mathematics

Article
On the Bessel Solution of Kepler’s Equation
Riccardo Borghi

Departimento di Ingegneria Civile, Informatica e delle Tecnologie Aeronautiche, Università degli Studi “Roma
Tre”, 00146 Rome, Italy; riccardo.borghi@uniroma3.it

Abstract: Since its introduction in 1650, Kepler’s equation has never ceased to fascinate mathemati-
cians, scientists, and engineers. Over the course of five centuries, a large number of different solution
strategies have been devised and implemented. Among them, the one originally proposed by J. L.
Lagrange and later by F. W. Bessel still continue to be a source of mathematical treasures. Here, the
Bessel solution of the elliptic Kepler equation is explored from a new perspective offered by the theory
of the Stieltjes series. In particular, it has been proven that a complex Kapteyn series obtained directly
by the Bessel expansion is a Stieltjes series. This mathematical result, to the best of our knowledge, is
a new integral representation of the KE solution. Some considerations on possible extensions of our
results to more general classes of the Kapteyn series are also presented.

Keywords: gravitation; Kepler equation; mathematical physics; Kapteyn series; Stieltjes series

MSC: 44A60; 33C10; 30B50

1. Introduction
There is no other equation in the history of science that boasts more solution strategies
than the famous Kepler equation (hereafter KE). Since 1650, a huge number of different
strategies and algorithms have been devised and implemented to deal with KE. The classic
textbook by Colwell [1] lists and briefly describes most of these methods. However, since
1993 (the year of publication of Colwell’s book), this list has has grown so rapidly that it is
difficult to account for all of the strategies that have been developed. Just to gain an idea,
quoting some of the papers that have been published on the subject in the last five years
Citation: Borghi, R. On the Bessel would be sufficient [2–17].
Solution of Kepler’s Equation. In this paper, we focus on a semi-analytic approach to the solution of KE, originally
Mathematics 2024, 12, 154. https:// conceived by J. L. Lagrange and later finalised by F. W. Bessel via a Fourier series expansion,
doi.org/10.3390/math12010154 in which his famous functions appeared. The whole third chapter of [1] is devoted to an
account of the history of the infinite series solutions of KE, starting with an important
Academic Editor: Arkadiusz Jadczyk
theorem proven by Lagrange in 1770, to the series expansion proposed by T. Levi-Civita in
Received: 5 December 2023 1904. Today, the Bessel solution of KE has been abandoned for practical, computational
Revised: 29 December 2023 purposes, although it continues to arouse some interest, especially regarding its mathe-
Accepted: 1 January 2024 matical properties. The Bessel solution is the first example of a class of expansions called
Published: 3 January 2024 the Kapteyn series (hereafter KS) [18], which has gained considerable importance in
mathematics and theoretical physics, even in recent times [19–30].
In the present paper, the Bessel solution of KE is studied from the perspective of the
theory of so-called Stieltjes functions [31–33]. In particular, two main results are established:
Copyright: © 2024 by the author.
(i) a Kapteyn series obtained by a simple “complexification” of the Bessel series is proven
Licensee MDPI, Basel, Switzerland.
This article is an open access article
to be Stieltjes and (ii) thanks to such a (constructive) proof, to our knowledge, a new
distributed under the terms and
integral representation of the KE solution will be obtained. We also hope that, under this
conditions of the Creative Commons
perspective, the availability of a well-established convergence theory for Stieltjes series
Attribution (CC BY) license (https:// based on the Padé approximant [32,34], could provide new life to the Bessel solution, even
creativecommons.org/licenses/by/ as a computational tool.
4.0/).

Mathematics 2024, 12, 154. https://doi.org/10.3390/math12010154 https://www.mdpi.com/journal/mathematics

Mathematics 2024, 12, 154 2 of 11

2. Bessel’ Solution of Elliptic Kepler’s Equation

Consider the motion of a planet around the star S along an elliptic orbit with eccentric-
ity ϵ, as sketched in Figure 1 [35].

Q
\

P
A’ ψ ϕ
O S A

Figure 1. The geometry of Kepler’s Equation.

