= Solution
For a bound <Kepler orbit>, the <semimajor axis> is half the long diameter of the ellipse; if $r_p,r_{mathrm a}$ are its apsidal distances, $a=(r_p+r_{mathrm a})/2$. The <orbital eccentricity> is $e=(r_{mathrm a}-r_p)/(r_{mathrm a}+r_p)$, with $0\leq e<1$ and semiminor axis $b=a\sqrt{1-e^2}$. The gravitating centre is a focus, displaced by $ae$ from the ellipse's geometric centre.
Choose a reference plane, a positive normal and a fixed reference direction in that plane. The <ascending node> is where the orbit crosses toward positive normal displacement. The <longitude of ascending node> $\Omega$ is the reference-plane angle from the fixed direction to that node. The <argument of periapsis> $\omega$ is the angle from the ascending node to periapsis, measured in the <orbital plane> in the direction of motion. The <orbital inclination> $i$ is the angle between the reference normal and the orbital angular-momentum normal, with $0\leq i\leq\pi$. The diagram separates the ellipse's shape from these orientation angles.
\Image[/past-exam-of-the-mathematics-course-of-the-university-of-cambridge/2007/iii/paper-69-orbital-elements.png]
{title=Size and shape of a Kepler ellipse, and the node, periapsis and inclination angles in a fixed reference frame}
Let $\mu=GM$ and use actions per unit test-particle mass. In the prograde spherical-action chart underlying the stated <Hamiltonian>, $J_\phi=L_z$ and $J_\theta=L-L_z$, so $J_\theta+J_\phi=L$. The radial action then satisfies $J_r+L=\sqrt{\mu a}$, consistently with $H_K=-\mu/(2a)$. Define
$$
\boxed{J_1=J_r+J_\theta+J_\phi,\qquad J_2=J_\theta+J_\phi,\qquad J_3=J_\phi.}
$$
If $(w_r,w_\theta,w_\phi)$ are the old angles, a type-two generating function is
$$
F_2=(J_1-J_2)w_r+(J_2-J_3)w_\theta+J_3w_\phi.
$$
Differentiation with respect to the old angles recovers $J_r,J_\theta,J_\phi$, and differentiation with respect to the new actions gives the <canonical transformation>
$$
\boxed{w_1=w_r,\qquad w_2=w_\theta-w_r,\qquad w_3=w_\phi-w_\theta.}
$$
Its triangular integer action matrix has determinant one, so it preserves the angle periodicities. In the new coordinates,
$$
\boxed{H_K=-\frac{\mu^2}{2J_1^2},\qquad
\dot w_1=\frac{\mu^2}{J_1^3}=\sqrt{\frac{\mu}{a^3}},\qquad \dot w_2=\dot w_3=0.}
$$
These are the <Delaunay variables>:
$$
J_1=\mathcal L=\sqrt{\mu a},\qquad
J_2=\mathcal G=\sqrt{\mu a(1-e^2)},\qquad
J_3=\mathcal H=\mathcal G\cos i.
$$
With the usual choices of angle origins, $w_1=\ell$ is <mean anomaly>, $w_2=\omega$ is <argument of periapsis>, and $w_3=\Omega$ is <longitude of ascending node>. The generator $\mathcal G=L$ rotates the orbit within its plane, whereas $\mathcal H=L_z$ rotates that plane about the reference normal; this explains the angular identification. The fast angle obeys <Kepler's equation> $\ell=\mathcal E-e\sin\mathcal E$, where $\mathcal E$ is <eccentric anomaly>.
For retrograde orbits, signed $J_\phi=L_z$ gives $J_\theta=L-|L_z|$ and the spherical Hamiltonian uses $J_r+J_\theta+|J_\phi|$. The stated plus-sign form is therefore a prograde chart, not a global signed-action formula. The <Delaunay variables> still use $\mathcal H=L_z$; on the retrograde branch the generating function is $(\mathcal L-\mathcal G)w_r+(\mathcal G+\mathcal H)w_\theta+\mathcal H w_\phi$. At $e=0$, periapsis is undefined, and at $i=0$ or $\pi$ the node is undefined: these are angle-chart degeneracies, not physical singularities.
