poliastro/poliastro

View on GitHub
src/poliastro/core/propagation.py

Summary

Maintainability
D
2 days
Test Coverage
import numpy as np

from poliastro.core.angles import (
    D_to_nu,
    E_to_nu,
    F_to_nu,
    M_to_nu,
    _kepler_equation,
    _kepler_equation_prime,
    nu_to_M,
)
from poliastro.core.elements import coe2rv, rv2coe
from poliastro.core.stumpff import c2, c3

from ._jit import jit


def func_twobody(t0, u_, k, ad, ad_kwargs):
    """Differential equation for the initial value two body problem.

    This function follows Cowell's formulation.

    Parameters
    ----------
    t0 : float
        Time.
    u_ : ~numpy.ndarray
        Six component state vector [x, y, z, vx, vy, vz] (km, km/s).
    k : float
        Standard gravitational parameter.
    ad : function(t0, u, k)
        Non Keplerian acceleration (km/s2).
    ad_kwargs : optional
        perturbation parameters passed to ad.

    """
    ax, ay, az = ad(t0, u_, k, **ad_kwargs)

    x, y, z, vx, vy, vz = u_
    r3 = (x ** 2 + y ** 2 + z ** 2) ** 1.5

    du = np.array([vx, vy, vz, -k * x / r3 + ax, -k * y / r3 + ay, -k * z / r3 + az])
    return du


@jit
def mean_motion(k, r0, v0, tof):
    r"""Propagates orbit using mean motion. This algorithm depends on the geometric shape of the
    orbit.

    For the case of the strong elliptic or strong hyperbolic orbits:

    ..  math::

        M = M_{0} + \frac{\mu^{2}}{h^{3}}\left ( 1 -e^{2}\right )^{\frac{3}{2}}t

    .. versionadded:: 0.9.0


    Parameters
    ----------
    k : float
        Standar Gravitational parameter
    r0 : ~astropy.units.Quantity
        Initial position vector wrt attractor center.
    v0 : ~astropy.units.Quantity
        Initial velocity vector.
    tof : float
        Time of flight (s).

    Note
    ----
    This method takes initial :math:`\vec{r}, \vec{v}`, calculates classical orbit parameters,
    increases mean anomaly and performs inverse transformation to get final :math:`\vec{r}, \vec{v}`
    The logic is based on formulae (4), (6) and (7) from http://dx.doi.org/10.1007/s10569-013-9476-9

    """

    # get the initial true anomaly and orbit parameters that are constant over time
    p, ecc, inc, raan, argp, nu0 = rv2coe(k, r0, v0)

    # get the initial mean anomaly
    M0 = nu_to_M(nu0, ecc)
    # strong elliptic or strong hyperbolic orbits
    if np.abs(ecc - 1.0) > 1e-2:
        a = p / (1.0 - ecc ** 2)
        # given the initial mean anomaly, calculate mean anomaly
        # at the end, mean motion (n) equals sqrt(mu / |a^3|)
        M = M0 + tof * np.sqrt(k / np.abs(a ** 3))
        nu = M_to_nu(M, ecc)

    # near-parabolic orbit
    else:
        q = p * np.abs(1.0 - ecc) / np.abs(1.0 - ecc ** 2)
        # mean motion n = sqrt(mu / 2 q^3) for parabolic orbit
        M = M0 + tof * np.sqrt(k / 2.0 / (q ** 3))
        nu = M_to_nu(M, ecc)

    return coe2rv(k, p, ecc, inc, raan, argp, nu)


@jit
def kepler(k, r0, v0, tof, numiter):
    r"""Solves Kepler's Equation by applying a Newton-Raphson method.

