Skip to content

GitLab

Commit 6fc06906 authored by Uwe Schulzweida's avatar Uwe Schulzweida
Browse files

Remove FLDATATYPE and NFDATATYPE.

parent 00cce37d
Hide whitespace changes
Inline Side-by-side
src/grid_search.cc
View file @ 6fc06906
......@@ -20,10 +20,6 @@
#define PI2 (2.0*PI)
#define FLDATATYPE double
#define NFDATATYPE double
static GridsearchMethod gridsearch_method_nn(GridsearchMethod::kdtree);
......@@ -31,7 +27,7 @@ struct gsFull {
size_t n;
const double *plons;
const double *plats;
FLDATATYPE **pts;
double **pts;
};
......@@ -72,8 +68,8 @@ struct PointCloud
};
typedef nanoflann::KDTreeSingleIndexAdaptor<
nanoflann::L2_Simple_Adaptor<NFDATATYPE, PointCloud<NFDATATYPE> > ,
PointCloud<NFDATATYPE>,
nanoflann::L2_Simple_Adaptor<double, PointCloud<double> > ,
PointCloud<double>,
3 /* dim */
> nfTree_t;
......@@ -93,9 +89,8 @@ double cdo_default_search_radius(void)
return search_radius;
}
template <typename T>
static inline
void LLtoXYZ(double lon, double lat, T *restrict xyz)
void LLtoXYZ(double lon, double lat, double *restrict xyz)
{
double cos_lat = cos(lat);
xyz[0] = cos_lat * cos(lon);
......@@ -103,16 +98,14 @@ void LLtoXYZ(double lon, double lat, T *restrict xyz)
xyz[2] = sin(lat);
}
template <typename T>
static constexpr
T square(const T x)
double square(const double x)
{
return x*x;
}
template <typename T>
static constexpr
T distance(const T *restrict a, const T *restrict b)
double distance(const double *restrict a, const double *restrict b)
{
return square(a[0]-b[0]) + square(a[1]-b[1]) + square(a[2]-b[2]);
}
......@@ -201,7 +194,7 @@ void *gs_create_kdtree(size_t n, const double *restrict lons, const double *rest
for ( size_t i = 0; i < n; i++ )
{
kdata_t *restrict point = pointlist[i].point;
LLtoXYZ<kdata_t>(lons[i], lats[i], point);
LLtoXYZ(lons[i], lats[i], point);
for ( unsigned j = 0; j < 3; ++j )
{
min[j] = point[j] < min[j] ? point[j] : min[j];
......@@ -228,10 +221,10 @@ void *gs_create_kdtree(size_t n, const double *restrict lons, const double *rest
static
void *gs_create_nanoflann(size_t n, const double *restrict lons, const double *restrict lats, struct gridsearch *gs)
{
PointCloud<NFDATATYPE> *pointcloud = new PointCloud<NFDATATYPE>();
PointCloud<double> *pointcloud = new PointCloud<double>();
if ( cdoVerbose ) printf("nanoflann init 3D: n=%zu nthreads=%d\n", n, ompNumThreads);
NFDATATYPE min[3], max[3];
double min[3], max[3];
min[0] = min[1] = min[2] = 1e9;
max[0] = max[1] = max[2] = -1e9;
// Generating Point Cloud
......@@ -241,8 +234,8 @@ void *gs_create_nanoflann(size_t n, const double *restrict lons, const double *r
#endif
for ( size_t i = 0; i < n; i++ )
{
NFDATATYPE point[3];
LLtoXYZ<NFDATATYPE>(lons[i], lats[i], point);
double point[3];
LLtoXYZ(lons[i], lats[i], point);
pointcloud->pts[i].x = point[0];
pointcloud->pts[i].y = point[1];
pointcloud->pts[i].z = point[2];
......@@ -301,8 +294,8 @@ void *gs_create_full(size_t n, const double *restrict lons, const double *restri
{
struct gsFull *full = (struct gsFull *) Calloc(1, sizeof(struct gsFull));
FLDATATYPE **p = (FLDATATYPE **) Malloc(n*sizeof(FLDATATYPE *));
p[0] = (FLDATATYPE *) Malloc(3*n*sizeof(FLDATATYPE));
double **p = (double **) Malloc(n*sizeof(double *));
p[0] = (double *) Malloc(3*n*sizeof(double));
for ( size_t i = 1; i < n; i++ ) p[i] = p[0] + i*3;
#ifdef HAVE_OPENMP4
......@@ -310,7 +303,7 @@ void *gs_create_full(size_t n, const double *restrict lons, const double *restri
#endif
for ( size_t i = 0; i < n; i++ )
{
LLtoXYZ<FLDATATYPE>(lons[i], lats[i], p[i]);
LLtoXYZ(lons[i], lats[i], p[i]);
}
full->n = n;
......@@ -361,7 +354,7 @@ void gridsearch_delete(struct gridsearch *gs)
if ( gs->sinlon ) Free(gs->sinlon);
