Source code for pde.grids.cartesian

"""
Cartesian grids of arbitrary dimension.

.. codeauthor:: David Zwicker <david.zwicker@ds.mpg.de>
 
"""

from __future__ import annotations

import itertools
import warnings
from typing import List  # @UnusedImport
from typing import TYPE_CHECKING, Any, Dict, Generator, Sequence, Tuple, Union

import numpy as np

from ..tools.cuboid import Cuboid
from ..tools.docstrings import fill_in_docstring
from ..tools.plotting import plot_on_axes
from .base import DimensionError, GridBase, _check_shape

if TYPE_CHECKING:
    from .boundaries.axes import Boundaries, BoundariesData  # @UnusedImport


[docs]class CartesianGrid(GridBase): # lgtm [py/missing-equals] r""" d-dimensional Cartesian grid with uniform discretization for each axis The grids can be thought of as a collection of n-dimensional boxes, called cells, of equal length in each dimension. The bounds then defined the total volume covered by these cells, while the cell coordinates give the location of the box centers. We index the boxes starting from 0 along each dimension. Consequently, the cell :math:`i-\frac12` corresponds to the left edge of the covered interval and the index :math:`i+\frac12` corresponds to the right edge, when the dimension is covered by d boxes. In particular, the discretization along dimension :math:`k` is defined as .. math:: x^{(k)}_i &= x^{(k)}_\mathrm{min} + \left(i + \frac12\right) \Delta x^{(k)} \quad \text{for} \quad i = 0, \ldots, N^{(k)} - 1 \\ \Delta x^{(k)} &= \frac{x^{(k)}_\mathrm{max} - x^{(k)}_\mathrm{min}}{N^{(k)}} where :math:`N^{(k)}` is the number of cells along this dimension. Consequently, cells have dimension :math:`\Delta x^{(k)}` and cover the interval :math:`[x^{(k)}_\mathrm{min}, x^{(k)}_\mathrm{max}]`. """ cuboid: Cuboid def __init__( self, bounds: Sequence[Tuple[float, float]], shape: Union[int, Sequence[int]], periodic: Union[Sequence[bool], bool] = False, ): """ Args: bounds (list of tuple): Give the coordinate range for each axis. This should be a tuple of two number (lower and upper bound) for each axis. The length of `bounds` thus determines the grid dimension. shape (list): The number of support points for each axis. The length of `shape` needs to match the grid dimension. periodic (bool or list): Specifies which axes possess periodic boundary conditions. This is either a list of booleans defining periodicity for each individual axis or a single boolean value specifying the same periodicity for all axes. """ bounds_arr = np.array(bounds, ndmin=1, dtype=np.double) if bounds_arr.shape == (2,): raise ValueError( "`bounds with shape (2,) are ambiguous. Either use shape (1, 2) to set " "up a 1d system with two bounds or shape (2, 1) for a 2d system with " "only the upper bounds specified" ) if bounds_arr.ndim == 1 or bounds_arr.shape[1] == 1: # only set the upper bounds bounds_arr = np.atleast_1d(np.squeeze(bounds_arr)) self.cuboid = Cuboid(np.zeros_like(bounds_arr), bounds_arr, mutable=False) elif bounds_arr.ndim == 2 and bounds_arr.shape[1] == 2: # upper and lower bounds of the grid are given self.cuboid = Cuboid.from_bounds(bounds_arr, mutable=False) else: raise ValueError( f"Do not know how to interpret shape {bounds_arr.shape} for bounds" ) # handle the shape array shape = _check_shape(shape) if len(shape) == 1 and self.cuboid.dim > 1: shape = (int(shape[0]),) * self.cuboid.dim if self.cuboid.dim != len(shape): raise DimensionError("Dimension of `bounds` and `shape` are not compatible") # initialize the base class super().__init__() self._shape = _check_shape(shape) self.dim = len(self.shape) self.num_axes = self.dim if isinstance(periodic, (bool, np.bool_)): self.periodic = [bool(periodic)] * self.dim elif len(periodic) != self.dim: raise DimensionError( "Number of axes with specified periodicity does not match grid " f"dimension ({len(periodic)} != {self.dim})" ) else: self.periodic = list(periodic) if self.dim <= 3: self.axes = list("xyz"[: self.dim]) else: self.axes = [chr(97 + i) for i in range(self.dim)] # determine the coordinates p1, p2 = self.cuboid.corners axes_coords, discretization = [], [] for d in range(self.dim): num = self.shape[d] c, dc = np.linspace(p1[d], p2[d], num, endpoint=False, retstep=True) if self.shape[d] == 1: # correct for singular dimension dc = p2[d] - p1[d] c += dc / 2 axes_coords.append(c) discretization.append(dc) self._discretization = np.array(discretization) self._axes_coords = tuple(axes_coords) self._axes_bounds = tuple(self.cuboid.bounds) @property def state(self) -> Dict[str, Any]: """dict: the state of the grid""" return { "bounds": self.axes_bounds, "shape": self.shape, "periodic": self.periodic, }