Suppose P is the position of the planet at time t, which is measured by assuming the
pericenter A as the initial (i.e., at t = 0) position (motion is counter-clockwise). Upon intro-
ducing the mean angular speed ω = 2π/T, with T being the orbital period, the so-called
mean anomaly, say M, is defined by M = ωt. To retrieve the position of the planet along
the orbit at time t > 0, the angle ψ is first evaluated by solving the following trascenden-
tal equation:
M = ψ − ϵ sin ψ , (1)
which provides the position of point Q along the circumscribed circle (dotted in the figure).
The planet’s position P is then immediately obtained through the geometrical construction
sketched in Figure 1. Equation (1) is called the elliptic KE.
The Bessel solution of Equation (1) is provided by [1] (Ch. 3)

ψ = M + S(ϵ; M) , (2)

where function S(ϵ; M ) represents the Fourier series

∞
2 Jn (n ϵ)
S(ϵ; M ) = ∑ n
sin nM . (3)
n =1

Although the series converges for any ϵ ∈ [0, 1), its convergence turns out to be extremely
slow, especially when ϵ → 1. This fact, together with the considerable number of Bessel
function evaluations to be implemented, unavoidably resulted in Equations (2) and (3)
being abandoned as far as the practical use of KE was concerned.
Nevertheless, the Bessel series expansion (3) remained a subject of considerable interest
in both mathematics and theoretical physics. The scope of the present paper is to explore
why this occurred. To this end, we introduce the complex function S(ϵ; M), defined through
the Kapteyn series
∞
2 Jn (n ϵ)
S(ϵ; M) = ∑ exp(inM) , (4)
n =1
n

in such a way that S = Im{S}. The convergence of the series (4) can be proven, for example,
by invoking the following Bessel function asymptotics, valid for ϵ < 1 [36] (Sec. 8.4):

1 exp(nλ)
Jn (n ϵ) ∼ √ √ , n → ∞, (5)
2πχ n
Mathematics 2024, 12, 154 3 of 11

√
where χ = 1 − ϵ2 denotes the ellipse’s aspect ratio and λ is a negative parameter,
provided by
1 1−χ
λ=χ+ log . (6)
2 1+χ
The parameter λ will play a key role in the rest of the paper.
After substituting Equation (5) into Equation (4) and upon introducing the complex
parameter z = exp(λ + iM ), it can immediately be seen that the single terms of the series (4)
asymptotically approach those of the paradigmatic model series
∞
zn
∑ n3/2
, (7)
n =1

which, for |z| < 1, is nothing but the Dirichlet series representation of the so-called
polylogarithm function L3/2 (z), where
∞
zn
Lν (z) = ∑ nν
, |z| < 1 . (8)
n =1

For |z| ≥ 1, the series in Equation (8) can be analytically continuated to the whole complex
plane, besides the half-axis Re{z} > 1.
More importantly (for the scopes of the present paper), functions Lν (z) are examples
of Stieltjes function. For the reader’s convenience, the main definitions of Stieltjes functions
and of Stieltjes series will now be briefly summarized. Interested readers can found some
more useful references for instance in Ref. [37]. Moreover, the formalism of Allen et al. [38]
will be adopted. Consider then an increasing real-valued function µ(t) defined for t ∈ [0, ∞],
with infinitely many points of increase. The measure dµ is then positive on [0, ∞]. Assume
all moments Z ∞
µm = tm dµ , m ≥ 0, (9)
0
to be finite. Then, the formal power series
∞
∑ µm zm , (10)
m =0

is called a Stieltjes series. Such series turns out to be asymptotic, for z → 0, to the function
F (z) defined by
Z ∞
dµ
F (z) = , (11)
0 1 − zt
which is called Stieltjes integral. To give a significative example, the Stieltjes nature of
functions Lν will now be proved. To this end, it is sufficient to start from the following
integral representation [39] (25.12.11):
Z ∞
1 x ν−1 exp(− x )
Lν (z) = − x, (12)
Γ(ν) 0 exp(− x ) − 1/z .

which, on setting t = exp(− x ), gives

(− log t)ν−1 /Γ(ν)

Z 1
Lν (z) = z dt , (13)
0 1 − zt

thus proving that the function Lν (z)/z is a Stieltjes integral exactly of the form (11), with
the measure
Γ(ν, − log t)

 , 0 ≤ t ≤ 1,
Γ(ν)


µ(t) = (14)


1 t > 1,

Mathematics 2024, 12, 154 4 of 11

where symbols Γ(·) and Γ(·, ·) denote the gamma and the incomplete gamma functions,
respectively [39].
The above described “asymptotic connection” between series (4) and (7) via Equation (5)
would at first sight suggest the function S(ϵ; M) could be Stieltjes too. In the next section
such a conjecture will definitely be proved.