if ( gs->method_nn == GridsearchMethod::kdtree ) gs_destroy_kdtree(gs->search_container);
else if ( gs->method_nn == GridsearchMethod::nanoflann ) delete((PointCloud<NFDATATYPE> *)gs->pointcloud);
else if ( gs->method_nn == GridsearchMethod::nanoflann ) delete((PointCloud<double> *)gs->pointcloud);
else if ( gs->method_nn == GridsearchMethod::full ) gs_destroy_full(gs->search_container);
Free(gs);
......@@ -395,7 +388,7 @@ size_t gs_nearest_kdtree(void *search_container, double lon, double lat, double
kdata_t range = range0;
kdata_t query_pt[3];
LLtoXYZ<kdata_t>(lon, lat, query_pt);
LLtoXYZ(lon, lat, query_pt);
if ( !gs->extrapolate )
for ( unsigned j = 0; j < 3; ++j )
......@@ -457,11 +450,11 @@ size_t gs_nearest_nanoflann(void *search_container, double lon, double lat, doub
nfTree_t *nft = (nfTree_t *) search_container;
if ( nft == NULL ) return index;
NFDATATYPE range0 = gs_set_range(prange);
NFDATATYPE range = range0;
double range0 = gs_set_range(prange);
double range = range0;
NFDATATYPE query_pt[3];
LLtoXYZ<NFDATATYPE>(lon, lat, query_pt);
double query_pt[3];
LLtoXYZ(lon, lat, query_pt);
if ( !gs->extrapolate )
for ( unsigned j = 0; j < 3; ++j )
......@@ -469,8 +462,8 @@ size_t gs_nearest_nanoflann(void *search_container, double lon, double lat, doub
const size_t num_results = 1;
size_t ret_index;
NFDATATYPE out_dist_sqr;
nanoflann::KNNResultSet<NFDATATYPE> resultSet(range, num_results);
double out_dist_sqr;
nanoflann::KNNResultSet<double> resultSet(range, num_results);
resultSet.init(&ret_index, &out_dist_sqr);
nft->findNeighbors(resultSet, query_pt, nanoflann::SearchParams(10));
//printf("%zu %g\n", ret_index, out_dist_sqr);
......@@ -494,18 +487,18 @@ size_t gs_nearest_full(void *search_container, double lon, double lat, double *p
struct gsFull *full = (struct gsFull *) search_container;
if ( full == NULL ) return index;
FLDATATYPE range0 = gs_set_range(prange);
double range0 = gs_set_range(prange);
FLDATATYPE query_pt[3];
LLtoXYZ<FLDATATYPE>(lon, lat, query_pt);
double query_pt[3];
LLtoXYZ(lon, lat, query_pt);
size_t n = full->n;
size_t closestpt = n;
FLDATATYPE **pts = full->pts;
FLDATATYPE dist = FLT_MAX;
double **pts = full->pts;
double dist = FLT_MAX;
for ( size_t i = 0; i < n; i++ )
{
FLDATATYPE d = (float) distance<FLDATATYPE>(query_pt, pts[i]);
double d = (float) distance(query_pt, pts[i]);
if ( closestpt >= n || d < dist || (d<=dist && i < closestpt) )
{
dist = d;
......@@ -559,7 +552,7 @@ size_t gs_qnearest_kdtree(struct gridsearch *gs, double lon, double lat, double
struct pqueue *result = NULL;
kdata_t query_pt[3];
LLtoXYZ<kdata_t>(lon, lat, query_pt);
LLtoXYZ(lon, lat, query_pt);
if ( !gs->extrapolate )
for ( unsigned j = 0; j < 3; ++j )
......@@ -607,11 +600,11 @@ size_t gs_qnearest_nanoflann(struct gridsearch *gs, double lon, double lat, doub
nfTree_t *nft = (nfTree_t*) gs->search_container;
if ( nft == NULL ) return nadds;
NFDATATYPE range0 = gs_set_range(prange);
NFDATATYPE range = range0;
double range0 = gs_set_range(prange);
double range = range0;
NFDATATYPE query_pt[3];
LLtoXYZ<NFDATATYPE>(lon, lat, query_pt);
double query_pt[3];
LLtoXYZ(lon, lat, query_pt);
if ( !gs->extrapolate )
for ( unsigned j = 0; j < 3; ++j )
......@@ -619,12 +612,12 @@ size_t gs_qnearest_nanoflann(struct gridsearch *gs, double lon, double lat, doub
if ( gs )
{
std::vector<NFDATATYPE> out_dist_sqr(nnn);
std::vector<double> out_dist_sqr(nnn);
nadds = nft->knnRangeSearch(&query_pt[0], range, nnn, &adds[0], &out_dist_sqr[0]);
for ( size_t i = 0; i < nadds; ++i ) dist[i] = out_dist_sqr[i];
NFDATATYPE frange = range;
double frange = range;
if ( prange ) *prange = frange;
}
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment