def expand(self, **hints):
expr = self.args[0]
condition = self._condition
if not is_random(expr):
return expr
if isinstance(expr, Add):
return Add.fromiter(
Expectation(a, condition=condition).expand()
for a in expr.args)
expand_expr = _expand(expr)
if isinstance(expand_expr, Add):
return Add.fromiter(
Expectation(a, condition=condition).expand()
for a in expand_expr.args)
elif isinstance(expr, Mul):
rv = []
nonrv = []
for a in expr.args:
if is_random(a):
rv.append(a)
else:
nonrv.append(a)
return Mul.fromiter(nonrv) * Expectation(Mul.fromiter(rv),
condition=condition)
return self
def expand(self, **hints):
arg1 = self.args[0]
arg2 = self.args[1]
condition = self._condition
if arg1 == arg2:
return Variance(arg1, condition).expand()
if not is_random(arg1):
return S.Zero
if not is_random(arg2):
return S.Zero
arg1, arg2 = sorted([arg1, arg2], key=default_sort_key)
if isinstance(arg1, RandomSymbol) and isinstance(arg2, RandomSymbol):
return Covariance(arg1, arg2, condition)
coeff_rv_list1 = self._expand_single_argument(arg1.expand())
coeff_rv_list2 = self._expand_single_argument(arg2.expand())
addends = [
a * b * Covariance(*sorted([r1, r2], key=default_sort_key),
condition=condition)
for (a, r1) in coeff_rv_list1 for (b, r2) in coeff_rv_list2
]
return Add.fromiter(addends)
def expand(self, **hints):
arg = self.args[0]
condition = self._condition
if not is_random(arg):
return S.Zero
if isinstance(arg, RandomSymbol):
return self
elif isinstance(arg, Add):
rv = []
for a in arg.args:
if is_random(a):
rv.append(a)
variances = Add(
*map(lambda xv: Variance(xv, condition).expand(), rv))
map_to_covar = lambda x: 2 * Covariance(*x, condition=condition
).expand()
covariances = Add(
*map(map_to_covar, itertools.combinations(rv, 2)))
return variances + covariances
elif isinstance(arg, Mul):
nonrv = []
rv = []
for a in arg.args:
if is_random(a):
rv.append(a)
else:
nonrv.append(a**2)
if len(rv) == 0:
return S.Zero
return Mul.fromiter(nonrv) * Variance(Mul.fromiter(rv), condition)
# this expression contains a RandomSymbol somehow:
return self
def JointEigenDistribution(mat):
"""
Creates joint distribution of eigen values of matrices with random
expressions.
Parameters
==========
mat: Matrix
The matrix under consideration
Returns
=======
JointDistributionHandmade
Examples
========
>>> from sympy.stats import Normal, JointEigenDistribution
>>> from sympy import Matrix
>>> A = [[Normal('A00', 0, 1), Normal('A01', 0, 1)],
... [Normal('A10', 0, 1), Normal('A11', 0, 1)]]
>>> JointEigenDistribution(Matrix(A))
JointDistributionHandmade(-sqrt(A00**2 - 2*A00*A11 + 4*A01*A10 + A11**2)/2
+ A00/2 + A11/2, sqrt(A00**2 - 2*A00*A11 + 4*A01*A10 + A11**2)/2 + A00/2 + A11/2)
"""
eigenvals = mat.eigenvals(multiple=True)
if any(not is_random(eigenval) for eigenval in set(eigenvals)):
raise ValueError("Eigen values don't have any random expression, "
"joint distribution cannot be generated.")
