Source code for MMTK.Field

# This module defines scalar and vector fields in molecular systems
#
# Written by Konrad Hinsen
#

"""
Scalar and vector fields in molecular systems

This module defines field objects that are useful in the analysis
and visualization of collective motions in molecular systems. Atomic
quantities characterizing collective motions vary slowly in space, and
can be considered functions of position instead of values per atom.
Functions of position are called fields, and mathematical techniques
for the analysis of fields have proven useful in many branches of
physics. Fields can be described numerically by values on a
regular grid. In addition to permitting the application of vector
analysis methods to atomic quantities, the introduction of
fields is a valuable visualization aid, because information defined on
a coarse regular grid can be added to a picture of a molecular system
without overloading it.
"""

__docformat__ = 'restructuredtext'

from MMTK import Collections, ParticleProperties, Visualization
from Scientific.Visualization import Color
from Scientific.Geometry import Vector, TensorAnalysis
from Scientific import N


[docs]class AtomicField(object): """ A field whose values are determined by atomic quantities This is an abstract base class. To create field objects, use one of its subclasses. """ def __init__(self, system, grid_size, values): self.system = system if isinstance(grid_size, tuple): self.box, self.field, self.points, self.colors = grid_size else: if values.data_rank != 1: raise TypeError("data not a particle variable") self.box = Collections.PartitionedAtomCollection(grid_size, system) self._setField(values) def _setField(self, data): points = [] values = [] colors = [] default_color = Color.ColorByName('black') for min, max, atoms in self.box.partitions(): center = 0.5*(min+max) points.append(center.array) v = data.zero() total_weight = 0. color = default_color for a in atoms: d = a.position()-center weight = N.exp(-d*d/self.box.partition_size**2) v = v + weight*data[a] total_weight = total_weight + weight c = default_color try: c = a.color except AttributeError: pass if isinstance(c, str): c = Color.ColorByName(c) color = color + c values.append(v/total_weight) colors.append(color) min = N.minimum.reduce(points) max = N.maximum.reduce(points) axes = (N.arange(min[0], max[0]+1, self.box.partition_size), N.arange(min[1], max[1]+1, self.box.partition_size), N.arange(min[2], max[2]+1, self.box.partition_size)) array = N.zeros(tuple(map(len, axes)) + data.value_rank*(3,), N.Float) inside = N.zeros(tuple(map(len, axes)), N.Float) for p, v in zip(points, values): indices = N.floor((p-min)/self.box.partition_size+0.5) indices = tuple(indices.astype(N.Int)) array[indices] = v inside[indices] = 1. self.field = self.field_class(axes, array, data.zero()) inside = TensorAnalysis.ScalarField(axes, inside).gradient().length() self.points = [] self.colors = [] for i in range(len(points)): p = points[i] test = 0 try: test = apply(inside, tuple(p)) > 1.e-10 except ValueError: pass if test: self.points.append(p) self.colors.append(colors[i]) def __call__(self, point): return self.field(point)
[docs] def particleValues(self): """ :returns: the values of the field at the positions of the atoms :rtype: :class:`~MMTK.ParticleProperties.ParticleProperty` """ universe = self.system.universe() rank = self.field.rank if rank == 0: v = ParticleProperties.ParticleScalar(universe) elif rank == 1: v = ParticleProperties.ParticleVector(universe) else: raise ValueError("no appropriate return type") for a in self.system.atomList(): v[a] = self.field(a.position()) return v
[docs] def writeToFile(self, filename, scale = 1., length_scale=1., color = None): """ Writes a graphical representation of the field to a VRML file. :param filename: the name of the destination file :type filename: str :param scale: scale factor applied to all field values :type scale: float :param color: the color for all graphics objects :type color: Scientific.Visualization.Color """ from Scientific.Visualization import VRML2 objects = self.graphicsObjects(scale=scale, length_scale=length_scale, color=color, graphics_module=VRML2) VRML2.Scene(objects).writeToFile(filename)
[docs] def view(self, scale = 1., length_scale = 1., color = None): """ Shows a graphical representation of the field using a VRML viewer. :param scale: scale factor applied to all field values :type scale: float :param color: the color for all graphics objects :type color: Scientific.Visualization.Color """ from Scientific.Visualization import VRML2 objects = self.graphicsObjects(scale=scale, length_scale=length_scale, color=color, graphics_module=VRML2) VRML2.Scene(objects).view()
