def test_isqrt():
from math import sqrt as _sqrt
limit = 17984395633462800708566937239551
assert int(_sqrt(limit)) == integer_nthroot(limit, 2)[0]
assert int(_sqrt(limit + 1)) != integer_nthroot(limit + 1, 2)[0]
assert isqrt(limit + 1) == integer_nthroot(limit + 1, 2)[0]
assert isqrt(limit + 1 + S.Half) == integer_nthroot(limit + 1, 2)[0]
def vonmisesvariate(self, mu, kappa):
random = self.random
if kappa <= 9.9999999999999995e-007:
return TWOPI * random()
a = 1.0 + _sqrt(1.0 + 4.0 * kappa * kappa)
b = (a - _sqrt(2.0 * a)) / 2.0 * kappa
r = (1.0 + b * b) / 2.0 * b
while 1:
u1 = random()
z = _cos(_pi * u1)
f = (1.0 + r * z) / (r + z)
c = kappa * (r - f)
u2 = random()
if u2 >= c * (2.0 - c):
pass
if not (u2 > c * _exp(1.0 - c)):
break
u3 = random()
if u3 > 0.5:
theta = mu % TWOPI + _acos(f)
else:
theta = mu % TWOPI - _acos(f)
return theta
def rgb_distance(r1, g1, b1, r2, g2, b2):
"""Calculates numerical distance between two colors in RGB color space.
The distance is calculated by CIE94 formula.
:params: Two colors with ``r, g, b`` values in ``0..1`` range
:returns: A number in ``0..100`` range. The lesser - the
closer colors are.
"""
# Formulae from wikipedia article re CIE94
L1, A1, B1 = xyz_to_lab(*rgb_to_xyz(r1, b1, g1))
L2, A2, B2 = xyz_to_lab(*rgb_to_xyz(r2, b2, g2))
dL = L1 - L2
C1 = _sqrt(A1 * A1 + B1 * B1)
C2 = _sqrt(A2 * A2 + B2 * B2)
dCab = C1 - C2
dA = A1 - A2
dB = B1 - B2
dEab = _sqrt(dL ** 2 + dA ** 2 + dB ** 2)
dHab = _sqrt(max(dEab ** 2 - dL ** 2 - dCab ** 2, 0.0))
dE = _sqrt((dL ** 2) + ((dCab / (1 + 0.045 * C1)) ** 2) + (
dHab / (1 + 0.015 * C1)) ** 2)
return dE
def vonmisesvariate(self, mu, kappa):
"""Circular data distribution.
mu is the mean angle, expressed in radians between 0 and 2*pi, and
kappa is the concentration parameter, which must be greater than or
equal to zero. If kappa is equal to zero, this distribution reduces
to a uniform random angle over the range 0 to 2*pi.
"""
random = self.random
if kappa <= 1e-06:
return TWOPI * random()
a = 1.0 + _sqrt(1.0 + 4.0 * kappa * kappa)
b = (a - _sqrt(2.0 * a)) / (2.0 * kappa)
r = (1.0 + b * b) / (2.0 * b)
while 1:
u1 = random()
z = _cos(_pi * u1)
f = (1.0 + r * z) / (r + z)
c = kappa * (r - f)
u2 = random()
if u2 < c * (2.0 - c) or u2 <= c * _exp(1.0 - c):
break
u3 = random()
if u3 > 0.5:
theta = mu % TWOPI + _acos(f)
else:
theta = mu % TWOPI - _acos(f)
return theta
def arcball(x, y):
h2 = x*x+y*y
if h2 > 1.:
h = _sqrt(h2)
v = x/h, y/h, 0.
else:
v = x, y, _sqrt(1.-h2)
return 0., v
def __parse_word(word):
"""
Read an string and determines if it is a word or a number
or a repetitive list os values such as 3*4.5=[4.5 , 4.5, 4.5]
Args:
word: An string that should be parsed
Returns:
result:
Value extracted (word, number or list)
kind:
'word' such as 'ntime', 'eV', '*1', 'angstrom', etc
'int' such as 1, 2, 3
'float' such as 4.5, 6.7, etc
'list' such as [4.5 , 4.5, 4.5]
"""
result = None
kind = None
if word[0].isalpha() and word[:4] != 'sqrt' and word[:5] != '-sqrt':
result = word
kind = 'word'
elif word[:4] == 'sqrt':
result = _sqrt(float(word[5:-1]))
kind = 'float'
elif word[:5] == '-sqrt':
result = -_sqrt(float(word[6:-1]))
kind = 'float'
elif word[0] == '*':
result = word
kind = 'word'
elif word.isdigit():
result = int(word)
kind = "int"
elif '*' in word:
splt = word.split('*')
if splt[0].isdigit():
mult = int(splt[0])
number, kind = string2number(splt[1])
if number is not None:
result = mult * [number]
kind = 'list'
else:
result = None
kind = None
else:
result, kind = string2number(word)
return result, kind
def get_sphere_mapping(x, y, width, height):
x = min([max([x, 0]), width])
y = min([max([y, 0]), height])
sr = _sqrt((width/2)**2 + (height/2)**2)
sx = ((x - width / 2) / sr)
sy = ((y - height / 2) / sr)
sz = 1.0 - sx**2 - sy**2
if sz > 0.0:
sz = _sqrt(sz)
return (sx, sy, sz)
else:
sz = 0
return norm((sx, sy, sz))
def _facmod(self, n, q):
res, N = 1, int(_sqrt(n))
# Exponent of prime p in n! is e_p(n) = [n/p] + [n/p**2] + ...
# for p > sqrt(n), e_p(n) < sqrt(n), the primes with [n/p] = m,
# occur consecutively and are grouped together in pw[m] for
# simultaneous exponentiation at a later stage
pw = [1]*N
m = 2 # to initialize the if condition below
for prime in sieve.primerange(2, n + 1):
if m > 1:
m, y = 0, n // prime
while y:
m += y
y //= prime
if m < N:
pw[m] = pw[m]*prime % q
else:
res = res*pow(prime, m, q) % q
for ex, bs in enumerate(pw):
if ex == 0 or bs == 1:
continue
if bs == 0:
return 0
res = res*pow(bs, ex, q) % q
return res
def gauss(self, mu, sigma):
# """Gaussian distribution.
# mu is the mean, and sigma is the standard deviation. This is
# slightly faster than the normalvariate() function.
# Not thread-safe without a lock around calls.
# """
# When x and y are two variables from [0, 1), uniformly
# distributed, then
#
# cos(2*pi*x)*sqrt(-2*log(1-y))
# sin(2*pi*x)*sqrt(-2*log(1-y))
#
# are two *independent* variables with normal distribution
# (mu = 0, sigma = 1).
# (Lambert Meertens)
# (corrected version; bug discovered by Mike Miller, fixed by LM)
# Multithreading note: When two threads call this function
# simultaneously, it is possible that they will receive the
# same return value. The window is very small though. To
# avoid this, you have to use a lock around all calls. (I
# didn't want to slow this down in the serial case by using a
# lock here.)
__random = self.random
z = self.gauss_next
self.gauss_next = None
if z is None:
x2pi = __random() * TWOPI
g2rad = _sqrt(-2.0 * _log(1.0 - __random()))
z = _cos(x2pi) * g2rad
self.gauss_next = _sin(x2pi) * g2rad
return mu + z*sigma
def vonmisesvariate(self, mu, kappa):
"""Circular data distribution.
mu is the mean angle, expressed in radians between 0 and 2*pi, and
kappa is the concentration parameter, which must be greater than or
equal to zero. If kappa is equal to zero, this distribution reduces
to a uniform random angle over the range 0 to 2*pi.