For the perturbed planet orbit, the radial and vertical frequencies mean small oscillations about a circular equatorial orbit of cylindrical radius $R$. Write $r=\sqrt{R^2+z^2}$ and retain the stated quadrupole potential. On the equator,
$$
\phi(R,0)=-\frac\mu R-\frac{\alpha^2\mu}{R^3},\qquad
\Omega_c^2=\frac1R\partial_R\phi=\frac\mu{R^3}+\frac{3\alpha^2\mu}{R^5}.
$$
At fixed axial <specific angular momentum> $L_z$, the radial effective potential is $\phi(R,0)+L_z^2/(2R^2)$. Its second derivative at circular balance $L_z=R^2\Omega_c$ gives the <radial epicyclic frequency>
$$
\boxed{\kappa^2=\partial_R^2\phi+3\Omega_c^2
=R\frac{d\Omega_c^2}{dR}+4\Omega_c^2
=\frac\mu{R^3}-\frac{3\alpha^2\mu}{R^5}.}
$$
Reflection symmetry decouples vertical motion to first order. Expanding the potential in $z$ gives
$$
\phi(R,z)=\phi(R,0)+\frac12\left(\frac\mu{R^3}+\frac{9\alpha^2\mu}{R^5}\right)z^2+O(z^4),
$$
so the <vertical epicyclic frequency> is
$$
\boxed{\nu^2=\frac\mu{R^3}+\frac{9\alpha^2\mu}{R^5}.}
$$
The circular orbit is radially stable only for $R^2>3\alpha^2$. For $\alpha\ne0$ in that domain, $\nu>\Omega_c>\kappa$. Vertical node crossings repeat at frequency $\nu$ while azimuth advances at $\Omega_c$, so the line of nodes has <nodal precession> rate $\Omega_c-\nu<0$. Radial periapsis passages repeat at $\kappa$, giving <apsidal precession> rate $\Omega_c-\kappa>0$. To leading quadrupole order, with $n=\sqrt{\mu/R^3}$,
$$
\boxed{\dot\Omega_{\mathrm{node}}\simeq-3n\frac{\alpha^2}{R^2},\qquad
\dot\varpi\simeq+3n\frac{\alpha^2}{R^2}.}
$$
This is <oblate-quadrupole epicyclic precession>. Here $\varpi$ denotes the nearly equatorial longitude of periapsis, not $\omega$ alone.
\b[The precession signs above have a small-eccentricity, small-inclination domain.] They are not a universal claim for every inclined Kepler ellipse. To see the distinction with the same <Delaunay variables>, average the perturbation over an unperturbed orbit. Since $z/r=\sin i\sin(\omega+f)$ and $dt=r^2df/\mathcal G$, the term linear in $e\cos f$ integrates to zero and the $\sin^2(\omega+f)$ average in the $r^{-3}$ integral is $1/2$. Thus
$$
\overline{H}_Q=\frac{\alpha^2\mu}{2a^3(1-e^2)^{3/2}}(1-3\cos^2i)
=\frac{\alpha^2\mu^4}{2\mathcal L^3\mathcal G^3}\left(1-\frac{3\mathcal H^2}{\mathcal G^2}\right).
$$
Differentiating at fixed other actions yields the first-order secular rates
$$
\dot\Omega=-\frac{3n\alpha^2\cos i}{a^2(1-e^2)^2},\qquad
\dot\omega=\frac{3n\alpha^2(5\cos^2i-1)}{2a^2(1-e^2)^2}.
$$
Their nearly equatorial sum reproduces the prograde $\dot\varpi$. For a polar orbit, however, the node is fixed and $\dot\omega<0$. This supplies a concrete counterexample to extending the requested forward periapsis precession beyond the intended epicyclic regime.
Back to article page
