def infer_inadequate(self, interaction_matrix):
"""
Who needs a description when you can have a picture?
COSY
H-------H
HSQC | | HSQC
C< INAD >C
implies => INAD bond between CC
"""
for i in range(len(interaction_matrix)):
cosy_i = [
x for x, y in enumerate(interaction_matrix[i])
if y == InteractionValues.COSY
]
hsqc_i = [
x for x, y in enumerate(interaction_matrix[i])
if y == InteractionValues.HSQC
]
b_cs = [x for x in cartesian(*[cosy_i, hsqc_i])]
for b_c in b_cs:
b = b_c[0]
c = b_c[1]
d_s = [
x for x, y in enumerate(interaction_matrix[b])
if y == InteractionValues.HSQC
]
b_ds = [x for x in cartesian([c], d_s)]
for b_d in b_ds:
if b_d[0] != b_d[1]:
interaction_matrix[b_d[0]][
b_d[1]] = InteractionValues.INAD
return interaction_matrix
def __init__(self, polarities, neutral, product_table):
self.P = polarities
self.N = neutral
self.table = product_table
self.is_monotone = True
self.is_final = True
self.is_declarative = True
self.is_dynamic = bool(set(self.P) > set(self.N) and self.N)
g = nx.DiGraph()
for a in self.table:
for prod in self.table[a].values():
if a != prod:
g.add_edge(a, prod)
try:
order = nx.topological_sort(g)
if set(self.N) != set(order[-len(self.N):]):
self.is_final = False
except nx.NetworkXUnfeasible:
self.is_monotone = False
self.is_final = False
for a, b, c in cartesian(self.table, repeat=3):
try:
x = self.table[self.table[a][b]][c]
y = self.table[self.table[b][c]][a]
if x != y:
self.is_declarative = False
break
except KeyError:
continue
def plot_stats(nucs, structure, param, violation_padding):
import matplotlib.pyplot as plt
stat_names = {"scale": {}, "violations": {'padding': violation_padding}}
stats = defaultdict(lambda: np.empty((len(nucs), 3), dtype='float'))
fig, axs = plt.subplots(len(stat_names), 1)
for (i, nuc), (stat, kwargs) in cartesian(enumerate(nucs),
stat_names.items()):
with HDFFile(nuc, "r") as f:
param_value = f['structures'][structure]['calculation'].attrs[
param]
if param_value == "particle_sizes":
param_value["particle_sizes"] = param_value["particle_sizes"][
-1]
stat_values = globals()[stat](f, structure, **kwargs)
stats[stat][i] = [
param_value,
np.mean(stat_values),
np.std(stat_values)
]
for ax, (stat_name, data) in zip(axs, stats.items()):
ax.set_ylabel(stat_name)
ax.set_xlabel(param)
data = np.sort(data, axis=0)
ax.errorbar(data.T[0], data.T[1], yerr=data.T[2])
fig.tight_layout()
plt.show()
def characterize_noise(args, adcs_sweep, seed, snr_noise, est_thres_s, ref, match_noise=False):
shape = np.shape(adcs_sweep)
shape = (shape[0], np.size(snr_noise), shape[1],)
real = np.stack((adcs_sweep,)*shape[1], axis=1)
calibrated = np.copy(real)
per_snr = np.zeros(shape)
cal_snr = np.zeros(shape)
nav_snr = np.zeros(shape)
confidence = args.confidence
sigmas = np.sqrt(2)*erfinv(confidence)
# Sweep thres
for idx in cartesian(*tuple(range(s) for s in shape[:-1])):
e_thres_s = est_thres_s[idx[0]]
noise = snr_noise[idx[1]]
idx = idx + (slice(None,),)
lsb_scale = (1 + 2*(e_thres_s*sigmas)) if match_noise else 1
mod_adcs = real[idx].tolist()
aargs = (noise, 0,) if ref else (0, noise,)
calibrated_, per_snr_, cal_snr_, nav_snr_ = \
characterize_point(args, mod_adcs, seed, lsb_scale, *aargs)
calibrated[idx] = np.array(calibrated_, dtype=object)
per_snr[idx] = np.array(per_snr_)
cal_snr[idx] = np.array(cal_snr_)
nav_snr[idx] = np.array(nav_snr_)
return real, calibrated, per_snr, cal_snr, nav_snr
def iter_stages_idx(self):
if len(self.shape) == 0:
yield tuple()
else:
for idx in cartesian(*tuple(range(ss) for ss in self.shape)):
yield idx
def __init__(self, neighbor_order, *args, **kwargs):
super(NthNeighborNodeHistogramCalculator,
self).__init__(*args, **kwargs)
self.labels_to_consider = kwargs.get('labels_to_consider',
self._gnx.graph["node_labels"])
self._num_classes = len(self.labels_to_consider)
self._neighbor_order = neighbor_order
