>(), mean ) );
double deviation = inner_product( zero_mean.begin(),zero_mean.end(), zero_mean.begin(), 0.0 );
deviation = sqrt( deviation / ( content.size() - 1 ) );
return deviation;
}
int main(int argc, char *argv[])
{
double a[5] = {1,2,3,4,5};
double b[5] = {1,2,3,4,5};
printf("The sum of the product of the two arrays is %f\n", inner_product(a,b,5));
getchar();
getchar();
return 0;
}
const vector DOK::conjugate_gradient(const vector& b, vector& x, size_t max_iter, double eps) const {
// vector x;
vector r = b - *this * x;
vector p = r;
double rsold = inner_product(r, r);
size_t i;
for(i = 0; i < max_iter && rsold > eps; ++i) {
vector Ap = *this * p;
double alpha = rsold / inner_product(p, Ap);
x += alpha * p;
r -= alpha * Ap;
double rsnew = inner_product(r, r);
p *= rsnew / rsold;
p += r;
rsold = rsnew;
}
if (i == max_iter)
printf("exce\n");
return x;
}
REAL CMetaModel::Test( const REAL* prInputs, const REAL* prOutputs, int count )
{
vector<REAL> vcOutputs( GetOutputs() );
//calculate the error function under the current weight
REAL rError = 0;
for( int i=0; i<count; i++ ){
Predict( prInputs + i*GetInputs(), &vcOutputs[0] );
rError += inner_product( vcOutputs.begin(), vcOutputs.end(), prOutputs, 0.0,
plus<REAL>(), minus_pow<REAL>(2.0) ) / GetOutputs();
}
return rError/count;
}
matrix1d calc_mean_var(matrix1d points) {
double sum = 0, mu = 0, sigma_sq = 0, sdev = 0, dev = 0;
sum = accumulate(points.begin(), points.end(), 0.0);
mu = sum / points.size();
sdev = inner_product(points.begin(), points.end(), points.begin(), 0.0);
sigma_sq = sdev/points.size() - mu*mu;
matrix1d m_and_v;
m_and_v.push_back(mu);
m_and_v.push_back(sigma_sq);
return m_and_v;
}
void CRegionalMetaModel::ComputeDistance( vector< vector<REAL> >& vcInputs, vector<REAL>& vcPtCen, dist_pair_vector& vcDists )
{
int nsize = vcInputs.size();
vcDists.clear();
vcDists.reserve( nsize );
for( int i=0; i<nsize; i++ ){
REAL dist = inner_product( vcInputs[i].begin(), vcInputs[i].end(), vcPtCen.begin(), 0.0,
plus<REAL>(), diff_sqr<REAL>() );
dist = sqrt(dist);
vcDists.push_back( dist_pair(i, dist) );
}
sort( vcDists.begin(), vcDists.end(), op_dist_less() );
}
static float _diff_disc_formfactor(Vertex *p, Element *e_src, Vertex *p_disc, float area, Vertex *normal, long process_id)
{
Vertex vec_sd ;
float dist_sq ;
float fnorm ;
float cos_s, cos_d, angle_factor ;
vec_sd.x = p_disc->x - p->x ;
vec_sd.y = p_disc->y - p->y ;
vec_sd.z = p_disc->z - p->z ;
dist_sq = vec_sd.x*vec_sd.x + vec_sd.y*vec_sd.y + vec_sd.z*vec_sd.z ;
fnorm = area / ((float)M_PI * dist_sq + area) ;
/* (2) Now, consider angle to the other patch from the normal. */
normalize_vector( &vec_sd, &vec_sd ) ;
cos_s = inner_product( &vec_sd, &e_src->patch->plane_equ.n ) ;
cos_d = -inner_product( &vec_sd, normal ) ;
angle_factor = cos_s * cos_d ;
/* Return the form factor */
return( fnorm * angle_factor ) ;
}
Field<dim> lbfgs(
const Functional& F,
const Gradient& dF,
const Field<dim>& phi_start,
const unsigned int m,
const double eps
)
{
double cost_old = std::numeric_limits<double>::infinity();
double cost = F(phi_start);
Field<dim> phi0(phi_start);
DualField<dim> df0 = dF(phi_start);
const auto& discretization = phi0.get_discretization();
// Create vectors for storing the last few differences of the guesses,
// differences of the gradients, and the inverses of their inner products
std::vector<Field<dim>> s(m, Field<dim>(discretization));
std::vector<DualField<dim>> y(m, DualField<dim>(discretization));
std::vector<double> rho(m);
// As an initial guess, we will assume that the Hessian inverse is a
// multiple of the inverse of the mass matrix
double gamma = 1.0;
for (unsigned int k = 0; std::abs(cost_old - cost)/cost > eps; ++k) {
