1. Introduction
It is well-known that the restricted three-body problem may serve as a basic dynamical model for the study of many real systems in the field of Celestial Mechanics. One of its benefits, in this field, is that it may be used for several applications in our Solar system and especially in the Earth-Moon one (see [
1,
2], among others) while another asset is that it may be utilized for revealing the orbital dynamics of a natural/artificial satellite or an asteroid (e.g., [
3,
4,
5]). In addition, it may be adopted for the consideration of the dynamical stability of an inner terrestrial planet which is in mean-motion resonance with an outer giant planet [
6]. For the study of the restricted three-body problem, several authors concentrated their work on periodic orbits because they play a key role on the exploration of its dynamics due to their immediate connection with the characterization of nearby orbits (see [
7,
8,
9,
10,
11] and references therein). Additionally, stable periodic orbits are important in planetary dynamics since they can host real planetary systems [
12]. These orbits are also strongly connected with the librational motion either in two or three-dimensions. For example, Voyatzis et al. [
13] determined resonant families of three-dimensional periodic orbits related to the dynamics of the Sun-Neptune and a trans-Neptunian object system in order to study the librations and the long-term evolution of orbits near them. Furthermore, several modifications of the restricted problem have been proposed in the past so as to make it more realistic to certain systems of Celestial Mechanics. Such modifications include the radiation and oblateness effects of the primaries [
14,
15,
16,
17] or the incorporation of some relativistic terms [
18,
19], while various works involve also a larger number for the primary bodies [
20,
21].
A special variant of the classical restricted three-body problem is the Hill one which was firstly proposed by Hill [
22] for the study of the moon’s motion. In this limiting case of the restricted problem the massless body is attracted by two primary bodies one of which is extremely larger from the second one, e.g., the Sun and the Earth. In the rotating coordinate system, the smaller primary is located at the origin while the positive O
x-axis points to the larger one which is always at infinite distance from the secondary. For this problem, Hénon [
23] explored numerically the network of families of simple planar periodic orbits together with their horizontal stability properties while the same author also studied the vertical stability of these families [
24]. The three-dimensional periodic solutions emanating from the two collinear equilibrium points of the Hill’s problem were determined by Zagouras and Markellos [
25]. For the same problem, Hénon [
26] searched for families of multiple planar periodic orbits while low-energy escaping trajectories were numerically investigated by means of the Poincaré maps by Villac and Scheeres [
27].
In the framework of the Hill’s problem where the radiation of the larger primary is also taken into account Kanavos et al. [
28] studied its equilibrium points and produced a general map of symmetric periodic orbits in the space of initial conditions while Papadakis [
29] presented families of symmetric periodic orbits in the case of regularized variables through the Levi-Civita coordinate transformation. In addition, for the same modified model, de Bustos et al. [
30] worked on the bifurcation analysis of the main families of simple periodic orbits. Also, Markellos et al. [
31] introduced the version of Hill’s problem with oblateness for which they determined the Hill stability of direct orbits around the smaller primary while the network of families of simple and multiple planar periodic orbits was computed by Perdiou et al. [
32] and Perdiou [
33], respectively. For the same problem, Papadakis [
34] determined the map of the basic families using the regularized equations of motion. In the case where both the oblateness and radiation are incorporated in the model Perdiou et al. [
35] presented the chart of the basic planar families together with their stability giving special attention to the stability of retrograde satellites whilst by the use of several numerical techniques Zotos [
36] revealed the fractal basins of attraction associated to the collinear equilibria in the complex plane.