self._relation_types = [
"".join(x) for x in cartesian(*(["io"] * self._neighbor_order))
]
self._print_name += "_%d" % (neighbor_order, )
counter = {i: 0 for i in range(self._num_classes)}
if self._gnx.is_directed():
self._features = {
node:
{rtype: counter.copy()
for rtype in self._relation_types}
for node in self._gnx
}
self._reg_features = {
node: {rtype: []
for rtype in self._relation_types}
for node in self._gnx
}
else:
self._features = {node: counter.copy() for node in self._gnx}
def __init__(self, polarities, neutral, product_table):
self.P = polarities
self.N = neutral
self.table = product_table
self.is_monotone = True
self.is_final = True
self.is_declarative = True
self.is_dynamic = bool(set(self.P) > set(self.N) and self.N)
g = nx.DiGraph()
for a in self.table:
for prod in self.table[a].values():
if a != prod:
g.add_edge(a, prod)
try:
order = nx.topological_sort(g)
if set(self.N) != set(order[-len(self.N):]):
self.is_final = False
except nx.NetworkXUnfeasible:
self.is_monotone = False
self.is_final = False
for a, b, c in cartesian(self.table, repeat=3):
try:
x = self.table[self.table[a][b]][c]
y = self.table[self.table[b][c]][a]
if x != y:
self.is_declarative = False
break
except KeyError:
continue
def characterize_uncertanty(args, adcs_sweep, seed, est_thres_s):
shape = np.shape(adcs_sweep)
shape = (
shape[0],
np.size(est_thres_s),
shape[1],
)
real = np.stack((adcs_sweep, ) * shape[1], axis=1)
calibrated = np.copy(real)
per_snr = np.zeros(shape)
cal_snr = np.zeros(shape)
nav_snr = np.zeros(shape)
confidence = args.confidence
sigmas = np.sqrt(2) * erfinv(confidence)
# Sweep thres
for idx in cartesian(*tuple(range(s) for s in shape[:-1])):
e_thres_s = est_thres_s[idx[1]]
idx = idx + (slice(None, ), )
lsb_scale = 1 + 2 * (e_thres_s * sigmas)
mod_adcs = real[idx].tolist()
calibrated_, per_snr_, cal_snr_, nav_snr_ = \
characterize_point(args, mod_adcs, seed, lsb_scale, 0, 0)
calibrated[idx] = np.array(calibrated_, dtype=object)
per_snr[idx] = np.array(per_snr_)
cal_snr[idx] = np.array(cal_snr_)
nav_snr[idx] = np.array(nav_snr_)
return real, calibrated, per_snr, cal_snr, nav_snr
def sweep_parameters(self, sweep_dicts):
def sweep(dct):
sw_type = dct.get("type", "linear")
if sw_type == "linear":
values = np.linspace(dct["start"], dct["end"], dct["samples"])
elif sw_type == "log":
values = np.logspace(dct["start"], dct["end"], dct["samples"])
else:
raise ValueError(
"sweep type {} not recognized.".format(sw_type))
def gen_dict(value):
copy_keys = (
"parameter",
"index",
)
result = {key: dct[key] for key in copy_keys}
result["value"] = value
return result
return [gen_dict(value) for value in values]
values_axes = tuple(sweep(dct) for dct in sweep_dicts)
shape = tuple(len(axis) for axis in values_axes)
ins = np.zeros(shape + self.shape, dtype=int)
stages = np.zeros(shape + self.shape, dtype=object)
for idx in cartesian(*tuple(range(ss) for ss in shape)):
val = tuple(vals[ii] for ii, vals in zip(idx, values_axes))
this_stages = copy.deepcopy(self.stages)
in_ = np.array(self.ins)
new_val = tuple()
for vall in val:
if vall["parameter"] == "test":
in_[(
Ellipsis,
vall["index"],
slice(None),
)] = vall["value"]
else:
new_val = (vall, ) + new_val
for sub_idx in this_stages.iter_idx():
this_stages[idx] = this_stages[idx].create_modified(new_val)
ins[idx + (Ellipsis, )] = in_
stages[idx + (Ellipsis, )] = this_stages
return values_axes, StageTestbench(stages.tolist(),
ins,
self.configuration_sequence,
shape=shape + self.shape)
def iter_idx(self):
if self.dims == 0:
yield tuple()
else:
for idx in cartesian(*tuple(
range(ss) for ss in np.shape(self.data))):
yield idx
def ambiguous(self):
from itertools import product as cartesian
try:
new_bases = self.ambiguous_bases.get(self[0], (self[0], ))
except IndexError:
yield Sequence('')
else:
yield from map(
Sequence('').join, cartesian(new_bases, self[1:].ambiguous))