[docs] @classmethod def from_state(cls, state: Dict[str, Any]) -> CartesianGrid: # type: ignore """create a field from a stored `state`. Args: state (dict): The state from which the grid is reconstructed. """ state_copy = state.copy() obj = cls( bounds=state_copy.pop("bounds"), shape=state_copy.pop("shape"), periodic=state_copy.pop("periodic"), ) if state_copy: raise ValueError(f"State items {state_copy.keys()} were not used") return obj
@property def volume(self) -> float: """float: total volume of the grid""" return float(self.cuboid.volume) @property def cell_volume_data(self): """size associated with each cell""" return tuple(self.discretization)
[docs] def iter_mirror_points( self, point: np.ndarray, with_self: bool = False, only_periodic: bool = True ) -> Generator: """generates all mirror points corresponding to `point` Args: point (:class:`~numpy.ndarray`): the point within the grid with_self (bool): whether to include the point itself only_periodic (bool): whether to only mirror along periodic axes Returns: A generator yielding the coordinates that correspond to mirrors """ point = np.asanyarray(point, dtype=np.double) # find all offsets of the individual axes offsets = [] for i in range(self.dim): if only_periodic and not self.periodic[i]: offsets.append([0]) else: s = self.cuboid.size[i] offsets.append([-s, 0, s]) # produce the respective mirrored points for offset in itertools.product(*offsets): if with_self or np.linalg.norm(offset) != 0: yield point + offset
[docs] def get_random_point( self, *, boundary_distance: float = 0, coords: str = "cartesian", rng: np.random.Generator = None, cartesian: bool = None, ) -> np.ndarray: """return a random point within the grid Args: boundary_distance (float): The minimal distance this point needs to have from all boundaries. coords (str): Determines the coordinate system in which the point is specified. Valid values are `cartesian`, `cell`, and `grid`; see :meth:`~pde.grids.base.GridBase.transform`. rng (:class:`~numpy.random.Generator`): Random number generator (default: :func:`~numpy.random.default_rng()`) Returns: :class:`~numpy.ndarray`: The coordinates of the point """ if cartesian is not None: # deprecated on 2022-03-11 warnings.warn("Argument `cartesian` is deprecated. Use `coords` instead") coords = "cartesian" if cartesian else "grid" if rng is None: rng = np.random.default_rng() # handle the boundary distance cuboid = self.cuboid if boundary_distance != 0: if any(cuboid.size <= 2 * boundary_distance): raise RuntimeError("Random points would be too close to boundary") cuboid = cuboid.buffer(-boundary_distance) # create random point point = cuboid.pos + rng.random(self.dim) * cuboid.size if coords == "cartesian" or coords == "grid": return point # type: ignore elif coords == "cell": return self.transform(point, "grid", "cell") else: raise ValueError(f"Unknown coordinate system `{coords}`")
[docs] def get_line_data(self, data: np.ndarray, extract: str = "auto") -> Dict[str, Any]: """return a line cut through the given data Args: data (:class:`~numpy.ndarray`): The values at the grid points extract (str): Determines which cut is done through the grid. Possible choices are (default is `cut_0`): * `cut_#`: return values along the axis specified by # and use the mid point along all other axes. * `project_#`: average values for all axes, except axis #. Here, # can either be a zero-based index (from 0 to dim-1) or a letter denoting the axis. Returns: A dictionary with information about the line cut, which is convenient for plotting. """ if data.shape[-self.dim :] != self.shape: raise ValueError( f"Shape {data.shape} of the data array is not compatible with grid " f"shape {self.shape}" ) def _get_axis(axis): """determine the axis from a given specifier""" try: axis = int(axis) except ValueError: try: axis = self.axes.index(axis) except ValueError: raise ValueError(f"Axis `{axis}` not defined") return axis if extract == "auto": extract = "cut_0" # use a cut along first axis if extract.startswith("cut_"): # consider a cut along a given axis axis = _get_axis(extract[4:]) data_y = data rank = data.ndim - self.dim # rank of data for ax in reversed(range(self.dim)): if ax != axis: mid_point = self.shape[ax] // 2 data_y = np.take(data_y, mid_point, axis=ax + rank) label_y = f"Cut along {self.axes[axis]}" elif extract.startswith("project_"): # consider a projection along a given axis axis = _get_axis(extract[8:]) avg_axes = [ax - self.dim for ax in range(self.dim) if ax != axis] data_y = data.mean(axis=tuple(avg_axes)) label_y = f"Projection onto {self.axes[axis]}" else: raise ValueError(f"Unknown extraction method `{extract}`") if self.dim == 1: label_y = "" # return the data with the respective labels return { "data_x": self.axes_coords[axis], "data_y": data_y, "extent_x": self.axes_bounds[axis], "label_x": self.axes[axis], "label_y": label_y, }