"""
random = self.random
if kappa <= 1e-06:
return TWOPI * random()
s = 0.5 / kappa
r = s + _sqrt(1.0 + s * s)
while 1:
u1 = random()
z = _cos(_pi * u1)
d = z / (r + z)
u2 = random()
if u2 < 1.0 - d * d or u2 <= (1.0 - d) * _exp(d):
break
q = 1.0 / r
f = (q + z) / (1.0 + q * z)
u3 = random()
if u3 > 0.5:
theta = (mu + _acos(f)) % TWOPI
else:
theta = (mu - _acos(f)) % TWOPI
return theta
def _swing(cls, n):
if n < 33:
return cls._small_swing[n]
else:
N, primes = int(_sqrt(n)), []
for prime in sieve.primerange(3, N + 1):
p, q = 1, n
while True:
q //= prime
if q > 0:
if q & 1 == 1:
p *= prime
else:
break
if p > 1:
primes.append(p)
for prime in sieve.primerange(N + 1, n//3 + 1):
if (n // prime) & 1 == 1:
primes.append(prime)
L_product = R_product = 1
for prime in sieve.primerange(n//2 + 1, n + 1):
L_product *= prime
for prime in primes:
R_product *= prime
return L_product*R_product
def vonmisesvariate(self, mu, kappa):
"""Circular data distribution.
mu is the mean angle, expressed in radians between 0 and 2*pi, and
kappa is the concentration parameter, which must be greater than or
equal to zero. If kappa is equal to zero, this distribution reduces
to a uniform random angle over the range 0 to 2*pi.
"""
# mu: mean angle (in radians between 0 and 2*pi)
# kappa: concentration parameter kappa (>= 0)
# if kappa = 0 generate uniform random angle
# Based upon an algorithm published in: Fisher, N.I.,
# "Statistical Analysis of Circular Data", Cambridge
# University Press, 1993.
# Thanks to Magnus Kessler for a correction to the
# implementation of step 4.
random = self.random
if kappa <= 1e-6:
return TWOPI * random()
a = 1.0 + _sqrt(1.0 + 4.0 * kappa * kappa)
b = (a - _sqrt(2.0 * a)) / (2.0 * kappa)
r = (1.0 + b * b) / (2.0 * b)
while 1:
u1 = random()
z = _cos(_pi * u1)
f = (1.0 + r * z) / (r + z)
c = kappa * (r - f)
u2 = random()
if not (u2 >= c * (2.0 - c) and u2 > c * _exp(1.0 - c)):
break
u3 = random()
if u3 > 0.5:
theta = (mu % TWOPI) + _acos(f)
else:
theta = (mu % TWOPI) - _acos(f)
return theta
def eval(cls, n, k):
n, k = map(sympify, (n, k))
if k.is_Number:
if k.is_Integer:
if k < 0:
return S.Zero
elif k == 0 or n == k:
return S.One
elif n.is_Integer and n >= 0:
n, k = int(n), int(k)
if k > n:
return S.Zero
elif k > n // 2:
k = n - k
M, result = int(_sqrt(n)), 1
for prime in sieve.primerange(2, n+1):
if prime > n - k:
result *= prime
elif prime > n // 2:
continue
elif prime > M:
if n % prime < k % prime:
result *= prime
else:
N, K = n, k
exp = a = 0
while N > 0:
a = int((N % prime) < (K % prime + a))
N, K = N // prime, K // prime
exp = a + exp
if exp > 0:
result *= prime**exp
return C.Integer(result)
else:
result = n - k + 1
for i in xrange(2, k+1):
result *= n-k+i
result /= i
return result
elif k.is_negative:
return S.Zero
elif (n - k).simplify().is_negative:
return S.Zero
else:
d = n - k
if d.is_Integer:
return cls.eval(n, d)
def is_prime2(num):
'''Tests if a given number is prime. Written with a map.'''
if num == 2:
return True
elif num % 2 == 0 or num <= 1:
return False
root = _ceil(_sqrt(num))
return all(map(lambda div: False if num % div == 0 else True,
range(3, root+1, 2)))
def is_prime(integer):
"""Returns True if ``integer`` is a prime, otherwise False."""
assert integer < primes[-1] ** 2
integer = -integer if integer < 0 else integer
limit = _floor(_sqrt(integer)) + 1
for i in _takewhile(lambda elem: elem < limit, primes):
if integer % i == 0:
return False
return integer > 1
def is_prime3(num):
'''Tests if a given number is prime. Written with reduce.'''
if num == 2:
return True
elif num % 2 == 0 or num <= 1:
return False
root = _ceil(_sqrt(num))
return _reduce(lambda acc, d: False if not acc or num % d == 0 else True,
range(3, root+1, 2), True)
def sqrt(S):
"""Convenience function for taking square roots of PowerSeries.
This can also replace the ``math.sqrt`` function, extending it to
take a PowerSeries as an argument.
"""
from math import sqrt as _sqrt
if isinstance(S, PowerSeries):
return S.squareroot()
return _sqrt(S)
def gauss(self, mu, sigma):
random = self.random
z = self.gauss_next
self.gauss_next = None
if z is None:
x2pi = random() * TWOPI
g2rad = _sqrt(-2.0 * _log(1.0 - random()))
z = _cos(x2pi) * g2rad
self.gauss_next = _sin(x2pi) * g2rad
return mu + z * sigma
def gammavariate(self, alpha, beta):
"""Gamma distribution. Not the gamma function!
Conditions on the parameters are alpha > 0 and beta > 0.
The probability distribution function is:
x ** (alpha - 1) * math.exp(-x / beta)
pdf(x) = --------------------------------------
math.gamma(alpha) * beta ** alpha
"""
if alpha <= 0.0 or beta <= 0.0:
raise ValueError, "gammavariate: alpha and beta must be > 0.0"
random = self.random
if alpha > 1.0:
ainv = _sqrt(2.0 * alpha - 1.0)
bbb = alpha - LOG4
ccc = alpha + ainv
while 1:
u1 = random()
if not 1e-07 < u1 < 0.9999999:
continue
u2 = 1.0 - random()
v = _log(u1 / (1.0 - u1)) / ainv
x = alpha * _exp(v)
z = u1 * u1 * u2
r = bbb + ccc * v - x
if r + SG_MAGICCONST - 4.5 * z >= 0.0 or r >= _log(z):
return x * beta
else:
if alpha == 1.0:
u = random()
while u <= 1e-07:
u = random()
return -_log(u) * beta
while 1:
u = random()
b = (_e + alpha) / _e
p = b * u
if p <= 1.0:
x = p ** (1.0 / alpha)
else:
x = -_log((b - p) / alpha)
u1 = random()
if p > 1.0:
if u1 <= x ** (alpha - 1.0):
break
elif u1 <= _exp(-x):
break
return x * beta
def _eval(self, n, k):
# n.is_Number and k.is_Integer and k != 1 and n != k
from sympy.functions.elementary.exponential import log
from sympy.core import N
if k.is_Integer:
if n.is_Integer and n >= 0:
n, k = int(n), int(k)
if k > n:
return S.Zero
elif k > n // 2:
k = n - k
if HAS_GMPY:
from sympy.core.compatibility import gmpy
return Integer(gmpy.bincoef(n, k))
prime_count_estimate = N(n / log(n))
# if the number of primes less than n is less than k, use prime sieve method
# otherwise it is more memory efficient to compute factorials explicitly
if prime_count_estimate < k:
M, result = int(_sqrt(n)), 1
for prime in sieve.primerange(2, n + 1):
if prime > n - k:
result *= prime
elif prime > n // 2:
continue
elif prime > M:
if n % prime < k % prime:
result *= prime
else:
N, K = n, k
exp = a = 0
while N > 0:
a = int((N % prime) < (K % prime + a))
N, K = N // prime, K // prime
exp = a + exp
if exp > 0:
result *= prime**exp
else:
result = ff(n, k) / factorial(k)
return Integer(result)
else:
d = result = n - k + 1
for i in range(2, k + 1):
d += 1
result *= d
result /= i
return result
def is_prime(num):
'''Tests if a given number is prime. Written procedurally.'''
