def eval_rational(self, dx=None, dy=None, n=15):
"""
Return a Rational approximation of ``self`` that has real
and imaginary component approximations that are within ``dx``
and ``dy`` of the true values, respectively. Alternatively,
``n`` digits of precision can be specified.
The interval is refined with bisection and is sure to
converge. The root bounds are updated when the refinement
is complete so recalculation at the same or lesser precision
will not have to repeat the refinement and should be much
faster.
The following example first obtains Rational approximation to
1e-8 accuracy for all roots of the 4-th order Legendre
polynomial. Since the roots are all less than 1, this will
ensure the decimal representation of the approximation will be
correct (including rounding) to 6 digits:
>>> from sympy import legendre_poly, Symbol
>>> x = Symbol("x")
>>> p = legendre_poly(4, x, polys=True)
>>> r = p.real_roots()[-1]
>>> r.eval_rational(10**-8).n(6)
0.861136
It is not necessary to a two-step calculation, however: the
decimal representation can be computed directly:
>>> r.evalf(17)
0.86113631159405258
"""
dy = dy or dx
if dx:
rtol = None
dx = dx if isinstance(dx, Rational) else Rational(str(dx))
dy = dy if isinstance(dy, Rational) else Rational(str(dy))
else:
# 5 binary (or 2 decimal) digits are needed to ensure that
# a given digit is correctly rounded
# prec_to_dps(dps_to_prec(n) + 5) - n <= 2 (tested for
# n in range(1000000)
rtol = S(10)**-(n + 2) # +2 for guard digits
interval = self._get_interval()
while True:
if self.is_real:
if rtol:
dx = abs(interval.center * rtol)
interval = interval.refine_size(dx=dx)
c = interval.center
real = Rational(c)
imag = S.Zero
if not rtol or interval.dx < abs(c * rtol):
break
elif self.is_imaginary:
if rtol:
dy = abs(interval.center[1] * rtol)
dx = 1
interval = interval.refine_size(dx=dx, dy=dy)
c = interval.center[1]
imag = Rational(c)
real = S.Zero
if not rtol or interval.dy < abs(c * rtol):
break
else:
if rtol:
dx = abs(interval.center[0] * rtol)
dy = abs(interval.center[1] * rtol)
interval = interval.refine_size(dx, dy)
c = interval.center
real, imag = map(Rational, c)
if not rtol or (interval.dx < abs(c[0] * rtol)
and interval.dy < abs(c[1] * rtol)):
break
# update the interval so we at least (for this precision or
# less) don't have much work to do to recompute the root
self._set_interval(interval)
return real + I * imag
def test_imag():
I = S('I')
assert mcode(I) == "1i"
assert mcode(5 * I) == "5i"
assert mcode((S(3) / 2) * I) == "3*1i/2"
assert mcode(3 + 4 * I) == "3 + 4i"
def pde_1st_linear_constant_coeff_homogeneous(eq, func, order, match,
solvefun):
r"""
Solves a first order linear homogeneous
partial differential equation with constant coefficients.
The general form of this partial differential equation is
.. math:: a \frac{df(x,y)}{dx} + b \frac{df(x,y)}{dy} + c f(x,y) = 0
where `a`, `b` and `c` are constants.
The general solution is of the form::
>>> from sympy.solvers import pdsolve
>>> from sympy.abc import x, y, a, b, c
>>> from sympy import Function, pprint
>>> f = Function('f')
>>> u = f(x,y)
>>> ux = u.diff(x)
>>> uy = u.diff(y)
>>> genform = a*ux + b*uy + c*u
>>> pprint(genform)
d d
a*--(f(x, y)) + b*--(f(x, y)) + c*f(x, y)
dx dy
>>> pprint(pdsolve(genform))
-c*(a*x + b*y)
---------------
2 2
a + b
f(x, y) = F(-a*y + b*x)*e
Examples
========
>>> from sympy.solvers.pde import (
... pde_1st_linear_constant_coeff_homogeneous)
>>> from sympy import pdsolve
>>> from sympy import Function, diff, pprint
>>> from sympy.abc import x,y
>>> f = Function('f')
>>> pdsolve(f(x,y) + f(x,y).diff(x) + f(x,y).diff(y))
Eq(f(x, y), F(x - y)*exp(-x/2 - y/2))
>>> pprint(pdsolve(f(x,y) + f(x,y).diff(x) + f(x,y).diff(y)))
x y
- - - -
2 2
f(x, y) = F(x - y)*e
References
==========
- Viktor Grigoryan, "Partial Differential Equations"
Math 124A - Fall 2010, pp.7
"""
# TODO : For now homogeneous first order linear PDE's having
# two variables are implemented. Once there is support for
# solving systems of ODE's, this can be extended to n variables.
f = func.func
x = func.args[0]
y = func.args[1]
b = match[match['b']]
c = match[match['c']]
d = match[match['d']]
return Eq(
f(x, y),
exp(-S(d) / (b**2 + c**2) * (b * x + c * y)) * solvefun(c * x - b * y))
def _expr_big_minus(cls, z, n):
return HyperRep_asin1._expr_big_minus(z, n) \
/HyperRep_power1._expr_big_minus(S(1)/2, z, n)
def _expr_small_minus(cls, z):
return log(S(1) / 2 + sqrt(1 + z) / 2)
def delta(self):
""" A quantity related to the convergence region of the integral,
c.f. references. """
return len(self.bm) + len(self.an) - S(len(self.ap) + len(self.bq)) / 2
def _expr_big_minus(cls, z, n):
return S(-1)**n * (asinh(sqrt(z)) / sqrt(z) + n * pi * I / sqrt(z))
def gauss_gen_laguerre(n, alpha, n_digits):
r"""
Computes the generalized Gauss-Laguerre quadrature [1]_ points and weights.
The generalized Gauss-Laguerre quadrature approximates the integral:
.. math::
\int_{0}^\infty x^{\alpha} e^{-x} f(x)\,dx \approx \sum_{i=1}^n w_i f(x_i)
The nodes `x_i` of an order `n` quadrature rule are the roots of `L^{\alpha}_n`
and the weights `w_i` are given by:
.. math::
w_i = \frac{\Gamma(\alpha+n)}{n \Gamma(n) L^{\alpha}_{n-1}(x_i) L^{\alpha+1}_{n-1}(x_i)}
Parameters
==========
n : the order of quadrature
alpha : the exponent of the singularity, `\alpha > -1`
n_digits : number of significant digits of the points and weights to return
Returns
=======
(x, w) : the ``x`` and ``w`` are lists of points and weights as Floats.
The points `x_i` and weights `w_i` are returned as ``(x, w)``
tuple of lists.
Examples
========
>>> from sympy import S
>>> from sympy.integrals.quadrature import gauss_gen_laguerre
>>> x, w = gauss_gen_laguerre(3, -S.Half, 5)
>>> x
[0.19016, 1.7845, 5.5253]
>>> w
[1.4493, 0.31413, 0.00906]
>>> x, w = gauss_gen_laguerre(4, 3*S.Half, 5)
>>> x
[0.97851, 2.9904, 6.3193, 11.712]
>>> w
[0.53087, 0.67721, 0.11895, 0.0023152]
See Also
========
gauss_legendre, gauss_laguerre, gauss_hermite, gauss_chebyshev_t, gauss_chebyshev_u, gauss_jacobi
References
==========
.. [1] http://en.wikipedia.org/wiki/Gauss%E2%80%93Laguerre_quadrature
.. [2] http://people.sc.fsu.edu/~jburkardt/cpp_src/gen_laguerre_rule/gen_laguerre_rule.html
"""
x = Dummy("x")
p = laguerre_poly(n, x, alpha=alpha, polys=True)
p1 = laguerre_poly(n - 1, x, alpha=alpha, polys=True)
p2 = laguerre_poly(n - 1, x, alpha=alpha + 1, polys=True)
xi = []
w = []
for r in p.real_roots():
if isinstance(r, RootOf):
r = r.eval_rational(S(1) / 10**(n_digits + 2))
xi.append(r.n(n_digits))
w.append((gamma(alpha + n) /
(n * gamma(n) * p1.subs(x, r) * p2.subs(x, r))).n(n_digits))
return xi, w
def gauss_jacobi(n, alpha, beta, n_digits):
r"""
Computes the Gauss-Jacobi quadrature [1]_ points and weights.
