Esempio n. 1
0
    def get_phi(self, k2):
        """Get the solution of the Helmholtz equation for a Gaussian."""
        #   This should lead to very big errors
        r = np.sqrt(self.r2)
        k = sqrt(k2)
        sigma = 1. / sqrt(2 * self.a)
        p = sigma * k / sqrt(2)
        i = 1j
        rhop = r / sqrt(2) / sigma + i * p
        rhom = r / sqrt(2) / sigma - i * p
        # h = np.sin(k * r)

        # the gpaw-Gaussian is sqrt(4 * pi) times a 3D normalized Gaussian
        return sqrt(4 * pi) * exp(-p**2) / r / 2 * (
            np.cos(k * r) * (cerf(rhop) + cerf(rhom)) + i * np.sin(k * r) *
            (2 + cerf(rhop) - cerf(rhom)))
Esempio n. 2
0
    def get_phi(self, k2):
        """Get the solution of the Helmholtz equation for a Gaussian."""
#   This should lead to very big errors
        r = np.sqrt(self.r2)
        k = sqrt(k2)
        sigma = 1. / sqrt(2 * self.a)
        p = sigma * k / sqrt(2)
        i = 1j
        rhop = r / sqrt(2) / sigma + i * p
        rhom = r / sqrt(2) / sigma - i * p
        h = np.sin(k * r)

        # the gpaw-Gaussian is sqrt(4 * pi) times a 3D normalized Gaussian
        return sqrt(4 * pi) * exp(-p**2) / r / 2 * (
            np.cos(k * r) * (cerf(rhop) + cerf(rhom)) +
            i * np.sin(k * r) * (2 + cerf(rhop) - cerf(rhom)))
Esempio n. 3
0
# 1           0.842701
# 3           0.999978
# I           0. + 1.65043 I
# 1 + I       1.31615 + 0.190453 I
# 0.3 + 3*I   1467.69 - 166.561 I
# 3 - 0.3*I   1.00001 - 0.0000228553 I
values = [
    [ 1, 0.84270079295+0j],
    [ 3, 0.999977909503+0j],
    [ 1j, 1.6504257588j],
    [ 1+1j, 1.3161512816979477+0.19045346923783471j ],
    [ 0.3 + 3j, 1467.69028322-166.560924526j ],
    [ 3 + 0.3j, 0.99997602085736015+2.1863701577230078e-06j]]

maxerror = 1.e-10

for test in values:
    z, res = test
    try:
        r = z.real
    except:
        r = z
    error = abs(res / cerf(z) - 1.)
    if error < maxerror:
        print 'z=', z, ' ok (error=', error, ')'
    else:
        print z, res, cerf(z), erf(r), error
        string = ('error for erf(' + str(z) +') = ' + str(error) + 
                  ' > ' + str(maxerror)) 
        assert(error < maxerror), string
Esempio n. 4
0
# Mathematica says:
#   z             erf(z)
# 1           0.842701
# 3           0.999978
# I           0. + 1.65043 I
# 1 + I       1.31615 + 0.190453 I
# 0.3 + 3*I   1467.69 - 166.561 I
# 3 - 0.3*I   1.00001 - 0.0000228553 I
values = [[1, 0.84270079295 + 0j], [3, 0.999977909503 + 0j],
          [1j, 1.6504257588j],
          [1 + 1j, 1.3161512816979477 + 0.19045346923783471j],
          [0.3 + 3j, 1467.69028322 - 166.560924526j],
          [3 + 0.3j, 0.99997602085736015 + 2.1863701577230078e-06j]]

maxerror = 1.e-10

for test in values:
    z, res = test
    try:
        r = z.real
    except:
        r = z
    error = abs(res / cerf(z) - 1.)
    if error < maxerror:
        print('z=', z, ' ok (error=', error, ')')
    else:
        print(z, res, cerf(z), erf(r), error)
        string = ('error for erf(' + str(z) + ') = ' + str(error) + ' > ' +
                  str(maxerror))
        assert (error < maxerror), string