if num == 2:
return True
elif num % 2 == 0 or num <= 1:
return False
count = 3
root = _sqrt(num)
while count <= root:
if num % count == 0: # If anything divides evenly, it isn't prime.
return False
count += 2
return True
def vonmisesvariate(self, mu, kappa):
random = self.random
if kappa <= 1e-06:
return TWOPI * random()
a = 1.0 + _sqrt(1.0 + 4.0 * kappa * kappa)
b = (a - _sqrt(2.0 * a)) / (2.0 * kappa)
r = (1.0 + b * b) / (2.0 * b)
while 1:
u1 = random()
z = _cos(_pi * u1)
f = (1.0 + r * z) / (r + z)
c = kappa * (r - f)
u2 = random()
if u2 < c * (2.0 - c) or u2 <= c * _exp(1.0 - c):
break
u3 = random()
if u3 > 0.5:
theta = mu % TWOPI + _acos(f)
else:
theta = mu % TWOPI - _acos(f)
return theta
def vonmisesvariate(self, mu, kappa):
"""Circular data distribution.
mu is the mean angle, expressed in radians between 0 and 2*pi, and
kappa is the concentration parameter, which must be greater than or
equal to zero. If kappa is equal to zero, this distribution reduces
to a uniform random angle over the range 0 to 2*pi.
"""
# mu: mean angle (in radians between 0 and 2*pi)
# kappa: concentration parameter kappa (>= 0)
# if kappa = 0 generate uniform random angle
# Based upon an algorithm published in: Fisher, N.I.,
# "Statistical Analysis of Circular Data", Cambridge
# University Press, 1993.
# Thanks to Magnus Kessler for a correction to the
# implementation of step 4.
random = self.random
if kappa <= 1e-6:
return TWOPI * random()
s = 0.5 / kappa
r = s + _sqrt(1.0 + s * s)
while 1:
u1 = random()
z = _cos(_pi * u1)
d = z / (r + z)
u2 = random()
if u2 < 1.0 - d * d or u2 <= (1.0 - d) * _exp(d):
break
q = 1.0 / r
f = (q + z) / (1.0 + q * z)
u3 = random()
if u3 > 0.5:
theta = (mu + _acos(f)) % TWOPI
else:
theta = (mu - _acos(f)) % TWOPI
return theta
def gammavariate(self, alpha, beta):
if alpha <= 0.0 or beta <= 0.0:
raise ValueError('gammavariate: alpha and beta must be > 0.0')
random = self.random
if alpha > 1.0:
ainv = _sqrt(2.0*alpha - 1.0)
bbb = alpha - LOG4
ccc = alpha + ainv
u1 = random()
if not 1e-07 < u1 < 0.9999999:
continue
u2 = 1.0 - random()
v = _log(u1/(1.0 - u1))/ainv
x = alpha*_exp(v)
z = u1*u1*u2
r = bbb + ccc*v - x
#ERROR: Unexpected statement: 517 BINARY_MULTIPLY | 518 RETURN_VALUE
if r + SG_MAGICCONST - 4.5*z >= 0.0 or r >= _log(z):
return x*beta
continue
else:
if alpha == 1.0:
u = random()
while u <= 1e-07:
u = random()
return -_log(u)*beta
while True:
u = random()
b = (_e + alpha)/_e
p = b*u
if p <= 1.0:
x = p**(1.0/alpha)
else:
x = -_log((b - p)/alpha)
u1 = random()
if p > 1.0:
if u1 <= x**(alpha - 1.0):
break
continue
if u1 <= _exp(-x):
break
elif u1 <= _exp(-x):
break
return x*beta
def _test_generator(n, funccall):
import time
print n, 'times', funccall
code = compile(funccall, funccall, 'eval')
sum = 0.0
sqsum = 0.0
smallest = 10000000000.0
largest = -10000000000.0
t0 = time.time()
for i in range(n):
x = eval(code)
sum = sum + x
sqsum = sqsum + x * x
smallest = min(x, smallest)
largest = max(x, largest)
t1 = time.time()
print round(t1 - t0, 3), 'sec,'avg = sum / nstddev = _sqrt(sqsum / n - avg * avg), 'avg %g, stddev %g, min %g, max %g' % (avg, stddev, smallest, largest)
def gauss(self, mu, sigma):
"""Gaussian distribution.
mu is the mean, and sigma is the standard deviation. This is
slightly faster than the normalvariate() function.
Not thread-safe without a lock around calls.
"""
random = self.random
z = self.gauss_next
self.gauss_next = None
if z is None:
x2pi = random() * TWOPI
g2rad = _sqrt(-2.0 * _log(1.0 - random()))
z = _cos(x2pi) * g2rad
self.gauss_next = _sin(x2pi) * g2rad
return mu + z * sigma
def gammavariate(self, alpha, beta):
if alpha <= 0.0 or beta <= 0.0:
raise ValueError, 'gammavariate: alpha and beta must be > 0.0'
random = self.random
if alpha > 1.0:
ainv = _sqrt(2.0 * alpha - 1.0)
bbb = alpha - LOG4
ccc = alpha + ainv
while 1:
u1 = random()
if not 1e-07 < u1 < 0.9999999:
continue
u2 = 1.0 - random()
v = _log(u1 / (1.0 - u1)) / ainv
x = alpha * _exp(v)
z = u1 * u1 * u2
r = bbb + ccc * v - x
if r + SG_MAGICCONST - 4.5 * z >= 0.0 or r >= _log(z):
return x * beta
else:
if alpha == 1.0:
u = random()
while u <= 1e-07:
u = random()
return -_log(u) * beta
while 1:
u = random()
b = (_e + alpha) / _e
p = b * u
if p <= 1.0:
x = p ** (1.0 / alpha)
else:
x = -_log((b - p) / alpha)
u1 = random()
if p > 1.0:
if u1 <= x ** (alpha - 1.0):
break
elif u1 <= _exp(-x):
break
return x * beta
def triangular(self, low=0.0, high=1.0, mode=None):
"""Triangular distribution.