    If the position of a body along its orbit wants to be computed
    for an specific time, it can be solved by terms of the Kepler's Equation:

    .. math::
        E = M + e\sin{E}

    In this case, the equation is written in terms of the Universal Anomaly:

    .. math::

        \sqrt{\mu}\Delta t = \frac{r_{o}v_{o}}{\sqrt{\mu}}\chi^{2}C(\alpha \chi^{2}) + (1 - \alpha r_{o})\chi^{3}S(\alpha \chi^{2}) + r_{0}\chi

    This equation is solved for the universal anomaly by applying a Newton-Raphson numerical method.
    Once it is solved, the Lagrange coefficients are returned:

    .. math::

        \begin{align}
            f &= 1 \frac{\chi^{2}}{r_{o}}C(\alpha \chi^{2}) \\
            g &= \Delta t - \frac{1}{\sqrt{\mu}}\chi^{3}S(\alpha \chi^{2}) \\
            \dot{f} &= \frac{\sqrt{\mu}}{rr_{o}}(\alpha \chi^{3}S(\alpha \chi^{2}) - \chi) \\
            \dot{g} &= 1 - \frac{\chi^{2}}{r}C(\alpha \chi^{2}) \\
        \end{align}

    Lagrange coefficients can be related then with the position and velocity vectors:

    .. math::
        \begin{align}
            \vec{r} &= f\vec{r_{o}} + g\vec{v_{o}} \\
            \vec{v} &= \dot{f}\vec{r_{o}} + \dot{g}\vec{v_{o}} \\
        \end{align}

    Parameters
    ----------

    k: float
        Standard gravitational parameter
    r0: ~numpy.array
        Initial position vector
    v0: ~numpy.array
        Initial velocity vector
    numiter: int
        Number of iterations

    Returns
    -------
    f: float
        First Lagrange coefficient
    g: float
        Second Lagrange coefficient
    fdot: float
        Derivative of the first coefficient
    gdot: float
        Derivative of the second coefficient


    Note
    ----
    The theoretical procedure is explained in section 3.7 of Curtis in really
    deep detail. For analytical example, check in the same book for example 3.6.
    """

    # Cache some results
    dot_r0v0 = np.dot(r0, v0)
    norm_r0 = np.dot(r0, r0) ** 0.5
    sqrt_mu = k ** 0.5
    alpha = -np.dot(v0, v0) / k + 2 / norm_r0

    # First guess
    if alpha > 0:
        # Elliptic orbit
        xi_new = sqrt_mu * tof * alpha
    elif alpha < 0:
        # Hyperbolic orbit
        xi_new = (
            np.sign(tof)
            * (-1 / alpha) ** 0.5
            * np.log(
                (-2 * k * alpha * tof)
                / (
                    dot_r0v0
                    + np.sign(tof) * np.sqrt(-k / alpha) * (1 - norm_r0 * alpha)
                )
            )
        )
    else:
        # Parabolic orbit
        # (Conservative initial guess)
        xi_new = sqrt_mu * tof / norm_r0

    # Newton-Raphson iteration on the Kepler equation
    count = 0
    while count < numiter:
        xi = xi_new
        psi = xi * xi * alpha
        c2_psi = c2(psi)
        c3_psi = c3(psi)
        norm_r = (
            xi * xi * c2_psi
            + dot_r0v0 / sqrt_mu * xi * (1 - psi * c3_psi)
            + norm_r0 * (1 - psi * c2_psi)
        )
        xi_new = (
            xi
            + (
                sqrt_mu * tof
                - xi * xi * xi * c3_psi
                - dot_r0v0 / sqrt_mu * xi * xi * c2_psi
                - norm_r0 * xi * (1 - psi * c3_psi)
            )
            / norm_r
        )
        if abs(xi_new - xi) < 1e-7:
            break
        else:
            count += 1
    else:
        raise RuntimeError("Maximum number of iterations reached")

    # Compute Lagrange coefficients
    f = 1 - xi ** 2 / norm_r0 * c2_psi
    g = tof - xi ** 3 / sqrt_mu * c3_psi

    gdot = 1 - xi ** 2 / norm_r * c2_psi
    fdot = sqrt_mu / (norm_r * norm_r0) * xi * (psi * c3_psi - 1)

    return f, g, fdot, gdot


@jit
def mikkola(k, r0, v0, tof, rtol=None):
    """ Raw algorithm for Mikkola's Kepler solver.

