def KS_two_samp_test(ref_vals, sim_vals, alpha):
assert not np.any(np.isnan(ref_vals))
assert not np.any(np.isnan(sim_vals))
assert not np.any(np.isnan([alpha]))
assert ref_vals.ndim == 1
assert sim_vals.ndim == 1
assert isinstance(alpha, float)
assert 0 < alpha < 1
ref_vals = ref_vals.copy()
sim_vals = sim_vals.copy()
ref_vals.sort()
sim_vals.sort()
n_ref = ref_vals.shape[0]
n_sim = sim_vals.shape[0]
ref_probs = rankdata(ref_vals) / (n_ref + 1)
sim_probs = rankdata(sim_vals) / (n_sim + 1)
c_alpha = (-0.5 * mlog((1 - alpha) * 0.5))**0.5
ks_lim = c_alpha * (((n_ref + n_sim) / (n_ref * n_sim))**0.5)
sim_probs_in_ref = np.interp(sim_vals, ref_vals, ref_probs)
max_diff = max(np.abs(sim_probs_in_ref - sim_probs))
if max_diff > ks_lim:
ret_val = 'reject'
else:
ret_val = 'accept'
return ret_val
def _make_graph(self):
phone_book = [[] for _ in range(self.world_size)]
for rank in range(self.world_size):
group = phone_book[rank]
for i in range(0, int(mlog(self.world_size - 1, 2)) + 1):
if i == 0:
f_peer = self._rotate_forward(rank, 1)
b_peer = self._rotate_backward(rank, 1)
else:
f_peer = self._rotate_forward(rank, 1 + 2 ** i)
b_peer = self._rotate_backward(rank, 1 + 2 ** i)
# create directory for non-passive peers
if not self.is_passive(rank) and (
self.is_passive(f_peer) and self.is_passive(b_peer)):
if f_peer not in group:
group.append(f_peer) # forward peer...
if b_peer not in group:
group.append(b_peer) # then backward peer
# create directory for passive peers
elif self.is_passive(rank) and (
not (self.is_passive(f_peer) or self.is_passive(b_peer))):
if b_peer not in group:
group.append(b_peer) # backward peer...
if f_peer not in group:
group.append(f_peer) # then forward peer
return phone_book
def sqNear0(lf):
assert(lf>=0)
d=int(mlog(lf,10))
r0=10**((d)//2)
fl=lf*10**(2*10)//r0**2
fl=msqrt(fl)
r1=r0*int(fl*10**10)//10**20
return(r1)
def logLR ( x, y ):
"""Calculate log10 Likelihood between H1 ( enriched ) and H0 (
chromatin bias ). Then set minus sign for depletion.
"""
key_value = ( x, y )
if logLR_dict.has_key(key_value):
return logLR_dict[key_value]
else:
if x > y:
s = (x*(mlog(x+1)-mlog(y+1))+y-x)*LOG10_E
elif x < y:
s = (-1*x*(mlog(x+1)-mlog(y+1))-y+x)*LOG10_E
else:
s = 0
logLR_dict[key_value] = s
return s
def log_dd(r, e):
'''Compute the log.
'''
rr, re = mlog(r), 0.0
ur, ue = exp_dd(rr, re)
tmpr, tmpe = add_dd(ur, ue, -r, -e)
denr, dene = add_dd(ur, ue, r, e)
tmpr, tmpe = div_dd(tmpr, tmpe, denr, dene)
tmpr, tmpe = mul_dd(2.0, 0.0, tmpr, tmpe)
return add_dd(rr, re, -tmpr, -tmpe)
def _make_graph(self):
for rank in range(self.world_size):
for i in range(
0,
int(mlog(self.world_size - 1, self._peers_per_itr + 1)) +
1):
for j in range(1, self._peers_per_itr + 1):
distance_to_neighbor = j * ((self._peers_per_itr + 1)**i)
f_peer = self._rotate_forward(rank, distance_to_neighbor)
self._add_peers(rank, [f_peer])
def _make_graph(self):
phone_book = [[] for _ in range(self.world_size)]
for rank in range(self.world_size):
group = phone_book[rank]
for i in range(0, int(mlog(self.world_size - 1, 2)) + 1):
f_peer = self._rotate_forward(rank, 2 ** i)
if f_peer not in group:
group.append(f_peer)
b_peer = self._rotate_backward(rank, 2 ** i)
if b_peer not in group:
group.append(b_peer)
return phone_book
def entropy(self, Y):
if Y.shape[0] == 0:
return 0
if all(Y == Y[0]):
return 0
labels = np.unique(Y)
label_probs = []
for l in labels:
label_probs.append(np.mean(Y == l))
entropy = 0
for p in label_probs:
entropy -= p * mlog(p, len(labels))
return entropy
def _make_graph(self):
for rank in range(self.world_size):
for i in range(0, int(mlog(self.world_size - 1, 2)) + 1):