def location_set(q):
# location_set is an iterative which enumerates the elements of the rectangle [0,q) (in Z^d)
# the map is essentially just q-adic representation of the numbers [0,prod(q))
if type(q) == int:
q = (q, )
#return [tuple(reversed(k)) for k in cartesian(*[range(qj) for qj in reversed(q)])];
return [
tuple(reversed(k))
for k in cartesian(*[range(qj) for qj in reversed(q)])
]
def characterize(args, adcs, seed):
with misc.push_random_state():
np.random.seed(seed)
seeds = [np.random.randint(0, 4294967296) for _ in range(3)]
assert not args.relative_snr_ref, "TODO implement relative"
assert not args.relative_snr_thres, "TODO implement relative"
min_snr_ref_v = args.min_snr_ref_v
min_snr_thres_v = args.min_snr_thres_v
fsr = adcs[0].stages[0].meta.fsr
n_bits = int(np.sum([np.floor(stage.meta.n_bits) for stage in adcs[0].stages]))
n_bits += int(gen.infer_thres_bits(adcs[0].tail)[0])
lsb = gen.compute_lsb(n_bits, *fsr)
if min_snr_ref_v is None:
min_snr_ref_v = lsb/2
if min_snr_thres_v is None:
min_snr_thres_v = lsb/2
snr_ref_inv = np.linspace(0, min_snr_ref_v, args.samples_snr_ref)
snr_thres_inv = np.linspace(0, min_snr_thres_v, args.samples_snr_thres)
snr_ref = np.power((fsr[1] - fsr[0]), 2)/np.power(snr_ref_inv, 2)
snr_thres = np.power((fsr[1] - fsr[0]), 2)/np.power(snr_thres_inv, 2)
snr_ref[0] = 0
snr_thres[0] = 0
real_thres_s = data.at_least_ndarray(args.real_thres_s)
shape = (np.size(real_thres_s), np.size(adcs),)
adcs_sweep = np.tile(np.array(adcs, dtype=object), shape[:-1] + (1,))
# Sweep thres
with misc.push_random_state():
np.random.seed(seed)
for idx in cartesian(*tuple(range(s) for s in shape)):
r_thres_s = real_thres_s[idx[0]]
adcs_sweep[idx] = copy.deepcopy(adcs_sweep[idx])
for stage in adcs_sweep[idx].stages:
r_thres_s_local = r_thres_s * stage.meta.lsb
stage._thres = np.random.normal(stage.thres, r_thres_s_local)
adcs_sweep[idx]._tail = np.random.normal(adcs_sweep[idx].tail, r_thres_s)
uncertain = characterize_uncertanty(args, adcs_sweep, seed, real_thres_s)
ref = characterize_noise(args, adcs_sweep, seed, snr_ref, real_thres_s, ref=True)
thres = characterize_noise(args, adcs_sweep, seed, snr_thres, real_thres_s, ref=False)
return uncertain, ref, thres
def array_location_set(k):
# returns an iterative which enumerates the indices of a matrix from the bottom left corner, reading left->right, bottom->top, consistent with R^n enumeration
# the 2nd to last coordinate needs to the x-axis, the last coordinate is the y-axis but counts backwards from num_rows
if type(k) == int:
return [(j, ) for j in range(k)]
elif len(k) == 1:
return [(j, ) for j in range(k[0])]
else:
return [
tuple(reversed((j[-1], ) + (-j[-2] - 1, ) + j[:-2]))
for j in cartesian(*[range(kj) for kj in reversed(k)])
]
def generate(a, b):
grammar = a.grammar
a, b = [x.node.values() for x in [a, b]]
pairs = filter_non_viable(cartesian(a, b), grammar)
intra = map(lambda s: cartesian(s, repeat=2), (a, b))
intra = filter_non_viable(chain(*intra), grammar)
viable_pairs = set(map(frozenset, pairs + intra))
stacks = chain.from_iterable(
combinations(pairs, n) for n in range(1, len(pairs)+1))
implic = tuple(chain(*(implications(p, grammar) for p in pairs)))
for stack in map(list, stacks):
stack_ = set(map(frozenset, stack))
if any(stack_ & {x, i} == {x} for x, i in implic):
continue
if any(not viable_pairs & {s} for s in indirect(stack)):
continue
yield stack
def __init__(self, *args, **kwargs):
super(DirectedNthNeighborNodeHistogramCalculator,
self).__init__(*args, **kwargs)
counter = {i: 0 for i in range(self._num_classes)}
self._print_name += "_%d" % (self._neighbor_order, )
self._relation_types = [
"".join(x) for x in cartesian(*(["io"] * self._neighbor_order))
]
self._features = {
node: {rtype: counter.copy()
for rtype in self._relation_types}
for node in self._gnx
}