3. A Constructive Proof That S(ϵ; M ) Is a Stieltjes Series

We are now going to prove the following:

Theorem 1. Function S(ϵ; M) defined through the series (4) is a Stieltjes function.

Proof of Theorem 1. First of all, the definition of S(ϵ; M ) given in Equation (4) must be
recast as follows:
∞
2 Jn (n ϵ)
S(ϵ; M) = ∑ exp(−λn) zn , (15)
n =1
n

with z = exp(λ + iM ). What we are going to show is that, similarly as for Equation (13), S
can be written as Z 1
ρ(t) dt
S(ϵ; M) = z , (16)
0 1 − zt

where the measure dµ = ρ(t)dt represents the proof goal. Accordingly, the following
infinite system must necessarily hold:
Z 1
2 Jn (n ϵ)
exp(−λn) = tn−1 ρ(t) dt , n ≥ 1. (17)
n 0

The key tool of our proof is Watson’s integral representation of Jn (nϵ), namely [40]

1
Z π
Jn (n ϵ) = exp[−n F (θ; ϵ)] dθ , 0 ≤ ϵ ≤ 1, (18)
π 0

where
p p
θ+ θ 2 − ϵ2 sin2 θ θ 2 − ϵ2 sin2 θ (19)
F (θ; ϵ) = log − , θ ∈ [0, π ] .
ϵ sin θ tan θ
Function F (θ; ϵ) has features which reveal very important for our scopes. One of them is
 q 
F 0; 1 − χ2 + λ = 0 , 0 ≤ χ ≤ 1, (20)

which suggests Equation (18) to be recast as

1
Z π
Jn (n ϵ) exp(−λ n) = exp[−n Gχ (θ )] dθ , (21)
π 0
 p 
where function Gχ (θ ) is defined as Gχ (θ ) = F θ; 1 − χ2 + λ.
Another property of F (θ; ϵ) is that ∂F (θ; ϵ)/∂θ > 0 for θ ∈ (0, π ] [40]. Then, in or-
der for Equation (21) to be “tuned” with Equation (17), the new integration variable
t = exp[− Gχ (θ )] is introduced, in such a way that

dt dGχ (θ )
= − exp[− Gχ (θ )] < 0, (22)
dθ dθ
for all values of θ except for θ = 0 and θ = π, where t′ (θ ) = 0. Accordingly, function
t = t(θ ) can be uniquely inverted into the function θ = θ (t), formally defined by

θ (t) = F −1 [− log t − λ] , t ∈ [0, 1] , (23)

Mathematics 2024, 12, 154 5 of 11

where the symbol F −1 denotes the inverse of F (θ; 1 − χ2 ) with respect to θ. Although
p

its explicit analytical expression cannot be found, θ (t) can be easily numerically com-
puted for any values of t and χ, for instance on using Mathematica’s native command
InverseFunction. In Figure 2, the behaviour of θ (t), evaluated according to Equation (23),
is plotted for different values of the aspect ratio χ.

3.0

2.5

2.0
θ(t)

1.5

1.0

0.5

0.0
0.0 0.2 0.4 0.6 0.8 1.0

t
Figure 2. Behaviour of the function θ (t), according to Equation (23), for χ = 1/10 (solid curve),
χ = 1/2 (dashed curve), and χ = 1 (dotted curve).

Now, we are going to prove that the function θ (t) actually coincides (apart some
constant proportional factors), with the density ρ(t) into Equation (16). To this end, it is
sufficient to change the integration variable in Equation (21) from θ to t, which yields
Z 1
1
Jn (n ϵ) exp(−λ n) = − tn θ ′ (t) dt . (24)
π 0

On performing the t-integration by parts we have

Z 1 Z 1
1 n n n
Jn (n ϵ) exp(−λ n) = − [t θ (t)]10 + tn−1 θ (t)dt = tn−1 θ (t)dt , (25)
π π 0 π 0

or, equivalently,
Z 1
2Jn (n ϵ) exp(−λ n) 2θ (t)
= t n −1 dt , n ≥ 1, (26)
n 0 π

which, once compared to Equation (17), leads to ρ(t) = 2θ (t)/π.