    Parameters
    ----------
    k : ~astropy.units.Quantity
        Standard gravitational parameter of the attractor.
    r : ~astropy.units.Quantity
        Position vector.
    v : ~astropy.units.Quantity
        Velocity vector.
    tofs : ~astropy.units.Quantity
        Array of times to propagate.
    rtol: float
        This method does not require of tolerance since it is non iterative.

    Returns
    -------
    rr : ~astropy.units.Quantity
        Propagated position vectors.
    vv : ~astropy.units.Quantity

    Note
    ----
    Original paper: https://doi.org/10.1007/BF01235850
    """

    # Solving for the classical elements
    p, ecc, inc, raan, argp, nu = rv2coe(k, r0, v0)
    M0 = nu_to_M(nu, ecc, delta=0)
    a = p / (1 - ecc ** 2)
    n = np.sqrt(k / np.abs(a) ** 3)
    M = M0 + n * tof

    # Solve for specific geometrical case
    if ecc < 1.0:
        # Equation (9a)
        alpha = (1 - ecc) / (4 * ecc + 1 / 2)
    else:
        alpha = (ecc - 1) / (4 * ecc + 1 / 2)

    beta = M / 2 / (4 * ecc + 1 / 2)

    # Equation (9b)
    if beta >= 0:
        z = (beta + np.sqrt(beta ** 2 + alpha ** 3)) ** (1 / 3)
    else:
        z = (beta - np.sqrt(beta ** 2 + alpha ** 3)) ** (1 / 3)

    s = z - alpha / z

    # Apply initial correction
    if ecc < 1.0:
        ds = -0.078 * s ** 5 / (1 + ecc)
    else:
        ds = 0.071 * s ** 5 / (1 + 0.45 * s ** 2) / (1 + 4 * s ** 2) / ecc

    s += ds

    # Solving for the true anomaly
    if ecc < 1.0:
        E = M + ecc * (3 * s - 4 * s ** 3)
        f = E - ecc * np.sin(E) - M
        f1 = 1.0 - ecc * np.cos(E)
        f2 = ecc * np.sin(E)
        f3 = ecc * np.cos(E)
        f4 = -f2
        f5 = -f3
    else:
        E = 3 * np.log(s + np.sqrt(1 + s ** 2))
        f = -E + ecc * np.sinh(E) - M
        f1 = -1.0 + ecc * np.cosh(E)
        f2 = ecc * np.sinh(E)
        f3 = ecc * np.cosh(E)
        f4 = f2
        f5 = f3

    # Apply Taylor expansion
    u1 = -f / f1
    u2 = -f / (f1 + 0.5 * f2 * u1)
    u3 = -f / (f1 + 0.5 * f2 * u2 + (1.0 / 6.0) * f3 * u2 ** 2)
    u4 = -f / (
        f1 + 0.5 * f2 * u3 + (1.0 / 6.0) * f3 * u3 ** 2 + (1.0 / 24.0) * f4 * (u3 ** 3)
    )
    u5 = -f / (
        f1
        + f2 * u4 / 2
        + f3 * (u4 * u4) / 6.0
        + f4 * (u4 * u4 * u4) / 24.0
        + f5 * (u4 * u4 * u4 * u4) / 120.0
    )

    E += u5

    if ecc < 1.0:
        nu = E_to_nu(E, ecc)
    else:
        if ecc == 1.0:
            # Parabolic
            nu = D_to_nu(E)
        else:
            # Hyperbolic
            nu = F_to_nu(E, ecc)

    return coe2rv(k, p, ecc, inc, raan, argp, nu)


@jit
def markley(k, r0, v0, tof):
    """ Solves the kepler problem by a non iterative method. Relative error is
    around 1e-18, only limited by machine double-precission errors.

