Publications of the Astronomical Society of Australia (PASA), Vol. 34, e024, 12 pages (2017).

© Astronomical Society of Australia 2017; published by Cambridge University Press.

doi:10.1017/pasa.2017.17

Dynamical Monte Carlo Simulations of 3-D Galactic Systems in

Axisymmetric and Triaxial Potentials

Ali Taani1,3 and Juan C. Vallejo2

1 Applied Science Department, Aqaba University College, Al-Balqa Applied University, P.O. Box 1199, Aqaba, Jordan
2 European Space Astronomy Centre, PO Box 78, E-28691 Villanueva de la Canada, Madrid, Spain
3 Email: ali82taani@gmail.com

(R ECEIVED January 20, 2017; ACCEPTED April 1, 2017)

Abstract
We describe the dynamical behavior of isolated old (≥ 1Gyr) objects-like Neutron Stars (NSs). These objects are evolved
under smooth, time-independent, gravitational potentials, axisymmetric and with a triaxial dark halo. We analysed the
geometry of the dynamics and applied the Poincaré section for comparing the influence of different birth velocities. The
inspection of the maximal asymptotic Lyapunov (λ) exponent shows that dynamical behaviors of the selected orbits are
nearly the same as the regular orbits with 2-DOF, both in axisymmetric and triaxial when (φ, qz )= (0,0). Conversely, a
few chaotic trajectories are found with a rotated triaxial halo when (φ, qz )= (90, 1.5). The tube orbits preserve direction of
their circulation around either the long or short axis as appeared in the triaxial potential, even when every initial condition
leads to different orientations. The Poincaré section shows that there are 2-D invariant tori and invariant curves (islands)
around stable periodic orbits that bound to the surface of 3-D tori. The regularity of several prototypical orbits offer the
means to identify the phase-space regions with localized motions and to determine their environment in different models,
because they can occupy significant parts of phase-space depending on the potential. This is of particular importance in
Galactic Dynamics.
Keywords: Neutron Stars: general – galaxies: galactic potentials – galaxy: disk – galaxies: kinematics and dynamics –
Hamiltonian systems – ordered and chaotic motion – stars: statistics

1 INTRODUCTION Treves et al. 2000; Pavlov, Sanwal, & Teter 2004; De Luca
2008; Halpern & Gotthelf 2010) . As a consequence, little
Neutron stars (NSs) manifest the most extreme values of is known about their physical and statistical properties. One
many stellar and physical parameters including spin pe- would hope that the isolated old NSs may be detected as
riod, orbital parameters, magnetic field, and kick velocity. soft X-ray sources (0.5–2 keV) in the eRosita all-sky sur-
These are typically old systems (1–10 Gyr) and relatively vey (Merloni et al. 2012; Doroshenko et al. 2014). However,
short timescales (∼ 106 yr) (e.g. Yakovlev & Pethick 2004; the estimation of the pulsar velocities depends on the direct
Lorimer 2008) and show a concentration towards the Galactic distance measurements (Hobbs et al. 2005) which can be
centre bulge (which is at ∼8 kpc) and in the globular clus- obtained by the dispersion measure and a Galactic density
ters (Caranicolas & Zotos 2009; Taani et al. 2012a; Kalamkar model (Paczyński 1990; Hansen & Phinney 1997; Cordes &
2013; Taani 2016). Chernoff 1998; Lorimer 2008; Sartore et al. 2010). These
It is generally believed that Millisecond Pulsar (MSPs) studies give a mean birth velocity 100–500 km s−1 , with pos-
are very old NSs spun up owing to mass accretion during the sibly a significant population having v ≥ 1 000 km s−1 . Ar-
phase of mass exchange in binaries (Alpar et al. 1982). These zoumanian, Chernoff, and Cordes (2002) favour a bimodal
systems are detectable as active radio pulsars only when they pulsar velocity distribution, with peaks around 100 and 500
recycled (spun up). While the isolated old NSs have not been km s−1 . However, the mechanisms of high velocity are still
identified (Haberl 2007; Kaplan 2008), and we define them open questions (Wang & Han 2010).
here by their steady flux, predominantly thermal X-ray emis- Isolated old NSs have attracted much attention because
sion, lack of optical or radio counterparts, and the absence of the hope that their properties could be used to constrain
of a surrounding pulsar wind nebula (e.g. Ostriker, Rees, & the poorly understood behaviour. Studying their orbital dy-
Silk 1970; Shvartsman 1971; Neuhäuser & Trümper 1999; namics in known gravitational potential is very significant to
1