The uniqueness of the function ρ(t) can further be proved by invoking Calerman’s
theorem for the Hausdorff problem, i.e., by showing that
∞
∑ µ−
m
1/2m
= +∞ . (27)
m =0

In order to prove the divergence of the above series, it is worth starting from the following
inequality, proved by Siegel [41]:

Jn (nϵ) exp(−λn) ≤ 1 , n ≥ 1, (28)

according to which we have

n
≥ n, n ≥ 1, (29)
Jn (nϵ) exp(−λn)

and then  1/2n

n
lim ≥ lim n1/2n = 1 , (30)
n→∞ Jn (nϵ) exp(−λn) n→∞

which leads to Equation (27). Our proof is now complete. S(ϵ; M ) is a Stieltjes function.
Mathematics 2024, 12, 154 6 of 11

4. A New Integral Representation of KE’s Solution

The proof given in Section 3 is the main result of the present paper. Here, an interesting
byproduct of such proof is now illustrated. In particular, the “Stieltjesness” of S(ϵ; M ) will
be employed to conceive a new, up to our knowledge, integral representation of the KE
solution. To this end, Equation (16) is first recast as follows:
Z 1
2 θ (t)
S(ϵ; M) = − dt , (31)
π 0 t − 1/z
so that formal partial integration gives

1 1 2 1
    
2 1
Z
S(ϵ; M) = − θ (t) log t − + log t − θ ′ (t) dt . (32)
π z 0 π 0 z

The change of variable t = exp[− Gχ (θ )] then yields

   
1 2 1
Z π
S(ϵ; M) = 2 log − − log exp[− Gχ (θ )] − dθ , (33)
z π 0 z

which can be rearranged into a simpler form with a few work. First of all, we have

z exp[− Gχ (θ )] − 1
   
1 2
Z π
S(ϵ; M) = 2 log − − log dθ , (34)
z π 0 z

or, equivalently,
 
1 2 1
Z π
S(ϵ; M) = 2 log − − log(z exp[− Gχ (θ )] − 1) dθ − 2 log =
z π 0 z
(35)
2
Z π
= 2iπ − log(exp[− F (θ; ϵ) + iM ] − 1) dθ .
π 0

To retrieve the KE solution, it is sufficient to recall that S(ϵ; M ) = Im{S(ϵ; M)}, which
gives at once
i
Z π
S(ϵ; M ) = 2π + dθ {log(exp[− F (θ; ϵ) + iM ] − 1) − log(exp[− F (θ; ϵ) − iM ] − 1)} , (36)
π 0
that can also be recast as
i 1 − exp[− F (θ; ϵ) + iM ]
Z π
S(ϵ; M) = log dθ , (37)
π 0 1 − exp[− F (θ; ϵ) − iM ]

where the term 2π in Equation (36) must be removed, due to the phase wrapping of the
principal branch of the natural logarithm, −π < arg{log z} ≤ π.
Equation (37) is the other main result of the present paper. Compared with other,
formally similar integral representations of the KE solution, for instance that in [24],
Equation (37) is expected to provide computational advantages due to the monotonic de-
creasing behaviour of the function exp[− F (θ; ϵ)] within the integration interval θ ∈ [0, π ].
Just to provide an idea, a single but significant numerical experiment will now be
carried out by using only native commands of Mathematica. A case of particular interest is
that of nearly parabolic orbits around the periapsis, which correspond to ϵ → 1. Accurate
numerical solutions for such a critical regime can be found, for instance, in [12,42]. In
particular, in [42], it was emphasized how solving the elliptic KE for a given couple (ϵ, M )
resulted in a bad conditioned problem in the neighbourhood of (1, 0). To provide an
example of the computational use of the new integral representation in Equation (37), we
considered the sole case of a parabolic orbit, i.e., ϵ = 1.
In Figure 3, the relative error, evaluated with respect to the “exact KE solution”, is
plotted against M ∈ (0, π ). In the above experiments, such an “exact value” was evaluated
Mathematics 2024, 12, 154 7 of 11

by solving Equation (1) via Mathematica’s native command FindRoot with the parameter
WorkingPrecision set to 50 and an initial guess of ψ = π. Function S(ϵ; M) was evaluated
by implementing Equation (36) using the native Mathematica command NIntegrate with
different degrees of accuracy, measured by the parameter WorkingPrecision, which was
set to 10 (a), 15 (b), 20 (c), and 25 (d).