The Gauss-Jacobi quadrature of the first kind approximates the integral:
.. math::
\int_{-1}^1 (1-x)^\alpha (1+x)^\beta f(x)\,dx \approx \sum_{i=1}^n w_i f(x_i)
The nodes `x_i` of an order `n` quadrature rule are the roots of `P^{(\alpha,\beta)}_n`
and the weights `w_i` are given by:
.. math::
w_i = -\frac{2n+\alpha+\beta+2}{n+\alpha+\beta+1}\frac{\Gamma(n+\alpha+1)\Gamma(n+\beta+1)}
{\Gamma(n+\alpha+\beta+1)(n+1)!} \frac{2^{\alpha+\beta}}{P'_n(x_i)
P^{(\alpha,\beta)}_{n+1}(x_i)}
Parameters
==========
n : the order of quadrature
alpha : the first parameter of the Jacobi Polynomial, `\alpha > -1`
beta : the second parameter of the Jacobi Polynomial, `\beta > -1`
n_digits : number of significant digits of the points and weights to return
Returns
=======
(x, w) : the ``x`` and ``w`` are lists of points and weights as Floats.
The points `x_i` and weights `w_i` are returned as ``(x, w)``
tuple of lists.
Examples
========
>>> from sympy import S
>>> from sympy.integrals.quadrature import gauss_jacobi
>>> x, w = gauss_jacobi(3, S.Half, -S.Half, 5)
>>> x
[-0.90097, -0.22252, 0.62349]
>>> w
[1.7063, 1.0973, 0.33795]
>>> x, w = gauss_jacobi(6, 1, 1, 5)
>>> x
[-0.87174, -0.5917, -0.2093, 0.2093, 0.5917, 0.87174]
>>> w
[0.050584, 0.22169, 0.39439, 0.39439, 0.22169, 0.050584]
See Also
========
gauss_legendre, gauss_laguerre, gauss_hermite, gauss_gen_laguerre, gauss_chebyshev_t, gauss_chebyshev_u
References
==========
.. [1] http://en.wikipedia.org/wiki/Gauss%E2%80%93Jacobi_quadrature
.. [2] http://people.sc.fsu.edu/~jburkardt/cpp_src/jacobi_rule/jacobi_rule.html
.. [3] http://people.sc.fsu.edu/~jburkardt/cpp_src/gegenbauer_rule/gegenbauer_rule.html
"""
x = Dummy("x")
p = jacobi_poly(n, alpha, beta, x, polys=True)
pd = p.diff(x)
pn = jacobi_poly(n + 1, alpha, beta, x, polys=True)
xi = []
w = []
for r in p.real_roots():
if isinstance(r, RootOf):
r = r.eval_rational(S(1) / 10**(n_digits + 2))
xi.append(r.n(n_digits))
w.append(
(-(2 * n + alpha + beta + 2) / (n + alpha + beta + S.One) *
(gamma(n + alpha + 1) * gamma(n + beta + 1)) /
(gamma(n + alpha + beta + S.One) * gamma(n + 2)) *
2**(alpha + beta) / (pd.subs(x, r) * pn.subs(x, r))).n(n_digits))
return xi, w
def gauss_legendre(n, n_digits):
r"""
Computes the Gauss-Legendre quadrature [1]_ points and weights.
The Gauss-Legendre quadrature approximates the integral:
.. math::
\int_{-1}^1 f(x)\,dx \approx \sum_{i=1}^n w_i f(x_i)
The nodes `x_i` of an order `n` quadrature rule are the roots of `P_n`
and the weights `w_i` are given by:
.. math::
w_i = \frac{2}{\left(1-x_i^2\right) \left(P'_n(x_i)\right)^2}
Parameters
==========
n : the order of quadrature
n_digits : number of significant digits of the points and weights to return
Returns
=======
(x, w) : the ``x`` and ``w`` are lists of points and weights as Floats.
The points `x_i` and weights `w_i` are returned as ``(x, w)``
tuple of lists.
Examples
========
>>> from sympy.integrals.quadrature import gauss_legendre
>>> x, w = gauss_legendre(3, 5)
>>> x
[-0.7746, 0, 0.7746]
>>> w
[0.55556, 0.88889, 0.55556]
>>> x, w = gauss_legendre(4, 5)
>>> x
[-0.86114, -0.33998, 0.33998, 0.86114]
>>> w
[0.34786, 0.65215, 0.65215, 0.34786]
See Also
========
gauss_laguerre, gauss_gen_laguerre, gauss_hermite, gauss_chebyshev_t, gauss_chebyshev_u, gauss_jacobi
References
==========
.. [1] http://en.wikipedia.org/wiki/Gaussian_quadrature
.. [2] http://people.sc.fsu.edu/~jburkardt/cpp_src/legendre_rule/legendre_rule.html
"""
x = Dummy("x")
p = legendre_poly(n, x, polys=True)
pd = p.diff(x)
xi = []
w = []
for r in p.real_roots():
if isinstance(r, RootOf):
r = r.eval_rational(S(1) / 10**(n_digits + 2))
xi.append(r.n(n_digits))
w.append((2 / ((1 - r**2) * pd.subs(x, r)**2)).n(n_digits))
return xi, w
def gauss_hermite(n, n_digits):
r"""
Computes the Gauss-Hermite quadrature [1]_ points and weights.
The Gauss-Hermite quadrature approximates the integral:
.. math::
\int_{-\infty}^{\infty} e^{-x^2} f(x)\,dx \approx \sum_{i=1}^n w_i f(x_i)
The nodes `x_i` of an order `n` quadrature rule are the roots of `H_n`
and the weights `w_i` are given by:
.. math::
w_i = \frac{2^{n-1} n! \sqrt{\pi}}{n^2 \left(H_{n-1}(x_i)\right)^2}
Parameters
==========
n : the order of quadrature
n_digits : number of significant digits of the points and weights to return
Returns
=======
(x, w) : the ``x`` and ``w`` are lists of points and weights as Floats.
The points `x_i` and weights `w_i` are returned as ``(x, w)``
tuple of lists.