Continuous distribution bounded by given lower and upper limits,
and having a given mode value in-between.
http://en.wikipedia.org/wiki/Triangular_distribution
"""
u = self.random()
try:
c = 0.5 if mode is None else (mode - low) / (high - low)
except ZeroDivisionError:
return low
if u > c:
u = 1.0 - u
c = 1.0 - c
low, high = high, low
return low + (high - low) * _sqrt(u * c)
def _test_generator(n, func, args):
print(n, 'times', func.__name__)
total = 0.0
sqsum = 0.0
smallest = 1e10
largest = -1e10
t0 = time.time()
for i in range(n):
x = func(*args)
total += x
sqsum = sqsum + x*x
smallest = min(x, smallest)
largest = max(x, largest)
t1 = time.time()
print(round(t1-t0, 3), 'sec,')
avg = total/n
stddev = _sqrt(sqsum/n - avg*avg)
print('avg %g, stddev %g, min %g, max %g' %
(avg, stddev, smallest, largest))
def _test_generator(n, func, args):
import time
print(n, "times", func.__name__)
total = 0.0
sqsum = 0.0
smallest = 1e10
largest = -1e10
t0 = time.time()
for i in range(n):
x = func(*args)
total += x
sqsum = sqsum + x * x
smallest = min(x, smallest)
largest = max(x, largest)
t1 = time.time()
print(round(t1 - t0, 3), "sec,", end=" ")
avg = total / n
stddev = _sqrt(sqsum / n - avg * avg)
print("avg %g, stddev %g, min %g, max %g\n" % (avg, stddev, smallest, largest))
def cdf(x):
return (1.0 + _erf(x / _sqrt(2.0))) / 2.0
def mag(a):
return _sqrt(a[0]**2 + a[1]**2 + a[2]**2)
"""Random variable generators.
from math import sqrt as _sqrt
from math import factorial as _fact
# pi = {n*sk}^{-1}
n = _sqrt(8) / (99**2)
sk = 0
for k in range(5):
sk += ((_fact(4 * k)) / (_fact(k)**4)) * ((26390 * k + 1103) /
(396**(4 * k)))
PI = 1 / (n * sk)
def reset_parameters(self):
self.weight.data.normal_(0, 1 * (_sqrt(1. / self.in_features)))
if self.bias is not None:
self.bias.data.zero_()
def gammavariate(self, alpha, beta):
# beta times standard gamma
ainv = _sqrt(2.0 * alpha - 1.0)
return beta * self.stdgamma(alpha, ainv, alpha - LOG4, alpha + ainv)
def gammavariate(self, alpha, beta):
"""Gamma distribution. Not the gamma function!
Conditions on the parameters are alpha > 0 and beta > 0.
The probability distribution function is:
x ** (alpha - 1) * math.exp(-x / beta)
pdf(x) = --------------------------------------
math.gamma(alpha) * beta ** alpha
"""
# alpha > 0, beta > 0, mean is alpha*beta, variance is alpha*beta**2
# Warning: a few older sources define the gamma distribution in terms
# of alpha > -1.0
if alpha <= 0.0 or beta <= 0.0:
raise ValueError('gammavariate: alpha and beta must be > 0.0')
random = self.random
if alpha > 1.0:
# Uses R.C.H. Cheng, "The generation of Gamma
# variables with non-integral shape parameters",
# Applied Statistics, (1977), 26, No. 1, p71-74
ainv = _sqrt(2.0 * alpha - 1.0)
bbb = alpha - LOG4
ccc = alpha + ainv
while 1:
u1 = random()
if not 1e-7 < u1 < .9999999:
continue
u2 = 1.0 - random()
v = _log(u1 / (1.0 - u1)) / ainv
x = alpha * _exp(v)
z = u1 * u1 * u2
r = bbb + ccc * v - x
if r + SG_MAGICCONST - 4.5 * z >= 0.0 or r >= _log(z):
return x * beta
elif alpha == 1.0:
# expovariate(1)
u = random()
while u <= 1e-7:
u = random()
return -_log(u) * beta
else: # alpha is between 0 and 1 (exclusive)
# Uses ALGORITHM GS of Statistical Computing - Kennedy & Gentle
while 1:
u = random()
b = (_e + alpha) / _e
p = b * u
if p <= 1.0:
x = p**(1.0 / alpha)
else:
x = -_log((b - p) / alpha)
u1 = random()
if p > 1.0:
if u1 <= x**(alpha - 1.0):
break
elif u1 <= _exp(-x):
break
return x * beta
def mysqrt(x):
if x < 0:
raise ValueError("sqrt of negativ number")
return math._sqrt(x)
def binomialvariate(self, n=1, p=0.5):
"""Binomial random variable.
Gives the number of successes for *n* independent trials
with the probability of success in each trial being *p*:
sum(random() < p for i in range(n))
Returns an integer in the range: 0 <= X <= n
"""
# Error check inputs and handle edge cases
if n < 0:
raise ValueError("n must be non-negative")
if p <= 0.0 or p >= 1.0:
if p == 0.0:
return 0
if p == 1.0:
return n
raise ValueError("p must be in the range 0.0 <= p <= 1.0")
random = self.random
# Fast path for a common case
if n == 1:
return _index(random() < p)
# Exploit symmetry to establish: p <= 0.5
if p > 0.5:
return n - self.binomialvariate(n, 1.0 - p)
if n * p < 10.0:
# BG: Geometric method by Devroye with running time of O(np).
# https://dl.acm.org/doi/pdf/10.1145/42372.42381
x = y = 0
c = _log(1.0 - p)
if not c:
return x
while True:
y += _floor(_log(random()) / c) + 1
if y > n:
return x
x += 1
# BTRS: Transformed rejection with squeeze method by Wolfgang Hörmann
# https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.47.8407&rep=rep1&type=pdf
assert n * p >= 10.0 and p <= 0.5
setup_complete = False
spq = _sqrt(n * p *
(1.0 - p)) # Standard deviation of the distribution
b = 1.15 + 2.53 * spq
a = -0.0873 + 0.0248 * b + 0.01 * p
c = n * p + 0.5
vr = 0.92 - 4.2 / b
while True:
u = random()
v = random()
u -= 0.5
us = 0.5 - _fabs(u)
k = _floor((2.0 * a / us + b) * u + c)
if k < 0 or k > n:
continue
# The early-out "squeeze" test substantially reduces
# the number of acceptance condition evaluations.
if us >= 0.07 and v <= vr:
return k
# Acceptance-rejection test.
# Note, the original paper errorneously omits the call to log(v)
# when comparing to the log of the rescaled binomial distribution.
if not setup_complete:
alpha = (2.83 + 5.1 / b) * spq
lpq = _log(p / (1.0 - p))
m = _floor((n + 1) * p) # Mode of the distribution
h = _lgamma(m + 1) + _lgamma(n - m + 1)
setup_complete = True # Only needs to be done once
v *= alpha / (a / (us * us) + b)
if _log(v) <= h - _lgamma(k + 1) - _lgamma(n - k +
1) + (k - m) * lpq:
return k
def gf_ddf_shoup(f, p, K):
"""
Kaltofen-Shoup: Deterministic Distinct Degree Factorization
Given a monic square-free polynomial ``f`` in ``GF(p)[x]``, computes
partial distinct degree factorization ``f_1,...,f_d`` of ``f`` where
``deg(f_i) != deg(f_j)`` for ``i != j``. The result is returned as a
list of pairs ``(f_i, e_i)`` where ``deg(f_i) > 0`` and ``e_i > 0``
is an argument to the equal degree factorization routine.
This algorithm is an improved version of Zassenhaus algorithm for
large ``deg(f)`` and modulus ``p`` (especially for ``deg(f) ~ lg(p)``).