def generate(a, b):
grammar = a.grammar
a, b = [x.node.values() for x in [a, b]]
pairs = filter_non_viable(cartesian(a, b), grammar)
intra = map(lambda s: cartesian(s, repeat=2), (a, b))
intra = filter_non_viable(chain(*intra), grammar)
viable_pairs = set(map(frozenset, pairs + intra))
stacks = chain.from_iterable(
combinations(pairs, n) for n in range(1,
len(pairs) + 1))
implic = tuple(chain(*(implications(p, grammar) for p in pairs)))
for stack in map(list, stacks):
stack_ = set(map(frozenset, stack))
if any(stack_ & {x, i} == {x} for x, i in implic):
continue
if any(not viable_pairs & {s} for s in indirect(stack)):
continue
yield stack
def letterCombinations(self, digits: str) -> List[str]:
if not digits:
return []
phone = {
'2': 'abc',
'3': 'def',
'4': 'ghi',
'5': 'jkl',
'6': 'mno',
'7': 'pqrs',
'8': 'tuv',
'9': 'wxyz'
}
return [''.join(i) for i in cartesian(*[phone[d] for d in digits])]
def __init__(self, neighbor_order, *args, **kwargs):
super(NthNeighborNodeEdgeHistogramCalculator,
self).__init__(*args, **kwargs)
# self._num_classes = len(self._gnx.graph["node_labels"])
self._neighbor_order = neighbor_order
self._relation_types = [
"".join(x) for x in cartesian(*(["io"] * self._neighbor_order))
]
self._print_name += "_%d" % (neighbor_order, )
counter = {i: 0 for i in self._gnx.graph["edge_labels"]}
self._features = {
node: {rtype: counter.copy()
for rtype in self._relation_types}
for node in self._gnx
}
def __call__(self, data):
supported_types = set([OPINION_BROKEN.short, OPINION_ISSUE.short, OPINION_PRAISE.short])
for key, value in recombined(data):
self.comments_in += 1
m_id, ts, type, product, version, platform, locale, manufacturer, device, url, message = value.split(
"\t", 10
)
if not url or type not in supported_types:
continue
app = "<%s>" % product
site = normalize_url(url)
out_keys = cartesian((version,), (site,), (app, platform, None), (type,))
out_value = (m_id, message)
self.comments_out += 1
for out_key in out_keys:
yield (out_key, out_value)
def __call__(self, data):
supported_types = set(
[OPINION_BROKEN.short, OPINION_ISSUE.short, OPINION_PRAISE.short])
for key, value in recombined(data):
self.comments_in += 1
m_id, ts, type, product, version, platform, locale, \
manufacturer, device, url, message = value.split('\t', 10)
if not url or type not in supported_types: continue
app = '<%s>' % product
site = normalize_url(url)
out_keys = cartesian((version, ), (site, ), (app, platform, None),
(type, ))
out_value = (m_id, message)
self.comments_out += 1
for out_key in out_keys:
yield (out_key, out_value)
def prepickle_data(dat, path_context, directory, name):
real, calibrated, per_snr, cal_snr, nav_snr = dat
real_str = np.empty_like(real)
calibrated_str = np.empty_like(calibrated)
for idx in cartesian(*tuple(range(s) for s in np.shape(real))):
lname = "{}[{}]".format(name, ','.join(str(ii) for ii in idx))
real_location = data.DataLocation(path_context, directory, lname + ".real", ".json")
real_str[idx] = data.save(real[idx], real_location)
calibrated_location = data.DataLocation(path_context, directory, lname + ".calibrated", ".json")
calibrated_str[idx] = data.save(calibrated[idx], calibrated_location)
return real_str, calibrated_str, per_snr, cal_snr, nav_snr
def decompose(self):
def sum_eq(a, b, c):
return tuple(map(add, a, b)) == c
length = ceil(log(len(self.P), 2))
configs = cartesian([0, 1], repeat=length)
prob = constraint.Problem()
prob.addVariables(self.P, list(configs))
prob.addConstraint(constraint.AllDifferentConstraint())
for p1 in self.table:
for p2 in self.table[p1]:
prob.addConstraint(sum_eq, (p1, p2, self.table[p1][p2]))
return prob.getSolution()
def decompose(self):
def sum_eq(a, b, c):
return tuple(map(add, a, b)) == c
length = ceil(log(len(self.P), 2))
configs = cartesian([0, 1], repeat=length)
prob = constraint.Problem()
