def _apply_KDTree(self, group):
"""Selection using KDTree but periodic = True not supported.
(KDTree routine is ca 15% slower than the distance matrix one)
"""
sys_indices = numpy.array([a.number for a in self._group_atoms_list])
sys_coor = Selection.coord[sys_indices]
if self.periodic:
pass # or warn? -- no periodic functionality with KDTree search
from MDAnalysis.KDTree.NeighborSearch import CoordinateNeighborSearch
CNS = CoordinateNeighborSearch(sys_coor) # cache the KDTree for this selection/frame?
found_indices = CNS.search(self.ref, self.cutoff)
res_atoms = [self._group_atoms_list[i] for i in found_indices] # make list numpy array and use fancy indexing?
return set(res_atoms)
def _apply_KDTree(self, group):
"""KDTree based selection is about 7x faster than distmat for typical problems.
Limitations: always ignores periodicity
"""
sel_atoms = self.sel._apply(group) # group is wrong, should be universe (?!)
sys_atoms_list = [a for a in (self._group_atoms - sel_atoms)] # list needed for back-indexing
sel_indices = numpy.array([a.number for a in sel_atoms], dtype=int)
sys_indices = numpy.array([a.number for a in sys_atoms_list], dtype=int)
sel_coor = Selection.coord[sel_indices]
sys_coor = Selection.coord[sys_indices]
from MDAnalysis.KDTree.NeighborSearch import CoordinateNeighborSearch
# Can we optimize search by using the larger set for the tree?
CNS = CoordinateNeighborSearch(sys_coor) # cache the KDTree for this selection/frame?
found_indices = CNS.search_list(sel_coor, self.cutoff)
res_atoms = [sys_atoms_list[i] for i in found_indices] # make list numpy array and use fancy indexing?
return set(res_atoms)
def _apply_KDTree(self, group):
"""Selection using KDTree but periodic = True not supported.
(KDTree routine is ca 15% slower than the distance matrix one)
"""
sys_indices = numpy.array([a.number for a in self._group_atoms_list])
sys_coor = Selection.coord[sys_indices]
if self.periodic:
pass # or warn? -- no periodic functionality with KDTree search
from MDAnalysis.KDTree.NeighborSearch import CoordinateNeighborSearch
CNS = CoordinateNeighborSearch(
sys_coor) # cache the KDTree for this selection/frame?
found_indices = CNS.search(self.ref, self.cutoff)
res_atoms = [self._group_atoms_list[i] for i in found_indices
] # make list numpy array and use fancy indexing?
return set(res_atoms)
def _apply_KDTree(self, group):
"""Selection using KDTree but periodic = True not supported.
(KDTree routine is ca 15% slower than the distance matrix one)
"""
sys_indices = numpy.array([a.number for a in self._group_atoms_list])
sys_coor = Selection.coord[sys_indices]
sel_atoms = self.sel._apply(group) # group is wrong, should be universe (?!)
sel_CoG = AtomGroup(sel_atoms).centerOfGeometry()
self.ref = numpy.array((sel_CoG[0], sel_CoG[1], sel_CoG[2]))
if self.periodic:
pass # or warn? -- no periodic functionality with KDTree search
from MDAnalysis.KDTree.NeighborSearch import CoordinateNeighborSearch
CNS = CoordinateNeighborSearch(sys_coor) # cache the KDTree for this selection/frame?
found_ExtIndices = CNS.search(self.ref, self.exRadius)
found_IntIndices = CNS.search(self.ref, self.inRadius)
found_indices = list(set(found_ExtIndices) - set(found_IntIndices))
res_atoms = [self._group_atoms_list[i] for i in found_indices]
return set(res_atoms)
def _apply_KDTree(self, group):
"""Selection using KDTree but periodic = True not supported.
(KDTree routine is ca 15% slower than the distance matrix one)
"""
sys_indices = numpy.array([a.number for a in self._group_atoms_list])
sys_coor = Selection.coord[sys_indices]
sel_atoms = self.sel._apply(
group) # group is wrong, should be universe (?!)
sel_CoG = AtomGroup(sel_atoms).centerOfGeometry()
self.ref = numpy.array((sel_CoG[0], sel_CoG[1], sel_CoG[2]))
if self.periodic:
pass # or warn? -- no periodic functionality with KDTree search
from MDAnalysis.KDTree.NeighborSearch import CoordinateNeighborSearch
CNS = CoordinateNeighborSearch(
sys_coor) # cache the KDTree for this selection/frame?
found_ExtIndices = CNS.search(self.ref, self.exRadius)
found_IntIndices = CNS.search(self.ref, self.inRadius)
found_indices = list(set(found_ExtIndices) - set(found_IntIndices))
res_atoms = [self._group_atoms_list[i] for i in found_indices]
return set(res_atoms)
def _apply_KDTree(self, group):
"""KDTree based selection is about 7x faster than distmat for typical problems.
Limitations: always ignores periodicity
"""
sel_atoms = self.sel._apply(
group) # group is wrong, should be universe (?!)
sys_atoms_list = [a for a in (self._group_atoms - sel_atoms)
] # list needed for back-indexing
sel_indices = numpy.array([a.number for a in sel_atoms], dtype=int)
sys_indices = numpy.array([a.number for a in sys_atoms_list],
dtype=int)
sel_coor = Selection.coord[sel_indices]
sys_coor = Selection.coord[sys_indices]
from MDAnalysis.KDTree.NeighborSearch import CoordinateNeighborSearch
# Can we optimize search by using the larger set for the tree?
CNS = CoordinateNeighborSearch(
sys_coor) # cache the KDTree for this selection/frame?
found_indices = CNS.search_list(sel_coor, self.cutoff)
res_atoms = [sys_atoms_list[i] for i in found_indices
] # make list numpy array and use fancy indexing?
return set(res_atoms)