Examples
========
>>> from sympy.polys.domains import ZZ
>>> from sympy.polys.galoistools import gf_ddf_shoup, gf_from_dict
>>> f = gf_from_dict({6: ZZ(1), 5: ZZ(-1), 4: ZZ(1), 3: ZZ(1), 1: ZZ(-1)}, 3, ZZ)
>>> gf_ddf_shoup(f, 3, ZZ)
[([1, 1, 0], 1), ([1, 1, 0, 1, 2], 2)]
References
==========
1. [Kaltofen98]_
2. [Shoup95]_
3. [Gathen92]_
"""
n = gf_degree(f)
k = int(_ceil(_sqrt(n//2)))
h = gf_pow_mod([K.one, K.zero], int(p), f, p, K)
U = [[K.one, K.zero], h] + [K.zero]*(k - 1)
for i in xrange(2, k + 1):
U[i] = gf_compose_mod(U[i - 1], h, f, p, K)
h, U = U[k], U[:k]
V = [h] + [K.zero]*(k - 1)
for i in xrange(1, k):
V[i] = gf_compose_mod(V[i - 1], h, f, p, K)
factors = []
for i, v in enumerate(V):
h, j = [K.one], k - 1
for u in U:
g = gf_sub(v, u, p, K)
h = gf_mul(h, g, p, K)
h = gf_rem(h, f, p, K)
g = gf_gcd(f, h, p, K)
f = gf_quo(f, g, p, K)
for u in reversed(U):
h = gf_sub(v, u, p, K)
F = gf_gcd(g, h, p, K)
if F != [K.one]:
factors.append((F, k*(i + 1) - j))
g, j = gf_quo(g, F, p, K), j - 1
if f != [K.one]:
factors.append((f, gf_degree(f)))
return factors
TEXT_STYLE_POLYGONS = 1
TEXT_STYLE_LABELS = 2
PATH_TYPE_NORMAL = 0
PATH_TYPE_ROUNDED = 1
PATH_TYPE_EXTENDED = 2
PATH_TYPES = [PATH_TYPE_NORMAL, PATH_TYPE_ROUNDED, PATH_TYPE_EXTENDED]
GDSII_MAX_COORDINATES = 200
NORTH = Coord2(0.0, 1.0)
SOUTH = Coord2(0.0, -1.0)
EAST = Coord2(1.0, 0.0)
WEST = Coord2(-1.0, 0.0)
_sqrt2_2 = _sqrt(0.5)
NORTHEAST = Coord2(
_sqrt2_2,
_sqrt2_2,
)
NORTHWEST = Coord2(
-_sqrt2_2,
_sqrt2_2,
)
SOUTHEAST = Coord2(
_sqrt2_2,
-_sqrt2_2,
)
SOUTHWEST = Coord2(
-_sqrt2_2,
-_sqrt2_2,
def _s():
s0 = Fraction.from_float(_sqrt(self.zero))
yield s0
for term in (self.tail * (s0 + S).reciprocal()):
yield term
def _eval_Mod(self, q):
n, k = self.args
if any(x.is_integer is False for x in (n, k, q)):
raise ValueError("Integers expected for binomial Mod")
if all(x.is_Integer for x in (n, k, q)):
n, k = map(int, (n, k))
aq, res = abs(q), 1
# handle negative integers k or n
if k < 0:
return 0
if n < 0:
n = -n + k - 1
res = -1 if k % 2 else 1
# non negative integers k and n
if k > n:
return 0
isprime = aq.is_prime
aq = int(aq)
if isprime:
if aq < n:
# use Lucas Theorem
N, K = n, k
while N or K:
res = res * binomial(N % aq, K % aq) % aq
N, K = N // aq, K // aq
else:
# use Factorial Modulo
d = n - k
if k > d:
k, d = d, k
kf = 1
for i in range(2, k + 1):
kf = kf * i % aq
df = kf
for i in range(k + 1, d + 1):
df = df * i % aq
res *= df
for i in range(d + 1, n + 1):
res = res * i % aq
res *= pow(kf * df % aq, aq - 2, aq)
res %= aq
else:
# Binomial Factorization is performed by calculating the
# exponents of primes <= n in `n! /(k! (n - k)!)`,
# for non-negative integers n and k. As the exponent of
# prime in n! is e_p(n) = [n/p] + [n/p**2] + ...
# the exponent of prime in binomial(n, k) would be
# e_p(n) - e_p(k) - e_p(n - k)
M = int(_sqrt(n))
for prime in sieve.primerange(2, n + 1):
if prime > n - k:
res = res * prime % aq
elif prime > n // 2:
continue
elif prime > M:
if n % prime < k % prime:
res = res * prime % aq
else:
N, K = n, k
exp = a = 0
while N > 0:
a = int((N % prime) < (K % prime + a))
N, K = N // prime, K // prime
exp += a
if exp > 0:
res *= pow(prime, exp, aq)
res %= aq
return Integer(res % q)
def _phase(self, k_y, x, y):
"""Phase function."""
return x * _sqrt(self._k**2 - k_y**2) + k_y * y
def __abs__(_self):
'Return the sqrt of the sum of the squares of all elements'
return _sqrt(sum(e * e for e in _self))
from math import sqrt as _sqrt # Taking formula of Time period of a pendulum, whose time # period already is 2s and length is 1m; we can solve the # equation to find that g = pi^2 _g = 9.81 # acceleration due to gravtiy on earth PI = _sqrt(_g)
def hypot(x, y):
arg = x * x + y * y
if arg.is_Rational:
return _sqrt(arg)
return sqrt(arg)
def _draw_arrow(self,
x1,
y1,
x2,
y2,
Dx,
Dy,
label="",
width=1.0,
arrow_curvature=1.0,
color="grey",
patchA=None,
patchB=None,
shrinkA=0,
shrinkB=0,
arrow_label_size=None):
"""
Draws a slightly curved arrow from (x1,y1) to (x2,y2).
Will allow the given patches at start end end.
"""
# set arrow properties
dist = _sqrt(((x2 - x1) / float(Dx))**2 + ((y2 - y1) / float(Dy))**2)
arrow_curvature *= 0.075 # standard scale
rad = arrow_curvature / (dist)
tail_width = width
head_width = max(0.5, 2 * width)
head_length = head_width
self.ax.annotate(
"",
xy=(x2, y2),
xycoords='data',
xytext=(x1, y1),
textcoords='data',
arrowprops=dict(
arrowstyle='simple,head_length=%f,head_width=%f,tail_width=%f'
% (head_length, head_width, tail_width),
color=color,
shrinkA=shrinkA,
shrinkB=shrinkB,
patchA=patchA,
patchB=patchB,
connectionstyle="arc3,rad=%f" % -rad),
zorder=0)
# weighted center position
center = _np.array([0.55 * x1 + 0.45 * x2, 0.55 * y1 + 0.45 * y2])
v = _np.array([x2 - x1, y2 - y1]) # 1->2 vector
vabs = _np.abs(v)
vnorm = _np.array([v[1], -v[0]]) # orthogonal vector
vnorm = _np.divide(vnorm, _np.linalg.norm(vnorm)) # normalize
# cross product to determine the direction into which vnorm points
z = _np.cross(v, vnorm)
if z < 0:
vnorm *= -1
offset = 0.5 * arrow_curvature * \
((vabs[0] / (vabs[0] + vabs[1]))
* Dx + (vabs[1] / (vabs[0] + vabs[1])) * Dy)
ptext = center + offset * vnorm
self.ax.text(ptext[0],
ptext[1],
label,
size=arrow_label_size,
horizontalalignment='center',
verticalalignment='center',
zorder=1)
from types import MethodType as _MethodType, BuiltinMethodType as _BuiltinMethodType
from math import log as _log, exp as _exp, pi as _pi, e as _e, ceil as _ceil
from math import sqrt as _sqrt, acos as _acos, cos as _cos, sin as _sin
from os import urandom as _urandom
from _collections_abc import Set as _Set, Sequence as _Sequence
from hashlib import sha512 as _sha512
__all__ = [
"Random", "seed", "random", "uniform", "randint", "choice", "sample",
"randrange", "shuffle", "normalvariate", "lognormvariate", "expovariate",
"vonmisesvariate", "gammavariate", "triangular", "gauss", "betavariate",
"paretovariate", "weibullvariate", "getstate", "setstate", "getrandbits",
"SystemRandom"
]
NV_MAGICCONST = 4 * _exp(-0.5) / _sqrt(2.0)
TWOPI = 2.0 * _pi
LOG4 = _log(4.0)
SG_MAGICCONST = 1.0 + _log(4.5)
BPF = 53 # Number of bits in a float
RECIP_BPF = 2**-BPF
# Translated by Guido van Rossum from C source provided by
# Adrian Baddeley. Adapted by Raymond Hettinger for use with
# the Mersenne Twister and os.urandom() core generators.
import _random
class Random(_random.Random):
"""Random number generator base class used by bound module functions.