return JointDistributionHandmade(*eigenvals)
def pmf(self, x):
x = sympify(x)
if not (x.is_number or x.is_Symbol or is_random(x)):
raise ValueError("'x' expected as an argument of type 'number', 'Symbol', or "
"'RandomSymbol' not %s" % (type(x)))
cond = Ge(x, 1) & Le(x, self.sides) & Contains(x, S.Integers)
return Piecewise((S.One/self.sides, cond), (S.Zero, True))
def rv(symbol, cls, *args):
args = list(map(sympify, args))
dist = cls(*args)
dist.check(*args)
pspace = SingleDiscretePSpace(symbol, dist)
if any(is_random(arg) for arg in args):
pspace = JointPSpace(symbol, CompoundDistribution(dist))
return pspace.value
def _expand_single_argument(cls, expr):
# return (coefficient, random_symbol) pairs:
if isinstance(expr, RandomSymbol):
return [(S.One, expr)]
elif isinstance(expr, Add):
outval = []
for a in expr.args:
if isinstance(a, Mul):
outval.append(cls._get_mul_nonrv_rv_tuple(a))
elif is_random(a):
outval.append((S.One, a))
return outval
elif isinstance(expr, Mul):
return [cls._get_mul_nonrv_rv_tuple(expr)]
elif is_random(expr):
return [(S.One, expr)]
def pmf(self, x):
x = sympify(x)
if not (x.is_number or x.is_Symbol or is_random(x)):
raise ValueError("'x' expected as an argument of type 'number', 'Symbol', or "
"'RandomSymbol' not %s" % (type(x)))
cond1 = Eq(x, 1) & x.is_integer
cond2 = Ge(x, 1) & Le(x, self.k) & x.is_integer
return Piecewise((1/self.k, cond1), (1/(x*(x - 1)), cond2), (S.Zero, True))
def pmf(self, x):
n, p = self.n, self.p
x = sympify(x)
if not (x.is_number or x.is_Symbol or is_random(x)):
raise ValueError("'x' expected as an argument of type 'number', 'Symbol', or "
"'RandomSymbol' not %s" % (type(x)))
cond = Ge(x, 0) & Le(x, n) & Contains(x, S.Integers)
return Piecewise((binomial(n, x) * p**x * (1 - p)**(n - x), cond), (S.Zero, True))
def _get_mul_nonrv_rv_tuple(cls, m):
rv = []
nonrv = []
for a in m.args:
if is_random(a):
rv.append(a)
else:
nonrv.append(a)
return (Mul.fromiter(nonrv), Mul.fromiter(rv))
def rv(symbol, cls, *args):
args = list(map(sympify, args))
dist = cls(*args)
dist.check(*args)
pspace = SingleDiscretePSpace(symbol, dist)
if any(is_random(arg) for arg in args):
from sympy.stats.compound_rv import CompoundPSpace, CompoundDistribution
pspace = CompoundPSpace(symbol, CompoundDistribution(dist))
return pspace.value
def expand(self, **hints):
arg = self.args[0]
condition = self._condition
if not is_random(arg):
return ZeroMatrix(*self.shape)
if isinstance(arg, RandomSymbol):
return self
elif isinstance(arg, Add):
rv = []
for a in arg.args:
if is_random(a):
rv.append(a)
variances = Add(
*map(lambda xv: Variance(xv, condition).expand(), rv))
map_to_covar = lambda x: 2 * Covariance(*x, condition=condition
).expand()
covariances = Add(
*map(map_to_covar, itertools.combinations(rv, 2)))
return variances + covariances
elif isinstance(arg, (Mul, MatMul)):
nonrv = []
rv = []
for a in arg.args:
if is_random(a):
rv.append(a)
else:
nonrv.append(a)
if len(rv) == 0:
return ZeroMatrix(*self.shape)
# Avoid possible infinite loops with MatMul:
if len(nonrv) == 0:
return self
# Variance of many multiple matrix products is not implemented:
if len(rv) > 1:
return self
return Mul.fromiter(nonrv) * Variance(
Mul.fromiter(rv),
condition) * (Mul.fromiter(nonrv)).transpose()
# this expression contains a RandomSymbol somehow:
return self
def expand(self, **hints):
expr = self.args[0]
condition = self._condition
if not is_random(expr):
return expr
if isinstance(expr, Add):
return Add.fromiter(
Expectation(a, condition=condition).expand()
for a in expr.args)
expand_expr = _expand(expr)
if isinstance(expand_expr, Add):
return Add.fromiter(
Expectation(a, condition=condition).expand()
for a in expand_expr.args)
elif isinstance(expr, (Mul, MatMul)):
rv = []
nonrv = []
postnon = []
for a in expr.args:
if is_random(a):
if rv:
rv.extend(postnon)
else:
nonrv.extend(postnon)
postnon = []
rv.append(a)
elif a.is_Matrix:
postnon.append(a)
else:
nonrv.append(a)
# In order to avoid infinite-looping (MatMul may call .doit() again),
# do not rebuild
if len(nonrv) == 0:
return self
return Mul.fromiter(nonrv) * Expectation(
Mul.fromiter(rv), condition=condition) * Mul.fromiter(postnon)
return self
def rv(name, cls, *args, **kwargs):
args = list(map(sympify, args))
dist = cls(*args)
if kwargs.pop('check', True):
dist.check(*args)
pspace = SingleFinitePSpace(name, dist)
if any(is_random(arg) for arg in args):
from sympy.stats.compound_rv import CompoundPSpace, CompoundDistribution
pspace = CompoundPSpace(name, CompoundDistribution(dist))
return pspace.value
def __new__(cls, expr, condition=None, **kwargs):
expr = _sympify(expr)
if condition is None:
if not is_random(expr):
return expr
obj = Expr.__new__(cls, expr)
else:
condition = _sympify(condition)
obj = Expr.__new__(cls, expr, condition)
obj._condition = condition
return obj
def expand(self, **hints):
arg1 = self.args[0]
arg2 = self.args[1]
condition = self._condition
if arg1 == arg2:
return VarianceMatrix(arg1, condition).expand()
if not is_random(arg1) or not is_random(arg2):
return ZeroMatrix(*self.shape)
if isinstance(arg1, RandomSymbol) and isinstance(arg2, RandomSymbol):
return CrossCovarianceMatrix(arg1, arg2, condition)
coeff_rv_list1 = self._expand_single_argument(arg1.expand())
coeff_rv_list2 = self._expand_single_argument(arg2.expand())
addends = [a*CrossCovarianceMatrix(r1, r2, condition=condition)*b.transpose()
for (a, r1) in coeff_rv_list1 for (b, r2) in coeff_rv_list2]
return Add.fromiter(addends)
def pdf(self, x, evaluate=False):
dist = self.args[0]
randoms = [rv for rv in dist.args if is_random(rv)]
if isinstance(dist, SingleFiniteDistribution):
y = Dummy('y', integer=True, negative=False)
expr = dist.pmf(y)
else:
y = Dummy('y')
expr = dist.pdf(y)
for rv in randoms:
expr = self._marginalise(expr, rv, evaluate)
return Lambda(y, expr)(x)
def __new__(cls, expr, condition=None, **kwargs):
expr = _sympify(expr)
if expr.is_Matrix:
from sympy.stats.symbolic_multivariate_probability import ExpectationMatrix
return ExpectationMatrix(expr, condition)
if condition is None:
if not is_random(expr):
return expr
obj = Expr.__new__(cls, expr)
else:
condition = _sympify(condition)
obj = Expr.__new__(cls, expr, condition)
obj._condition = condition
return obj
def pmf(self, x):
x = sympify(x)
if not (x.is_number or x.is_Symbol or is_random(x)):
raise ValueError("'x' expected as an argument of type 'number', 'Symbol', or "
"'RandomSymbol' not %s" % (type(x)))
cond1 = Eq(x, 1) & x.is_integer
cond2 = Ge(x, 1) & Le(x, self.k) & x.is_integer
rho = Piecewise((Rational(1, self.k), cond1), (Rational(1, x*(x-1)), cond2), (S.Zero, True))
cond1 = Ge(x, 1) & Le(x, round(self.k/self.R)-1)
cond2 = Eq(x, round(self.k/self.R))
tau = Piecewise((self.R/(self.k * x), cond1), (self.R * log(self.R/self.delta)/self.k, cond2), (S.Zero, True))
return (rho + tau)/self.Z
def expand(self, **hints):
expr = self.args[0]
condition = self._condition
if not is_random(expr):
return expr
if isinstance(expr, Add):
return Add(*[
Expectation(a, condition=condition).expand() for a in expr.args
])
elif isinstance(expr, Mul):
if isinstance(_expand(expr), Add):
return Expectation(_expand(expr)).expand()
rv = []
nonrv = []
for a in expr.args:
if is_random(a):
rv.append(a)
else:
nonrv.append(a)
return Mul(*nonrv) * Expectation(Mul(*rv), condition=condition)
return self
def doit(self, **hints):
deep = hints.get('deep', True)
condition = self._condition
expr = self.args[0]
numsamples = hints.get('numsamples', False)
for_rewrite = not hints.get('for_rewrite', False)
if deep:
expr = expr.doit(**hints)
if not is_random(expr) or isinstance(
expr, Expectation): # expr isn't random?
return expr
if numsamples: # Computing by monte carlo sampling?
evalf = hints.get('evalf', True)
return sampling_E(expr,
condition,
numsamples=numsamples,
evalf=evalf)
if expr.has(RandomIndexedSymbol):
return pspace(expr).compute_expectation(expr, condition)
# Create new expr and recompute E
if condition is not None: # If there is a condition
return self.func(given(expr, condition)).doit(**hints)
