---
title: 'Eccentric Velocity Divergence: Astrophysical Dynamics'
url: https://www.emergentmind.com/topics/eccentric-velocity-divergence
type: topic
---

# Eccentric Velocity Divergence: Astrophysical Dynamics

Eccentric velocity divergence denotes a family of context-dependent phenomena in which eccentricity, or a model that attributes observations to eccentric motion, generates a systematic velocity mismatch, harmonic residual, or compressional term. In radial-velocity exoplanet analyses it is explicitly quantified as the difference between a single eccentric Keplerian signal and a two-planet near-\(2\!:\!1\) circular alternative, and it is central to the problem of model degeneracy [1503.07769]. In close-in planetary systems it also appears as an apparent eccentricity induced by stellar tides, because a disk-integrated tidal radial-velocity term at twice the orbital frequency is absorbed by a Keplerian fit as a nonzero \(e\) with \(\omega\approx270^\circ\) [1912.01360]. In continuum kinematics the relevant issue is the role of velocity-field divergence in the evolution of eccentric shapes: for a material ellipse in a linear flow, isotropic divergence cancels out of \(\dot e\), whereas in eccentric discs the divergence of the orbital flow drives compressional “breathing” dynamics [1802.02079]; [1409.6487]. In eccentric binary black-hole mergers, the expression has been used for the oscillatory departure of the final recoil velocity from its quasi-circular value as the initial eccentricity varies [2311.08822].

## 1. Formal senses of the term

The available sources indicate that “eccentric velocity divergence” is not a single standardized invariant, but a recurring descriptor for velocity departures tied to eccentric dynamics or to incorrect eccentric modeling. The common structure is that eccentricity introduces nontrivial harmonics, phasing shifts, or compressive terms that can mimic other mechanisms, conceal additional degrees of freedom, or modulate final observables.

| Context | Velocity quantity | Defining feature |
|---|---|---|
| RV model comparison | \(\Delta RV(t_i)\) | Mismatch between single eccentric and near-\(2\!:\!1\) circular models |
| Tidal RV in hot-Jupiter hosts | \(v_{\rm tide}(t)\) | Second harmonic misread as orbital eccentricity |
| Ellipse and disc kinematics | \(\delta\) or \(\nabla\!\cdot\!u\) | Isotropic divergence preserves shape; orbital divergence can drive compression |
| Eccentric BBH mergers | \(V_f(e_0)-V_f^{\rm circ}\) | Oscillatory recoil departure as \(e_0\) varies |

This taxonomy also separates two distinct meanings of “divergence.” In the exoplanet literature it is a mismatch between competing velocity models, while in fluid and disc dynamics it is the literal divergence operator acting on a velocity field. The sources further suggest that confusion arises most often when a second harmonic or compressional effect is absorbed into a simpler eccentric parameterization.