prob.addVariables(self.P, list(configs))
prob.addConstraint(constraint.AllDifferentConstraint())
for p1 in self.table:
for p2 in self.table[p1]:
prob.addConstraint(sum_eq, (p1, p2, self.table[p1][p2]))
return prob.getSolution()
def spacedRotations(D, N):
from math import pi, sin, cos, sqrt
from itertools import product as cartesian, repeat
from .util import frange
if D == 2:
yield from zip(frange(-pi, pi, 2*pi/N))
elif D == 3:
# Ken Shoemake
# Graphics Gems III, pp 124-132
from .quaternion import Quaternion
for X, *theta in cartesian(frange(0, 1, 1/N), *repeat(frange(0, 2*pi, 2*pi/N), 2)):
R = (sqrt(1-X), sqrt(X))
yield Quaternion(sin(theta[0]) * R[0], cos(theta[0]) * R[0],
sin(theta[1]) * R[1], cos(theta[1]) * R[1]).axis_angle
else:
raise NotImplementedError("Only defined for D in [2..3], not {}"
.format(D))
def postpickle_data(dat, path_context):
if dat is None:
return None
real_str, calibrated_str, per_snr, cal_snr, nav_snr = dat
real = np.empty_like(real_str)
calibrated = np.empty_like(calibrated_str)
memo = {}
for idx in cartesian(*tuple(range(s) for s in np.shape(real_str))):
real_path = real_str[idx]
real_dl = data.DataLocation.Parse(real_path, path_context, None_on_fail=False)
real[idx] = data.load(gen.PipeParameters, real_dl, memo)
calibrated_path = calibrated_str[idx]
calibrated_dl = data.DataLocation.Parse(calibrated_path, path_context, None_on_fail=False)
calibrated[idx] = data.load(gen.PipeParameters, calibrated_dl, memo)
return real, calibrated, per_snr, cal_snr, nav_snr
def spacedRotations(D, N):
from math import pi, sin, cos, sqrt
from itertools import product as cartesian, repeat
from .util import frange
if D == 2:
yield from zip(frange(-pi, pi, 2 * pi / N))
elif D == 3:
# Ken Shoemake
# Graphics Gems III, pp 124-132
from .quaternion import Quaternion
for X, *theta in cartesian(frange(0, 1, 1 / N),
*repeat(frange(0, 2 * pi, 2 * pi / N), 2)):
R = (sqrt(1 - X), sqrt(X))
yield Quaternion(
sin(theta[0]) * R[0],
cos(theta[0]) * R[0],
sin(theta[1]) * R[1],
cos(theta[1]) * R[1]).axis_angle
else:
raise NotImplementedError(
"Only defined for D in [2..3], not {}".format(D))
from itertools import islice, product as cartesian, starmap
from coherent_point_drift.geometry import rigidXform, rotationMatrix, RMSD, randomRotations
from coherent_point_drift.align import globalAlignment
import matplotlib.pyplot as plt
from collections import defaultdict
import numpy as np
fig, axs = plt.subplots(2, 1)
for ndim, ax in zip((2, 3), axs):
ax.set_title("{}D".format(ndim))
nstepss = range(1, 8)
pointss = np.random.random((10, 12, ndim))
rotations = starmap(rotationMatrix, islice(randomRotations(ndim), 10))
rmsds = defaultdict(list)
for nsteps, points, rotation in cartesian(nstepss, pointss, rotations):
degraded = rigidXform(points, R=rotation)
xform = globalAlignment(points, degraded, w=0.1, nsteps=nsteps)
rmsds[nsteps].append(RMSD(points, rigidXform(degraded, *xform)))
labels = sorted(rmsds.keys())
ax.violinplot([rmsds[i] for i in labels], labels)
fig.tight_layout()
plt.show()
def test_matrix_alignment(self):
block = InterfaceBlock('Foo', layout=BlockLayout.std140)
for r, c in cartesian(range(2, 5), repeat=2):
mat = 'mat{}x{}'.format(r, c)
i = InterfaceBlockMember(block, 'foo', mat)
self.assertEqual(i.alignment, 16)
def __init__(self,
stages,
ins,
configuration_sequence,
shape=None,
data_location=None):
super().__init__(data_location)
NestedStages = data.nested_lists_of(gen.StageParameters)
NestedSequences = data.nested_lists_of(gen.ConfigurationSequence)
if not isinstance(configuration_sequence, NestedSequences):
conf_shape = np.shape(configuration_sequence)
configuration_sequence = NestedSequences(configuration_sequence,
len(conf_shape))
conf_shape = np.shape(configuration_sequence.data)
if not isinstance(stages, NestedStages):
shape = default(shape, np.shape(stages))
dims = len(shape)