Furthermore, in the framework of the classical spatial Hill three-body problem, Batkhin and Batkhina [
37] investigated the families of spatial periodic orbits which bifurcate from the Vertical Critical (VC) orbits of the basic families of simple planar periodic solutions and formed a network that connect those planar orbits with the determined spatial ones. In the same vein, our aim here is to numerically explore all the families of three-dimensional periodic orbits (up to their natural termination), which emerge through their bifurcations from the Vertical Self-Resonant (VSR) orbits of the previous mentioned basic planar families. In all the considered cases, we find that each VSR orbit gives rise to two branches of families of three-dimensional periodic orbits whose type of symmetry depends on their own multiplicity. In particular, if the multiplicity of the detected spatial orbits is odd, the member orbits of the one branch is of axisymmetric type while the members of the second one possess the plane symmetric type. In the case where the spotted spatial orbits have been ascertained to have even multiplicity, both the generated branches are constituted by doubly symmetric periodic orbits. This pattern was firstly observed by Robin and Markellos [
38] for the vertical branches of the basic family of retrograde satellites in the circular restricted three-body problem. Our results focus to the VSR orbits which give rise to three-dimensional periodic orbits with multiplicity three and four which means that the generated spatial families consist of orbits having the triple or quadruple the period comparably to that of the VSR orbit, respectively. Twenty four such families are found, six of which consist of axisymmetric orbits, six of plane symmetric ones and twelve families are composed by doubly symmetric periodic orbits.
Our work is structured as follows. In
Section 2, we recall the equations of motion of the Hill’s problem and compute the families of simple planar symmetric periodic orbits together with their stability properties. Specifically, we determine accurately the VSR periodic orbits at which families of three-dimensional periodic orbits bifurcate where their orbits have the triple or quadruple multiplicity (period) with respect to the multiplicity of the detected planar VSR orbits. In
Section 3, the families of spatial periodic orbits that bifurcate from the planar VSR orbits are computed and presented. Finally, in
Section 4, we summarize our work and conclude.
3. Spatial Periodic Orbits
The VSR orbits give rise to families of three-dimensional periodic orbits which may possess all possible types of symmetry. In particular, the multiplicity
q, as it is defined in (
12), determines the symmetry properties of the generated spatial periodic orbits. These orbits possess, at least initially, the period
where
T is the whole period of a VSR orbit. Additionally, at each VSR orbit two branches of spatial periodic orbits bifurcate, i.e., from VSR orbits families always occur in pairs. This has been firstly identified by Robin and Markellos [
38] who also described the mechanism where the spatial periodic orbits branch out from the plane. They pointed out that at a VSR orbit, i.e.,
exactly two branches of three-dimensional periodic orbits bifurcate while their symmetry properties depend on their own multiplicity. Specifically, in case that the generated spatial periodic orbits are of:
Odd multiplicity q, i.e., their period is with T being the VSR orbit’s period, two such families branch out from the corresponding VSR orbit; one family consists of axis-axis symmetric periodic orbits while the members of the other family are plane-plane symmetric. More precisely, each branch has one of the following symmetries:
- (a)
O Ox symmetry, i.e., they are axisymmetric spatial periodic orbits. They start on the Ox-axis with initial conditions of the form while at their half period return on this axis with final conditions of the form .
- (b)
O O symmetry, i.e., they are plane symmetric spatial periodic orbits. They start on the O plane with initial conditions of the form while at their half period return on this plane perpendicularly with final conditions .
Even multiplicity q, i.e., their period is with T being the VSR orbit’s period, two families branch out from the corresponding VSR orbit; both of them consist of orbits which are doubly symmetric according to the following characteristics:
- (a)
O O symmetry, i.e., they are doubly symmetric spatial periodic orbits. They start on the Ox-axis with initial conditions of the form while at the quarter of their period return perpendicularly on the O plane with the final conditions .
- (b)
O Ox symmetry, i.e., they are also doubly symmetric spatial periodic orbits. However, now they start on the O plane with initial conditions of the form while at the quarter of their period return on the Ox-axis with the final conditions respectively.