    Parameters
    ----------
    k : float
        Standar Gravitational parameter
    r0 : array
        Initial position vector wrt attractor center.
    v0 : array
        Initial velocity vector.
    tof : float
        Time of flight.

    Returns
    -------
    rf: array
        Final position vector
    vf: array
        Final velocity vector

    Note
    ----
    The following algorithm was taken from http://dx.doi.org/10.1007/BF00691917.
    """

    # Solve first for eccentricity and mean anomaly
    p, ecc, inc, raan, argp, nu = rv2coe(k, r0, v0)

    M0 = nu_to_M(nu, ecc, delta=0)
    a = p / (1 - ecc ** 2)
    n = np.sqrt(k / a ** 3)
    M = M0 + n * tof

    # Range between -pi and pi
    M = M % (2 * np.pi)
    if M > np.pi:
        M = -(2 * np.pi - M)

    # Equation (20)
    alpha = (3 * np.pi ** 2 + 1.6 * (np.pi - np.abs(M)) / (1 + ecc)) / (np.pi ** 2 - 6)

    # Equation (5)
    d = 3 * (1 - ecc) + alpha * ecc

    # Equation (9)
    q = 2 * alpha * d * (1 - ecc) - M ** 2

    # Equation (10)
    r = 3 * alpha * d * (d - 1 + ecc) * M + M ** 3

    # Equation (14)
    w = (np.abs(r) + np.sqrt(q ** 3 + r ** 2)) ** (2 / 3)

    # Equation (15)
    E = (2 * r * w / (w ** 2 + w * q + q ** 2) + M) / d

    # Equation (26)
    f0 = _kepler_equation(E, M, ecc)
    f1 = _kepler_equation_prime(E, M, ecc)
    f2 = ecc * np.sin(E)
    f3 = ecc * np.cos(E)
    f4 = -f2

    # Equation (22)
    delta3 = -f0 / (f1 - 0.5 * f0 * f2 / f1)
    delta4 = -f0 / (f1 + 0.5 * delta3 * f2 + 1 / 6 * delta3 ** 2 * f3)
    delta5 = -f0 / (
        f1 + 0.5 * delta4 * f2 + 1 / 6 * delta4 ** 2 * f3 + 1 / 24 * delta4 ** 3 * f4
    )

    E += delta5
    nu = E_to_nu(E, ecc)

    return coe2rv(k, p, ecc, inc, raan, argp, nu)


@jit
def pimienta(k, r0, v0, tof):
    """ Raw algorithm for Adonis' Pimienta and John L. Crassidis 15th order
    polynomial Kepler solver.

    Parameters
    ----------
    k : float
        Standar Gravitational parameter
    r0 : array
        Initial position vector wrt attractor center.
    v0 : array
        Initial velocity vector.
    tof : float
        Time of flight.

    Returns
    -------
    rf: array
        Final position vector
    vf: array
        Final velocity vector

    Note
    ----
    This algorithm was drived from the original paper: http://hdl.handle.net/10477/50522
    """

    # TODO: implement hyperbolic case

    # Solve first for eccentricity and mean anomaly
    p, ecc, inc, raan, argp, nu = rv2coe(k, r0, v0)

    M0 = nu_to_M(nu, ecc, delta=0)
    semi_axis_a = p / (1 - ecc ** 2)
    n = np.sqrt(k / np.abs(semi_axis_a) ** 3)
    M = M0 + n * tof

    # Equation (32a), (32b), (32c) and (32d)
    c3 = 5 / 2 + 560 * ecc
    a = 15 * (1 - ecc) / c3
    b = -M / c3
    y = np.sqrt(b ** 2 / 4 + a ** 3 / 27)

    # Equation (33)
    x_bar = (-b / 2 + y) ** (1 / 3) - (b / 2 + y) ** (1 / 3)

    # Coefficients from equations (34a) and (34b)
    c15 = 3003 / 14336 + 16384 * ecc
    c13 = 3465 / 13312 - 61440 * ecc
    c11 = 945 / 2816 + 92160 * ecc
    c9 = 175 / 384 - 70400 * ecc
    c7 = 75 / 112 + 28800 * ecc
    c5 = 9 / 8 - 6048 * ecc