## 2. Radial-velocity mismatch between eccentric and resonant-planet models

For extrasolar radial-velocity data, the basic degeneracy is between a single eccentric planet and two near-\(2\!:\!1\) circular planets. The competing models are
\[
RV_1(t)=K[\cos(\nu(t)+\omega)+e\cos\omega]+V_0
\]
and
\[
RV_2(t)=K_1\cos(2\pi t/P_1+\phi_1)+K_2\cos(2\pi t/P_2+\phi_2)+V_0,
\]
with \(P_2\approx2P_1\). If \(P_2=2P_1\) and \(e_2=0\) are enforced, both models have six free parameters; if \(P_2\) is free, the two-planet circular model has seven parameters, while a full two-Keplerian fit has eleven. The instantaneous eccentric velocity divergence is
\[
\Delta RV(t_i)=RV_1(t_i)-RV_2(t_i),
\]
and the associated rms mismatch is
\[
{\rm rms}[\Delta RV]=\sqrt{\frac{1}{N}\sum_{i=1}^N[\Delta RV(t_i)]^2 }.
\]
For observed data \(\{RV_{\rm obs}(t_i),\sigma_i\}\), model preference is assessed with
\[
\chi^2=\sum_i \frac{[RV_{\rm obs}(t_i)-RV_{\rm model}(t_i)]^2}{\sigma_i^2},
\qquad
p(\chi^2)=1-F_{\chi^2}(\chi^2;{\rm DoF}),
\]
with \({\rm DoF}=N-N_{\rm par}\). The degeneracy criterion is that a single-Keplerian fit is “indistinguishable” from a \(2\!:\!1\) circular pair whenever
\[
\sigma_{\rm Kep}\equiv {\rm rms}({\rm data}-{\rm best\!-\!Keplerian})\le {\rm rms}_{\rm model}(e_{\rm fit}).
\]
The simulation protocol comprised a dense-sampling experiment with 1000 equally spaced points over \(t\in[0,P_1)\), \(P_1\equiv1\), \(P_2=0.5\), amplitude-ratio grid \(\hat K_2/K_1=i/8\), \(i=0\ldots16\), and 12 phase shifts \(\Delta\phi=0^\circ,30^\circ,\ldots,330^\circ\), as well as a sparse-sampling experiment using the 20 observation times of HD 27894 from Moutou et al. 2005. Up to \(e\approx0.6\), a bona fide \(2\!:\!1\) circular pair can masquerade as a single eccentric orbit, with \({\rm rms}_{\rm model}/K\) rising from \(\lesssim0.01\) at \(e\lesssim0.1\) to \(\approx0.6\) at \(e\approx0.6\). In the January 2015 EOD literature survey, 254 single-planet, \(e>0\), non-transiting systems with reported \(K\) formed the version 1 sample, and \(74\%\) (\(188/254\)) lay above the \({\rm rms}_{\rm model}\) curve, implying that a \(2\!:\!1\) circular pair could not be excluded. In the version 2 subset of 187 systems with published \(\chi^2\) and \(N\), 101 (\(54\%\)) had \(p(\chi^2_{\rm Kep})<0.05\) and 72 (\(39\%\)) had \(p(\chi^2_{\rm Kep})<10^{-3}\), providing confidence thresholds for rejecting a single-Keplerian description [1503.07769].

HD 27894 serves as the concrete case study. The original dataset consisted of 20 RVs with \(\sigma_{\rm obs}\approx1.8\,{\rm m\,s^{-1}}\) instrumental uncertainty. A one-planet Keplerian fit yielded \(P_1=17.990\pm0.008\) d, \(e_1=0.049\pm0.008\), \(K_1=58.13\pm0.50\,{\rm m\,s^{-1}}\), RMS \(=4.21\,{\rm m\,s^{-1}}\), \(\chi_\nu^2=8.71\), and \(p(\chi^2)=2.7\times10^{-19}\). A forced \(2\!:\!1\) resonant circular-orbit solution with \(P_2=P_1/2\) gave \(P_1=17.9915\pm0.0071\) d, \(P_2=8.9957\) d, \(K_2=3.00\pm0.51\,{\rm m\,s^{-1}}\), RMS \(=4.17\,{\rm m\,s^{-1}}\), \(\chi_\nu^2=8.58\), and \(p(\chi^2)=5.9\times10^{-19}\), and was explicitly described as indistinguishable. Allowing a free circular inner period produced \(P_2=8.276\pm0.035\) d, \(K_2=5.17\pm0.62\,{\rm m\,s^{-1}}\), RMS \(=3.20\,{\rm m\,s^{-1}}\), \(\chi_\nu^2=5.44\), and an \(F\)-test significance of \(83\%\). A full two-Keplerian fit gave the best formal \(\chi^2\), with RMS \(=2.42\,{\rm m\,s^{-1}}\), \(\chi_\nu^2=2.52\), \(p(\chi^2)=6.9\times10^{-3}\), and \(e_2=0.532\pm0.076\), but this high-eccentricity inner solution was dynamically unstable unless \(e_2\lesssim0.3\). For \(M_\star=0.80\,M_\odot\), the single-planet interpretation implied \(m_1\sin i=0.645\,M_{\rm Jup}\) and \(a_1\approx0.125\,{\rm AU}\), whereas the circular two-planet interpretation implied an inner candidate with \(m_2\sin i\approx14\,M_\oplus\) and \(a_2\approx0.074\,{\rm AU}\). Dynamical integrations with SyMBA over \(10\,000\) yr showed that the free-\(P_2\) circular two-planet solution was stable with \(e_1,e_2\lesssim10^{-3}\ldots10^{-2}\), while the full two-Keplerian solution was violently unstable unless the inner eccentricity was constrained below \(0.3\). The largest \(\Delta RV(t)\) values occur at specific orbital phases; for \(K_2/K_1=0.5\), the quoted phases are \(\phi\approx0.18,0.40,0.60,0.82\) in \(P_1\)-folded phase, so concentrated follow-up there maximizes leverage.