Downloaded from https:/www.cambridge.org/core. Cambridge University Press, on 20 Jun 2017 at 15:23:44, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms
. https://doi.org/10.1017/pasa.2017.17
 2 Taani and Vallejo

our understanding the Galactic gravitational field, as well as 2 SIMULATION AND NUMERICAL SETUP
the evolution of the Galactic disk structure itself. Paczyński
(1990) (hereafter P90) simulated the motion of NSs in a galac- 2.1. Galactic gravitational potential
tic potential and calculated the NS space density distribution.
The equations presented in this paper describe how could
In the same work, P90 also suggested a simplified expression
the motion and position of NSs be affected by the Galactic
for the gravitational potential which is still often applied in
gravitational potential. We will follow papers I and II in their
the simulation of NS distribution.
procedures to introduce the P90 Galactic gravitational po-
This paper is the third in a series of papers. Wei et al.
tential. This model is time independent and very simple. The
(2010a, hereafter Paper I) used this potential, and they also
integration of the orbit, using this model, is very rapid and
adopted a Galactic distribution with one-component initial
can achieve high numerical precision.
random velocity models. Wei et al. (2010b) investigated the
It is noteworthy to mention here that the P90 model is
above gravitational potential of the Galactic disk and several
taken to be a homogeneous function of the density, and ig-
prototypical orbits of stars, and found that all of the orbits
nore the interstellar friction. This is a reliable approxima-
are symmetric with respect to the galactic plane. Taani et al.
tion for our axisymmetric model, because the steady state
(2012b, hereafter Paper II) constructed a phenomenological
distribution of old NSs depends only weakly on the non-
model with the same gravitational potential, but under a two-
homogeneous part of the galactic potential (Frei, Huang, &
component Maxwellian initial random velocity distribution
Paczyński 1992). Using P90 may not be a good approxima-
following P90 and Faucher-Giguère & Kaspi (2006). The
tion when studying non-axisymmetric models, because ro-
conclusion was that there are some non-symmetric orbits.
tating non-axisymmetric components (like bars or spirals)
When the motion ranges in the vertical direction and hence
can introduce resonances (Patsis et al. 2002). This model
becomes larger than the one in the radial direction, the or-
is time independent and very simple. The integration of the
bits become more regular. It was also found that the irregular
orbit, using this model, is very rapid and can achieve high
character of the motion of NSs increases when the vertical
numerical precision. This model combines three axisymmet-
direction becomes larger than radial direction. We mean here
ric potential-density forms to produce a model of the mat-
when the motion is out of the disk plane. The large Galac-
ter distribution and its gravitational potential of the Milky
tic radial expansion (understood as R ∼ 15 Kpc) could give
Way. The basic components of the Milky Way are the visi-
hints to the distribution of NS progenitors. The majority of
ble disk and spheroid, and the invisible dark matter halo. For
them (80%) falls within R ≤25 kpc from the Galactic rotation
the bulge (spheroid), we adopt a Plummer sphere (Plummer
axis.
1911) in order to increase the central mass of the galaxy. For
In the present paper, we extend and complete the study of
the disk, we adopt Miyamoto–Nagai potential (Miyamoto &
the old NS sample by describing the dynamical orbital evolu-
Nagai 1975) which is added in order to reproduce the scale
tion of the isolated old NSs. These isolated NSs are evolved
length of the disk corresponding to n = 0. However, Evans
under a smooth, time-independent, 3-D axisymmetric gravi-
& Bowden (2014) and Evans & Williams (2014) presented
tational potential that represents a non-rotated galactic disk.
new analytical families of axisymmetric dark matter haloes,
In addition, we extend the purpose of the present study to
based on interesting modifications of the Miyamoto–Nagai
explore the orbital dynamics of realistic triaxial potential in
potential, called Miyamoto–Nagai sequence. And for the dark
a systematic way. We focus on plotting the 3-D trajectories
matter halo, we adopt a logarithmic (modified sphere) poten-
and their 2-D projections under a variety of initial conditions.
tial, which produces a flat rotation curve at large radius, since
These initial conditions are obtained by performing Monte
for strongly flattened systems it is more natural to work in
Carlo simulations to develop perturbation approximations of
cylindrical coordinates R, z, φ rather than spherical r, θ , φ
the 3-D orbits.
(Binney, Merrifield, & Shu 1988; Flynn, Sommer-Larsen, &
Our main objective is to investigate the regular or chaotic
Christensen 1996).
nature of the computed trajectories. It is noteworthy to men-
tion that Evans (1994) found some semi-stochastic orbits in  = sph + disk + halo (1)
triaxial potentials, because the chaos was tentatively associ- where sph , halo , and disk define the spheroid, halo, and
ated with linear instability of the short- and intermediate- disk components, respectively.
axis orbits, while Goodman & Schwarzschild (1981) dis- The first two components are described by the same law.
cussed the importance of stochastic orbits in triaxial For the disk component, we follow a Miyamoto–Nagai po-
models. tential:
The structure of the paper is as follows. First, we in-
GMdisk
troduce the model and the Monte Carlo technique we disk (R, z) = −  (2)
have used. Then, we will analyse the dynamical proper- R2 + [A + (z2 + B2 )1/2 ]2
ties of a set of selected initial conditions, including the where it depends on two variables, namely R, which denotes
Poincaré sections, Lyapunov Asymptotic Exponents, and the radial distance perpendicular to the Galactic central axis,
dark triaxial haloes case. Finally, we will end with some R2 = x 2 + y2 and z, which denotes the vertical distance from
conclusions. the Galactic plane. The parameter A is a measure of the radial
PASA, 34, e024 (2017)
doi:10.1017/pasa.2017.17