Examples
========
>>> from sympy.integrals.quadrature import gauss_hermite
>>> x, w = gauss_hermite(3, 5)
>>> x
[-1.2247, 0, 1.2247]
>>> w
[0.29541, 1.1816, 0.29541]
>>> x, w = gauss_hermite(6, 5)
>>> x
[-2.3506, -1.3358, -0.43608, 0.43608, 1.3358, 2.3506]
>>> w
[0.00453, 0.15707, 0.72463, 0.72463, 0.15707, 0.00453]
See Also
========
gauss_legendre, gauss_laguerre, gauss_gen_laguerre, gauss_chebyshev_t, gauss_chebyshev_u, gauss_jacobi
References
==========
.. [1] http://en.wikipedia.org/wiki/Gauss-Hermite_Quadrature
.. [2] http://people.sc.fsu.edu/~jburkardt/cpp_src/hermite_rule/hermite_rule.html
.. [3] http://people.sc.fsu.edu/~jburkardt/cpp_src/gen_hermite_rule/gen_hermite_rule.html
"""
x = Dummy("x")
p = hermite_poly(n, x, polys=True)
p1 = hermite_poly(n - 1, x, polys=True)
xi = []
w = []
for r in p.real_roots():
if isinstance(r, RootOf):
r = r.eval_rational(S(1) / 10**(n_digits + 2))
xi.append(r.n(n_digits))
w.append(((2**(n - 1) * factorial(n) * sqrt(pi)) /
(n**2 * p1.subs(x, r)**2)).n(n_digits))
return xi, w
def test_deltasummation():
ds = deltasummation
assert ds(x, (j, 1, 0)) == 0
assert ds(x, (j, 1, 3)) == 3*x
assert ds(x + y, (j, 1, 3)) == 3*(x + y)
assert ds(x*y, (j, 1, 3)) == 3*x*y
assert ds(KD(i, j), (k, 1, 3)) == 3*KD(i, j)
assert ds(x*KD(i, j), (k, 1, 3)) == 3*x*KD(i, j)
assert ds(x*y*KD(i, j), (k, 1, 3)) == 3*x*y*KD(i, j)
n = symbols('n', integer=True, nonzero=True)
assert ds(KD(n, 0), (n, 1, 3)) == 0
# return unevaluated, until it gets implemented
assert ds(KD(i**2, j**2), (j, -oo, oo)) == \
Sum(KD(i**2, j**2), (j, -oo, oo))
assert Piecewise((KD(i, k), And(S(1) <= i, i <= 3)), (0, True)) == \
ds(KD(i, j)*KD(j, k), (j, 1, 3)) == \
ds(KD(j, k)*KD(i, j), (j, 1, 3))
assert ds(KD(i, k), (k, -oo, oo)) == 1
assert ds(KD(i, k), (k, 0, oo)) == Piecewise((1, i >= 0), (0, True))
assert ds(KD(i, k), (k, 1, 3)) == \
Piecewise((1, And(S(1) <= i, i <= 3)), (0, True))
assert ds(k*KD(i, j)*KD(j, k), (k, -oo, oo)) == j*KD(i, j)
assert ds(j*KD(i, j), (j, -oo, oo)) == i
assert ds(i*KD(i, j), (i, -oo, oo)) == j
assert ds(x, (i, 1, 3)) == 3*x
assert ds((i + j)*KD(i, j), (j, -oo, oo)) == 2*i
assert ds(KD(i, j), (j, 1, 3)) == \
Piecewise((1, And(S(1) <= i, i <= 3)), (0, True))
assert ds(KD(i, j), (j, 1, 1)) == Piecewise((1, Eq(i, 1)), (0, True))
assert ds(KD(i, j), (j, 2, 2)) == Piecewise((1, Eq(i, 2)), (0, True))
assert ds(KD(i, j), (j, 3, 3)) == Piecewise((1, Eq(i, 3)), (0, True))
assert ds(KD(i, j), (j, 1, k)) == \
Piecewise((1, And(S(1) <= i, i <= k)), (0, True))
assert ds(KD(i, j), (j, k, 3)) == \
Piecewise((1, And(k <= i, i <= 3)), (0, True))
assert ds(KD(i, j), (j, k, l)) == \
Piecewise((1, And(k <= i, i <= l)), (0, True))
assert ds(x*KD(i, j), (j, 1, 3)) == \
Piecewise((x, And(S(1) <= i, i <= 3)), (0, True))
assert ds(x*KD(i, j), (j, 1, 1)) == Piecewise((x, Eq(i, 1)), (0, True))
assert ds(x*KD(i, j), (j, 2, 2)) == Piecewise((x, Eq(i, 2)), (0, True))
assert ds(x*KD(i, j), (j, 3, 3)) == Piecewise((x, Eq(i, 3)), (0, True))
assert ds(x*KD(i, j), (j, 1, k)) == \
Piecewise((x, And(S(1) <= i, i <= k)), (0, True))
assert ds(x*KD(i, j), (j, k, 3)) == \
Piecewise((x, And(k <= i, i <= 3)), (0, True))
assert ds(x*KD(i, j), (j, k, l)) == \
Piecewise((x, And(k <= i, i <= l)), (0, True))
assert ds((x + y)*KD(i, j), (j, 1, 3)) == \
Piecewise((x + y, And(S(1) <= i, i <= 3)), (0, True))
assert ds((x + y)*KD(i, j), (j, 1, 1)) == \
Piecewise((x + y, Eq(i, 1)), (0, True))
assert ds((x + y)*KD(i, j), (j, 2, 2)) == \
Piecewise((x + y, Eq(i, 2)), (0, True))
assert ds((x + y)*KD(i, j), (j, 3, 3)) == \
Piecewise((x + y, Eq(i, 3)), (0, True))
assert ds((x + y)*KD(i, j), (j, 1, k)) == \
Piecewise((x + y, And(S(1) <= i, i <= k)), (0, True))
assert ds((x + y)*KD(i, j), (j, k, 3)) == \
Piecewise((x + y, And(k <= i, i <= 3)), (0, True))
assert ds((x + y)*KD(i, j), (j, k, l)) == \
Piecewise((x + y, And(k <= i, i <= l)), (0, True))
assert ds(KD(i, k) + KD(j, k), (k, 1, 3)) == piecewise_fold(
Piecewise((1, And(S(1) <= i, i <= 3)), (0, True)) +
Piecewise((1, And(S(1) <= j, j <= 3)), (0, True)))
assert ds(KD(i, k) + KD(j, k), (k, 1, 1)) == piecewise_fold(
Piecewise((1, Eq(i, 1)), (0, True)) +
Piecewise((1, Eq(j, 1)), (0, True)))
assert ds(KD(i, k) + KD(j, k), (k, 2, 2)) == piecewise_fold(
Piecewise((1, Eq(i, 2)), (0, True)) +
Piecewise((1, Eq(j, 2)), (0, True)))
assert ds(KD(i, k) + KD(j, k), (k, 3, 3)) == piecewise_fold(
Piecewise((1, Eq(i, 3)), (0, True)) +
Piecewise((1, Eq(j, 3)), (0, True)))
assert ds(KD(i, k) + KD(j, k), (k, 1, l)) == piecewise_fold(
Piecewise((1, And(S(1) <= i, i <= l)), (0, True)) +
Piecewise((1, And(S(1) <= j, j <= l)), (0, True)))
assert ds(KD(i, k) + KD(j, k), (k, l, 3)) == piecewise_fold(
Piecewise((1, And(l <= i, i <= 3)), (0, True)) +
Piecewise((1, And(l <= j, j <= 3)), (0, True)))
assert ds(KD(i, k) + KD(j, k), (k, l, m)) == piecewise_fold(
Piecewise((1, And(l <= i, i <= m)), (0, True)) +
Piecewise((1, And(l <= j, j <= m)), (0, True)))
assert ds(x*KD(i, k) + KD(j, k), (k, 1, 3)) == piecewise_fold(
Piecewise((x, And(S(1) <= i, i <= 3)), (0, True)) +
Piecewise((1, And(S(1) <= j, j <= 3)), (0, True)))
assert ds(x*KD(i, k) + KD(j, k), (k, 1, 1)) == piecewise_fold(
Piecewise((x, Eq(i, 1)), (0, True)) +
Piecewise((1, Eq(j, 1)), (0, True)))
assert ds(x*KD(i, k) + KD(j, k), (k, 2, 2)) == piecewise_fold(
Piecewise((x, Eq(i, 2)), (0, True)) +
Piecewise((1, Eq(j, 2)), (0, True)))
assert ds(x*KD(i, k) + KD(j, k), (k, 3, 3)) == piecewise_fold(
Piecewise((x, Eq(i, 3)), (0, True)) +
Piecewise((1, Eq(j, 3)), (0, True)))