## 3. Apparent eccentricity from tidal radial velocities

A different eccentric velocity divergence arises when a circular planetary orbit excites stellar tides whose radial-velocity signature is then misread as orbital eccentricity. In the equilibrium-tide approximation, to leading order in
\[
\epsilon \equiv \frac{M_p}{M_\star}\left(\frac{R_\star}{a}\right)^3,
\]
the stellar surface displacement is
\[
\xi_r(\theta,\phi,t)=R_\star\,\epsilon\,P_2(\cos\theta)\cos[2(\phi-nt)],
\]
and the local radial velocity of the tidal bulge is
\[
v_r(\theta,\phi,t)\approx 2nR_\star\epsilon P_2(\cos\theta)\sin[2(\phi-nt)].
\]
After integration over the visible stellar disk, the surviving disk-averaged signal has the form
\[
v_{\rm tide}(t)=K_{\rm tide}\sin(2nt+\phi_2),
\qquad
K_{\rm tide}=\beta\,\frac{M_p}{M_\star}\left(\frac{R_\star}{a}\right)^3 nR_\star,
\]
with \(\beta\) of order unity. By contrast, the small-eccentricity Keplerian stellar reflex velocity,
\[
v_{\rm orb}(t)=K_{\rm orb}[\cos(\omega+f(t))+e\cos\omega],
\]
contains a second harmonic of amplitude \((e/2)K_{\rm orb}\) when expanded to first order in \(e\). A circular orbit plus \(v_{\rm tide}\) is therefore misinterpreted as an apparently eccentric orbit with
\[
e_{\rm app}\simeq \frac{2K_{\rm tide}}{K_{\rm orb}},
\qquad
\omega_{\rm app}\simeq270^\circ.
\]
The source material further states that \(e_{\rm app}\) is on the order of \(10^{-2}\)–\(10^{-1}\) for hot Jupiters [1912.01360].

WASP-12 b is the exemplar. The dataset combined 17 new HARPS-N radial velocities with literature SOPHIE, HIRES, and HARPS-N points; in-transit observations were corrected for the Rossiter–McLaughlin effect, and a stellar jitter of \(7.4\,{\rm m\,s^{-1}}\) was added in quadrature to formal errors to obtain \(\chi_\nu^2\approx1\). The circular fit, with 8 free parameters including per-dataset zero-points and the known orbital-decay \(dP/dE\), had \(\chi^2_{\rm circ}=186.0\) for \(\nu=124\). The eccentric fit, with two additional parameters \(e\) and \(\omega\), had \(\chi^2_{\rm ecc}=142.5\) for \(\nu=122\), so \(\Delta\chi^2=43.5\) for \(\Delta\nu=2\), with \(F\)-test significance \(\gg5\sigma\). The BIC values were \(304.8\) for the circular model and \(291.9\) for the eccentric model, giving \(\Delta{\rm BIC}=12.9\) and odds \(\approx e^{\Delta{\rm BIC}/2}=6.2\times10^2\) in favor of the eccentric solution. The best-fit Keplerian parameters were \(e=0.035\pm0.006\) and \(\omega=270.7^\circ\pm0.6^\circ\), corresponding to a \(5.8\sigma\) detection. A combined orbital-plus-tide model,
\[
v(t)=\gamma'+K_{\rm orb}\sin[2\pi(\phi-\phi_0)]+K_{\rm tide}\sin[4\pi(\phi-\phi_0)],
\]
fit with an MCMC of \(100\times10^4\) steps and burn-in \(10^3\), returned \(K_{\rm orb}=220.0\pm1.3\,{\rm m\,s^{-1}}\), \(K_{\rm tide}=7.5\pm1.2\,{\rm m\,s^{-1}}\), \(\phi_0\approx0.001\), and \(\gamma'\approx0\). Because \(\phi_0\) and \(\gamma'\) are consistent with zero, the tidal component alone explains the apparent eccentricity. The same \(\ell=2\) tide is predicted to produce an ellipsoidal flux modulation
\[
\Delta F_{\rm ellip}=\alpha_{\rm ellip}\frac{M_p}{M_\star}\left(\frac{R_\star}{a}\right)^3\sin^2 i,
\]
with amplitude \(\approx80\) ppm for WASP-12 b, peaking at orbital phases \(0.25\) and \(0.75\). The cited diagnostics for distinguishing true eccentricity from tide-induced eccentric velocity divergence are the detection of this \(2n\) photometric signal, transit-to-occultation timing consistency, inspection of higher harmonics, line-profile and bisector variability, and joint RV-plus-photometry fitting.