Can be used like:
from parmed.constants import *
"""
from __future__ import division
from math import pi as _pi, sqrt as _sqrt, log10 as _log10
__all__ = ['AMBER_ELECTROSTATIC', 'AMBER_POINTERS', 'NATOM', 'NTYPES', 'NBONH',
'MBONA', 'NTHETH', 'MTHETA', 'NPHIH', 'MPHIA', 'NHPARM', 'NPARM',
'NEXT', 'NRES', 'NBONA', 'NTHETA', 'NPHIA', 'NUMBND', 'NUMANG',
'NPTRA', 'NATYP', 'NPHB', 'IFPERT', 'NBPER', 'NGPER', 'NDPER',
'MBPER', 'MGPER', 'MDPER', 'IFBOX', 'NMXRS', 'IFCAP', 'NUMEXTRA',
'NCOPY', 'NNB', 'RAD_TO_DEG', 'DEG_TO_RAD']
AMBER_ELECTROSTATIC = 18.2223
CHARMM_ELECTROSTATIC = _sqrt(332.0716)
AMBER_POINTERS = """
NATOM : total number of atoms
NTYPES : total number of distinct atom types
NBONH : number of bonds containing hydrogen
MBONA : number of bonds not containing hydrogen
NTHETH : number of angles containing hydrogen
MTHETA : number of angles not containing hydrogen
NPHIH : number of dihedrals containing hydrogen
MPHIA : number of dihedrals not containing hydrogen
NHPARM : currently not used
NPARM : currently not used
NEXT : number of excluded atoms
NRES : number of residues
NBONA : MBONA + number of constraint bonds
def distance(first, second):
"""returns the Euclidian distance between first and second"""
dist = 0
for i in xrange(dim):
dist += (first.features[i] - second.features[i])**2
return _sqrt(dist)
def __init__(self, x, params, called=False):
"""..."""
self._W_y = params['W_y']
self._norm = 2 * _sqrt(math.pi) / self._W_y
super().__init__(x, params, called)
def isqrt(n):
"""Return the largest integer less than or equal to sqrt(n)."""
if n < 17984395633462800708566937239552:
return int(_sqrt(n))
return integer_nthroot(int(n), 2)[0]
def sqrt(x):
if x > 0:
return _sqrt(x)
else:
raise MathError(translate('MathErrors', _errors['mde']))
from math import sqrt as _sqrt, pi # noinspection PyUnresolvedReferences from scipy.constants import c as clight from scipy.constants import physical_constants as _cst from scipy.constants import e as qe e_mass = 1.0e+06 * _cst['electron mass energy equivalent in MeV'][0] # eV p_mass = 1.0e+06 * _cst['proton mass energy equivalent in MeV'][0] # eV _e_radius = _cst['classical electron radius'][0] _hbar_c = _cst['Planck constant over 2 pi times c in MeV fm'][0] Cgamma = 4.0 * pi * _e_radius / 3.0 / pow(e_mass, 3) # m/eV^3 Cq = 55 / 32 / _sqrt(3) * _hbar_c / e_mass * 1.0e-9 # m
def __isnumber(word):
"""
Take a string and try to convert into a number
of list of numbers.
It try to returns integer values always that is
is possible.
Examples:
'3' => 3
'3.14' => 3.14
'1.5e-5' => 0.000015
'1.5d-5' => 0.000015
'1.5E-5' => 0.000015
'1.5D-5' => 0.000015
'1/2' => 0.5
'sqrt(3.0)' => 1.732050
Args:
word:
An string that should be converted into a number
Returns:
number:
The value extracted
kind:
The kind of value
"""
number = None
kind = None
try:
number = int(word)
kind = 'int'
except ValueError:
try:
number = float(word)
kind = 'float'
except ValueError:
if 'd' in word:
word = word.replace('d', 'e')
elif 'D' in word:
word = word.replace('D', 'e')
try:
number = float(word)
kind = 'float'
except ValueError:
if '/' in word:
splt = word.split('/')
if splt[0].isdigit() and splt[1].isdigit():
number = float(splt[0]) / float(splt[1])
kind = 'float'
elif word[:4] == 'sqrt':
number = _sqrt(float(word[5:-1]))
kind = 'float'
elif word[:5] == '-sqrt':
number = -_sqrt(float(word[6:-1]))
kind = 'float'
else:
print('ERROR: "%s" is not a number\n' % word)
return number, kind
def metrics(returns,
benchmark=None,
rf=0.,
display=True,
mode='basic',
sep=False,
compounded=True,
periods_per_year=252,
prepare_returns=True,
match_dates=False,
**kwargs):
win_year, _ = _get_trading_periods(periods_per_year)
if benchmark is not None \
and isinstance(benchmark, _pd.DataFrame) and len(benchmark.columns) > 1:
raise ValueError("`benchmark` must be a pandas Series, "
"but a multi-column DataFrame was passed")
blank = ['']
if isinstance(returns, _pd.DataFrame):
if len(returns.columns) > 1:
raise ValueError(
"`returns` needs to be a Pandas Series or one column DataFrame. multi colums DataFrame was passed"
)
returns = returns[returns.columns[0]]
if prepare_returns:
returns = _utils._prepare_returns(returns)
df = _pd.DataFrame({"returns": returns})
if benchmark is not None:
blank = ['', '']
benchmark = _utils._prepare_benchmark(benchmark, returns.index, rf)
if match_dates is True:
returns, benchmark = _match_dates(returns, benchmark)
df["returns"] = returns
df["benchmark"] = benchmark
df = df.fillna(0)
# pct multiplier
pct = 100 if display or "internal" in kwargs else 1
if kwargs.get("as_pct", False):
pct = 100
# return df
dd = _calc_dd(df,
display=(display or "internal" in kwargs),
as_pct=kwargs.get("as_pct", False))
metrics = _pd.DataFrame()
s_start = {'returns': df['returns'].index.strftime('%Y-%m-%d')[0]}
s_end = {'returns': df['returns'].index.strftime('%Y-%m-%d')[-1]}
s_rf = {'returns': rf}
if "benchmark" in df:
s_start['benchmark'] = df['benchmark'].index.strftime('%Y-%m-%d')[0]
s_end['benchmark'] = df['benchmark'].index.strftime('%Y-%m-%d')[-1]
s_rf['benchmark'] = rf
metrics['Start Period'] = _pd.Series(s_start)
metrics['End Period'] = _pd.Series(s_end)
metrics['Risk-Free Rate %'] = _pd.Series(s_rf)
metrics['Time in Market %'] = _stats.exposure(df,
prepare_returns=False) * pct
metrics['~'] = blank
if compounded:
metrics['Cumulative Return %'] = (_stats.comp(df) * pct).map(
'{:,.2f}'.format)
else:
metrics['Total Return %'] = (df.sum() * pct).map('{:,.2f}'.format)
metrics['CAGR﹪%'] = _stats.cagr(df, rf, compounded) * pct
metrics['~~~~~~~~~~~~~~'] = blank
metrics['Sharpe'] = _stats.sharpe(df, rf, win_year, True)
if mode.lower() == 'full':
metrics['Smart Sharpe'] = _stats.smart_sharpe(df, rf, win_year, True)
metrics['Sortino'] = _stats.sortino(df, rf, win_year, True)
if mode.lower() == 'full':
metrics['Smart Sortino'] = _stats.smart_sortino(df, rf, win_year, True)
metrics['Sortino/√2'] = metrics['Sortino'] / _sqrt(2)
if mode.lower() == 'full':
metrics['Smart Sortino/√2'] = metrics['Smart Sortino'] / _sqrt(2)
metrics['Omega'] = _stats.omega(df, rf, 0., win_year)
metrics['~~~~~~~~'] = blank
metrics['Max Drawdown %'] = blank
metrics['Longest DD Days'] = blank
if mode.lower() == 'full':
ret_vol = _stats.volatility(
df['returns'], win_year, True, prepare_returns=False) * pct
if "benchmark" in df:
bench_vol = _stats.volatility(
df['benchmark'], win_year, True, prepare_returns=False) * pct
metrics['Volatility (ann.) %'] = [ret_vol, bench_vol]