cur = stages
root_arr = stages
# Check a valid 0,0,0...
for dd in range(dims):
assert isinstance(cur, (
tuple,
list,
))
cur = cur[0]
# Check elements
for idx in cartesian(*tuple(range(ss) for ss in shape)):
assert isinstance(misc.getitem(stages, idx),
gen.StageParameters)
# Check regularity
def rec_check(lst, shape):
valid = (not isinstance(lst, list)
and len(shape) == 0) or len(lst) == shape[0]
if len(shape) > 1:
sub_shape = shape[1:]
valid = valid and all(
[rec_check(llst, sub_shape) for llst in lst])
return valid
assert rec_check(stages, shape)
else:
stages_shape = np.shape(stages.data)
shape = default(shape, stages_shape)
assert shape == stages_shape
dims = len(shape)
root_arr = stages.data
ref_element = misc.getitem(root_arr, (0, ) * dims)
ins = data.at_least_ndarray(ins)
if len(np.shape(ins)) == 1:
ins = ins[..., np.newaxis]
# Broadcast ins
if len(np.shape(ins)) == 2:
ins = ins[(np.newaxis, ) * dims + (Ellipsis, )]
ins = np.tile(ins, shape + (
1,
1,
))
assert len(np.shape(ins)) == dims + 2
if np.size(ins, -1) != ref_element.meta.n_diff:
cm = ref_element.meta.common_mode
ins = np.concatenate((
cm - ins,
cm + ins,
), axis=1)
# All meta the same
for idx in cartesian(*tuple(range(ss) for ss in shape)):
c_element = misc.getitem(root_arr, idx)
assert c_element.meta == ref_element.meta
self._stages = NestedStages.EnsureIsInstance(stages)
self._shape = shape
self._ins = ins
self._configuration_sequence = configuration_sequence
output_file_name_base_str = "output/" + input_file_name[
input_file_name.rfind("/") + 1:input_file_name.index(".wav")] + "/"
try:
mkdir(path=output_file_name_base_str)
except (FileExistsError):
pass
window_sizes = [int(sys.argv[2])
] #[16] #[4, 8, 16] #[int(sys.argv[2])] #[4, 8, 16]
max_depth_sizes = [int(sys.argv[3])
] #[16] #[8, 16] #[int(sys.argv[3])] #[16, 4, 8, 32]
random_state_sizes = [0] #[2, 1, 0]
num_estimators_sizes = [int(sys.argv[4])] #[100, 50, 200, 500]
for window_size, MAX_DEPTH, RANDOM_STATE, N_ESTIMATORS in cartesian(
window_sizes, max_depth_sizes, random_state_sizes,
num_estimators_sizes):
print("\n\n-----NEW TRIAL-----")
print((window_size, MAX_DEPTH, RANDOM_STATE, N_ESTIMATORS))
gc.collect()
output_file_name_restored = output_file_name_base_str + "w_" + str(
window_size) + "_c_di_do_md_" + str(MAX_DEPTH) + "_rs_" + str(
RANDOM_STATE) + "_ne_" + str(
N_ESTIMATORS
) + ".wav" #distorted input, distorted output, center window
#print(output_file_name_restored)
# c = 0
# print(output_file_name_restored[:output_file_name_restored.index(".wav")] + "_chan_" + str(c) + ".rf")
# exit()
def iter_idx(self):
return cartesian(self.iter_conf_idx(), self.iter_stages_idx())
def calib(meta, args, interlace, use_full_range=None):
n_caps = meta.n_caps
n_refs = meta.n_refs
n_diff = meta.n_diff
n_cs = (n_caps - 1) // 2
n_cf = n_caps - n_cs
use_full_range = misc.default(use_full_range, n_cs < 2)
ds_samples = args.samples
if args.n_test > 0:
raise ValueError("Minimal does not support test inputs.")
comb_cs = misc.iterate_combinations(n_caps, n_cs)
if args.full:
comb_cs = [tuple(misc.iterate_permutations(cs)) for cs in comb_cs]
comb_cs = [elem for tlp in comb_cs for elem in tlp]
slice_ = slice(None) if use_full_range else slice(1, -1)
comb_refs = gen.ds_map(n_cs, n_refs, n_cs * (n_refs - 1) + 1)
comb_refs = np.transpose(comb_refs[:, slice_], (
1,
0,
2,
))
comb_refs = comb_refs.tolist()
comb_refs = [(
comb_refs[ii],
comb_refs[jj],
) for ii in range(len(comb_refs)) for jj in range(ii + 1, len(comb_refs))]
comb_cs = list(comb_cs)
comb_refs = list(comb_refs)
even_configs = []
even_ins = []
ics = []
with misc.push_random_state():
seed = None if args.seed is None else int(args.seed)
np.random.seed(seed)
for cs_ii, refs_ii in cartesian(comb_cs, comb_refs):
even_configs.append(gen.Configuration(meta, cs_ii))
top_ii, bot_ii = refs_ii
if args.inputs == "":
sub_seed = np.random.randint(0, 4294967296)
even_ins.append(
gen.InternalRandom(meta, np.size(cs_ii), sub_seed))
else:
top = gen.InternalDC(meta, top_ii)
bot = gen.InternalDC(meta, bot_ii)
even_ins.append(gen.ACCombinator(meta, top, bot, args.period))
inv = [[n_refs - 1 - iii for iii in ii] for ii in top_ii]
inv = inv + [[n_refs // 2, n_refs - n_refs // 2][:n_diff]
] * (n_cf - n_cs)
ics.append(gen.InitialCondition(meta, inv))
if interlace:
n_cs_h = n_cs // 2
assert n_cs_h > 0, "Not enough capacitors to decrease bits."