assert ds(x*KD(i, k) + KD(j, k), (k, 1, l)) == piecewise_fold(
Piecewise((x, And(S(1) <= i, i <= l)), (0, True)) +
Piecewise((1, And(S(1) <= j, j <= l)), (0, True)))
assert ds(x*KD(i, k) + KD(j, k), (k, l, 3)) == piecewise_fold(
Piecewise((x, And(l <= i, i <= 3)), (0, True)) +
Piecewise((1, And(l <= j, j <= 3)), (0, True)))
assert ds(x*KD(i, k) + KD(j, k), (k, l, m)) == piecewise_fold(
Piecewise((x, And(l <= i, i <= m)), (0, True)) +
Piecewise((1, And(l <= j, j <= m)), (0, True)))
assert ds(x*(KD(i, k) + KD(j, k)), (k, 1, 3)) == piecewise_fold(
Piecewise((x, And(S(1) <= i, i <= 3)), (0, True)) +
Piecewise((x, And(S(1) <= j, j <= 3)), (0, True)))
assert ds(x*(KD(i, k) + KD(j, k)), (k, 1, 1)) == piecewise_fold(
Piecewise((x, Eq(i, 1)), (0, True)) +
Piecewise((x, Eq(j, 1)), (0, True)))
assert ds(x*(KD(i, k) + KD(j, k)), (k, 2, 2)) == piecewise_fold(
Piecewise((x, Eq(i, 2)), (0, True)) +
Piecewise((x, Eq(j, 2)), (0, True)))
assert ds(x*(KD(i, k) + KD(j, k)), (k, 3, 3)) == piecewise_fold(
Piecewise((x, Eq(i, 3)), (0, True)) +
Piecewise((x, Eq(j, 3)), (0, True)))
assert ds(x*(KD(i, k) + KD(j, k)), (k, 1, l)) == piecewise_fold(
Piecewise((x, And(S(1) <= i, i <= l)), (0, True)) +
Piecewise((x, And(S(1) <= j, j <= l)), (0, True)))
assert ds(x*(KD(i, k) + KD(j, k)), (k, l, 3)) == piecewise_fold(
Piecewise((x, And(l <= i, i <= 3)), (0, True)) +
Piecewise((x, And(l <= j, j <= 3)), (0, True)))
assert ds(x*(KD(i, k) + KD(j, k)), (k, l, m)) == piecewise_fold(
Piecewise((x, And(l <= i, i <= m)), (0, True)) +
Piecewise((x, And(l <= j, j <= m)), (0, True)))
assert ds((x + y)*(KD(i, k) + KD(j, k)), (k, 1, 3)) == piecewise_fold(
Piecewise((x + y, And(S(1) <= i, i <= 3)), (0, True)) +
Piecewise((x + y, And(S(1) <= j, j <= 3)), (0, True)))
assert ds((x + y)*(KD(i, k) + KD(j, k)), (k, 1, 1)) == piecewise_fold(
Piecewise((x + y, Eq(i, 1)), (0, True)) +
Piecewise((x + y, Eq(j, 1)), (0, True)))
assert ds((x + y)*(KD(i, k) + KD(j, k)), (k, 2, 2)) == piecewise_fold(
Piecewise((x + y, Eq(i, 2)), (0, True)) +
Piecewise((x + y, Eq(j, 2)), (0, True)))
assert ds((x + y)*(KD(i, k) + KD(j, k)), (k, 3, 3)) == piecewise_fold(
Piecewise((x + y, Eq(i, 3)), (0, True)) +
Piecewise((x + y, Eq(j, 3)), (0, True)))
assert ds((x + y)*(KD(i, k) + KD(j, k)), (k, 1, l)) == piecewise_fold(
Piecewise((x + y, And(S(1) <= i, i <= l)), (0, True)) +
Piecewise((x + y, And(S(1) <= j, j <= l)), (0, True)))
assert ds((x + y)*(KD(i, k) + KD(j, k)), (k, l, 3)) == piecewise_fold(
Piecewise((x + y, And(l <= i, i <= 3)), (0, True)) +
Piecewise((x + y, And(l <= j, j <= 3)), (0, True)))
assert ds((x + y)*(KD(i, k) + KD(j, k)), (k, l, m)) == piecewise_fold(
Piecewise((x + y, And(l <= i, i <= m)), (0, True)) +
Piecewise((x + y, And(l <= j, j <= m)), (0, True)))
assert ds(x*y + x*KD(i, j), (j, 1, 3)) == \
Piecewise((3*x*y + x, And(S(1) <= i, i <= 3)), (3*x*y, True))
assert ds(x*y + x*KD(i, j), (j, 1, 1)) == \
Piecewise((x*y + x, Eq(i, 1)), (x*y, True))
assert ds(x*y + x*KD(i, j), (j, 2, 2)) == \
Piecewise((x*y + x, Eq(i, 2)), (x*y, True))
assert ds(x*y + x*KD(i, j), (j, 3, 3)) == \
Piecewise((x*y + x, Eq(i, 3)), (x*y, True))
assert ds(x*y + x*KD(i, j), (j, 1, k)) == \
Piecewise((k*x*y + x, And(S(1) <= i, i <= k)), (k*x*y, True))
assert ds(x*y + x*KD(i, j), (j, k, 3)) == \
Piecewise(((4 - k)*x*y + x, And(k <= i, i <= 3)), ((4 - k)*x*y, True))
assert ds(x*y + x*KD(i, j), (j, k, l)) == Piecewise(
((l - k + 1)*x*y + x, And(k <= i, i <= l)), ((l - k + 1)*x*y, True))
assert ds(x*(y + KD(i, j)), (j, 1, 3)) == \
Piecewise((3*x*y + x, And(S(1) <= i, i <= 3)), (3*x*y, True))
assert ds(x*(y + KD(i, j)), (j, 1, 1)) == \
Piecewise((x*y + x, Eq(i, 1)), (x*y, True))
assert ds(x*(y + KD(i, j)), (j, 2, 2)) == \
Piecewise((x*y + x, Eq(i, 2)), (x*y, True))
assert ds(x*(y + KD(i, j)), (j, 3, 3)) == \
Piecewise((x*y + x, Eq(i, 3)), (x*y, True))
assert ds(x*(y + KD(i, j)), (j, 1, k)) == \
Piecewise((k*x*y + x, And(S(1) <= i, i <= k)), (k*x*y, True))
assert ds(x*(y + KD(i, j)), (j, k, 3)) == \
Piecewise(((4 - k)*x*y + x, And(k <= i, i <= 3)), ((4 - k)*x*y, True))
assert ds(x*(y + KD(i, j)), (j, k, l)) == Piecewise(
((l - k + 1)*x*y + x, And(k <= i, i <= l)), ((l - k + 1)*x*y, True))
assert ds(x*(y + 2*KD(i, j)), (j, 1, 3)) == \
Piecewise((3*x*y + 2*x, And(S(1) <= i, i <= 3)), (3*x*y, True))
assert ds(x*(y + 2*KD(i, j)), (j, 1, 1)) == \
Piecewise((x*y + 2*x, Eq(i, 1)), (x*y, True))
assert ds(x*(y + 2*KD(i, j)), (j, 2, 2)) == \
Piecewise((x*y + 2*x, Eq(i, 2)), (x*y, True))
assert ds(x*(y + 2*KD(i, j)), (j, 3, 3)) == \
Piecewise((x*y + 2*x, Eq(i, 3)), (x*y, True))
assert ds(x*(y + 2*KD(i, j)), (j, 1, k)) == \
Piecewise((k*x*y + 2*x, And(S(1) <= i, i <= k)), (k*x*y, True))
assert ds(x*(y + 2*KD(i, j)), (j, k, 3)) == Piecewise(
((4 - k)*x*y + 2*x, And(k <= i, i <= 3)), ((4 - k)*x*y, True))
assert ds(x*(y + 2*KD(i, j)), (j, k, l)) == Piecewise(
((l - k + 1)*x*y + 2*x, And(k <= i, i <= l)), ((l - k + 1)*x*y, True))
assert ds((x + y)*(y + KD(i, j)), (j, 1, 3)) == Piecewise(
(3*(x + y)*y + x + y, And(S(1) <= i, i <= 3)), (3*(x + y)*y, True))
assert ds((x + y)*(y + KD(i, j)), (j, 1, 1)) == \
Piecewise(((x + y)*y + x + y, Eq(i, 1)), ((x + y)*y, True))
assert ds((x + y)*(y + KD(i, j)), (j, 2, 2)) == \
Piecewise(((x + y)*y + x + y, Eq(i, 2)), ((x + y)*y, True))
assert ds((x + y)*(y + KD(i, j)), (j, 3, 3)) == \
Piecewise(((x + y)*y + x + y, Eq(i, 3)), ((x + y)*y, True))
assert ds((x + y)*(y + KD(i, j)), (j, 1, k)) == Piecewise(
(k*(x + y)*y + x + y, And(S(1) <= i, i <= k)), (k*(x + y)*y, True))
assert ds((x + y)*(y + KD(i, j)), (j, k, 3)) == Piecewise(