metrics['R^2'] = _stats.r_squared(df['returns'],
df['benchmark'],
prepare_returns=False)
else:
metrics['Volatility (ann.) %'] = [ret_vol]
metrics['Calmar'] = _stats.calmar(df, prepare_returns=False)
metrics['Skew'] = _stats.skew(df, prepare_returns=False)
metrics['Kurtosis'] = _stats.kurtosis(df, prepare_returns=False)
metrics['~~~~~~~~~~'] = blank
metrics['Expected Daily %%'] = _stats.expected_return(
df, prepare_returns=False) * pct
metrics['Expected Monthly %%'] = _stats.expected_return(
df, aggregate='M', prepare_returns=False) * pct
metrics['Expected Yearly %%'] = _stats.expected_return(
df, aggregate='A', prepare_returns=False) * pct
metrics['Kelly Criterion %'] = _stats.kelly_criterion(
df, prepare_returns=False) * pct
metrics['Risk of Ruin %'] = _stats.risk_of_ruin(df,
prepare_returns=False)
metrics['Daily Value-at-Risk %'] = -abs(
_stats.var(df, prepare_returns=False) * pct)
metrics['Expected Shortfall (cVaR) %'] = -abs(
_stats.cvar(df, prepare_returns=False) * pct)
metrics['~~~~~~'] = blank
metrics['Gain/Pain Ratio'] = _stats.gain_to_pain_ratio(df, rf)
metrics['Gain/Pain (1M)'] = _stats.gain_to_pain_ratio(df, rf, "M")
# if mode.lower() == 'full':
# metrics['GPR (3M)'] = _stats.gain_to_pain_ratio(df, rf, "Q")
# metrics['GPR (6M)'] = _stats.gain_to_pain_ratio(df, rf, "2Q")
# metrics['GPR (1Y)'] = _stats.gain_to_pain_ratio(df, rf, "A")
metrics['~~~~~~~'] = blank
metrics['Payoff Ratio'] = _stats.payoff_ratio(df, prepare_returns=False)
metrics['Profit Factor'] = _stats.profit_factor(df, prepare_returns=False)
metrics['Common Sense Ratio'] = _stats.common_sense_ratio(
df, prepare_returns=False)
metrics['CPC Index'] = _stats.cpc_index(df, prepare_returns=False)
metrics['Tail Ratio'] = _stats.tail_ratio(df, prepare_returns=False)
metrics['Outlier Win Ratio'] = _stats.outlier_win_ratio(
df, prepare_returns=False)
metrics['Outlier Loss Ratio'] = _stats.outlier_loss_ratio(
df, prepare_returns=False)
# returns
metrics['~~'] = blank
comp_func = _stats.comp if compounded else _np.sum
today = df.index[-1] # _dt.today()
metrics['MTD %'] = comp_func(
df[df.index >= _dt(today.year, today.month, 1)]) * pct
d = today - _td(3 * 365 / 12)
metrics['3M %'] = comp_func(
df[df.index >= _dt(d.year, d.month, d.day)]) * pct
d = today - _td(6 * 365 / 12)
metrics['6M %'] = comp_func(
df[df.index >= _dt(d.year, d.month, d.day)]) * pct
metrics['YTD %'] = comp_func(df[df.index >= _dt(today.year, 1, 1)]) * pct
d = today - _td(12 * 365 / 12)
metrics['1Y %'] = comp_func(
df[df.index >= _dt(d.year, d.month, d.day)]) * pct
d = today - _td(3 * 365)
metrics['3Y (ann.) %'] = _stats.cagr(
df[df.index >= _dt(d.year, d.month, d.day)], 0., compounded) * pct
d = today - _td(5 * 365)
metrics['5Y (ann.) %'] = _stats.cagr(
df[df.index >= _dt(d.year, d.month, d.day)], 0., compounded) * pct
d = today - _td(10 * 365)
metrics['10Y (ann.) %'] = _stats.cagr(
df[df.index >= _dt(d.year, d.month, d.day)], 0., compounded) * pct
metrics['All-time (ann.) %'] = _stats.cagr(df, 0., compounded) * pct
# best/worst
if mode.lower() == 'full':
metrics['~~~'] = blank
metrics['Best Day %'] = _stats.best(df, prepare_returns=False) * pct
metrics['Worst Day %'] = _stats.worst(df, prepare_returns=False) * pct
metrics['Best Month %'] = _stats.best(
df, aggregate='M', prepare_returns=False) * pct
metrics['Worst Month %'] = _stats.worst(
df, aggregate='M', prepare_returns=False) * pct
metrics['Best Year %'] = _stats.best(
df, aggregate='A', prepare_returns=False) * pct
metrics['Worst Year %'] = _stats.worst(
df, aggregate='A', prepare_returns=False) * pct
# dd
metrics['~~~~'] = blank
for ix, row in dd.iterrows():
metrics[ix] = row
metrics['Recovery Factor'] = _stats.recovery_factor(df)
metrics['Ulcer Index'] = _stats.ulcer_index(df)
metrics['Serenity Index'] = _stats.serenity_index(df, rf)
# win rate
if mode.lower() == 'full':
metrics['~~~~~'] = blank
metrics['Avg. Up Month %'] = _stats.avg_win(
df, aggregate='M', prepare_returns=False) * pct
metrics['Avg. Down Month %'] = _stats.avg_loss(
df, aggregate='M', prepare_returns=False) * pct
metrics['Win Days %%'] = _stats.win_rate(df,
prepare_returns=False) * pct
metrics['Win Month %%'] = _stats.win_rate(
df, aggregate='M', prepare_returns=False) * pct
metrics['Win Quarter %%'] = _stats.win_rate(
df, aggregate='Q', prepare_returns=False) * pct
metrics['Win Year %%'] = _stats.win_rate(
df, aggregate='A', prepare_returns=False) * pct
if "benchmark" in df:
metrics['~~~~~~~'] = blank
greeks = _stats.greeks(df['returns'],
df['benchmark'],
win_year,
prepare_returns=False)
metrics['Beta'] = [str(round(greeks['beta'], 2)), '-']
metrics['Alpha'] = [str(round(greeks['alpha'], 2)), '-']
# prepare for display
for col in metrics.columns:
try:
metrics[col] = metrics[col].astype(float).round(2)
if display or "internal" in kwargs:
metrics[col] = metrics[col].astype(str)
except Exception:
pass
if (display or "internal" in kwargs) and "%" in col:
metrics[col] = metrics[col] + '%'
try:
metrics['Longest DD Days'] = _pd.to_numeric(
metrics['Longest DD Days']).astype('int')
metrics['Avg. Drawdown Days'] = _pd.to_numeric(
metrics['Avg. Drawdown Days']).astype('int')
if display or "internal" in kwargs:
metrics['Longest DD Days'] = metrics['Longest DD Days'].astype(str)
metrics['Avg. Drawdown Days'] = metrics[
'Avg. Drawdown Days'].astype(str)
except Exception:
metrics['Longest DD Days'] = '-'
metrics['Avg. Drawdown Days'] = '-'
if display or "internal" in kwargs:
metrics['Longest DD Days'] = '-'
metrics['Avg. Drawdown Days'] = '-'
metrics.columns = [
col if '~' not in col else '' for col in metrics.columns
]
metrics.columns = [
col[:-1] if '%' in col else col for col in metrics.columns
]
metrics = metrics.T
if "benchmark" in df:
metrics.columns = ['Strategy', 'Benchmark']
else:
metrics.columns = ['Strategy']
if display:
print(_tabulate(metrics, headers="keys", tablefmt='simple'))
return None
if not sep:
metrics = metrics[metrics.index != '']
return metrics
from math import sqrt as _sqrt
_upper_limit = 100
PI = 0
for k in range(0, _upper_limit + 1):
PI += ((-1)**k) / ((3**k) * (2 * k + 1))
PI *= _sqrt(12)
def norm(v):
return _sqrt(dot(v, v))
def plot_network(self,
state_sizes=None,
state_scale=1.0,
state_colors='#ff5500',
state_labels='auto',
arrow_scale=1.0,
arrow_curvature=1.0,
arrow_labels='weights',
arrow_label_format='%10.2f',
max_width=12,
max_height=12,
figpadding=0.2,
xticks=False,
yticks=False,
show_frame=False,
**textkwargs):
"""
Draws a network using discs and curved arrows.