odd_configs = []
odd_ins = []
for conf, in_ in zip(even_configs, even_ins):
left = (n_cs - n_cs_h) // 2
cs_range = range(left, left + n_cs_h)
mask = np.zeros((n_cs, ), dtype=bool)
mask[cs_range] = 1
odd_configs.append(
gen.Configuration(conf.meta, conf.cs[cs_range, :]))
odd_ins.append(gen.InputMask(in_.meta, in_, mask))
else:
odd_ins = even_ins
odd_configs = even_configs
conf_sets = []
parity = 0
for samples in ds_samples:
if parity == 0:
configs = even_configs
inputs = even_ins
else:
configs = odd_configs
inputs = odd_ins
conf_sets.append(gen.ConfigurationSet(samples, inputs, configs))
parity = (parity + 1) % 2
if args.ic == "clear":
ics = [gen.InitialCondition.Discharged(meta, n_cf)] * len(odd_ins)
elif args.ic == "precharge":
pass
else:
raise ValueError("ic type {} not supported".format(args.ic))
return gen.ConfigurationSequence(ics, conf_sets * args.loop)
def __init__(self, value):
self.prefix = self.__prefixes__[self.name]
self.machine_type = self.__machine_types__[self.name]
self.scalar_type = self
self.opaque = False
scalar_doc = Scalar.__doc__
Scalar = Enum('Scalar', ((s, s) for s in scalar_types), type=Scalar)
Scalar.__doc__ = scalar_doc
floating_point_scalars = { Scalar.float, Scalar.double }
sampler_dims = range(1, 4)
sampler_data_types = {Scalar.float, Scalar.int, Scalar.uint}
sampler_types = [ "{}sampler{}D".format(scalar_type.prefix, ndim)
for scalar_type, ndim in cartesian(sampler_data_types, sampler_dims) ]
class Sampler(str, BasicType, Enum):
'''The GLSL sampler types.
Scalars difine the following attributes:
*opaque*
Whether the datatype is an opaque type (:py:obj:`True`)
'''
__ndims__ = { "{}sampler{}D".format(scalar_type.prefix, ndim): ndim
for scalar_type, ndim in cartesian(sampler_data_types, sampler_dims) }
def __init__(self, value):
self.ndim = self.__ndims__[self.name]
self.opaque = True
sampler_doc = Sampler.__doc__
def recreate(eff,
caps,
refs,
thres,
ins,
common_mode,
c_seq,
cache,
scalar=None,
limit_samples=None):
fun = sims.Simulator
scalar = default(scalar, len(np.shape(eff)) == 0)
limit_samples = default(limit_samples, c_seq.samples) + 1
samples = 0
cache = dict(cache)
cache["data"] = fun.simulate_setup(eff,
caps,
refs,
thres,
ins,
common_mode,
c_seq,
scalar=scalar)
data = cache["data"]
seq_idx = cache["seq_idx"]
u_history = []
# shape (..., n_conf, n_diff,)
u = fun.init_seq(seq_idx, data)
u_history.append(u)
samples += 1
# Simulate each configuration set
dct_idx, dct_ext, dct_n = get(data, "indexing", "extended", "n")
n_conf, n_diff = get(dct_n, "n_conf", "n_diff")
eff, cm = get(dct_ext, "eff", "cm")
base_shape, base_len = get(dct_idx, "base_shape", "base_len")
c_sets = c_seq.configuration_sets
def multi_idx_filter(idx_tuple, sub_idx):
return tuple(ii[sub_idx] if hasattr(ii, "__getitem__") else ii
for ii in idx_tuple)
for ii_set, c_set, data_trans_du in zip(range(len(c_sets)), c_sets,
cache["sets"]):
set_data, trans_idx, du_idx = data_trans_du
if set_data["previous"] is not None:
u = fun.transition_step(trans_idx, u[-1:, ...], set_data, data)
u_history.append(u)
samples += 1
r_du = (1 - eff) * cm
n_u = np.empty((c_set.ds_samples, ) + base_shape + (
n_conf,
n_diff,
))
local_du_idx = dict(du_idx)
# used in u
du = fun.du_compute(du_idx, c_set, set_data, data)
for idx in cartesian(*tuple(range(ss) for ss in base_shape)):
local_idx = tuple(slice(ii, ii + 1)
for ii in idx) + (Ellipsis, )
ext_local_idx = (slice(None), ) + local_idx
local_du_idx["r_ref"] = local_du_idx["r_ref"][ext_local_idx]
local_du_idx["m_in_ref"] = multi_idx_filter(
local_du_idx["m_in_ref"], ext_local_idx)
local_du_idx["m_in_ins"] = multi_idx_filter(
local_du_idx["m_in_ins"], ext_local_idx)
zi = u[(slice(-1, None), ) + local_idx] * eff[local_idx]
local_u = lfilter([1], [1, -eff[local_idx].item()],
du[ext_local_idx] +
r_du[(np.newaxis, ) + local_idx],
zi=zi,
axis=0)[0]
n_u[ext_local_idx] = local_u
u = n_u
u_history.append(u)
samples += np.size(u, 0)
if samples >= limit_samples:
break
u_history = np.concatenate(u_history, axis=0)
u_history = u_history[:limit_samples, ...]