## 4. Divergence and shape evolution of a fluid ellipse

In linear two-dimensional flow, the issue is not observational degeneracy but whether the velocity-field divergence alters the eccentricity of an advected ellipse. Writing the ellipse as
\[
\frac1{a^2(t)}\bigl[x\cos\theta(t)+y\sin\theta(t)\bigr]^2+
\frac1{b^2(t)}\bigl[-x\sin\theta(t)+y\cos\theta(t)\bigr]^2=1,
\]
with eccentricity
\[
e(t)=\sqrt{1-\bigl[b(t)/a(t)\bigr]^2},
\]
the velocity gradient is decomposed as
\[
\nabla{\bf u}=
\begin{pmatrix}
u_x & u_y\\
v_x & v_y
\end{pmatrix},
\]
with divergence \(\delta=u_x+v_y\), vorticity \(\zeta=v_x-u_y\), normal strain \(\gamma_1=u_x-v_y\), and shear strain \(\gamma_2=v_x+u_y\). The strain magnitude and orientation are
\[
\gamma=\sqrt{\gamma_1^2+\gamma_2^2},
\qquad
\alpha=\tfrac12\arctan\!\bigl(\gamma_2/\gamma_1\bigr).
\]
In the ellipse-aligned frame, the principal strains are
\[
\tilde\gamma_1=\gamma\cos2(\alpha-\theta),
\qquad
\tilde\gamma_2=\gamma\sin2(\alpha-\theta),
\]
and the axis evolution equations are
\[
\frac{d}{dt}\ln a=\frac12\delta+\frac12\tilde\gamma_1,
\qquad
\frac{d}{dt}\ln b=\frac12\delta-\frac12\tilde\gamma_1.
\]
Thus \(\delta\) appears symmetrically in both axis equations as an isotropic stretch rate, while \(\tilde\gamma_1\) is the differential stretching along the principal axes [1802.02079].

Differentiating \(e^2=1-b^2/a^2\) yields
\[
\dot e=\frac{1-e^2}{2e}\,\gamma\,\cos2(\alpha-\theta).
\]
The source explicitly emphasizes that \(\delta\) has completely cancelled out of \(\dot e\). Pure divergence, with \(\delta\neq0\) but \(\gamma_1=\gamma_2=0\), gives
\[
\dot a/a=\dot b/b=\delta/2,
\qquad
\dot e=0,
\]
so the degree of ellipticity is frozen. Incompressible flow, with \(\delta=0\), gives
\[
\dot a/a=\tilde\gamma_1/2,
\qquad
\dot b/b=-\tilde\gamma_1/2,
\]
and all shape change is strain-driven. The physical interpretation is correspondingly sharp: isotropic area stretch multiplies both \(a\) and \(b\) by the same factor \(e^{\tfrac12\int\delta\,dt}\), preserving \(b/a\), whereas \(\tilde\gamma_1\) alone changes the shape. This use of “divergence” is therefore almost the complement of the exoplanet RV usage: the divergence exists in the axis equations but drops out of the eccentricity equation.