( a) (b )
10-8
relative error

10-9

relative error
10-7
10-10

10-11
10-10
10-12

10-13
10-13
10-14
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

M M
(c) (d)
10-16 10-21
relative error

relative error
10-22
10-18
10-23

10-20
10-24

10-22 10-25

0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

M M
Figure 3. Behaviour of the relative error against M ∈ (0, π ). In the above experiments, the “exact
value” of the KE solution was evaluated by solving Equation (1) via Mathematica’s native command
FindRoot with the parameter WorkingPrecision set to 50 and the initial guess of ψ = π. Function
S(ϵ; M) was evaluated by implementing Equation (36) through the native Mathematica command
NIntegrate with different degrees of accuracy, measured by the parameter WorkingPrecision, which
was set to 10 (a), 15 (b), 20 (c), and 25 (d).

5. Discussions
Here, the classical Bessel solution of Kepler’s equation was revisited from a new
perspective, offered by the beautiful theory of the Stieltjes series. In particular, after intro-
ducing a “complexified” version of the original Kapteyn series (3), its Stieltjes nature was
mathematically proven. Our principal tool was Watson’s integral representation of Bessel
functions. As a potentially interesting by-product of our analysis, to our knowledge, a new
integral representation of the KE solution was achieved.
Exploring the Stieltjes nature of Equation (4) could also reveal new interesting aspects
concerning the more general framework of the Kapteyn series. The need to develop new
methods to achieve the summability of the Kapteyn series of both the first and second kind
have already been invoked in the recent past [25]. In this respect, we believe the results
presented here could be of some help.
zm
Just to gain an idea, consider again the Kapteyn series ∑ Jm (mϵ), but suppose
m ≥1
m
that |z| > exp(−λ). In that case, the series diverges. However, the analysis carried out
in Section 3 is also still valid for z belonging to the whole complex plane, provided that
|arg(z)| < π. In other words, our Kapteyn series can be continued to a Stieltjes function
according to

zm 1
Z π
∑ m
Jm (mϵ) ∼ iπ −
π 0
log(z exp[− F (θ; ϵ)] − 1) dθ , |arg(z)| < π . (38)
m ≥1
Mathematics 2024, 12, 154 8 of 11

which lives, for ϵ ∈ [0, 1), on the whole complex plane, besides the infinite interval (1, ∞).
Proving that a divergent power series is a Stieltjes series also implies that such a series
would be Padé summable. To quote an important example, in [43], numerical evidence for
the factorially divergent perturbation expansion for energy eigenvalues of the PT-symmetric
Hamiltonian H (λ) = p2 + 1/4x2 + iλx3 being a Stieltjes series and thus Padé summable,
were provided. Such conjecture was later rigorously proven in [44]. The Padé summability
of our Kapteyn series also implies that suitable resummation algorithms, the most well-
known being Wynn’s ϵ algorithm, can successfully be employed to retrieve the correct value
provided in Equation (38). Padé approximants remain the most favourite mathematical tool
to resume divergent and slowly convergent series in theoretical physics. This is because a
solid convergence theory has been established for them, especially in the case of Stieltjes
asymptotic series [31–33]. In particular, if the input data are the partial sums of a Stieltjes
series, it can be rigorously proven that certain subsequences of the Padé table converge to
the value of the corresponding Stieltjes function [32,34] (Chapter 5). A typical feature of
Wynn’s ϵ algorithm is that only the input of the numerical values of a finite substring of
the partial sequence of the series is required to achieve a meaningful resummation.
In the case of the Stieltjes series, however, important additional structural information
about the behaviour of the series terms is available. For example, it is known that the
magnitudes of the truncation errors are bounded by the first term neglected in the partial
sum, and they also have the same sign patterns. In the seminal 1973 paper by Levin [45], a
new class of sequence transformations was introduced that used such valuable a priori
structural information to improve the efficiency and convergence speed of the transformed
sequences. For a general discussion on the construction principles of Levin-type sequence
transformations, readers are encouraged to go through the paper by Weniger [46], as well
as through Ref. [47], which contains a slightly upgraded and formally improved version of
the former. To provide a simple example, Table 1 shows the performance of a particular
Levin-type transformation, the so-called Weniger δ-transformation [46], in the resummation
of the Kapteyn series on the left side of Equation (38) for ϵ = 9/10 and z = 10 exp(iπ/3).
The convergence to the value obtained by numerically computing the integral on the
right side of Equation (38) is evident.

Table 1. Resummation, via Weniger δ-transformation [46], of the (divergent) Kapteyn series
zm
∑ m Jm (mϵ), evaluated for z = 10 exp(iπ/3). The value obtained by numerically computing the
m ≥1
integral into the right side of Equation (38) is: −1.001838. . . + 1.238765. . . i.