const Field<dim> p = -lbfgs_two_loop(df0, s, y, rho, gamma);
const Field<dim> phi1 = line_search(F, phi0, df0, p, eps);
const DualField<dim> df1 = dF(phi1);
s[0] = phi1 - phi0;
y[0] = df1 - df0;
const double cos_angle = inner_product(y[0], s[0]);
const double norm_y = norm(y[0]);
rho[0] = 1.0 / cos_angle;
gamma = cos_angle / (norm_y * norm_y);
std::rotate(s.begin(), s.begin() + 1, s.end());
std::rotate(y.begin(), y.begin() + 1, y.end());
std::rotate(rho.begin(), rho.begin() + 1, rho.end());
phi0 = phi1;
df0 = df1;
cost_old = cost;
cost = F(phi0);
}
return phi0;
}
inline Real mixed_product (const T1& v1,
const T2& v2,
const T3& v3,
T4& temp)
{
// sanity checks
cf3_assert(v1.size() == 3);
cf3_assert(v2.size() == 3);
cf3_assert(v3.size() == 3);
cf3_assert(temp.size() == 3);
cross_product(v1, v2, temp);
return inner_product(v3, temp);
}
// calculate linear interpolation of relative frequency estimates based on joint count and marginal counts
//unused for now; enable in config?
double LinearInterpolationFromCounts(vector<float> &joint_counts, vector<float> &marginals, vector<float> &multimodelweights)
{
vector<float> p(marginals.size());
for (size_t i=0; i < marginals.size(); i++) {
if (marginals[i] != 0) {
p[i] = joint_counts[i]/marginals[i];
}
}
double p_weighted = inner_product(p.begin(), p.end(), multimodelweights.begin(), 0.0);
return p_weighted;
}
void update_vector(double *b, double **Q, int rows, int columns, double *gamma) {
int i, k;
double *u, factor;
u = malloc((rows) * sizeof(double));
for (k = 0; k < columns; k ++) {
u[0] = 1;
for (i = k + 1; i < rows; i ++)
u[i - k] = Q[i][k];
factor = gamma[k]*inner_product(u, b, rows, k);
for (i = k; i < rows; i ++)
b[i] -= factor*u[i - k];
}
free(u);
}
//数值算法
void stl_numeric_algo(){
double array[] = {-2,2,4,4,5};
double array1[] = {2,3,4,5,6};
vector<double> a(array,array + sizeof(array)/sizeof(double));
vector<double> b(array1,array1 + sizeof(array1)/sizeof(double));
vector<double> result(10);
cout<<"accumulate func: "<<accumulate(a.begin(), a.end(), 0, minus<double>())<<endl;
cout<<"inner product func: "<<inner_product(a.begin(),a.end(),b.begin(),0.0,plus<double>(),multiplies<double>())<<endl;
adjacent_difference(a.begin(),a.end(),result.begin());
cout<<"adjacent_difference : "<<result<<endl;
partial_sum(result.begin(),result.end(),result.begin());
cout<<"partial_sum : "<<result<<endl;
ostream_iterator<double> myout(cout, " ");
copy(result.begin(), result.end(), myout);
}
//------------------------------------------------------------------------------------
void calcShadowMatrix(const float groundplane[],
const float lightpos[],
float shadowMatrix[])
{
float dot = inner_product(groundplane, groundplane+4, lightpos, 0.f);
for(int y = 0; y < 4;++y) {
for(int x = 0; x < 4; ++x) {
shadowMatrix[y*4+x] = - groundplane[y]*lightpos[x];
if (x == y) shadowMatrix[y*4+x] += dot;
}
}
}
virtual double
energy( const typename domain_space_type::element_type & de,
const typename dual_image_space_type::element_type & ie ) const
{
auto matrix = M_backend->newMatrix( _test=this->dualImageSpace(), _trial=this->domainSpace(), _pattern=M_pattern );
sumAllMatrices( matrix, true );
matrix->close();
vector_ptrtype _v1( M_backend->newVector( _test=de.functionSpace() ) );
*_v1 = de;_v1->close();
vector_ptrtype _v2( M_backend->newVector( _test=ie.functionSpace() ) );
*_v2 = ie;
vector_ptrtype _v3( M_backend->newVector( _test=ie.functionSpace() ) );
M_backend->prod( matrix, _v1, _v3 );
return inner_product( _v2, _v3 );
}
int main(int argc, char** argv)
{
_();
vec_as_string();
vec_fill_and_remove();
vec_not1();
vec_bind2nd();
verse_iterator_find_if();
reverse_iterator_find_if();
