def __init__(self,
npts=4,
retry=1,
tol=8,
memtol=1,
map=None,
archive=None,
sprayer=None,
seeker=None,
traj=False,
disp=False):
#XXX: better not to use klepto as default? just dict and false cache?
from klepto.archives import dict_archive as _archive
from mystic.solvers import BuckshotSolver
from mystic.solvers import PowellDirectionalSolver
from mystic.pools import SerialPool as Pool
from __builtin__ import map as _map
self.archive = _archive(cached=False) if archive is None else archive
self.sprayer = BuckshotSolver if sprayer is None else sprayer
self.seeker = PowellDirectionalSolver if seeker is None else seeker
self.map = Pool().map if map in (None, _map) else map
self.traj = traj
self.disp = disp
self.npts = npts # number of solvers
self.retry = retry # max consectutive retries w/o a cache 'miss'
self.tol = tol # rounding precision
self.memtol = memtol # memoization rounding precision
self._allSolvers = []
self._inv = False # self-awareness: am I inverted (maximizing)?
return
def __init__(self,
npts=4,
retry=1,
tol=8,
memtol=1,
memsize=None,
map=None,
archive=None,
cache=None,
sprayer=None,
seeker=None,
traj=False,
disp=False,
repeat=0):
"""searcher, which searches for all minima of a response surface
Input:
npts - number of solvers in the ensemble
retry - max consectutive retries w/o a cache 'miss'
tol - rounding precision for the minima comparator
memtol - rounding precision for memoization
memsize - maximum size of cache to hold in memory
map - map used for spawning solvers in the ensemble
archive - the sampled point archive(s)
cache - the trajectory cache(s)
sprayer - the mystic.ensemble instance
seeker - the mystic.solvers instance
traj - if True, save the parameter trajectories
disp - if True, be verbose
repeat - number of times to repeat the search
"""
#XXX: better not to use klepto as default? just dict and false cache?
from klepto.archives import dict_archive as _archive
from mystic.solvers import BuckshotSolver #XXX: or SparsitySolver?
from mystic.solvers import PowellDirectionalSolver
from mystic.pools import SerialPool as Pool
import sys
if (sys.hexversion >= 0x30000f0):
from builtins import map as _map
else:
from __builtin__ import map as _map
del sys
self.archive = _archive(cached=False) if archive is None else archive
self.cache = _archive(cached=False) if cache is None else cache
self.sprayer = BuckshotSolver if sprayer is None else sprayer
self.seeker = PowellDirectionalSolver if seeker is None else seeker
self.map = Pool().map if map in (None, _map) else map
self.traj = traj
self.disp = disp
self.npts = npts # number of solvers
self.retry = retry # max consectutive retries w/o a cache 'miss'
self.repeat = repeat # number of times to repeat the search
self.tol = tol # rounding precision
self.memtol = memtol # memoization rounding precision
self.memsize = memsize # max in-memory cache size
self._allSolvers = []
self._inv = False # self-awareness: am I inverted (maximizing)?
return
solvers = [fmin_powell, fmin, diffev, diffev2]
# test solver one-liners
for solver in solvers:
test_one_liner(solver)
classes = [NelderMeadSimplexSolver, PowellDirectionalSolver, \
DifferentialEvolutionSolver, DifferentialEvolutionSolver2]
# nested solver one-liners (using inner solver class, not nested one-liner)
for solver in classes:
nested = BuckshotSolver
test_nested_solver(nested, solver)
# solver multi-line
for solver in classes:
test_multi_liner(solver)
# nested solver multi-line
for solver in classes:
nested = BuckshotSolver
test_inner_solver(nested, solver)
# solver with mapper
#from pathos.pools import ProcessPool as Pool
from mystic.pools import SerialPool as Pool
map = Pool(5).map
for solver in classes:
nested = BuckshotSolver
test_mapped_solver(nested, solver, map)
# EOF