The thicknesses and labels of the arrows are taken from the off-diagonal matrix elements
in A.
"""
# Set the default values for the text dictionary
from matplotlib import pyplot as _plt
textkwargs.setdefault('size', None)
textkwargs.setdefault('horizontalalignment', 'center')
textkwargs.setdefault('verticalalignment', 'center')
textkwargs.setdefault('color', 'black')
# remove the temporary key 'arrow_label_size' as it cannot be parsed by plt.text!
arrow_label_size = textkwargs.pop('arrow_label_size',
textkwargs['size'])
if self.pos is None:
self.layout_automatic()
# number of nodes
n = len(self.pos)
# get bounds and pad figure
xmin = _np.min(self.pos[:, 0])
xmax = _np.max(self.pos[:, 0])
Dx = xmax - xmin
xmin -= Dx * figpadding
xmax += Dx * figpadding
Dx *= 1 + figpadding
ymin = _np.min(self.pos[:, 1])
ymax = _np.max(self.pos[:, 1])
Dy = ymax - ymin
ymin -= Dy * figpadding
ymax += Dy * figpadding
Dy *= 1 + figpadding
# sizes of nodes
if state_sizes is None:
state_sizes = 0.5 * state_scale * \
min(Dx, Dy)**2 * _np.ones(n) / float(n)
else:
state_sizes = 0.5 * state_scale * \
min(Dx, Dy)**2 * state_sizes / (_np.max(state_sizes) * float(n))
# automatic arrow rescaling
arrow_scale *= 1.0 / \
(_np.max(self.A - _np.diag(_np.diag(self.A))) * _sqrt(n))
# size figure
if (Dx / max_width > Dy / max_height):
figsize = (max_width, Dy * (max_width / Dx))
else:
figsize = (Dx / Dy * max_height, max_height)
if self.ax is None:
logger.debug("creating new figure")
fig = _plt.figure(None, figsize=figsize)
self.ax = fig.add_subplot(111)
else:
fig = self.ax.figure
window_extend = self.ax.get_window_extent()
axes_ratio = window_extend.height / window_extend.width
data_ratio = (ymax - ymin) / (xmax - xmin)
q = axes_ratio / data_ratio
if q > 1.0:
ymin *= q
ymax *= q
else:
xmin /= q
xmax /= q
if not xticks:
self.ax.get_xaxis().set_ticks([])
if not yticks:
self.ax.get_yaxis().set_ticks([])
# show or suppress frame
self.ax.set_frame_on(show_frame)
# set node labels
if state_labels is None:
pass
elif isinstance(state_labels, str) and state_labels == 'auto':
state_labels = [str(i) for i in _np.arange(n)]
else:
if len(state_labels) != n:
raise ValueError(
"length of state_labels({}) has to match length of states({})."
.format(len(state_labels), n))
# set node colors
if state_colors is None:
state_colors = '#ff5500' # None is not acceptable
if isinstance(state_colors, str):
state_colors = [state_colors] * n
if isinstance(state_colors, list) and not len(state_colors) == n:
raise ValueError(
"Mistmatch between nstates and nr. state_colors (%u vs %u)" %
(n, len(state_colors)))
try:
colorscales = _types.ensure_ndarray(state_colors,
ndim=1,
kind='numeric')
colorscales /= colorscales.max()
state_colors = [
_plt.cm.binary(int(256.0 * colorscales[i])) for i in range(n)
]
except AssertionError:
# assume we have a list of strings now.
logger.debug("could not cast 'state_colors' to numeric values.")
# set arrow labels
if isinstance(arrow_labels, _np.ndarray):
L = arrow_labels
if isinstance(arrow_labels[0, 0], str):
arrow_label_format = '%s'
elif isinstance(arrow_labels,
str) and arrow_labels.lower() == 'weights':
L = self.A[:, :]
elif arrow_labels is None:
L = _np.empty(_np.shape(self.A), dtype=object)
L[:, :] = ''
arrow_label_format = '%s'
else:
raise ValueError('invalid arrow labels')
# draw circles
circles = []
for i in range(n):
# choose color
c = _plt.Circle(self.pos[i],
radius=_sqrt(0.5 * state_sizes[i]) / 2.0,
color=state_colors[i],
zorder=2)
circles.append(c)
self.ax.add_artist(c)
# add annotation
if state_labels is not None:
self.ax.text(self.pos[i][0],
self.pos[i][1],
state_labels[i],
zorder=3,
**textkwargs)
assert len(circles) == n, "%i != %i" % (len(circles), n)
# draw arrows
for i in range(n):
for j in range(i + 1, n):
if (abs(self.A[i, j]) > 0):
self._draw_arrow(self.pos[i, 0],
self.pos[i, 1],
self.pos[j, 0],
self.pos[j, 1],
Dx,
Dy,
label=arrow_label_format % L[i, j],
width=arrow_scale * self.A[i, j],
arrow_curvature=arrow_curvature,
patchA=circles[i],
patchB=circles[j],
shrinkA=3,
shrinkB=0,
arrow_label_size=arrow_label_size)
if (abs(self.A[j, i]) > 0):
self._draw_arrow(self.pos[j, 0],
self.pos[j, 1],
self.pos[i, 0],
self.pos[i, 1],
Dx,
Dy,
label=arrow_label_format % L[j, i],
width=arrow_scale * self.A[j, i],
arrow_curvature=arrow_curvature,
patchA=circles[j],
patchB=circles[i],
shrinkA=3,
shrinkB=0,
arrow_label_size=arrow_label_size)
# plot
self.ax.set_xlim(xmin, xmax)
self.ax.set_ylim(ymin, ymax)
return fig