inner_product();
descending_sort();
deque_push_erase();
deque_invalidate();
deque_assign();
return EXIT_SUCCESS;
}
double correlation(const Cit &start1, const Cit &end1, const Cit &start2) {
const auto end2 = start2 + distance(start1, end1);
const auto mean1 = accumulate(start1, end1, 0.0) / distance(start1, end1);
const auto mean2 = accumulate(start2, end2, 0.0) / distance(start2, end2);
const auto numerator = inner_product(
start1, end1, start2, 0.0, plus<double>(),
[=](double a, double b) noexcept { return (a - mean1) * (b - mean2); });
const auto squaredError1 =
accumulate(start1, end1, 0.0, [mean1](double sum, double x) noexcept {
return sum + (x - mean1) * (x - mean1);
});
const auto squaredError2 =
accumulate(start2, end2, 0.0, [mean2](double sum, double x) noexcept {
return sum + (x - mean2) * (x - mean2);
});
const auto denominator = sqrt(squaredError1) * sqrt(squaredError2);
return numerator / denominator;
}
int main(void)
{
double a[N], b[N], n = 0, sum, *p;
printf("Enter %d elements for array A: ", N);
for (p = a; p < a + N; p++){
scanf("%lf", p);
n++;
}
printf("Enter %d elements for array B: ", N);
for (p = b; p < b + N; p++){
scanf("%lf", p);
}
sum = inner_product(a, b, n);
printf("%lf\n", sum);
return 0;
}
double entropy(const Matrix3_3 &matrix) {
array<double, 3> H;
for (int i = 0; i < 3; i++) {
const auto _row = row(matrix, i);
array<double, 3> multipliedByLog;
transform(begin(_row), end(_row), begin(multipliedByLog),
[](double x) { return x * log(x) / log(2); });
H[i] = accumulate(begin(multipliedByLog), end(multipliedByLog), 0.0);
}
array<double, 3> P;
for (int i = 0; i < 3; i++) {
const auto column = col(matrix, i);
P[i] = accumulate(begin(column), end(column), 0.0);
}
return -inner_product(begin(P), end(P), begin(H), 0.0);
}
void polarization(double * function_1, double left_endpoint, double right_endpoint, int num_points, double * integral){
int flag = 1;
int ii;
double beta = 1.0;
//defining the size of the interval(s). note this is num_points+1 because the interval can't start at zero.
double h = (right_endpoint - left_endpoint)/(num_points+1.0);
//allocating memory for new vector for X * some arbitrary vector.
double * psi_hat;
psi_hat = (double*)malloc(num_points*sizeof(double));
X_operator(beta, num_points, right_endpoint, left_endpoint, psi_hat, function_1, flag);
inner_product(&function_1[0], &psi_hat[0], left_endpoint, right_endpoint, num_points, integral);
free(psi_hat);
}
void Do_Statistics( std::vector < std::vector <double>> data) {
std::cout << "Doing Stats" << std::endl;
std::string rchange = "flat";
std::ofstream statisticsOutputFile;
statisticsOutputFile.open("Statistic_Output_" + rchange +".txt");
std::vector<int> binCount;
std::vector<double> typeAverage;
std::vector<int> weights;
double maxCellNumber = data.at(0).at(1);
for(int i =0; i < data.size(); i++){
weights.push_back(i);
}
//Type average is the dot product between the 0:number_muts
//Bin count is counting the non zero bins
for(int rows=0; rows < data.size(); rows++){
auto countt = std::count(data.at(rows).begin()+1, data.at(rows).end(), 0);
binCount.push_back(data.at(rows).size() - 1 - countt);
double dotProduct = inner_product(data.at(rows).begin()+1, data.at(rows).end(), weights.begin(), 0)/maxCellNumber;
typeAverage.push_back(dotProduct);
}
//wirting to a file the r$i is for ggplot
for(int j = 0; j < binCount.size(); j++){
statisticsOutputFile << data.at(j).at(0) << "\t" << typeAverage.at(j) << "\t" << binCount.at(j) << "\t" << rchange << std::endl;
}
statisticsOutputFile.close();
}
//the compute distance is undertaken an overhaull. I prefere to compute the probability instead of the
//euclidean distance.