((4 - k)*(x + y)*y + x + y, And(k <= i, i <= 3)),
((4 - k)*(x + y)*y, True))
assert ds((x + y)*(y + KD(i, j)), (j, k, l)) == Piecewise(
((l - k + 1)*(x + y)*y + x + y, And(k <= i, i <= l)),
((l - k + 1)*(x + y)*y, True))
assert ds((x + KD(i, k))*(y + KD(i, j)), (j, 1, 3)) == piecewise_fold(
Piecewise((KD(i, k) + x, And(S(1) <= i, i <= 3)), (0, True)) +
3*(KD(i, k) + x)*y)
assert ds((x + KD(i, k))*(y + KD(i, j)), (j, 1, 1)) == piecewise_fold(
Piecewise((KD(i, k) + x, Eq(i, 1)), (0, True)) +
(KD(i, k) + x)*y)
assert ds((x + KD(i, k))*(y + KD(i, j)), (j, 2, 2)) == piecewise_fold(
Piecewise((KD(i, k) + x, Eq(i, 2)), (0, True)) +
(KD(i, k) + x)*y)
assert ds((x + KD(i, k))*(y + KD(i, j)), (j, 3, 3)) == piecewise_fold(
Piecewise((KD(i, k) + x, Eq(i, 3)), (0, True)) +
(KD(i, k) + x)*y)
assert ds((x + KD(i, k))*(y + KD(i, j)), (j, 1, k)) == piecewise_fold(
Piecewise((KD(i, k) + x, And(S(1) <= i, i <= k)), (0, True)) +
k*(KD(i, k) + x)*y)
assert ds((x + KD(i, k))*(y + KD(i, j)), (j, k, 3)) == piecewise_fold(
Piecewise((KD(i, k) + x, And(k <= i, i <= 3)), (0, True)) +
(4 - k)*(KD(i, k) + x)*y)
assert ds((x + KD(i, k))*(y + KD(i, j)), (j, k, l)) == piecewise_fold(
Piecewise((KD(i, k) + x, And(k <= i, i <= l)), (0, True)) +
(l - k + 1)*(KD(i, k) + x)*y)
def equivalence_hypergeometric(A, B, func):
# This method for finding the equivalence is only for 2F1 type.
# We can extend it for 1F1 and 0F1 type also.
x = func.args[0]
# making given equation in normal form
I1 = factor(cancel(A.diff(x) / 2 + A**2 / 4 - B))
# computing shifted invariant(J1) of the equation
J1 = factor(cancel(x**2 * I1 + S(1) / 4))
num, dem = J1.as_numer_denom()
num = powdenest(expand(num))
dem = powdenest(expand(dem))
# this function will compute the different powers of variable(x) in J1.
# then it will help in finding value of k. k is power of x such that we can express
# J1 = x**k * J0(x**k) then all the powers in J0 become integers.
def _power_counting(num):
_pow = {0}
for val in num:
if val.has(x):
if isinstance(val, Pow) and val.as_base_exp()[0] == x:
_pow.add(val.as_base_exp()[1])
elif val == x:
_pow.add(val.as_base_exp()[1])
else:
_pow.update(_power_counting(val.args))
return _pow
pow_num = _power_counting((num, ))
pow_dem = _power_counting((dem, ))
pow_dem.update(pow_num)
_pow = pow_dem
k = gcd(_pow)
# computing I0 of the given equation
I0 = powdenest(simplify(factor(((J1 / k**2) - S(1) / 4) / ((x**k)**2))),
force=True)
I0 = factor(cancel(powdenest(I0.subs(x, x**(S(1) / k)), force=True)))
# Before this point I0, J1 might be functions of e.g. sqrt(x) but replacing
# x with x**(1/k) should result in I0 being a rational function of x or
# otherwise the hypergeometric solver cannot be used. Note that k can be a
# non-integer rational such as 2/7.
if not I0.is_rational_function(x):
return None
num, dem = I0.as_numer_denom()
max_num_pow = max(_power_counting((num, )))
dem_args = dem.args
sing_point = []
dem_pow = []
# calculating singular point of I0.
for arg in dem_args:
if arg.has(x):
if isinstance(arg, Pow):
# (x-a)**n
dem_pow.append(arg.as_base_exp()[1])
sing_point.append(
list(roots(arg.as_base_exp()[0], x).keys())[0])
else:
# (x-a) type
dem_pow.append(arg.as_base_exp()[1])
sing_point.append(list(roots(arg, x).keys())[0])
dem_pow.sort()
# checking if equivalence is exists or not.
if equivalence(max_num_pow, dem_pow) == "2F1":
return {'I0': I0, 'k': k, 'sing_point': sing_point, 'type': "2F1"}
else:
return None
def test_one_is_composite():
assert S(1).is_composite is False
def roots_quintic(f):
"""
Calculate exact roots of a solvable quintic
"""
result = []
coeff_5, coeff_4, p, q, r, s = f.all_coeffs()
# Eqn must be of the form x^5 + px^3 + qx^2 + rx + s
if coeff_4:
return result
if coeff_5 != 1:
l = [p / coeff_5, q / coeff_5, r / coeff_5, s / coeff_5]
if not all(coeff.is_Rational for coeff in l):
return result
f = Poly(f / coeff_5)
elif not all(coeff.is_Rational for coeff in (p, q, r, s)):
return result
quintic = PolyQuintic(f)
# Eqn standardized. Algo for solving starts here
if not f.is_irreducible:
return result
f20 = quintic.f20
# Check if f20 has linear factors over domain Z
if f20.is_irreducible:
return result
# Now, we know that f is solvable
for _factor in f20.factor_list()[1]:
if _factor[0].is_linear:
theta = _factor[0].root(0)
break
d = discriminant(f)
delta = sqrt(d)
# zeta = a fifth root of unity
zeta1, zeta2, zeta3, zeta4 = quintic.zeta
T = quintic.T(theta, d)
tol = S(1e-10)
alpha = T[1] + T[2] * delta
alpha_bar = T[1] - T[2] * delta
beta = T[3] + T[4] * delta
beta_bar = T[3] - T[4] * delta
disc = alpha**2 - 4 * beta
disc_bar = alpha_bar**2 - 4 * beta_bar
l0 = quintic.l0(theta)
Stwo = S(2)
l1 = _quintic_simplify((-alpha + sqrt(disc)) / Stwo)
l4 = _quintic_simplify((-alpha - sqrt(disc)) / Stwo)
l2 = _quintic_simplify((-alpha_bar + sqrt(disc_bar)) / Stwo)
l3 = _quintic_simplify((-alpha_bar - sqrt(disc_bar)) / Stwo)
order = quintic.order(theta, d)
test = (order * delta.n()) - ((l1.n() - l4.n()) * (l2.n() - l3.n()))
# Comparing floats
if not comp(test, 0, tol):
l2, l3 = l3, l2
# Now we have correct order of l's
R1 = l0 + l1 * zeta1 + l2 * zeta2 + l3 * zeta3 + l4 * zeta4
R2 = l0 + l3 * zeta1 + l1 * zeta2 + l4 * zeta3 + l2 * zeta4
R3 = l0 + l2 * zeta1 + l4 * zeta2 + l1 * zeta3 + l3 * zeta4
R4 = l0 + l4 * zeta1 + l3 * zeta2 + l2 * zeta3 + l1 * zeta4
Res = [None, [None] * 5, [None] * 5, [None] * 5, [None] * 5]
Res_n = [None, [None] * 5, [None] * 5, [None] * 5, [None] * 5]
sol = Symbol('sol')