    # Precompute x_bar powers, equations (35a) to (35d)
    x_bar2 = x_bar ** 2
    x_bar3 = x_bar2 * x_bar
    x_bar4 = x_bar3 * x_bar
    x_bar5 = x_bar4 * x_bar
    x_bar6 = x_bar5 * x_bar
    x_bar7 = x_bar6 * x_bar
    x_bar8 = x_bar7 * x_bar
    x_bar9 = x_bar8 * x_bar
    x_bar10 = x_bar9 * x_bar
    x_bar11 = x_bar10 * x_bar
    x_bar12 = x_bar11 * x_bar
    x_bar13 = x_bar12 * x_bar
    x_bar14 = x_bar13 * x_bar
    x_bar15 = x_bar14 * x_bar

    # Function f and its derivatives are given by all the (36) equation set
    f = (
        c15 * x_bar15
        + c13 * x_bar13
        + c11 * x_bar11
        + c9 * x_bar9
        + c7 * x_bar7
        + c5 * x_bar5
        + c3 * x_bar3
        + 15 * (1 - ecc) * x_bar
        - M
    )
    f1 = (
        15 * c15 * x_bar14
        + 13 * c13 * x_bar12
        + 11 * c11 * x_bar10
        + 9 * c9 * x_bar8
        + 7 * c7 * x_bar6
        + 5 * c5 * x_bar4
        + 3 * c3 * x_bar2
        + 15 * (1 - ecc)
    )
    f2 = (
        210 * c15 * x_bar13
        + 156 * c13 * x_bar11
        + 110 * c11 * x_bar9
        + 72 * c9 * x_bar7
        + 42 * c7 * x_bar5
        + 20 * c5 * x_bar3
        + 6 * c3 * x_bar
    )
    f3 = (
        2730 * c15 * x_bar12
        + 1716 * c13 * x_bar10
        + 990 * c11 * x_bar8
        + 504 * c9 * x_bar6
        + 210 * c7 * x_bar4
        + 60 * c5 * x_bar2
        + 6 * c3
    )
    f4 = (
        32760 * c15 * x_bar11
        + 17160 * c13 * x_bar9
        + 7920 * c11 * x_bar7
        + 3024 * c9 * x_bar5
        + 840 * c7 * x_bar3
        + 120 * c5 * x_bar
    )
    f5 = (
        360360 * c15 * x_bar10
        + 154440 * c13 * x_bar8
        + 55440 * c11 * x_bar6
        + 15120 * c9 * x_bar4
        + 2520 * c7 * x_bar2
        + 120 * c5
    )
    f6 = (
        3603600 * c15 * x_bar9
        + 1235520 * c13 * x_bar7
        + 332640 * c11 * x_bar5
        + 60480 * c9 * x_bar3
        + 5040 * c7 * x_bar
    )
    f7 = (
        32432400 * c15 * x_bar8
        + 8648640 * c13 * x_bar6
        + 1663200 * c11 * x_bar4
        + 181440 * c9 * x_bar2
        + 5040 * c7
    )
    f8 = (
        259459200 * c15 * x_bar7
        + 51891840 * c13 * x_bar5
        + 6652800 * c11 * x_bar3
        + 362880 * c9 * x_bar
    )
    f9 = (
        1.8162144e9 * c15 * x_bar6
        + 259459200 * c13 * x_bar4
        + 19958400 * c11 * x_bar2
        + 362880 * c9
    )
    f10 = (
        1.08972864e10 * c15 * x_bar5
        + 1.0378368e9 * c13 * x_bar3
        + 39916800 * c11 * x_bar
    )
    f11 = 5.4486432e10 * c15 * x_bar4 + 3.1135104e9 * c13 * x_bar2 + 39916800 * c11
    f12 = 2.17945728e11 * c15 * x_bar3 + 6.2270208e9 * c13 * x_bar
    f13 = 6.53837184 * c15 * x_bar2 + 6.2270208e9 * c13
    f14 = 1.307674368e13 * c15 * x_bar
    f15 = 1.307674368e13 * c15

