Eccentricity-enhanced near-resonant transit-timing variation (source code)

= Eccentricity-enhanced near-resonant transit-timing variation
{title2=$|A_t|/P_t\simeq3\mu_2\alpha|f|e/(2\pi|\Delta|)$}

Near a first-order $(j+1):j$ <orbital resonance>, a weak periodic <resonant orbital perturbation> can accumulate a large <transit-timing variation>. Let subscripts 1 and 2 denote inner and outer <planets>, define $\mu_2=M_2/M_\star\ll1$, $\alpha=a_1/a_2$, and write
$$
\frac{P_2}{P_1}=\frac{j+1}{j}(1+\Delta),\qquad
\omega=(j+1)n_2-jn_1=-\frac{jn_1\Delta}{1+\Delta}.
$$
Here $n_i=2\pi/P_i$ are <mean motions>. Approximate the inner <planet>'s disturbing potential per unit <mass> by
$$
\mathcal R=\frac{GM_2}{a_2}f(\alpha)e\cos\phi,\qquad
\phi=(j+1)\lambda_2-j\lambda_1-\varpi_1,
$$
where $f$ is the dimensionless coefficient of this chosen resonant harmonic, $e$ is its effectively constant free <orbital eccentricity>, $\lambda_i$ are <mean longitudes> and $\varpi_1$ is <longitude of periapsis>. Neglect precession and suppose the <resonant argument> circulates approximately uniformly.

The leading semimajor-axis equation gives $|\dot a_1|/a_1\sim2j\mu_2\alpha n_1|f|e$. Since $\delta n_1/n_1=-3\delta a_1/(2a_1)$, one integration supplies a factor $1/|\omega|$ in the <mean motion> perturbation, and a second supplies it in the <mean longitude>. Converting the phase perturbation to a timing amplitude gives
$$
|A_t|\simeq\frac{3j\mu_2\alpha n_1|f|e}{\omega^2},\qquad
\frac{|A_t|}{P_t}\simeq\frac{3\mu_2\alpha|f|e(1+\Delta)}{2\pi|\Delta|},\qquad P_t=\frac{2\pi}{|\omega|}.
$$
Dropping $\Delta$ against one in the numerator gives the displayed near-resonant estimate. Its $e/\Delta^2$ enhancement assumes this term dominates direct phase and forced-eccentricity effects. Close enough to resonance, <resonant-argument libration> and changes of the free <orbital eccentricity> invalidate the expansion; the divergence does not predict arbitrarily large transit times.