[docs]class AtomicScalarField(AtomicField, Visualization.Viewable): """ Scalar field defined by atomic quantities For visualization, scalar fields are represented by a small cube on each grid point whose color indicates the field's value on a symmetric red-to-green color scale defined by the range of the field values. Additional keyword options exist for graphics object generation: - scale=factor, to multiply all field values by a factor - range=(min, max), to eliminate graphics objects for values that are smaller than min or larger than max """ def __init__(self, system, grid_size, values): """ :param system: any subset of a molecular system :param grid_size: the spacing of a cubic grid on which the field values are defined. The value for a point is obtained by averaging the atomic quantities over all atoms in a cube centered on the point. :type grid_size: float :param values: the atomic values that define the field :type values: :class:`~MMTK.ParticleProperties.ParticleScalar` """ if values is not None and values.value_rank != 0: raise TypeError("data not a vector field") self.field_class = TensorAnalysis.ScalarField AtomicField.__init__(self, system, grid_size, values)
[docs] def gradient(self): """ :returns: the gradient of the field :rtype: :class:`~MMTK.Field.AtomicVectorField` """ field = self.field.gradient() return AtomicVectorField(self.system, (self.box, field, self.points, self.colors), None)
[docs] def laplacian(self): """ :returns: the laplacian of the field :rtype: :class:`~MMTK.Field.AtomicScalarField` """ field = self.field.laplacian() return AtomicScalarField(self.system, (self.box, field, self.points, self.colors), None)
def _graphics(self, conf, distance_fn, model, module, options): scale = options.get('scale', 1.) range = options.get('range', (None, None)) length_scale = options.get('length_scale', 1.) lower, upper = range size = self.box.partition_size/10. objects = [] color_scale = module.SymmetricColorScale(1.) for p, c in zip(self.points, self.colors): p = tuple(p) v = apply(self.field, p) if (lower is None or v > lower) and (upper is None or v < upper): p = apply(Vector, p) v = scale*v if v < -1. or v > 1.: m = module.DiffuseMaterial('black') else: m = module.Material(diffuse_color = color_scale(scale*v)) objects.append(module.Cube(length_scale*p, length_scale*size, material = m)) return objects
[docs]class AtomicVectorField(AtomicField, Visualization.Viewable): """ Vector field defined by atomic quantities For visualization, scalar fields are represented by a small arrow on each grid point. The arrow starts at the grid point and represents the vector value at that point. Additional keyword options exist for graphics object generation: - scale=factor, to multiply all field values by a factor - diameter=number, to define the diameter of the arrow objects (default: 1.) - range=(min, max), to eliminate graphics objects for values that are smaller than min or larger than max - color=string, to define the color of the arrows by a color name """ def __init__(self, system, grid_size, values): """ :param system: any subset of a molecular system :param grid_size: the spacing of a cubic grid on which the field values are defined. The value for a point is obtained by averaging the atomic quantities over all atoms in a cube centered on the point. :type grid_size: float :param values: the atomic values that define the field :type values: :class:`~MMTK.ParticleProperties.ParticleVector` """ if values is not None and values.value_rank != 1: raise TypeError("data not a vector field") self.field_class = TensorAnalysis.VectorField AtomicField.__init__(self, system, grid_size, values)
[docs] def length(self): """ :returns: a field of the length of the field vectors :rtype: :class:`~MMTK.Field.AtomicScalarField` """ field = self.field.length() return AtomicScalarField(self.system, (self.box, field, self.points, self.colors), None)
[docs] def divergence(self): """ :returns: the divergence of the field :rtype: :class:`~MMTK.Field.AtomicScalarField` """ field = self.field.divergence() return AtomicScalarField(self.system, (self.box, field, self.points, self.colors), None)
[docs] def curl(self): """ :returns: the curl of the field :rtype: :class:`~MMTK.Field.AtomicVectorField` """ field = self.field.curl() return AtomicVectorField(self.system, (self.box, field, self.points, self.colors), None)
[docs] def laplacian(self): """ :returns: the laplacian of the field :rtype: :class:`~MMTK.Field.AtomicVectorField` """ field = self.field.curl() return AtomicVectorField(self.system, (self.box, field, self.points, self.colors), None)
def _graphics(self, conf, distance_fn, model, module, options): scale = options.get('scale', 1.) diameter = options.get('diameter', 1.) color = options.get('color', None) range = options.get('range', (None, None)) length_scale = options.get('length_scale', 1.) lower, upper = range size = diameter*self.box.partition_size/50. if color is not None: color = module.ColorByName(color) objects = [] materials = {} for p, c in zip(self.points, self.colors): p = tuple(p) v = apply(self.field, p) lv = v.length() if (lower is None or lv > lower) and (upper is None or lv < upper): p = apply(Vector, p) if color is not None: c = color try: m = materials[c] except KeyError: m = module.Material(diffuse_color = c) materials[c] = m objects.append(module.Arrow(length_scale*p, length_scale*(p+scale*v), length_scale*size, material = m)) return objects