if scalar:
assert np.size(u_history, 1) == 1
u_history = u_history[:, 0, ...]
return u_history
def build_report(ideal, real, ideal_title="IDEAL", real_title="REAL"):
assert ideal.meta == real.meta
title_len = 24
index_len = 6
ideal_len = 10
real_len = 10
error_len = 10
def title(header):
return "- {} ".format(header) + '-' * max(title_len - 3 - len(header),
0)
def index(idx):
assert len(idx) > 0
if len(idx) == 1:
return "{}".format(idx)
else:
return "({})".format(','.join([str(ii) for ii in idx]))
head = " {{:^{}}} {{:^{}}} {{:^{}}} {{:^{}}}".format(
index_len, ideal_len, real_len, error_len)
line = " {{:^{}}} {{:{}.{}f}} {{:{}.{}f}} {{:{}.{}f}}".format(
index_len, ideal_len, ideal_len - 3, real_len, real_len - 3, error_len,
error_len - 3)
cm, eff, cap, cs_CF, ref = compare_adcs(ideal, real)
result = []
result.append(title("CHARGE TRANSFER"))
result.append(head.format("INDEX", ideal_title, real_title, "ERROR (%)"))
for idx in cartesian(*tuple(range(ss) for ss in np.shape(eff))):
result.append(line.format('-', ideal.eff, real.eff, 100 * eff))
result.append('')
result.append(title("COMMON MODE"))
result.append(head.format("INDEX", ideal_title, real_title, "ERROR (LSB)"))
for idx in cartesian(*tuple(range(ss) for ss in np.shape(cm))):
r_cm = real.common_mode[idx]
i_cm = ideal.common_mode[idx]
result.append(line.format('-', i_cm, r_cm, cm[idx]))
result.append('')
result.append(title("CAPACITOR"))
result.append(head.format("INDEX", ideal_title, real_title, "ERROR (%)"))
for idx in cartesian(*tuple(range(ss) for ss in np.shape(cap))):
r_cap = real.caps[idx]
i_cap = ideal.caps[idx]
result.append(line.format(index(idx), i_cap, r_cap, 100 * cap[idx]))
result.append('')
result.append(title("CAPACITOR (Cs/CF)"))
result.append(head.format("INDEX", ideal_title, real_title, "ERROR (%)"))
for idx in cartesian(*tuple(range(ss) for ss in np.shape(cs_CF))):
r_cap = real.caps[idx] / np.sum(real.caps)
i_cap = ideal.caps[idx] / np.sum(ideal.caps)
result.append(line.format(index(idx), i_cap, r_cap, 100 * cs_CF[idx]))
result.append('')
result.append(title("REFERENCE"))
result.append(head.format("INDEX", ideal_title, real_title, "ERROR (LSB)"))
for idx in cartesian(*tuple(range(ss) for ss in np.shape(ref))):
r_ref = real.refs[idx]
i_ref = ideal.refs[idx]
result.append(line.format(index(idx), i_ref, r_ref, ref[idx]))
return '\n'.join(result)
return np.append(N, min(1, N[-1]))
N = list(reversed(get_N(I)))
#print( N);print(111)
return linregress(range(len(N)), list(map(log2, N)))[0]
zoom_in = lambda x, scale: scale * (x % (1. / scale))
def invert(A, t, mode='<='):
if mode == '<=': return (A <= t).astype(int)
else: return (A >= t).astype(int)
matrix_range = lambda m, n: map(np.array, cartesian(range(m), range(n)))
indices = lambda shape: cartesian(range(shape[0]), range(shape[1]))
pad0 = lambda a, n: np.pad(a, (0, n), mode='constant', constant_values=0)
def make_fractal_image(S, n, stopshort=False):
img = S
listS = S.tolist()
values = set.union(*(set(row) for row in listS))
if stopshort:
m = n // len(S)
else:
m = n
def products(max_val):
for a, b in cartesian(multiplicands(max_val), repeat=2):
yield a * b
