def eval_direct(self, target_boxes, neighbor_sources_starts,
neighbor_sources_lists, src_weights):
pot = self.potential_zeros()
from pyfmmlib import hpotgrad2dall_vec
for itgt_box, tgt_ibox in enumerate(target_boxes):
tgt_pslice = self._get_target_slice(tgt_ibox)
if tgt_pslice.stop - tgt_pslice.start == 0:
continue
tgt_result = np.zeros(tgt_pslice.stop - tgt_pslice.start, np.complex128)
start, end = neighbor_sources_starts[itgt_box:itgt_box+2]
for src_ibox in neighbor_sources_lists[start:end]:
src_pslice = self._get_source_slice(src_ibox)
if src_pslice.stop - src_pslice.start == 0:
continue
tmp_pot, _, _ = hpotgrad2dall_vec(
ifgrad=False, ifhess=False,
sources=self._get_sources(src_pslice),
charge=src_weights[src_pslice],
targets=self._get_targets(tgt_pslice), zk=self.helmholtz_k)
tgt_result += tmp_pot
pot[tgt_pslice] = tgt_result
return pot
def eval_direct(self, target_boxes, neighbor_sources_starts,
neighbor_sources_lists, src_weights):
pot = self.potential_zeros()
from pyfmmlib import hpotgrad2dall_vec
for itgt_box, tgt_ibox in enumerate(target_boxes):
tgt_pslice = self._get_target_slice(tgt_ibox)
if tgt_pslice.stop - tgt_pslice.start == 0:
continue
tgt_result = np.zeros(tgt_pslice.stop - tgt_pslice.start,
np.complex128)
start, end = neighbor_sources_starts[itgt_box:itgt_box + 2]
for src_ibox in neighbor_sources_lists[start:end]:
src_pslice = self._get_source_slice(src_ibox)
if src_pslice.stop - src_pslice.start == 0:
continue
tmp_pot, _, _ = hpotgrad2dall_vec(
ifgrad=False,
ifhess=False,
sources=self._get_sources(src_pslice),
charge=src_weights[src_pslice],
targets=self._get_targets(tgt_pslice),
zk=self.helmholtz_k)
tgt_result += tmp_pot
pot[tgt_pslice] = tgt_result
return pot
def test_pyfmmlib_fmm(ctx_getter):
logging.basicConfig(level=logging.INFO)
from pytest import importorskip
importorskip("pyfmmlib")
ctx = ctx_getter()
queue = cl.CommandQueue(ctx)
nsources = 3000
ntargets = 1000
dims = 2
dtype = np.float64
helmholtz_k = 2
sources = p_normal(queue, nsources, dims, dtype, seed=15)
targets = (
p_normal(queue, ntargets, dims, dtype, seed=18)
+ np.array([2, 0]))
sources_host = particle_array_to_host(sources)
targets_host = particle_array_to_host(targets)
from boxtree import TreeBuilder
tb = TreeBuilder(ctx)
tree, _ = tb(queue, sources, targets=targets,
max_particles_in_box=30, debug=True)
from boxtree.traversal import FMMTraversalBuilder
tbuild = FMMTraversalBuilder(ctx)
trav, _ = tbuild(queue, tree, debug=True)
trav = trav.get(queue=queue)
from pyopencl.clrandom import PhiloxGenerator
rng = PhiloxGenerator(queue.context, seed=20)
weights = rng.uniform(queue, nsources, dtype=np.float64).get()
#weights = np.ones(nsources)
logger.info("computing direct (reference) result")
from pyfmmlib import hpotgrad2dall_vec
ref_pot, _, _ = hpotgrad2dall_vec(ifgrad=False, ifhess=False,
sources=sources_host.T, charge=weights,
targets=targets_host.T, zk=helmholtz_k)
from boxtree.pyfmmlib_integration import Helmholtz2DExpansionWrangler
wrangler = Helmholtz2DExpansionWrangler(trav.tree, helmholtz_k, nterms=10)
from boxtree.fmm import drive_fmm
pot = drive_fmm(trav, wrangler, weights)
rel_err = la.norm(pot - ref_pot) / la.norm(ref_pot)
logger.info("relative l2 error: %g" % rel_err)
assert rel_err < 1e-5
def test_translations():
nterms = 15
zk = 3
rscale = 1
n = 40
# centered at the origin, extent [-.5,.5]
sources = np.random.uniform(size=(n, 2)) - 0.5
charges = np.random.uniform(size=n)
targets_center = np.array([10, 0])
targets = np.random.uniform(size=(n, 2)) - 0.5 + targets_center
from pyfmmlib import (h2dformmp, h2dmpmp_vec, h2dmploc_vec, h2dlocloc_vec,
h2dtaeval_vec, hpotgrad2dall_vec, h2dmpeval_vec)
ref_value, _, _ = hpotgrad2dall_vec(ifgrad=False,
ifhess=False,
sources=sources.T,
charge=charges,
targets=targets.T,
zk=zk)
# {{{ multipole 1
mp1_center = np.array([0, 0])
ier, mp1 = h2dformmp(zk, rscale, sources.T, charges, mp1_center, nterms)
assert ier == 0
mp1_value, _, _ = h2dmpeval_vec(zk,
rscale,
mp1_center,
mp1,
ztarg=targets.T,
ifgrad=False,
ifhess=False)
assert la.norm(mp1_value - ref_value) / la.norm(ref_value) < 1e-12
# }}}
# {{{ multipole 2
mp2_center = np.array([2, 0])