## 5. Orbital-flow divergence in eccentric astrophysical discs

In eccentric discs, the orbital velocity field can have a genuine nonzero divergence that enters the local mass, momentum, and energy budgets. Orbits are labeled by the semi-latus rectum
\[
\lambda=a(1-e^2),
\]
and in orbital coordinates the radius is
\[
r=R(\lambda,\phi)=\frac{\lambda}{1+e(\lambda)\cos(\phi-\omega(\lambda))}.
\]
The purely orbital motion has
\[
u^\lambda=0,\qquad u^\phi=\Omega(\lambda,\phi),\qquad u^z=0,
\]
with
\[
\Omega(\lambda,\phi)=\sqrt{\frac{GM}{\lambda^3}\,\bigl[1+e(\lambda)\cos(\phi-\omega(\lambda))\bigr]^2}.
\]
If \(J(\lambda,\phi)=R\,R_\lambda\) is the Jacobian, then the covariant divergence is
\[
\nabla\!\cdot\!u=\nabla_i u^i=\frac{1}{J}\partial_i(Ju^i),
\]
and because only the \(\phi\)-component survives,
\[
\nabla\!\cdot\!u=\frac{1}{J}\,\partial_\phi(J\Omega)\equiv\Delta(\lambda,\phi).
\]
With \(\theta=\phi-\omega(\lambda)\), \(e'=de/d\lambda\), and \(\omega'=d\omega/d\lambda\), the explicit expression quoted in the source is
\[
\Delta(\lambda,\phi)
=
\sqrt{\frac{GM}{\lambda^3}}\;
\frac{(1+e\cos\theta)\,[\,\lambda e'\sin\theta-\lambda e\,\omega'(\cos\theta+e)\,]}
{1+(e-\lambda e')\cos\theta-\lambda e\,\omega'\sin\theta}.
\]
In the untwisted small-\(e\), small-\(e'\) limit, this reduces to
\[
\Delta \simeq \sqrt{\frac{GM}{\lambda^3}}\;\lambda e' \sin\theta,
\]
so a nonzero eccentricity gradient \(\lambda e'\) is the exclusive source of \(\nabla\!\cdot\!u\) in that limit. By contrast, in a uniformly eccentric disc with \(e'=0\) and \(\omega'=0\), \(J\Omega={\rm const}\) and therefore \(\nabla\!\cdot\!u=0\), even though both \(J\) and \(\Omega\) vary with \(\phi\) [1409.6487].

This divergence forces vertical dynamics. The local vertical momentum and continuity equations are
\[
D\,v^\zeta=-\Phi_2(\lambda,\phi)\,\zeta-\frac{1}{\rho}\partial_\zeta p,
\qquad
D\,\rho=-\rho\,(\Delta+\partial_\zeta v^\zeta),
\]
where
\[
\Phi_2(\lambda,\phi)=\frac{GM}{R^3}
=\frac{GM}{\lambda^3}(1+e\cos\theta)^3.
\]
For the laminar ansatz
\[
v^\zeta=w(t)\zeta,\qquad
\rho=\hat\rho(t)\,F_\rho(\zeta/H(t)),\qquad
p=\hat p(t)\,F_p(\zeta/H(t)),
\]
the source gives the ODE system
\[
\dot H/H=w,\qquad
\dot w+w^2=-\Phi_2(t)+\frac{\hat p}{\hat\rho\,H^2},
\]
\[
\frac{\dot{\hat\rho}}{\hat\rho}=-(\Delta+w),\qquad
\frac{\dot{\hat p}}{\hat p}=-\gamma(\Delta+w).
\]
Thus \(\Delta\) acts as a background compression or expansion rate, performs \(PdV\) work through a \(-p\Delta\) term in the energy equation, and modulates the vertically integrated stresses that enter the global evolution of the surface density and complex eccentricity \(E\). The abstract further states that the laminar solutions can exhibit extreme compressional behaviour for eccentricities greater than about \(0.5\), especially in discs that behave isothermally.

