In all plots, blue shows the exact solution, green shows the arbitrary-precision approximation and red shows the classical Kepler/Newton approximation.
In this computation, we set

and

. Then the effective radial potential is:

,
where we introduce the normalized parameter
restricted to the range

. Expanding the potential around the minimum gives

.
The frequency

of small oscillations is then given by

.

.
This is the simplest exact form for zero-order precession. As

and

, then

and there is no precession. As

and

, then

. In terms of the effective potential, the divergence occurs as the local minimum nears the local maximum and the curve flattens out. The local max and min coincide when

. The Schwarzschild radius is at

. At the critical point,

gives the relative radius of the innermost stable circular orbit (isco), depicted as a dashed black line around the black hole in the orbit graphs. The prediction that

provides a test for all solutions: as

, the orbit should approach no closer to

than

.
Other authors [9-11] discuss aesthetically pleasing solutions that arise from a conspiracy of parameters, as occurs with

,

. It is possible to write an algorithm that operates on the approximate solution, a particular value for

and a rational number

,

, that returns an approximate solution for the

value, ultimately by finding the root of a polynomial equation. Acceptable values close to

can be found quickly using a linear procedure because
is a small number. For example for

and

,

. Doing a quick search, we obtain the following approximate "Magic Numbers":
 |  |  |  | ❘ |  |  |  |  | ❘ |  |  |  |  |
1/2 | 6/5 | 0.702 | 0.007 | ❘ | 1/20 | 2 | 0.252 | 0.009 | ❘ | 1/50 | 3 | 0.113 | 0.024 |
1/2 | 5/4 | 0.652 | 0.012 | ❘ | 2/20 | 2 | 0.259 | 0.035 | ❘ | 2/50 | 3 | 0.118 | 0.089 |
1/2 | 4/3 | 0.584 | 0.024 | ❘ | 3/20 | 2 | 0.269 | 0.074 | ❘ | 3/50 | 3 | 0.126 | 0.181 |
1/2 | 3/2 | 0.489 | 0.069 | ❘ | 4/20 | 2 | 0.283 | 0.119 | ❘ | 4/50 | 3 | 0.136 | 0.286 |
Exploring these sequences shows interesting properties of closed orbits. We see that each figure has

-fold dihedral symmetry with

crossing points.
From a mathematical perspective, it would be interesting to extend the precision measurement of

values, prove the empirical crossing formula and give a physical characterization of the crossing points using the full equations of motion, but physicists will be more interested in comparing predictions with data. Planetary precession calculations often use only the zero-order precession terms. For example, Mercury has

,

,

arcseconds per Julian century.
Substituting

, this Demonstration shows that a Kepler/Newton approximation is almost good enough to describe an isolated Mercury-Sun system on the order of one year. In more extreme gravity, the zero-order approximation fails, and higher-order energy dependence needs to be considered. This is an exciting opportunity, as humankind continues to develop incredible new experiments in black hole astronomy [2, 6].
[2] J. A. Wheeler,
A Journey into Gravity and Spacetime, New York: W. H. Freeman and Co., 1990 pp. 168–183.
[3] R. M. Wald,
General Relativity, Chicago: University of Chicago Press, 1984 pp. 139–143.
[5] M. Abramowitz and I. A. Stegun (eds.),
Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables, New York: Dover Publications, 1965. (Sep 28, 2016)
people.math.sfu.ca/~cbm/aands/page_627.htm.
[6] G. Scharf. "Schwarzschild Geodesics in Terms of Elliptic Functions and the Related Red Shift."
arxiv.org/abs/1101.1207.
[9] P. Erdös, "Spiraling the Earth with C. G. J. Jacobi,"
American Journal of Physics,
68(10), 2000 pp. 888–895.
doi:10.1119/1.1285882.