__copyright__ = "Copyright (C) 2017 Andreas Kloeckner"
__license__ = """
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.
"""
import numpy as np
import pyopencl as cl
from pytools import memoize_in
from meshmode.dof_array import flatten
from sumpy.fmm import UnableToCollectTimingData
__doc__ = """
.. autoclass:: PotentialSource
.. autoclass:: PointPotentialSource
.. autoclass:: LayerPotentialSourceBase
"""
[docs]class PotentialSource:
"""
.. automethod:: preprocess_optemplate
.. method:: op_group_features(expr)
Return a characteristic tuple by which operators that can be
executed together can be grouped.
*expr* is a subclass of
:class:`pytential.symbolic.primitives.IntG`.
"""
[docs] def preprocess_optemplate(self, name, discretizations, expr):
return expr
@property
def real_dtype(self):
raise NotImplementedError
@property
def complex_dtype(self):
raise NotImplementedError
def get_p2p(self, actx, kernels):
raise NotImplementedError
class _SumpyP2PMixin:
def get_p2p(self, actx, target_kernels, source_kernels=None):
@memoize_in(actx, (_SumpyP2PMixin, "p2p"))
def p2p(target_kernels, source_kernels):
if any(knl.is_complex_valued for knl in target_kernels):
value_dtype = self.complex_dtype
else:
value_dtype = self.real_dtype
from sumpy.p2p import P2P
return P2P(actx.context,
target_kernels, exclude_self=False, value_dtypes=value_dtype,
source_kernels=source_kernels)
return p2p(target_kernels, source_kernels)
# {{{ point potential source
[docs]class PointPotentialSource(_SumpyP2PMixin, PotentialSource):
"""
.. attribute:: nodes
An :class:`pyopencl.array.Array` of shape ``[ambient_dim, ndofs]``.
.. attribute:: ndofs
.. automethod:: cost_model_compute_potential_insn
.. automethod:: exec_compute_potential_insn
"""
def __init__(self, nodes):
self._nodes = nodes
@property
def points(self):
from warnings import warn
warn("'points' has been renamed to nodes().",
DeprecationWarning, stacklevel=2)
return self._nodes
[docs] def nodes(self):
return self._nodes
@property
def real_dtype(self):
return self._nodes.dtype
@property
def ndofs(self):
for coord_ary in self._nodes:
return coord_ary.shape[0]
@property
def complex_dtype(self):
return {
np.float32: np.complex64,
np.float64: np.complex128
}[self.real_dtype.type]
@property
def ambient_dim(self):
return self._nodes.shape[0]
def op_group_features(self, expr):
from pytential.utils import sort_arrays_together
# since IntGs with the same source kernels and densities calculations
# for P2E and E2E are the same and only differs in E2P depending on the
# target kernel, we group all IntGs with same source kernels and densities.
# sorting is done to avoid duplicates as the order of the sum of source
# kernels does not matter.
result = (
expr.source,
*sort_arrays_together(expr.source_kernels, expr.densities, key=str),
expr.target_kernel,
)
return result
[docs] def cost_model_compute_potential_insn(self, actx, insn, bound_expr,
evaluate, costs):
raise NotImplementedError
[docs] def exec_compute_potential_insn(self, actx, insn, bound_expr, evaluate,
return_timing_data):
if return_timing_data:
from warnings import warn
warn(
"Timing data collection not supported.",
category=UnableToCollectTimingData)
p2p = None
kernel_args = {}
for arg_name, arg_expr in insn.kernel_arguments.items():
kernel_args[arg_name] = evaluate(arg_expr)
strengths = [evaluate(density) for density in insn.densities]
# FIXME: Do this all at once
results = []
for o in insn.outputs:
target_discr = bound_expr.places.get_discretization(
o.target_name.geometry, o.target_name.discr_stage)
# no on-disk kernel caching
if p2p is None:
p2p = self.get_p2p(actx, source_kernels=insn.source_kernels,
target_kernels=insn.target_kernels)
from arraycontext import thaw
evt, output_for_each_kernel = p2p(actx.queue,
flatten(thaw(target_discr.nodes(), actx), strict=False),
self._nodes,
strengths, **kernel_args)
from meshmode.discretization import Discretization
result = output_for_each_kernel[o.target_kernel_index]
if isinstance(target_discr, Discretization):
from meshmode.dof_array import unflatten
result = unflatten(actx, target_discr, result)
results.append((o.name, result))
timing_data = {}
return results, timing_data
# }}}
# {{{ layer potential source
def _entry_dtype(ary):
from meshmode.dof_array import DOFArray
if isinstance(ary, DOFArray):
# the "normal case"
return ary.entry_dtype
elif isinstance(ary, np.ndarray):
if ary.dtype.char == "O":
from pytools import single_valued
return single_valued(_entry_dtype(entry) for entry in ary.flat)
else:
return ary.dtype
elif isinstance(ary, cl.array.Array):
# for "unregularized" layer potential sources
return ary.dtype
else:
raise TypeError(f"unexpected type '{type(ary)}' in _entry_dtype")
[docs]class LayerPotentialSourceBase(_SumpyP2PMixin, PotentialSource):
"""A discretization of a layer potential using panel-based geometry, with
support for refinement and upsampling.
.. rubric:: Discretization data
.. attribute:: density_discr
.. attribute:: cl_context
.. attribute:: ambient_dim
.. attribute:: dim
.. attribute:: real_dtype
.. attribute:: complex_dtype
"""
def __init__(self, density_discr):
self.density_discr = density_discr
@property
def ambient_dim(self):
return self.density_discr.ambient_dim
@property
def _setup_actx(self):
return self.density_discr._setup_actx
@property
def dim(self):
return self.density_discr.dim
@property
def cl_context(self):
return self._setup_actx.context
@property
def real_dtype(self):
return self.density_discr.real_dtype
@property
def complex_dtype(self):
return self.density_discr.complex_dtype
# {{{ fmm setup helpers
def get_fmm_kernel(self, kernels):
fmm_kernel = None
from sumpy.kernel import TargetTransformationRemover
for knl in kernels:
candidate_fmm_kernel = TargetTransformationRemover()(knl)
if fmm_kernel is None:
fmm_kernel = candidate_fmm_kernel
else:
assert fmm_kernel == candidate_fmm_kernel
return fmm_kernel
def get_fmm_output_and_expansion_dtype(self, kernels, strengths):
if any(knl.is_complex_valued for knl in kernels) or \
_entry_dtype(strengths).kind == "c":
return self.complex_dtype
else:
return self.real_dtype
def get_fmm_expansion_wrangler_extra_kwargs(
self, actx, target_kernels, tree_user_source_ids, arguments, evaluator):
# This contains things like the Helmholtz parameter k or
# the normal directions for double layers.
queue = actx.queue
def reorder_sources(source_array):
if isinstance(source_array, cl.array.Array):
return (source_array
.with_queue(queue)
[tree_user_source_ids]
.with_queue(None))
else:
return source_array
kernel_extra_kwargs = {}
source_extra_kwargs = {}
from sumpy.tools import gather_arguments, gather_source_arguments
from pytools.obj_array import obj_array_vectorize
for func, var_dict in [
(gather_arguments, kernel_extra_kwargs),
(gather_source_arguments, source_extra_kwargs),
]:
for arg in func(target_kernels):
var_dict[arg.name] = obj_array_vectorize(
reorder_sources,
flatten(evaluator(arguments[arg.name]), strict=False))
return kernel_extra_kwargs, source_extra_kwargs
# }}}
# }}}
# vim: foldmethod=marker