## 6. Oscillatory recoil divergence in eccentric binary black-hole mergers

In numerical-relativity studies of nonspinning eccentric binary black holes, the final recoil velocity shows a systematic oscillatory departure from its circular-limit value as the initial eccentricity \(e_0\) varies. The cited summary represents the numerical data by the ansatz
\[
V_f(e_0)=V_f^{\rm circ}
+A_V(q,D_{\rm ini})\,
\sin\!\bigl[\omega_V(q,D_{\rm ini})\,e_0+\phi_V(q,D_{\rm ini})\bigr]\,
e^{-\beta_V(q,D_{\rm ini})\,e_0},
\]
where \(V_f^{\rm circ}\) is the circular-limit recoil, \(A_V\) is the oscillation amplitude, \(\omega_V\) sets the “wavelength” in \(e_0\), \(\phi_V\) is a phase offset, and \(\beta_V\) is a mild envelope-decay parameter. For equal masses \(q=1\) at \(D_{\rm ini}=11.3\,M\), the quoted least-squares fit is approximately
\[
V_f^{\rm circ}\simeq1.75\times10^{-3},\quad
A_V\simeq5\times10^{-4},\quad
\omega_V\simeq60,\quad
\phi_V\simeq0.3,\quad
\beta_V\simeq0.8.
\]
The same summary states that this corresponds to kicks of order \(200\)–\(400\)\,km/s, oscillating by \(\pm50\)–\(100\)\,km/s as \(e_0\) varies from 0 to 1, with oscillation amplitudes typically \(20\%\)–\(30\%\) of \(V_f^{\rm circ}\), a period \(\Delta e_0\approx2\pi/\omega_V\sim0.08\)–0.12, and the first maximum around \(e_0\approx0.05\)–0.1 [2311.08822].

The physical mechanism is a phasing effect tied to the number of gravitational-wave cycles,
\[
N_{\rm waves}(e_0)=\frac{\Delta\Phi_{22}}{4\pi},
\qquad
\Delta\Phi_{22}=\Phi_{22}(t_{\rm merger})-\Phi_{22}(t_0+t_{\rm relax}).
\]
Each time \(N_{\rm waves}\) crosses an integer, the plunge phasing shifts by \(\pi\), changing the sign or magnitude of the net linear-momentum flux. The source describes this as a “phasing resonance” that alternately enhances and suppresses the recoil, producing a quasi-sinusoidal wriggle in \(V_f(e_0)\). Correlations with other remnant quantities form a spiral-like internal fine structure. Defining
\[
Z(e_0)=\bigl[L_{\rm peak}(e_0)-L_{\rm peak}^{\rm circ}\bigr]
+i\bigl[V_f(e_0)-V_f^{\rm circ}\bigr]
\equiv R(e_0)e^{i\theta(e_0)},
\]
the locus winds outward from the circular reference as \(e_0\) increases, and similar spirals appear in \(\{L_{\rm peak},M_f\}\), \(\{M_f,\alpha_f\}\), and \(\{V_f,\alpha_f\}\). The cited implications are astrophysical: in globular clusters, galactic nuclei, and AGN disks, where dynamical interactions can produce mergers with \(e_0\gtrsim0.2\), retention probabilities and population-synthesis recoil distributions can differ by tens to hundreds of km/s from quasi-circular predictions, and the multi-observable spiral may serve as a consistency check for residual eccentricity at merger.

Across these domains, eccentric velocity divergence is therefore best understood as a technically heterogeneous but structurally related concept. In RV exoplanet work it measures how a wrong eccentric model absorbs resonant or tidal harmonics; in continuum mechanics it clarifies when velocity divergence is shape-neutral and when it is compressive; and in compact-object mergers it captures an oscillatory strong-field departure from circular-limit recoil. A recurring lesson in all cases is that eccentricity should not be treated as a single scalar nuisance parameter: it reorganizes the harmonic content, phase structure, and compressional dynamics of the underlying system.

Source: https://www.emergentmind.com/topics/eccentric-velocity-divergence