# Simplifying improves performance a lot for exact expressions
R1 = _quintic_simplify(R1)
R2 = _quintic_simplify(R2)
R3 = _quintic_simplify(R3)
R4 = _quintic_simplify(R4)
# Solve imported here. Causing problems if imported as 'solve'
# and hence the changed name
from sympy.solvers.solvers import solve as _solve
a, b = symbols('a b', cls=Dummy)
_sol = _solve(sol**5 - a - I * b, sol)
for i in range(5):
_sol[i] = factor(_sol[i])
R1 = R1.as_real_imag()
R2 = R2.as_real_imag()
R3 = R3.as_real_imag()
R4 = R4.as_real_imag()
for i, currentroot in enumerate(_sol):
Res[1][i] = _quintic_simplify(currentroot.subs({a: R1[0], b: R1[1]}))
Res[2][i] = _quintic_simplify(currentroot.subs({a: R2[0], b: R2[1]}))
Res[3][i] = _quintic_simplify(currentroot.subs({a: R3[0], b: R3[1]}))
Res[4][i] = _quintic_simplify(currentroot.subs({a: R4[0], b: R4[1]}))
for i in range(1, 5):
for j in range(5):
Res_n[i][j] = Res[i][j].n()
Res[i][j] = _quintic_simplify(Res[i][j])
r1 = Res[1][0]
r1_n = Res_n[1][0]
for i in range(5):
if comp(im(r1_n * Res_n[4][i]), 0, tol):
r4 = Res[4][i]
break
# Now we have various Res values. Each will be a list of five
# values. We have to pick one r value from those five for each Res
u, v = quintic.uv(theta, d)
testplus = (u + v * delta * sqrt(5)).n()
testminus = (u - v * delta * sqrt(5)).n()
# Evaluated numbers suffixed with _n
# We will use evaluated numbers for calculation. Much faster.
r4_n = r4.n()
r2 = r3 = None
for i in range(5):
r2temp_n = Res_n[2][i]
for j in range(5):
# Again storing away the exact number and using
# evaluated numbers in computations
r3temp_n = Res_n[3][j]
if (comp((r1_n * r2temp_n**2 + r4_n * r3temp_n**2 - testplus).n(),
0, tol) and comp(
(r3temp_n * r1_n**2 + r2temp_n * r4_n**2 -
testminus).n(), 0, tol)):
r2 = Res[2][i]
r3 = Res[3][j]
break
if r2:
break
else:
return [] # fall back to normal solve
# Now, we have r's so we can get roots
x1 = (r1 + r2 + r3 + r4) / 5
x2 = (r1 * zeta4 + r2 * zeta3 + r3 * zeta2 + r4 * zeta1) / 5
x3 = (r1 * zeta3 + r2 * zeta1 + r3 * zeta4 + r4 * zeta2) / 5
x4 = (r1 * zeta2 + r2 * zeta4 + r3 * zeta1 + r4 * zeta3) / 5
x5 = (r1 * zeta1 + r2 * zeta2 + r3 * zeta3 + r4 * zeta4) / 5
result = [x1, x2, x3, x4, x5]
# Now check if solutions are distinct
saw = set()
for r in result:
r = r.n(2)
if r in saw:
# Roots were identical. Abort, return []
# and fall back to usual solve
return []
saw.add(r)
return result
def gauss_laguerre(n, n_digits):
r"""
Computes the Gauss-Laguerre quadrature [1]_ points and weights.
The Gauss-Laguerre quadrature approximates the integral:
.. math::
\int_0^{\infty} e^{-x} f(x)\,dx \approx \sum_{i=1}^n w_i f(x_i)
The nodes `x_i` of an order `n` quadrature rule are the roots of `L_n`
and the weights `w_i` are given by:
.. math::
w_i = \frac{x_i}{(n+1)^2 \left(L_{n+1}(x_i)\right)^2}
Parameters
==========
n : the order of quadrature
n_digits : number of significant digits of the points and weights to return
Returns
=======
(x, w) : the ``x`` and ``w`` are lists of points and weights as Floats.
The points `x_i` and weights `w_i` are returned as ``(x, w)``
tuple of lists.
Examples
========
>>> from sympy.integrals.quadrature import gauss_laguerre
>>> x, w = gauss_laguerre(3, 5)
>>> x
[0.41577, 2.2943, 6.2899]
>>> w
[0.71109, 0.27852, 0.010389]
>>> x, w = gauss_laguerre(6, 5)
>>> x
[0.22285, 1.1889, 2.9927, 5.7751, 9.8375, 15.983]
>>> w
[0.45896, 0.417, 0.11337, 0.010399, 0.00026102, 8.9855e-7]
See Also
========
gauss_legendre, gauss_gen_laguerre, gauss_hermite, gauss_chebyshev_t, gauss_chebyshev_u, gauss_jacobi
References
==========
.. [1] http://en.wikipedia.org/wiki/Gauss%E2%80%93Laguerre_quadrature
.. [2] http://people.sc.fsu.edu/~jburkardt/cpp_src/laguerre_rule/laguerre_rule.html
"""
x = Dummy("x")
p = laguerre_poly(n, x, polys=True)
p1 = laguerre_poly(n + 1, x, polys=True)
xi = []
w = []
for r in p.real_roots():
if isinstance(r, RootOf):
r = r.eval_rational(S(1) / 10**(n_digits + 2))
xi.append(r.n(n_digits))
w.append((r / ((n + 1)**2 * p1.subs(x, r)**2)).n(n_digits))
return xi, w
def gosper_term(f, n):
r"""
Compute Gosper's hypergeometric term for ``f``.
Suppose ``f`` is a hypergeometric term such that:
.. math::
s_n = \sum_{k=0}^{n-1} f_k
and `f_k` doesn't depend on `n`. Returns a hypergeometric
term `g_n` such that `g_{n+1} - g_n = f_n`.
Examples
========
>>> from sympy.concrete.gosper import gosper_term
>>> from sympy.functions import factorial
>>> from sympy.abc import n
>>> gosper_term((4*n + 1)*factorial(n)/factorial(2*n + 1), n)
(-n - 1/2)/(n + 1/4)
"""
r = hypersimp(f, n)
if r is None:
return None # 'f' is *not* a hypergeometric term
p, q = r.as_numer_denom()
A, B, C = gosper_normal(p, q, n)
B = B.shift(-1)
N = S(A.degree())
M = S(B.degree())
K = S(C.degree())
if (N != M) or (A.LC() != B.LC()):
D = set([K - max(N, M)])
elif not N:
D = set([K - N + 1, S(0)])
else:
D = set([K - N + 1, (B.nth(N - 1) - A.nth(N - 1))/A.LC()])
for d in set(D):
if not d.is_Integer or d < 0:
D.remove(d)
if not D:
return None # 'f(n)' is *not* Gosper-summable
d = max(D)
coeffs = symbols('c:%s' % (d + 1), cls=Dummy)
domain = A.get_domain().inject(*coeffs)
x = Poly(coeffs, n, domain=domain)
H = A*x.shift(1) - B*x - C
solution = solve(H.coeffs(), coeffs)
if solution is None:
return None # 'f(n)' is *not* Gosper-summable
x = x.as_expr().subs(solution)
for coeff in coeffs:
if coeff not in solution:
x = x.subs(coeff, 0)
if x is S.Zero:
return None # 'f(n)' is *not* Gosper-summable
else:
return B.as_expr()*x/C.as_expr()
def apart_full_decomposition(P, Q):
"""
Bronstein's full partial fraction decomposition algorithm.