void CRegionalMetaModel::ComputeDistance( vector< vector<REAL> >& mxInputs, vector<REAL>& vcPtCen, dist_pair_vector& vcDists )
{
int rows = mxInputs.size();
int cols = mxInputs.front().size();
/*compute the center of the hyperecllipsoid*/
vector<REAL> center(cols, 0.0);
for( vector< vector<REAL> >::iterator iter=mxInputs.begin(); iter!=mxInputs.end(); iter++ )
transform( center.begin(), center.end(), iter->begin(), center.begin(), plus<REAL>() );
transform( center.begin(), center.end(), center.begin(), bind2nd(divides<REAL>(), rows) );
/*compute the standard deviation of the columns*/
//compute sum( (x-u)*(x-u) ) for each column
vector<REAL> stds(cols, 0.0);
for( int i=0; i<rows; i++ )
for( int j=0; j<cols; j++ )stds[j] += sqr( center[j] - mxInputs[i][j] );
//using compose function operator to do "sqrt( sum/N-1 )"
transform( stds.begin(), stds.end(), stds.begin(),
compose_f_gx(
ptr_fun(sqrt), bind2nd(divides<double>(), rows-1)
)
);
/* compute the radius */
REAL alpha = 1.5;
vector<REAL> radius(cols);
transform( stds.begin(), stds.end(), radius.begin(), bind2nd(multiplies<REAL>(), alpha) );
for( i=0; i<cols; i++ )if( radius[i]<=0.0001 )radius[i]=0.0001;
/* compute the distance for each point */
vector<REAL> radius_sqr(cols);
transform( radius.begin(), radius.end(), radius.begin(), radius_sqr.begin(), multiplies<REAL>() );
for( i=0; i<rows; i++ ){
vector<REAL> temp(cols);
transform( mxInputs[i].begin(), mxInputs[i].end(), center.begin(), temp.begin(), diff_sqr<REAL>() );
REAL distance = inner_product( temp.begin(), temp.end(), radius_sqr.begin(), 0.0, plus<REAL>(), divides<REAL>() );
vcDists.push_back( dist_pair(i, distance) );
}
sort( vcDists.begin(), vcDists.end(), op_dist_less() );
}
static void set_init_cond_magnt(double *B)
{
for (int i = 0; i < N0; i++)
for (int k = 0; k < N2-1; k++)
for (int j = 0; j < N1-1; j++) {
const struct vector pos = {(double)i, (double)j+0.5, (double)k+0.5};
// othnormal component
const struct vector p_xyz = othnormal_position<c>(pos);
const struct vector b_xyz = init_cond_magnt(p_xyz);
// physical component vertical with the grid
// Covariant basic vector is a member of coordinate transformation
// matrix here.
const double b = inner_product(
covariant_basic_vector<c, c>(pos), b_xyz);
// Convert to physical length.
*magnt_flux<N1, N2>(i, j, k, B) =
b * vector_length(contravariant_basic_vector<c, c>(pos));
}
}
static void set_init_cond_elect(double *E)
{
for (int i = 0; i < N0-1; i++)
for (int k = 0; k < N2; k++)
for (int j = 0; j < N1; j++) {
const struct vector pos = {(double)i+0.5, (double)j, (double)k};
// othnormal component
const struct vector p_xyz = othnormal_position<c>(pos);
const struct vector e_xyz = init_cond_elect(p_xyz);
// unphysical components along to the grid.
// Contravariant basic vector is a member of coordinate transformation
// matrix here.
const double e = inner_product(
contravariant_basic_vector<c, c>(pos), e_xyz);
// Convert to physical length.
*elect_field<N1, N2>(i, j, k, E) =
e * vector_length(covariant_basic_vector<c, c>(pos));
}
}
// calculate a unit vector normal to each external face
void GeomData::calculate_extFaceVersor(pMesh theMesh) {
int row = 0;
double dij[3], norma;
int ndom = getNumDomains();
for (int i=0; i<ndom; i++) {
FIter fit = M_faceIter(theMesh);
while ( pFace face = FIter_next(fit) ) {
// only external boudanry faces
if (F_numRegions(face)==1) {
if (faceBelongToDomain(face,domainList[i])) {
getDij(face,domainList[i],dij);
norma = sqrt( inner_product(dij,dij,3) );
versor_ExtBdryElem[0].setValue(row,0,dij[0]/norma);
versor_ExtBdryElem[0].setValue(row,1,dij[1]/norma);
versor_ExtBdryElem[0].setValue(row,2,dij[2]/norma);
row++;
}
}
}
FIter_delete(fit);
}
}