[docs] def get_image_data(self, data: np.ndarray) -> Dict[str, Any]: """return a 2d-image of the data Args: data (:class:`~numpy.ndarray`): The values at the grid points Returns: A dictionary with information about the image, which is convenient for plotting. """ if data.shape[-self.dim :] != self.shape: raise ValueError( f"Shape {data.shape} of the data array is not compatible with grid " f"shape {self.shape}" ) if self.dim == 2: image_data = data elif self.dim == 3: image_data = data[:, :, self.shape[-1] // 2] else: raise NotImplementedError( "Creating images is only implemented for 2d and 3d grids" ) extent: List[float] = [] for c in self.axes_bounds[:2]: extent.extend(c) return { "data": image_data, "x": self.axes_coords[0], "y": self.axes_coords[1], "extent": extent, "label_x": self.axes[0], "label_y": self.axes[1], }
[docs] def point_to_cartesian( self, points: np.ndarray, *, full: bool = False ) -> np.ndarray: """convert coordinates of a point to Cartesian coordinates Args: points (:class:`~numpy.ndarray`): Points given in grid coordinates full (bool): Compatibility option not used in this method Returns: :class:`~numpy.ndarray`: The Cartesian coordinates of the point """ assert points.shape[-1] == self.dim, f"Point must have {self.dim} coordinates" return points
[docs] def point_from_cartesian(self, coords: np.ndarray) -> np.ndarray: """convert points given in Cartesian coordinates to this grid Args: coords (:class:`~numpy.ndarray`): Points in Cartesian coordinates. Returns: :class:`~numpy.ndarray`: Points given in the coordinates of the grid """ assert coords.shape[-1] == self.dim, f"Point must have {self.dim} coordinates" return coords
[docs] def polar_coordinates_real( self, origin: np.ndarray, *, ret_angle: bool = False ) -> Union[np.ndarray, Tuple[np.ndarray, np.ndarray, np.ndarray]]: """return polar coordinates associated with the grid Args: origin (:class:`~numpy.ndarray`): Coordinates of the origin at which the polar coordinate system is anchored. ret_angle (bool): Determines whether angles are returned alongside the distance. If `False` only the distance to the origin is returned for each support point of the grid. If `True`, the distance and angles are returned. For a 1d system system, the angle is defined as the sign of the difference between the point and the origin, so that angles can either be 1 or -1. For 2d systems and 3d systems, polar coordinates and spherical coordinates are used, respectively. """ origin = np.array(origin, dtype=np.double, ndmin=1) if len(origin) != self.dim: raise DimensionError("Dimensions are not compatible") # calculate the difference vector between all cells and the origin diff = self.difference_vector_real(origin, self.cell_coords) dist: np.ndarray = np.linalg.norm(diff, axis=-1) # get distance # determine distance and optionally angles for these vectors if ret_angle: if self.dim == 1: return dist, np.sign(diff)[..., 0] # type: ignore elif self.dim == 2: return dist, np.arctan2(diff[:, :, 0], diff[:, :, 1]) # type: ignore elif self.dim == 3: theta = np.arccos(diff[..., 2] / dist) phi = np.arctan2(diff[..., 0], diff[..., 1]) return dist, theta, phi else: raise NotImplementedError( f"Cannot calculate angles for dimension {self.dim}" ) else: return dist
[docs] def from_polar_coordinates( self, distance: np.ndarray, angle: np.ndarray, origin: np.ndarray = None ) -> np.ndarray: """convert polar coordinates to Cartesian coordinates This function is currently only implemented for 1d and 2d systems. Args: distance (:class:`~numpy.ndarray`): The radial distance angle (:class:`~numpy.ndarray`): The angle with respect to the origin origin (:class:`~numpy.ndarray`, optional): Sets the origin of the coordinate system. If omitted, the zero point is assumed as the origin. Returns: :class:`~numpy.ndarray`: The Cartesian coordinates corresponding to the given polar coordinates. """ distance = np.asarray(distance) angle = np.asarray(angle) if origin is None: origin = np.zeros(self.dim) else: origin = np.atleast_1d(origin) if self.dim == 1: diff = distance * angle coords = origin + diff[..., None] elif self.dim == 2: unit_vector = np.moveaxis(np.array([np.sin(angle), np.cos(angle)]), 0, -1) diff = distance[..., None] * unit_vector coords = origin + diff else: raise NotImplementedError( f"Cannot calculate coordinates for dimension {self.dim}" ) return self.normalize_point(coords, reflect=False)