Given a univariate rational function ``f``, performing only GCD
operations over the algebraic closure of the initial ground domain
of definition, compute full partial fraction decomposition with
fractions having linear denominators.
Note that no factorization of the initial denominator of ``f`` is
performed. The final decomposition is formed in terms of a sum of
:class:`RootSum` instances.
**References**
1. [Bronstein93]_
"""
f, x, U = P/Q, P.gen, []
u = Function('u')(x)
a = Dummy('a')
partial = S(0)
for d, n in Q.sqf_list_include(all=True):
b = d.as_expr()
U += [ u.diff(x, n-1) ]
h = cancel(f*b**n) / u**n
H, subs = [h], []
for j in range(1, n):
H += [ H[-1].diff(x) / j ]
for j in range(1, n+1):
subs += [ (U[j-1], b.diff(x, j) / j) ]
for j in range(0, n):
P, Q = cancel(H[j]).as_numer_denom()
for i in range(0, j+1):
P = P.subs(*subs[j-i])
Q = Q.subs(*subs[0])
P = Poly(P, x)
Q = Poly(Q, x)
G = P.gcd(d)
D = d.quo(G)
B, g = Q.half_gcdex(D)
b = (P * B.quo(g)).rem(D)
numer = b.as_expr()
denom = (x-a)**(n-j)
func = Lambda(a, numer.subs(x, a)/denom)
partial += RootSum(D, func, auto=False)
return partial
def _expr_big(cls, z, n):
return S(-1)**n * (
(S(1) / 2 - n) * pi / sqrt(z) + I * acosh(sqrt(z)) / sqrt(z))
def _expr_big(cls, z, n):
if n.is_even:
return (n - S(1)/2)*pi*I + log(sqrt(z)/2) + I*asin(1/sqrt(z))
else:
return (n - S(1)/2)*pi*I + log(sqrt(z)/2) - I*asin(1/sqrt(z))
def _expr_small_minus(cls, z):
return HyperRep_asin1._expr_small_minus(z) \
/HyperRep_power1._expr_small_minus(S(1)/2, z)
def test_del_operator():
# Tests for curl
assert delop ^ Vector.zero == Vector.zero
assert ((delop ^ Vector.zero).doit() == Vector.zero == curl(Vector.zero))
assert delop.cross(Vector.zero) == delop ^ Vector.zero
assert (delop ^ i).doit() == Vector.zero
assert delop.cross(2 * y**2 * j, doit=True) == Vector.zero
assert delop.cross(2 * y**2 * j) == delop ^ 2 * y**2 * j
v = x * y * z * (i + j + k)
assert ((delop ^ v).doit() == (-x * y + x * z) * i + (x * y - y * z) * j +
(-x * z + y * z) * k == curl(v))
assert delop ^ v == delop.cross(v)
assert (delop.cross(
2 * x**2 *
j) == (Derivative(0, C.y) - Derivative(2 * C.x**2, C.z)) * C.i +
(-Derivative(0, C.x) + Derivative(0, C.z)) * C.j +
(-Derivative(0, C.y) + Derivative(2 * C.x**2, C.x)) * C.k)
assert (delop.cross(2 * x**2 * j, doit=True) == 4 * x * k == curl(
2 * x**2 * j))
#Tests for divergence
assert delop & Vector.zero == S(0) == divergence(Vector.zero)
assert (delop & Vector.zero).doit() == S(0)
assert delop.dot(Vector.zero) == delop & Vector.zero
assert (delop & i).doit() == S(0)
assert (delop & x**2 * i).doit() == 2 * x == divergence(x**2 * i)
assert (delop.dot(v, doit=True) == x * y + y * z + z * x == divergence(v))
assert delop & v == delop.dot(v)
assert delop.dot(1/(x*y*z) * (i + j + k), doit=True) == \
- 1 / (x*y*z**2) - 1 / (x*y**2*z) - 1 / (x**2*y*z)
v = x * i + y * j + z * k
assert (delop & v == Derivative(C.x, C.x) + Derivative(C.y, C.y) +
Derivative(C.z, C.z))
assert delop.dot(v, doit=True) == 3 == divergence(v)
assert delop & v == delop.dot(v)
assert simplify((delop & v).doit()) == 3
#Tests for gradient
assert (delop.gradient(0, doit=True) == Vector.zero == gradient(0))
assert delop.gradient(0) == delop(0)
assert (delop(S(0))).doit() == Vector.zero
assert (delop(x) == (Derivative(C.x, C.x)) * C.i +
(Derivative(C.x, C.y)) * C.j + (Derivative(C.x, C.z)) * C.k)
assert (delop(x)).doit() == i == gradient(x)
assert (delop(x * y * z) == (Derivative(C.x * C.y * C.z, C.x)) * C.i +
(Derivative(C.x * C.y * C.z, C.y)) * C.j +
(Derivative(C.x * C.y * C.z, C.z)) * C.k)
assert (delop.gradient(x * y * z, doit=True) ==
y * z * i + z * x * j + x * y * k == gradient(x * y * z))
assert delop(x * y * z) == delop.gradient(x * y * z)
assert (delop(2 * x**2)).doit() == 4 * x * i
assert ((delop(a * sin(y) / x)).doit() == -a * sin(y) / x**2 * i +
a * cos(y) / x * j)
#Tests for directional derivative
assert (Vector.zero & delop)(a) == S(0)
assert ((Vector.zero & delop)(a)).doit() == S(0)
assert ((v & delop)(Vector.zero)).doit() == Vector.zero
assert ((v & delop)(S(0))).doit() == S(0)
assert ((i & delop)(x)).doit() == 1
assert ((j & delop)(y)).doit() == 1
assert ((k & delop)(z)).doit() == 1
assert ((i & delop)(x * y * z)).doit() == y * z
assert ((v & delop)(x)).doit() == x
assert ((v & delop)(x * y * z)).doit() == 3 * x * y * z
assert (v & delop)(x + y + z) == C.x + C.y + C.z
assert ((v & delop)(x + y + z)).doit() == x + y + z
assert ((v & delop)(v)).doit() == v
assert ((i & delop)(v)).doit() == i
assert ((j & delop)(v)).doit() == j
assert ((k & delop)(v)).doit() == k
assert ((v & delop)(Vector.zero)).doit() == Vector.zero
def _expr_small(cls, z):
return log(S(1) / 2 + sqrt(1 - z) / 2)
def test_linrec():
assert linrec(coeffs=[1, 1], init=[1, 1], n=20) == 10946
assert linrec(coeffs=[1, 2, 3, 4, 5], init=[1, 1, 0, 2], n=10) == 1040
assert linrec(coeffs=[0, 0, 11, 13], init=[23, 27], n=25) == 59628567384
assert linrec(coeffs=[0, 0, 1, 1, 2], init=[1, 5, 3], n=15) == 165
assert linrec(coeffs=[11, 13, 15, 17], init=[1, 2, 3, 4], n=70) == \
56889923441670659718376223533331214868804815612050381493741233489928913241
assert linrec(coeffs=[0]*55 + [1, 1, 2, 3], init=[0]*50 + [1, 2, 3], n=4000) == \
702633573874937994980598979769135096432444135301118916539
assert linrec(coeffs=[11, 13, 15, 17], init=[1, 2, 3, 4], n=10**4)
assert linrec(coeffs=[11, 13, 15, 17], init=[1, 2, 3, 4], n=10**5)
assert all(
linrec(coeffs=[1, 1], init=[0, 1], n=n) == fibonacci(n)
for n in range(95, 115))
assert all(
linrec(coeffs=[1, 1], init=[1, 1], n=n) == fibonacci(n + 1)
for n in range(595, 615))
a = [S(1) / 2, S(3) / 4, S(5) / 6, 7, S(8) / 9, S(3) / 5]
b = [1, 2, 8, S(5) / 7, S(3) / 7, S(2) / 9, 6]
x, y, z = symbols('x y z')
assert linrec(coeffs=a[:5], init=b[:4], n=80) == \