For the computation of these branches we can construct corrector-predictor algorithms based on the periodicity conditions. So, to compute a three-dimensional periodic orbit of type, e.g., O
O
double symmetry (case 2(a) above) we choose an initial state vector of the form
. In this case, at the quarter
of the period, namely the orbit meets the O
plane, the following conditions must be hold:
i.e., the orbit has to cross perpendicularly the O
plane in order the periodicity to be established. In general, these are not satisfied, so we seek appropriate corrections
and
of the initial conditions. Then, the resulting equations are expanded in Taylor series up to first order terms obtaining:
Since we have two simultaneous equations with three unknowns, we choose to fix one of them, e.g.,
and solve for obtaining the remaining corrections. In this case, these are:
where the partial derivatives
and
are elements of the variational matrix given in (
8). By iterating this process the three-dimensional periodic solution will be computed with the desired accuracy. If a spatial periodic orbit has been sought, i.e., conditions (
19) are fulfilled with a predetermined accuracy, a next such orbit existing in its neighbourhood can be predicted. To do so, we use the linearized system (
20) and slightly change
to an arbitrarily chosen small constant
in order to get the remaining predictions in the form:
where in system (
20) we have now set
and
to distinguish the prediction and correction steps. Also, the new terms appeared in (
22) are
and
.
The stability of a three-dimensional periodic orbit can be determined through the variational Equation (
7) and using the following formulas [
38]:
with
and
while
V is the variational matrix given in (
8), computed at the orbit’s period
T. In this case, the periodic orbit in question is stable if the defined parameters
Q of (
23) are reals for which it simultaneously holds
and
. Note that, we can also exploit the symmetry properties of the periodic orbit so as to determine the variational matrix at the half or the quarter of the period
i.e.,
or
depending on whether it possesses a simple or double symmetry, respectively. In particular, for the simple types of symmetry we get the variational matrix by using the following two formulas [
38]:
where the first one corresponds to the O
O
x type of symmetry and the second holds for the O
O
plane one. Finally, for the orbits of double symmetry we have that:
The first formula fits to the O
O
double symmetry while the second to the O
O
x one. In all the above mentioned cases,
M are the diagonal matrices
and
respectively. Obviously, the determination of the variational matrix through transformations (
24) and (
25) offers economy to the computations of the numerical integration.
So, by applying the above mentioned techniques we have determined twenty four families of three-dimensional periodic orbits, together with their stability properties, which bifurcate from the VSR orbits given in
Table 1. Note that, the corresponding families where the VC orbits, of the same table, generate have been discussed by Batkhin and Batkhina [
37], so we do not consider their study here. Specifically,
Table 1 contains twelve VSR periodic orbits; each one of these planar orbits possesses two vertical intersections with the O
x-axis (given in this table as
and
, respectively) so as a mirror configuration to be satisfied [
45]. Twelve families of three-dimensional periodic orbits branch from the first vertical intersection of a VSR orbit while the rest twelve spatial families are generated from the second intersection of the same VSR orbit. All these branches of spatial periodic orbits have been computed and followed to their natural end.
The planar Lyapunov family
a has two VSR orbits under consideration at which four families of spatial periodic orbits bifurcate. These are shown in
Figure 3 where we have plotted their characteristic curves in the space of their initial conditions. Families
and
bifurcate from the first perpendicular intersections of
and
VSR orbits, respectively, while the second such intersections give rise to the families
and
correspondingly. In the superscript, the first number follows the running number of the corresponding VSR orbit while the second one indicates that the period of the spatial periodic orbits is, at least initially,
or
where
T is the period of the VSR orbit, i.e., it describes the period commensurability between the planar and spatial orbits.
More precisely, in the first frame of
Figure 3 we present the characteristic curve of the planar family
a by giving its initial conditions
where the value of the Jacobi constant
is determined by (
3) for
and
(first perpendicular cuts of family’s orbits with the O
x-axis), while the second frame depicts the respective characteristic curve in the
plane, i.e., these conditions correspond to the second perpendicular cuts
and
of the planar orbits with the O
x-axis (obviously it holds
). Also, according to the multiplicity
q, family
consists of axisymmetric spatial periodic orbits and family
of plane symmetric ones while families
and
have spatial members which are doubly symmetric. All the computed families go to three-dimensional collision orbits, so we consider that this is their natural termination and stop calculating them.
In
Figure 4 the corresponding spatial branches which the planar family
g generates are presented. This family possesses four VSR orbits therefore, eight such branches bifurcate. For their names we follow the same terminology described in the previous paragraph. So,
gives rise to families
and
which consist of doubly symmetric spatial members and both of them terminate with co-planar periodic orbits. Families
and
bifurcate from
and due to their orbit’s multiplicity they are also comprised by orbits of double symmetry. Both of them goes to three-dimensional collision orbits. Finally, families
and
are produced by the VSR orbits
and
respectively. The spatial member orbits of branches
and
are axisymmetric while the corresponding members of families
and
are plane symmetric. The last four families terminate on the plane.