Order Partial Sum Sequence Weniger δ-Transformation

1 2.02 + 3.51 i 0.112240 + 1.211289 i
10 (4.4 − 10 i) × 108 −1.003096 + 1.238166 i
20 (−3.1 + 32 i) × 1018 −1.001839 + 1.238763 i
30 (7.7 + 10 i) × 1027 −1.001838 + 1.238765 i
... ... ...

Finally, before concluding, it is worth illustrating at least one experiment of the use of
the Weniger transformation to solve Kepler’s equation. Once again, the parabolic case will
be analysed, but now only “small values” of M, namely less than 1/10,  will be considered.
1 1
In Figure 4, the relative error is still plotted versus M ∈ , . Again, the “exact
1000 10
solution” was evaluated by solving Equation (1) via Mathematica’s command FindRoot
with the parameter WorkingPrecision set to 50 and an initial guess of ψ = π, but now the
function S(ϵ; M ) is thought of as the imaginary part of S(ϵ; M ). The latter is computed,
as done in Table 1, via the Weniger δ-transformation with different orders.
Mathematics 2024, 12, 154 9 of 11

0.001

10-7

relative error
10-11

10-15

10-19

0.00 0.02 0.04 0.06 0.08 0.10

M
 
1 1
Figure 4. The same as in Figure 3, but for M ∈ , . Note that now the function S(ϵ; M) is
1000 10
thought of as the imaginary part of S(ϵ; M), which is computed, similarly to that in Table 1, via the
Weniger δ-transformation with an order of 20 (black circles), 30 (open circles), 40 (open squares),
and 50 (black squares).

6. Conclusions
In common with almost any scientific problem which achieves a certain longevity
and whose literature exceeds a certain critical mass, the Kepler problem has acquired
luster and allure for the modern practitioner. Any new technique for the treatment
of trascendental equations should be applied to this illustrious case; any new insight,
however slight, lets its conceiver join an eminent list of contributors.

Nothing can illustrate the fascination that Kepler’s equation has held for nearly
400 years better than the above paragraph, which appears in the introduction to Colwell’s
book. Kepler’s problem continues to inspire mathematicians, scientists, and engineers to
search for further contributions to the subject. In the present paper, the Bessel solution of
KE has been studied from the perspective of Stieltjes function theory. In particular, it has
been proven that the Kapteyn series (3) obtained by the “complexification” of the Bessel
series solution (2) is a Stieltjes series. Moreover, thanks to our constructive proof, to our
knowledge, a new integral representation of the KE solution has been obtained. More
importantly, the limit of validity of such a representation extends far beyond that fixed by
the convergence circle of the Kapteyn series itself.
Stieltjes series are strictly connected with Padé summability via a well established con-
vergence theory. Divergent Stieltjes series can be resummed to the corresponding Stieltjes
integral values using Padé approximants. In the last 40 years, Levin-type transformations
have been shown to possess retrieving capabilities considerably superior to those of Padé,
as far as the resummation of the Stieltjes series is concerned. At present, the main theoretical
drawback of Levin-type transformations is the lack of a rigorous convergence theory, like
that available for Padé. Thus, we also hope that what is contained in the present paper
could provide a further contribution to the construction of such a convergence theory, also
in order to provide Levin-type transformations the scientific recognition they certainly
deserve, but that, unfortunately, in our opinion, still seems too far away.

Funding: This research received no external funding.

Data Availability Statement: The data presented in this study are available on request from the
corresponding author.
Acknowledgments: I am indebted with the anonymous reviewers for their criticisms and suggestions,
always aimed at improving the technical level of the paper. I wish to thank Gabriella Cincotti and
Turi Maria Spinozzi for their useful comments and help. Dedicated to Maurizio Giura, on his
eighty-fifth birthday.
Conflicts of Interest: The author declares no conflicts of interest.
Mathematics 2024, 12, 154 10 of 11