    # Solving g parameters defined by equations (37a), (37b), (37c) and (37d)
    g1 = 1 / 2
    g2 = 1 / 6
    g3 = 1 / 24
    g4 = 1 / 120
    g5 = 1 / 720
    g6 = 1 / 5040
    g7 = 1 / 40320
    g8 = 1 / 362880
    g9 = 1 / 3628800
    g10 = 1 / 39916800
    g11 = 1 / 479001600
    g12 = 1 / 6.2270208e9
    g13 = 1 / 8.71782912e10
    g14 = 1 / 1.307674368e12

    # Solving for the u_{i} and h_{i} variables defined by equation (38)
    u1 = -f / f1

    h2 = f1 + g1 * u1 * f2
    u2 = -f / h2

    h3 = f1 + g1 * u2 * f2 + g2 * u2 ** 2 * f3
    u3 = -f / h3

    h4 = f1 + g1 * u3 * f2 + g2 * u3 ** 2 * f3 + g3 * u3 ** 3 * f4
    u4 = -f / h4

    h5 = f1 + g1 * u4 * f2 + g2 * u4 ** 2 * f3 + g3 * u4 ** 3 * f4 + g4 * u4 ** 4 * f5
    u5 = -f / h5

    h6 = (
        f1
        + g1 * u5 * f2
        + g2 * u5 ** 2 * f3
        + g3 * u5 ** 3 * f4
        + g4 * u5 ** 4 * f5
        + g5 * u5 ** 5 * f6
    )
    u6 = -f / h6

    h7 = (
        f1
        + g1 * u6 * f2
        + g2 * u6 ** 2 * f3
        + g3 * u6 ** 3 * f4
        + g4 * u6 ** 4 * f5
        + g5 * u6 ** 5 * f6
        + g6 * u6 ** 6 * f7
    )
    u7 = -f / h7

    h8 = (
        f1
        + g1 * u7 * f2
        + g2 * u7 ** 2 * f3
        + g3 * u7 ** 3 * f4
        + g4 * u7 ** 4 * f5
        + g5 * u7 ** 5 * f6
        + g6 * u7 ** 6 * f7
        + g7 * u7 ** 7 * f8
    )
    u8 = -f / h8

    h9 = (
        f1
        + g1 * u8 * f2
        + g2 * u8 ** 2 * f3
        + g3 * u8 ** 3 * f4
        + g4 * u8 ** 4 * f5
        + g5 * u8 ** 5 * f6
        + g6 * u8 ** 6 * f7
        + g7 * u8 ** 7 * f8
        + g8 * u8 ** 8 * f9
    )
    u9 = -f / h9

    h10 = (
        f1
        + g1 * u9 * f2
        + g2 * u9 ** 2 * f3
        + g3 * u9 ** 3 * f4
        + g4 * u9 ** 4 * f5
        + g5 * u9 ** 5 * f6
        + g6 * u9 ** 6 * f7
        + g7 * u9 ** 7 * f8
        + g8 * u9 ** 8 * f9
        + g9 * u9 ** 9 * f10
    )
    u10 = -f / h10

    h11 = (
        f1
        + g1 * u10 * f2
        + g2 * u10 ** 2 * f3
        + g3 * u10 ** 3 * f4
        + g4 * u10 ** 4 * f5
        + g5 * u10 ** 5 * f6
        + g6 * u10 ** 6 * f7
        + g7 * u10 ** 7 * f8
        + g8 * u10 ** 8 * f9
        + g9 * u10 ** 9 * f10
        + g10 * u10 ** 10 * f11
    )
    u11 = -f / h11