if i == 0:
f_peer = self._rotate_forward(rank, 1)
b_peer = self._rotate_backward(rank, 1)
else:
f_peer = self._rotate_forward(rank, 1 + 2**i)
b_peer = self._rotate_backward(rank, 1 + 2**i)
# create directory for non-passive peers
if not self.is_passive(rank) and (self.is_passive(f_peer)
and self.is_passive(b_peer)):
self._add_peers(rank, [f_peer, b_peer])
# create directory for passive peers
elif self.is_passive(rank) and (not (self.is_passive(f_peer) or
self.is_passive(b_peer))):
self._add_peers(rank, [f_peer, b_peer])
def save_image(self, binary, extension, watermark):
"""
Save an image with given extension to the cache directory of the object
Keyword arguments:
binary: the binary image (raw from the GET request)
"""
max_num_files = int(mlog(self.cache_size, 10)//1 + 1)
filename_mini = "image_%0*d" % (max_num_files, self.image_counter)
filename_path = self.cache_dir + filename_mini
filename = filename_path + "." + extension
filename_orig = filename_path + "_orig." + extension
self.logger.info("Saving to %s", filename)
for to_remove in glob.glob(filename_path+"*.*"):
os.remove(to_remove)
image = Image.open(BytesIO(binary))
image_resized = resize_image(image, self.resolution)
self.logger.info("Resizing to resolution %s", self.resolution)
image_watermarked = add_watermark(
image_resized,
watermark,
self.resolution)
try:
image_watermarked.save(filename)
image.save(filename_orig)
except FileNotFoundError:
self.logger.critical("Impossible to save at location %s, "
"ensure the folder exists.", filename)
return False
except PermissionError:
self.logger.critical("Permission to write at location %s "
"was denied, ensure you have correct"
" rights.", filename)
return False
self.image_counter += 1
if self.image_counter == self.cache_size:
self.logger.info("Reached max cache size (%s), "
"starting to overwrite",
self.cache_size)
self.image_counter = 0
return filename
def isFibonacci(num):
from math import log as mlog
a = 2.07684408521711 * mlog(2.237 * num)
return (a % 1 * 100 > 90) or (a % 1 * 100 < 5)
def ll_normal_es_py(o,m,e):
"""Normal log likelihood for scalar average standard deviation."""
npt = o.size
return -npt*mlog(e) -0.5*npt*LOG_TWO_PI - 0.5*square(o-m).sum()/e**2
def sqNear1(lf):
le=int(mlog(lf,10)//2)*2
if le <100:
return(int(msqrt(lf)))
a=lf//10**(le-100)
return(int(msqrt(a))*10**((le-100)//2))
def get_k_c(kmer_max: int, count_max: int) -> Tuple[int, int]:
round_k = lambda x: 2**(int(ceil(mlog(2 * x) / mlog(2))))
round_c = lambda x: ceil(mlog(x + 1, 2))
return round_k(kmer_max), round_k(round_c(count_max) / 2)
def ll_normal_es_py(o, m, e):
"""Normal log likelihood for scalar average standard deviation."""
npt = o.size
return -npt * mlog(e) - 0.5 * npt * LOG_TWO_PI - 0.5 * square(
o - m).sum() / e**2
def _make_graph(self):
for rank in range(self.world_size):
for i in range(0, int(mlog(self.world_size - 1, 2)) + 1):
f_peer = self._rotate_forward(rank, 2**i)
b_peer = self._rotate_backward(rank, 2**i)
self._add_peers(rank, [f_peer, b_peer])
def GSPTIME(gsum, rk):
from math import log as mlog
lk = rk.uniform_pos()
tnr = (1.0 / gsum) * mlog(1.0 / lk)
return tnr
def isFibonacci2(num):
from math import log as mlog
a = (mlog(
((num * 5**(1 / 2) + (5 * num**2 + 4)**(1 / 2)) / 2), 1.618) + mlog(
((num * 5**(1 / 2) + (5 * num**2 - 4)**(1 / 2)) / 2), 1.618)) / 2
return (a % 1 * 100 > 99) or (a % 1 * 100 < 1)
def mctime(rtsum,rk):
from math import log as mlog
lk=rk.uniform_pos()
tnr=(1.0/rtsum)*mlog(1.0/lk)
return tnr
def GSPTIME(gsum,rk): from math import log as mlog lk=rk.uniform_pos() tnr=(1.0/gsum)*mlog(1.0/lk) return tnr