mp2 = h2dmpmp_vec(zk, rscale, mp1_center, mp1, rscale, mp2_center, nterms)
mp2_value, _, _ = h2dmpeval_vec(zk,
rscale,
mp2_center,
mp2,
ztarg=targets.T,
ifgrad=False,
ifhess=False)
assert la.norm(mp2_value - ref_value) / la.norm(ref_value) < 3e-5
# }}}
# {{{ local 1
loc1_center = targets_center - np.array([1, 0])
loc1 = h2dmploc_vec(zk, rscale, mp2_center, mp2, rscale, loc1_center,
nterms)
loc1_value, _, _ = h2dtaeval_vec(zk,
rscale,
loc1_center,
loc1,
ztarg=targets.T,
ifgrad=False,
ifhess=False)
assert la.norm(loc1_value - ref_value) / la.norm(ref_value) < 3e-5
# }}}
# {{{ local 2
loc2_center = targets_center
loc2 = h2dlocloc_vec(zk, rscale, loc1_center, loc1, rscale, loc2_center,
nterms)
loc2_value, _, _ = h2dtaeval_vec(zk,
rscale,
loc2_center,
loc2,
ztarg=targets.T,
ifgrad=False,
ifhess=False)
assert la.norm(loc2_value - ref_value) / la.norm(ref_value) < 1e-4
def test_translations():
nterms = 15
zk = 3
rscale = 1
n = 40
# centered at the origin, extent [-.5,.5]
sources = np.random.uniform(size=(n, 2)) - 0.5
charges = np.random.uniform(size=n)
targets_center = np.array([10, 0])
targets = np.random.uniform(size=(n, 2)) - 0.5 + targets_center
from pyfmmlib import (h2dformmp, h2dmpmp_vec, h2dmploc_vec,
h2dlocloc_vec, h2dtaeval_vec, hpotgrad2dall_vec, h2dmpeval_vec)
ref_value, _, _ = hpotgrad2dall_vec(ifgrad=False, ifhess=False,
sources=sources.T, charge=charges,
targets=targets.T, zk=zk)
# {{{ multipole 1
mp1_center = np.array([0, 0])
ier, mp1 = h2dformmp(zk, rscale, sources.T, charges, mp1_center, nterms)
assert ier == 0
mp1_value, _, _ = h2dmpeval_vec(zk, rscale, mp1_center, mp1, ztarg=targets.T,
ifgrad=False, ifhess=False)
assert la.norm(mp1_value - ref_value) / la.norm(ref_value) < 1e-12
# }}}
# {{{ multipole 2
mp2_center = np.array([2, 0])
mp2 = h2dmpmp_vec(zk, rscale, mp1_center, mp1, rscale, mp2_center, nterms)
mp2_value, _, _ = h2dmpeval_vec(zk, rscale, mp2_center, mp2, ztarg=targets.T,
ifgrad=False, ifhess=False)
assert la.norm(mp2_value - ref_value) / la.norm(ref_value) < 3e-5
# }}}
# {{{ local 1
loc1_center = targets_center - np.array([1, 0])
loc1 = h2dmploc_vec(zk, rscale, mp2_center, mp2, rscale, loc1_center, nterms)
loc1_value, _, _ = h2dtaeval_vec(zk, rscale, loc1_center, loc1,
ztarg=targets.T, ifgrad=False, ifhess=False)
assert la.norm(loc1_value - ref_value) / la.norm(ref_value) < 3e-5
# }}}
# {{{ local 2
loc2_center = targets_center
loc2 = h2dlocloc_vec(zk, rscale, loc1_center, loc1, rscale, loc2_center, nterms)
loc2_value, _, _ = h2dtaeval_vec(zk, rscale, loc2_center, loc2, ztarg=targets.T,
ifgrad=False, ifhess=False)
assert la.norm(loc2_value - ref_value) / la.norm(ref_value) < 1e-4
def test_pyfmmlib_fmm(ctx_getter):
logging.basicConfig(level=logging.INFO)
from pytest import importorskip
importorskip("pyfmmlib")
ctx = ctx_getter()
queue = cl.CommandQueue(ctx)
nsources = 3000
ntargets = 1000
dims = 2
dtype = np.float64
helmholtz_k = 2
sources = p_normal(queue, nsources, dims, dtype, seed=15)
targets = (p_normal(queue, ntargets, dims, dtype, seed=18) +
np.array([2, 0]))
sources_host = particle_array_to_host(sources)
targets_host = particle_array_to_host(targets)
from boxtree import TreeBuilder
tb = TreeBuilder(ctx)
tree, _ = tb(queue,
sources,
targets=targets,
max_particles_in_box=30,
debug=True)
from boxtree.traversal import FMMTraversalBuilder
tbuild = FMMTraversalBuilder(ctx)
trav, _ = tbuild(queue, tree, debug=True)
trav = trav.get(queue=queue)
from pyopencl.clrandom import RanluxGenerator
rng = RanluxGenerator(queue, seed=20)
weights = rng.uniform(queue, nsources, dtype=np.float64).get()
#weights = np.ones(nsources)
logger.info("computing direct (reference) result")
from pyfmmlib import hpotgrad2dall_vec
ref_pot, _, _ = hpotgrad2dall_vec(ifgrad=False,
ifhess=False,
sources=sources_host.T,
charge=weights,
targets=targets_host.T,
zk=helmholtz_k)
from boxtree.pyfmmlib_integration import Helmholtz2DExpansionWrangler
wrangler = Helmholtz2DExpansionWrangler(trav.tree, helmholtz_k, nterms=10)
from boxtree.fmm import drive_fmm
pot = drive_fmm(trav, wrangler, weights)
rel_err = la.norm(pot - ref_pot) / la.norm(ref_pot)
logger.info("relative l2 error: %g" % rel_err)
assert rel_err < 1e-5