Rational(1726244235456268979436592226626304376013002142588105090705187189,
1960143456748895967474334873705475211264)
assert linrec(coeffs=a[:4], init=b[:4], n=50) == \
Rational(368949940033050147080268092104304441, 504857282956046106624)
assert linrec(coeffs=a[3:], init=b[:3], n=35) == \
Rational(97409272177295731943657945116791049305244422833125109,
814315512679031689453125)
assert linrec(coeffs=[0]*60 + [S(2)/3, S(4)/5], init=b, n=3000) == \
26777668739896791448594650497024/S(48084516708184142230517578125)
raises(TypeError,
lambda: linrec(coeffs=[11, 13, 15, 17], init=[1, 2, 3, 4, 5], n=1))
raises(TypeError, lambda: linrec(coeffs=a[:4], init=b[:5], n=10000))
raises(ValueError, lambda: linrec(coeffs=a[:4], init=b[:4], n=-10000))
raises(TypeError, lambda: linrec(x, b, n=10000))
raises(TypeError, lambda: linrec(a, y, n=10000))
assert linrec(coeffs=[x, y, z], init=[1, 1, 1], n=4) == \
x**2 + x*y + x*z + y + z
assert linrec(coeffs=[1, 2, 1], init=[x, y, z], n=20) == \
269542*x + 664575*y + 578949*z
assert linrec(coeffs=[0, 3, 1, 2], init=[x, y], n=30) == \
58516436*x + 56372788*y
assert linrec(coeffs=[0]*50 + [1, 2, 3], init=[x, y, z], n=1000) == \
11477135884896*x + 25999077948732*y + 41975630244216*z
assert linrec(coeffs=[], init=[1, 1], n=20) == 0
def _expr_big_minus(cls, z, n):
if n.is_even:
return pi * I * n + log(S(1) / 2 + sqrt(1 + z) / 2)
else:
return pi * I * n + log(sqrt(1 + z) / 2 - S(1) / 2)
def _eval_rewrite_as_Piecewise(self, arg):
if arg.is_real:
return Piecewise((1, arg > 0), (S(1) / 2, Eq(arg, 0)), (0, True))
def as_real_imag(self):
from sympy import I
real = (S(1) / 2) * (self + self._eval_conjugate())
im = (self - self._eval_conjugate()) / (2 * I)
return (real, im)
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 S.Zero
if n < 0:
n = -n + k - 1
res = -1 if k % 2 else 1
# non negative integers k and n
if k > n:
return S.Zero
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 S(res % q)
def pde_1st_linear_constant_coeff(eq, func, order, match, solvefun):
r"""
Solves a first order linear partial differential equation
with constant coefficients.
The general form of this partial differential equation is
.. math:: a \frac{df(x,y)}{dx} + b \frac{df(x,y)}{dy} + c f(x,y) = G(x,y)
where `a`, `b` and `c` are constants and `G(x, y)` can be an arbitrary
function in `x` and `y`.
The general solution of the PDE is::
>>> from sympy.solvers import pdsolve
>>> from sympy.abc import x, y, a, b, c
>>> from sympy import Function, pprint
>>> f = Function('f')
>>> G = Function('G')
>>> u = f(x,y)
>>> ux = u.diff(x)
>>> uy = u.diff(y)
>>> genform = a*u + b*ux + c*uy - G(x,y)
>>> pprint(genform)
d d
a*f(x, y) + b*--(f(x, y)) + c*--(f(x, y)) - G(x, y)
dx dy
>>> pprint(pdsolve(genform, hint='1st_linear_constant_coeff_Integral'))
// b*x + c*y \
|| / |
|| | |
|| | a*xi |
|| | ------- |
|| | 2 2 |
|| | /b*xi + c*eta -b*eta + c*xi\ b + c |
|| | G|------------, -------------|*e d(xi)|
|| | | 2 2 2 2 | |
|| | \ b + c b + c / |
|| | |
|| / |
|| |
f(x, y) = ||F(eta) + -------------------------------------------------------|*
|| 2 2 |
\\ b + c /
<BLANKLINE>
\|
||
||
||
||
||
||
||
||
-a*xi ||
-------||
2 2||
b + c ||
e ||
||
/|eta=-b*y + c*x, xi=b*x + c*y
Examples
========
>>> from sympy.solvers.pde import pdsolve
>>> from sympy import Function, diff, pprint, exp
>>> from sympy.abc import x,y
>>> f = Function('f')
>>> eq = -2*f(x,y).diff(x) + 4*f(x,y).diff(y) + 5*f(x,y) - exp(x + 3*y)
>>> pdsolve(eq)
Eq(f(x, y), (F(4*x + 2*y) + exp(x/2 + 4*y)/15)*exp(x/2 - y))
References
==========
- Viktor Grigoryan, "Partial Differential Equations"
Math 124A - Fall 2010, pp.7
"""
# TODO : For now homogeneous first order linear PDE's having
# two variables are implemented. Once there is support for
# solving systems of ODE's, this can be extended to n variables.
xi, eta = symbols("xi eta")
f = func.func
x = func.args[0]
y = func.args[1]
b = match[match['b']]
c = match[match['c']]
d = match[match['d']]
e = -match[match['e']]
expterm = exp(-S(d) / (b**2 + c**2) * xi)
functerm = solvefun(eta)
solvedict = solve((b * x + c * y - xi, c * x - b * y - eta), x, y)
# Integral should remain as it is in terms of xi,
# doit() should be done in _handle_Integral.
genterm = (1 / S(b**2 + c**2)) * Integral(
(1 / expterm * e).subs(solvedict), (xi, b * x + c * y))
return Eq(
f(x, y),
Subs(expterm * (functerm + genterm), (eta, xi),
(c * x - b * y, b * x + c * y)))
def interpolate(data, x):
"""
Construct an interpolating polynomial for the data points
evaluated at point x (which can be symbolic or numeric).
Examples
========
>>> from sympy.polys.polyfuncs import interpolate
>>> from sympy.abc import a, b, x
A list is interpreted as though it were paired with a range starting
from 1:
>>> interpolate([1, 4, 9, 16], x)
x**2
This can be made explicit by giving a list of coordinates:
>>> interpolate([(1, 1), (2, 4), (3, 9)], x)
x**2
The (x, y) coordinates can also be given as keys and values of a
dictionary (and the points need not be equispaced):
>>> interpolate([(-1, 2), (1, 2), (2, 5)], x)
x**2 + 1
>>> interpolate({-1: 2, 1: 2, 2: 5}, x)
x**2 + 1
If the interpolation is going to be used only once then the
value of interest can be passed instead of passing a symbol:
>>> interpolate([1, 4, 9], 5)
25
Symbolic coordinates are also supported:
>>> [(i,interpolate((a, b), i)) for i in range(1, 4)]
[(1, a), (2, b), (3, -a + 2*b)]
"""
n = len(data)
if isinstance(data, dict):
if x in data:
return S(data[x])
X, Y = list(zip(*data.items()))
else:
if isinstance(data[0], tuple):
X, Y = list(zip(*data))
if x in X:
return S(Y[X.index(x)])
else:
if x in range(1, n + 1):
return S(data[x - 1])
Y = list(data)
X = list(range(1, n + 1))
try:
return interpolating_poly(n, x, X, Y).expand()
except ValueError:
d = Dummy()
return interpolating_poly(n, d, X, Y).expand().subs(d, x)