# A few known statements for efficiency
if expr.is_Add: # We know that E is Linear
return Add(*[
self.func(arg, condition).doit(
**hints) if not isinstance(arg, Expectation) else self.
func(arg, condition) for arg in expr.args
])
if expr.is_Mul:
if expr.atoms(Expectation):
return expr
if pspace(expr) == PSpace():
return self.func(expr)
# Otherwise case is simple, pass work off to the ProbabilitySpace
result = pspace(expr).compute_expectation(expr, evaluate=for_rewrite)
if hasattr(result, 'doit') and for_rewrite:
return result.doit(**hints)
else:
return result
def probability(self, condition, **kwargs):
z = Dummy('z', real=True)
cond_inv = False
if isinstance(condition, Ne):
condition = Eq(condition.args[0], condition.args[1])
cond_inv = True
# Univariate case can be handled by where
try:
domain = self.where(condition)
rv = [rv for rv in self.values if rv.symbol == domain.symbol][0]
# Integrate out all other random variables
pdf = self.compute_density(rv, **kwargs)
# return S.Zero if `domain` is empty set
if domain.set is S.EmptySet or isinstance(domain.set, FiniteSet):
return S.Zero if not cond_inv else S.One
if isinstance(domain.set, Union):
return sum(
Integral(pdf(z), (z, subset), **kwargs)
for subset in domain.set.args
if isinstance(subset, Interval))
# Integrate out the last variable over the special domain
return Integral(pdf(z), (z, domain.set), **kwargs)
# Other cases can be turned into univariate case
# by computing a density handled by density computation
except NotImplementedError:
from sympy.stats.rv import density
expr = condition.lhs - condition.rhs
if not is_random(expr):
dens = self.density
comp = condition.rhs
else:
dens = density(expr, **kwargs)
comp = 0
if not isinstance(dens, ContinuousDistribution):
from sympy.stats.crv_types import ContinuousDistributionHandmade
dens = ContinuousDistributionHandmade(dens,
set=self.domain.set)
# Turn problem into univariate case
space = SingleContinuousPSpace(z, dens)
result = space.probability(condition.__class__(space.value, comp))
return result if not cond_inv else S.One - result
def doit(self, **hints):
if not is_random(self.args[0]):
return self.args[0]
return self.rewrite(Expectation).doit(**hints)
def test_is_random():
X = Normal('X', 0, 1)
Y = Normal('Y', 0, 1)
a, b = symbols('a, b')
G = GaussianUnitaryEnsemble('U', 2)
B = BernoulliProcess('B', 0.9)
assert not is_random(a)
assert not is_random(a + b)
assert not is_random(a * b)
assert not is_random(Matrix([a**2, b**2]))
assert is_random(X)
assert is_random(X**2 + Y)
assert is_random(Y + b**2)
assert is_random(Y > 5)
assert is_random(B[3] < 1)
assert is_random(G)
assert is_random(X * Y * B[1])
assert is_random(Matrix([[X, B[2]], [G, Y]]))
assert is_random(Eq(X, 4))
def variance_prop(expr, consts=(), include_covar=False):
r"""Symbolically propagates variance (`\sigma^2`) for expressions.
This is computed as as seen in [1]_.
Parameters
==========
expr : Expr
A sympy expression to compute the variance for.
consts : sequence of Symbols, optional
Represents symbols that are known constants in the expr,
and thus have zero variance. All symbols not in consts are
assumed to be variant.
include_covar : bool, optional
Flag for whether or not to include covariances, default=False.
Returns
=======
var_expr : Expr
An expression for the total variance of the expr.
The variance for the original symbols (e.g. x) are represented
via instance of the Variance symbol (e.g. Variance(x)).
Examples
========
>>> from sympy import symbols, exp
>>> from sympy.stats.error_prop import variance_prop
>>> x, y = symbols('x y')
>>> variance_prop(x + y)
Variance(x) + Variance(y)
>>> variance_prop(x * y)
x**2*Variance(y) + y**2*Variance(x)
>>> variance_prop(exp(2*x))
4*exp(4*x)*Variance(x)
References
==========
.. [1] https://en.wikipedia.org/wiki/Propagation_of_uncertainty
"""
args = expr.args
if len(args) == 0:
if expr in consts:
return S.Zero
elif is_random(expr):
return Variance(expr).doit()
elif isinstance(expr, Symbol):
return Variance(RandomSymbol(expr)).doit()
else:
return S.Zero
nargs = len(args)
var_args = list(map(variance_prop, args, repeat(consts, nargs),
repeat(include_covar, nargs)))
if isinstance(expr, Add):
var_expr = Add(*var_args)
if include_covar:
terms = [2 * Covariance(_arg0_or_var(x), _arg0_or_var(y)).expand() \
for x, y in combinations(var_args, 2)]
var_expr += Add(*terms)
elif isinstance(expr, Mul):
terms = [v/a**2 for a, v in zip(args, var_args)]
var_expr = simplify(expr**2 * Add(*terms))
if include_covar:
terms = [2*Covariance(_arg0_or_var(x), _arg0_or_var(y)).expand()/(a*b) \
for (a, b), (x, y) in zip(combinations(args, 2),
combinations(var_args, 2))]
var_expr += Add(*terms)
elif isinstance(expr, Pow):
b = args[1]
v = var_args[0] * (expr * b / args[0])**2
var_expr = simplify(v)
elif isinstance(expr, exp):
var_expr = simplify(var_args[0] * expr**2)
else:
# unknown how to proceed, return variance of whole expr.
var_expr = Variance(expr)
return var_expr
def _(x):
return is_random(x.base)
def _(x):
atoms = x.free_symbols
if len(atoms) == 1 and next(iter(atoms)) == x:
return False
return any([is_random(i) for i in atoms])