    h12 = (
        f1
        + g1 * u11 * f2
        + g2 * u11 ** 2 * f3
        + g3 * u11 ** 3 * f4
        + g4 * u11 ** 4 * f5
        + g5 * u11 ** 5 * f6
        + g6 * u11 ** 6 * f7
        + g7 * u11 ** 7 * f8
        + g8 * u11 ** 8 * f9
        + g9 * u11 ** 9 * f10
        + g10 * u11 ** 10 * f11
        + g11 * u11 ** 11 * f12
    )
    u12 = -f / h12

    h13 = (
        f1
        + g1 * u12 * f2
        + g2 * u12 ** 2 * f3
        + g3 * u12 ** 3 * f4
        + g4 * u12 ** 4 * f5
        + g5 * u12 ** 5 * f6
        + g6 * u12 ** 6 * f7
        + g7 * u12 ** 7 * f8
        + g8 * u12 ** 8 * f9
        + g9 * u12 ** 9 * f10
        + g10 * u12 ** 10 * f11
        + g11 * u12 ** 11 * f12
        + g12 * u12 ** 12 * f13
    )
    u13 = -f / h13

    h14 = (
        f1
        + g1 * u13 * f2
        + g2 * u13 ** 2 * f3
        + g3 * u13 ** 3 * f4
        + g4 * u13 ** 4 * f5
        + g5 * u13 ** 5 * f6
        + g6 * u13 ** 6 * f7
        + g7 * u13 ** 7 * f8
        + g8 * u13 ** 8 * f9
        + g9 * u13 ** 9 * f10
        + g10 * u13 ** 10 * f11
        + g11 * u13 ** 11 * f12
        + g12 * u13 ** 12 * f13
        + g13 * u13 ** 13 * f14
    )
    u14 = -f / h14

    h15 = (
        f1
        + g1 * u14 * f2
        + g2 * u14 ** 2 * f3
        + g3 * u14 ** 3 * f4
        + g4 * u14 ** 4 * f5
        + g5 * u14 ** 5 * f6
        + g6 * u14 ** 6 * f7
        + g7 * u14 ** 7 * f8
        + g8 * u14 ** 8 * f9
        + g9 * u14 ** 9 * f10
        + g10 * u14 ** 10 * f11
        + g11 * u14 ** 11 * f12
        + g12 * u14 ** 12 * f13
        + g13 * u14 ** 13 * f14
        + g14 * u14 ** 14 * f15
    )
    u15 = -f / h15

    # Solving for x
    x = x_bar + u15
    w = x - 0.01171875 * x ** 17 / (1 + ecc)

    # Solving for the true anomaly from eccentricity anomaly
    E = M + ecc * (
        -16384 * w ** 15
        + 61440 * w ** 13
        - 92160 * w ** 11
        + 70400 * w ** 9
        - 28800 * w ** 7
        + 6048 * w ** 5
        - 560 * w ** 3
        + 15 * w
    )

    nu = E_to_nu(E, ecc)

    return coe2rv(k, p, ecc, inc, raan, argp, nu)


@jit
def gooding(k, r0, v0, tof, numiter=150, rtol=1e-8):
    """ Solves the Elliptic Kepler Equation with a cubic convergence and
    accuracy better than 10e-12 rad is normally achieved. It is not valid for
    eccentricities equal or higher than 1.0.

    Parameters
    ----------
    k : float
        Standard gravitational parameter of the attractor.
    r : 1x3 vector
        Position vector.
    v : 1x3 vector
        Velocity vector.
    tof : float
        Time of flight.
    rtol: float
        Relative error for accuracy of the method.