References
1. Colwell, P. Solving Kepler’s Equation Over Three Centuries; Willmann-Bell: Richmond, VA, USA, 1993.
2. Ibrahim, R.; Saleh, A.R. Re-evaluation solution methods for Kepler’s equation of an elliptical orbit. Iraqi J. Sci. 2019, 60, 2269–2279.
[CrossRef]
3. Calvo, M.; Elipe, A.; Montijano, J.; Rández, L. A monotonic starter for solving the hyperbolic Kepler equation by Newton method.
Celest. Mech. Dyn. Astron. 2019, 131, 18. [CrossRef]
4. Tommasini, D.; Olivieri, D. Fast switch and spline scheme for accurate inversion of nonlinear functions: The new first choice
solution to Kepler’s equation. Appl. Math. Comput. 2020, 364, 124677. [CrossRef]
5. Abubekerov, M.K.; Gostev, N.Y. Solution of Kepler’s Equation with Machine Precision. Astron. Rep. 2020, 64, 1060–1066.
[CrossRef]
6. Sacchetti, A. Francesco Carlini: Kepler’s equation and the asymptotic solution to singular differential equations. Hist. Math. 2020,
53, 1–32. [CrossRef]
7. Tommasini, D.; Olivieri, D. Fast Switch and Spline Function Inversion Algorithm with Multistep Optimization and k-Vector
Search for Solving Kepler’s Equation in Celestial Mechanics. Mathematics 2020, 8, 2017. [CrossRef]
8. Zechmeister, M. Solving Kepler’ss equation with CORDIC double iterations. Mon. Not. R. Astron. Soc. 2021, 500, 109–117.
[CrossRef]
9. González-Gaxiola, O.; Hernández-Linares, S. An Efficient Iterative Method for Solving the Elliptical Kepler’s Equation. Int. J.
Appl. Comput. Math. 2021, 7, 42. [CrossRef]
10. Tommasini, D. Bivariate Infinite Series Solution of Kepler’s Equation. Mathematics 2021, 9, 785. [CrossRef]
11. Philcox, O.H.E.; Goodman, J.; Slepian, Z. Kepler’s Goat Herd: An exact solution to Kepler’s equation for elliptical orbits. Mon.
Not. R. Astron. Soc. 2021, 506, 6111–6116. [CrossRef]
12. Tommasini, D.; Olivieri, D. Two fast and accurate routines for solving the elliptic Kepler equation for all values of the eccentricity
and mean anomaly. Astron. Astrophys. 2022, 658, A196. [CrossRef]
13. Zhang, R.; Bian, S.; Li, H. Symbolic iteration method based on computer algebra analysis for Kepler’s equation. Sci. Rep. 2022,
12, 2957. [CrossRef] [PubMed]
14. Wu, B.; Zhou, Y.; Lim, C.W.; Zhong, H. A new solution approach via analytical approximation of the elliptic Kepler equation.
Acta Astronaut. 2023, 202, 303–310. [CrossRef]
15. Vavrukh, M.; Dzikovskyi, D.; Stelmakh, O. Analytical images of Kepler’s equation solutions and their applications. Math. Model.
Comput. 2023, 10, 351–358. [CrossRef]
16. Calvo, M.; Elipe, A.; Rández, L. On the integral solution of elliptic Kepler’s equation. Celest. Mech. Dyn. Astron. 2023, 135, 26.
[CrossRef]
17. Brown, M.T. An improved cubic approximation for K epler’s equation. Mon. Not. R. Astron. Soc. 2023, 525, 57–66. [CrossRef]
18. Kapteyn, W. Researches sur les functions de Fourier-Bessel. Ann. Sci. L’Ecole Norm. Sup. Ser. 1893, 3, 91–122.
19. Dominici, D. A new Kapteyn series. Integral Transform. Spec. Funct. 2007, 18, 409–418. [CrossRef]
20. Lerche, I.; Tautz, R. A note on summation of kapteyn series in astrophysical problems. Astrophys. J. 2007, 665, 1288–1291.
[CrossRef]
21. Pogány, T.K. Convergence of generalized Kapteyn expansion. Appl. Math. Comput. 2007, 190, 1844–1847. [CrossRef]
22. Lerche, I.; Tautz, R. Kapteyn series arising in radiation problems. J. Phys. A Math. Theor. 2008, 41, 035202. [CrossRef]
23. Lerche, I.; Tautz, R.; Citrin, D. Terahertz-sideband spectra involving Kapteyn series. J. Phys. A Math. Theor. 2009, 42, 365206.
[CrossRef]
24. Eisinberg, A.; Fedele, G.; Ferrise, A.; Frascino, D. On an integral representation of a class of Kapteyn (Fourier-Bessel) series:
Kepler’s equation, radiation problems and Meissel’s expansion. Appl. Math. Lett. 2010, 23, 1331–1335. [CrossRef]
25. Tautz, R.C.; Lerche, I.; Dominici, D. Methods for summing general Kapteyn series. J. Phys. A Math. Theor. 2011, 44, 385202.
[CrossRef]
26. Baricz, A.; Jankov, D.; Pogány, T.K. Integral representation of first kind Kapteyn series. J. Math. Phys. 2011, 52, 043518. [CrossRef]
27. Nikishov, A.I. Kapteyn series and photon emission. Bull. Lebedev Phys. Inst. 2014, 41, 332–338. [CrossRef]
28. Baricz, A.; Masirevic, D.J.; Pogány, T.K. Kapteyn Series. Lect. Notes Math. 2017, 2207, 87–111.
29. Xue, X.; Li, Z.; Man, Y.; Xing, S.; Liu, Y.; Li, B.; Wu, Q. Improved Massive MIMO RZF Precoding Algorithm Based on Truncated
Kapteyn Series Expansion. Information 2019, 10, 136. [CrossRef]
30. Bornemann, F. A Jentzsch-Theorem for Kapteyn, Neumann and General Dirichlet Series. Comput. Methods Funct. Theory 2023,
23, 723–739. [CrossRef]
31. Bender, C.M.; Orszag, S.A. Advanced Mathematical Methods for Scientists and Engineers; McGraw-Hill: New York, NY, USA, 1978.
32. Baker, G.A., Jr.; Graves-Morris, P. Padé Approximants, 2nd ed.; Cambridge U. P.: Cambridge, UK, 1996.
33. Widder, D.V. The Stieltjes transform. Trans. Am. Math. Soc. 1938, 43, 7–60. [CrossRef]
34. Brezinski, C. Extrapolation algorithms and Padé approximations: A historical survey. Appl. Numer. Math. 1996, 20, 299–318.
[CrossRef]
35. Orlando, F.; Farina, C.; Zarro, C.; Terra, P. Kepler’s equation and some of its pearls. Am. J. Phys. 2018, 86, 11. [CrossRef]
36. Watson, G.N. A Treatise on the Theory of Bessel Functions; Cambridge University Press: Cambridge, UK, 1944.
Mathematics 2024, 12, 154 11 of 11