[docs] @plot_on_axes() def plot(self, ax, **kwargs): r"""visualize the grid Args: {PLOT_ARGS} \**kwargs: Extra arguments are passed on the to the matplotlib plotting routines, e.g., to set the color of the lines """ if self.dim not in {1, 2}: raise NotImplementedError( f"Plotting is not implemented for grids of dimension {self.dim}" ) kwargs.setdefault("color", "k") xb = self.axes_bounds[0] for x in np.linspace(*xb, self.shape[0] + 1): ax.axvline(x, **kwargs) ax.set_xlim(*xb) ax.set_xlabel(self.axes[0]) if self.dim == 2: yb = self.axes_bounds[1] for y in np.linspace(*yb, self.shape[1] + 1): ax.axhline(y, **kwargs) ax.set_ylim(*yb) ax.set_ylabel(self.axes[1]) ax.set_aspect(1)
[docs] @fill_in_docstring def get_boundary_conditions( self, bc: "BoundariesData" = "auto_periodic_neumann", rank: int = 0 ) -> "Boundaries": """constructs boundary conditions from a flexible data format Args: bc (str or list or tuple or dict): The boundary conditions applied to the field. {ARG_BOUNDARIES} rank (int): The tensorial rank of the value associated with the boundary conditions. Raises: ValueError: If the data given in `bc` cannot be read PeriodicityError: If the boundaries are not compatible with the periodic axes of the grid. """ from .boundaries import Boundaries # @Reimport # get boundary conditions return Boundaries.from_data(self, bc, rank=rank)
[docs] def get_subgrid(self, indices: Sequence[int]) -> CartesianGrid: """return a subgrid of only the specified axes Args: indices (list): Indices indicating the axes that are retained in the subgrid Returns: :class:`CartesianGrid`: The subgrid """ subgrid = self.__class__( bounds=[self.axes_bounds[i] for i in indices], shape=tuple(self.shape[i] for i in indices), periodic=[self.periodic[i] for i in indices], ) subgrid.axes = [self.axes[i] for i in indices] return subgrid
[docs]class UnitGrid(CartesianGrid): r"""d-dimensional Cartesian grid with unit discretization in all directions The grids can be thought of as a collection of d-dimensional cells of unit length. The `shape` parameter determines how many boxes there are in each direction. The cells are enumerated starting with 0, so the last cell has index :math:`n-1` if there are :math:`n` cells along a dimension. A given cell :math:`i` extends from coordinates :math:`i` to :math:`i + 1`, so the midpoint is at :math:`i + \frac12`, which is the cell coordinate. Taken together, the cells covers the interval :math:`[0, n]` along this dimension. """ def __init__( self, shape: Sequence[int], periodic: Union[Sequence[bool], bool] = False ): """ Args: shape (list): The number of support points for each axis. The dimension of the grid is given by `len(shape)`. periodic (bool or list): Specifies which axes possess periodic boundary conditions. This is either a list of booleans defining periodicity for each individual axis or a single boolean value specifying the same periodicity for all axes. """ if isinstance(shape, int): shape = [shape] super().__init__([(0, s) for s in shape], shape, periodic) self.cuboid = Cuboid(np.zeros(self.dim), self.shape) self._discretization = np.ones(self.dim) # determine the cell center coordinates self._axes_coords = tuple(np.arange(n) + 0.5 for n in self.shape) self._axes_bounds = tuple(self.cuboid.bounds) @property def state(self) -> Dict[str, Any]: """dict: the state of the grid""" return {"shape": self.shape, "periodic": self.periodic}
[docs] @classmethod def from_state(cls, state: Dict[str, Any]) -> UnitGrid: # type: ignore """create a field from a stored `state`. Args: state (dict): The state from which the grid is reconstructed. """ state_copy = state.copy() obj = cls(shape=state_copy.pop("shape"), periodic=state_copy.pop("periodic")) if state_copy: raise ValueError(f"State items {state_copy.keys()} were not used") return obj
[docs] def to_cartesian(self) -> CartesianGrid: """convert unit grid to :class:`CartesianGrid`""" return CartesianGrid( self.cuboid.bounds, shape=self.shape, periodic=self.periodic )
[docs] def get_subgrid(self, indices: Sequence[int]) -> UnitGrid: """return a subgrid of only the specified axes Args: indices (list): Indices indicating the axes that are retained in the subgrid Returns: :class:`UnitGrid`: The subgrid """ subgrid = self.__class__( shape=[self.shape[i] for i in indices], periodic=[self.periodic[i] for i in indices], ) subgrid.axes = [self.axes[i] for i in indices] return subgrid
# temporarily define CartesianGridBase, but this is deprecated since 2022-03-22 CartesianGridBase = CartesianGrid