All families of three-dimensional periodic orbits which bifurcate from the VSR orbits of the planar family
and their members have initially multiplicities 3 and 4 are presented in
Figure 5. The six VSR orbits
and
of
which have been spotted indicate that, in total, twelve vertical branches intersect them. In particular, the branched families
and
, which are generated from the first vertical intersections of the VSR orbits
and
respectively, consist of three-dimensional members which admit the O
O
x axis symmetry. Moreover, the first two families terminate on the O
plane with coplanar orbits while the third one goes to a collision orbit in three-dimensions. The second vertical intersections of these VSR orbits give rise to the families
and
whose member orbits possess the O
O
plane symmetry. Furthermore, the first two families fall on planar periodic orbits and eventually cease to exist in three-dimensions while the rest one is led to spatial collision orbits. Besides, families
and
emerge from the VSR orbits
and
respectively, and they are constituted by orbits of the double symmetry O
O
. The first family terminate on the plane while the rest two go to collision orbits. Finally, the same VSR orbits give also rise to the branches
and
respectively, where the first two end on the plane while the third one goes, in its evolution, to collision orbits. The member orbits of the latter families are O
O
x doubly symmetric.
The terminations of the spatial families which were found to end on the physical plane O
are shown in
Figure 6 with the cyan, green and magenta coloured dots. In this figure, we have plotted the basic families of planar symmetric periodic orbits along both the positive and negative direction of the flow, i.e., for
(black continuous lines) and
(blue dotted lines), respectively, providing thus a chart with all vertical intersections of the respective planar orbits with the O
x-axis. In particular, the green dots correspond to the terminations of families
and
where they end up with degenerated orbits around the primary body located at the origin. Furthermore, the magenta dots indicate the planar terminations of families
and
. Specifically, family
ends up on a planar family where its members are of multiplicity five (magenta dot near
),
terminates on another family with planar periodic orbits of multiplicity three (magenta dot near the characteristic curve of
f) while
and
end up also on planar families with members of multiplicity three (the magenta dots which are shown to be located on the characteristic curves of
and
). The latter four cases of planar families have not been identified in our study since we have only considered the basic families of the Hill’s problem. Also, the two cyan dots on the characteristic curve of family
f with
i.e., family
correspond to the termination orbits of families
and
. However, the first family bifurcates from a VSR orbit of the family
of retrograde satellites with
in relation (
12) while the second one emanates from a VSR orbit with
, two cases which have not also been considered in our study. The remaining five cyan dots located on the characteristic curves of
and
represent the terminations of families
and
which return on VSR orbits of these planar families (existing in
Table 1) or on VSR orbits which are images of the latter ones under reflection in the origin and have the same value of the Jacobi constant
. In particular, families
and
end up on the two vertical intersections of the symmetrical VSR orbit of
. Also,
terminates on the positive crossing of the image planar orbit of
while
falls on the vertical intersection
of
. Finally,
falls on the vertical intersection
of
from which
originates, so these two families are essentially the same.
Table 2 incorporates initial conditions for sample members of the six computed spatial families whose orbits are axisymmetric and have been generated from the VSR orbits
and
. In particular, each entry involves the orbit’s half period
, the position and velocity components
which the orbit has initially on the O
x-axis as well as the value of the Jacobi constant. The last column indicates whether the family contains some stable parts
or not
. In
Table 3 we present sample members of the six families which bifurcate from the other vertical intersection of the same planar VSR orbits. Since their orbits are planar symmetric, we give now the position and velocity components
when the orbit starts perpendicularly from the O
plane. Finally, in the
Table 4 and
Table 5 we provide data for sample orbits of the families which the VSR orbits
and
generate from their two vertical intersections with the O
x-axis. The presented data correspond to the quarter of the orbits’ period
since these families consist of spatial orbits which are doubly symmetric.