37. Caliceti, E.; Meyer-Hermann, M.; Ribeca, P.; Surzhykov, A.; Jentschura, U.D. From useful algorithms for slowly convergent series
to physical predictions based on divergent perturbative expansions. Phys. Rep. 2007, 446, 1–96. [CrossRef]
38. Allen, G.D.; Chui, C.K.; Madych, W.R.; Narcowich, F.J.; Smith, P.W. Padé approximation of Stieltjes series. J. Approx. Theory 1975,
14, 302–316. [CrossRef]
39. Olver, F.W.J.; Olde Daalhuis, A.B.; Lozier, D.W.; Schneider, B.I.; Boisvert, R.F.; Clark, C.W.; Miller, B.R.; Saunders, B.V.; Cohl, H.S.;
McClain, M.A. (Eds.) NIST Digital Library of Mathematical Functions; Release 1.1.3 of 2021-09-15; NIST: Gaithersburg, MD, USA,
2022.
40. Watson, G.N. Bessel functions and Kapteyn series. Proc. Lond. Math. Soc. 1917, 16, 150–174. [CrossRef]
41. Siegel, K.M. An inequality involving Bessel functions of argument nearly equal to their order. Proc. Am. Math. Soc. 1953,
4, 858–859. [CrossRef]
42. Farnocchia, D.; Cioci, D.B.; Milani, A. Robust resolution of Kepler’s equation in all eccentricity regimes. Celest. Mech. Dyn. Astron.
2013, 116, 21–34. [CrossRef]
43. Bender, C.M.; Weniger, E.J. Numerical evidence that the perturbation expansion for a non-Hermitian P T -symmetric Hamiltonian
is Stieltjes. J. Math. Phys. 2001, 42, 2167–2183. [CrossRef]
44. Grecchi, V.; Maioli, M.; Martinez, A. Padé summability of the cubic oscillator. J. Phys. A 2009, 42, 425208. [CrossRef]
45. Levin, D. Development of non-linear transformations for improving convergence of sequences. Int. J. Comput. Math. B 1973,
3, 371–388. [CrossRef]
46. Weniger, E.J. Nonlinear sequence transformations for the acceleration of convergence and the summation of divergent series.
Comput. Phys. Rep. 1989, 10, 189–371. [CrossRef]
47. Borghi, R.; Weniger, E.J. Convergence analysis of the summation of the factorially divergent Euler series by Padé approximants
and the delta transformation. Appl. Numer. Math. 2015, 94, 149–178. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.