    Returns
    -------
    rr : 1x3 vector
        Propagated position vectors.
     vv : 1x3 vector

    Note
    ----
    Original paper for the algorithm: https://doi.org/10.1007/BF01238923
    """

    # Solve first for eccentricity and mean anomaly
    p, ecc, inc, raan, argp, nu = rv2coe(k, r0, v0)

    # TODO: parabolic and hyperbolic not implemented cases
    if ecc >= 1.0:
        raise NotImplementedError(
            "Parabolic/Hyperbolic cases still not implemented in gooding."
        )

    M0 = nu_to_M(nu, ecc, delta=0)
    semi_axis_a = p / (1 - ecc ** 2)
    n = np.sqrt(k / np.abs(semi_axis_a) ** 3)
    M = M0 + n * tof

    # Start the computation
    n = 0
    c = ecc * np.cos(M)
    s = ecc * np.sin(M)
    psi = s / np.sqrt(1 - 2 * c + ecc ** 2)
    f = 1.0
    while f ** 2 >= rtol and n <= numiter:
        xi = np.cos(psi)
        eta = np.sin(psi)
        fd = (1 - c * xi) + s * eta
        fdd = c * eta + s * xi
        f = psi - fdd
        psi = psi - f * fd / (fd ** 2 - 0.5 * f * fdd)
        n += 1

    E = M + psi
    nu = E_to_nu(E, ecc)

    return coe2rv(k, p, ecc, inc, raan, argp, nu)


@jit
def danby(k, r0, v0, tof, numiter=20, rtol=1e-8):
    """ Kepler solver for both elliptic and parabolic orbits based on Danby's
    algorithm.

    Parameters
    ----------
    k : float
        Standard gravitational parameter of the attractor.
    r : 1x3 vector
        Position vector.
    v : 1x3 vector
        Velocity vector.
    tof : float
        Time of flight.
    rtol: float
        Relative error for accuracy of the method.

    Returns
    -------
    rr : 1x3 vector
        Propagated position vectors.
    vv : 1x3 vector

    Note
    ----
    This algorithm was developed by Danby in his paper *The solution of Kepler
    Equation* with DOI: https://doi.org/10.1007/BF01686811
    """

    # Solve first for eccentricity and mean anomaly
    p, ecc, inc, raan, argp, nu = rv2coe(k, r0, v0)
    M0 = nu_to_M(nu, ecc, delta=0)
    semi_axis_a = p / (1 - ecc ** 2)
    n = np.sqrt(k / np.abs(semi_axis_a) ** 3)
    M = M0 + n * tof

    # Range mean anomaly
    xma = M - 2 * np.pi * np.floor(M / 2 / np.pi)

    if ecc == 0:
        # Solving for circular orbit
        nu = xma
        return coe2rv(k, p, ecc, inc, raan, argp, nu)

    elif ecc < 1.0:
        # For elliptical orbit
        E = xma + 0.85 * np.sign(np.sin(xma)) * ecc

    else:
        # For parabolic and hyperbolic
        E = np.log(2 * xma / ecc + 1.8)

    # Iterations begin
    n = 0
    while n <= numiter:

        if ecc < 1.0:
            s = ecc * np.sin(E)
            c = ecc * np.cos(E)
            f = E - s - xma
            fp = 1 - c
            fpp = s
            fppp = c
        else:
            s = ecc * np.sinh(E)
            c = ecc * np.cosh(E)
            f = s - E - xma
            fp = c - 1
            fpp = s
            fppp = c

        if np.abs(f) <= rtol:

            if ecc < 1.0:
                sta = np.sqrt(1 - ecc ** 2) * np.sin(E)
                cta = np.cos(E) - ecc
            else:
                sta = np.sqrt(ecc ** 2 - 1) * np.sinh(E)
                cta = ecc - np.cosh(E)

            nu = np.arctan2(sta, cta)
            return coe2rv(k, p, ecc, inc, raan, argp, nu)

        else:
            delta = -f / fp
            delta_star = -f / (fp + 0.5 * delta * fpp)
            deltak = -f / (fp + 0.5 * delta_star * fpp + delta_star ** 2 * fppp / 6)
            E = E + deltak
            n += 1

    raise ValueError("Maximum number of iterations has been reached.")