Downloaded from https:/www.cambridge.org/core. Cambridge University Press, on 20 Jun 2017 at 15:23:44, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms
. https://doi.org/10.1017/pasa.2017.17
 Dynamical Monte Carlo Simulations of 3-D Galactic Systems in Axisymmetric and Triaxial Potentials 3

scale length of the disc while the parameter B is a measure of

the disc thickness in the z direction. Galactic discs are much
larger in the radial than in the vertical directions, thus the
values A will be greater than those of the B parameter in our
models. The corresponding values for the disk component are
A = 3.7 kpc, B = 0.20 kpc, and Mdisk = 8.01 × 1010 M .
Using Poisson’s equation, ∇ 2 φ = 4π ρG, we can obtain
the expression for the disc density as

  √ √
B2 M AR2 + (A + 3 z2 + B2 )(A + z2 + B2 )2
ρ(R, z) = √ √
4π [R2 + (A + z2 + B2 )2 ]5/2 ( z2 + B2 )3/2
(3)
Note that when A = 0, the potential reduces to a spherical
potential on the galactic plane (z = 0). The spheroid compo-
nent of the Galactic gravitational potential is similar to the
above:

GMsph
sph (R, z) = −  (4) Figure 1. Rotation curve of the axisymmetric potential includes a
R2 + [A + (z2 + B2 )1/2 ]2 Miyamoto–Nagai disk and spheriod, triaxial halo, and averaged circular
velocity in the galactic plane.

The spheroid component, A = 0.0 kpc, B = 0.28 kpc,

and Msph = 1.12 × 1010 M . Here, we must point out, that
this potential is not intended to represent a black hole nor 2.2. NS initial velocities
any other compact object, but a dense and massive bulge.
The initial velocities of the NSs are calculated as the vec-
Therefore, we do not include any relativistic effects (Jung &
tor addition of three different velocities: (1) a Maxwellian
Zotos 2015).
distribution, (2) a constant kick, and (3) the circular rotation
Regarding the halo component of the Galactic gravitational
velocities at the birth place. The details on their calculation
potential, is given by the following equation:
can be found in Arzoumanian et al. (2002) and Hobbs et al.
   (2005) and are summarised here.
GMhalo 1 R2 + z 2 Concerning the density distribution of the total number of
halo = ln 1 +
rc 2 rc2 NSs which we used in this simulation, it is about ∼1 × 107 ,
√  and it distributes uniformly in the age ≥1 Gyr. This is equiva-
rc R2 + z 2
+ √ arctan (5) lent to a simulating birthrate of one NS per century (Kulkarni
R2 + z 2 rc
& van Kerkwijk 1998; Lyne & smith 2007). However, the
precise estimates of both the number and lifetime of the NS
where rhalo = 12 kpc and Mhalo = 5.0 × 1010 M . population are hard to obtain especially at larger distances
While in a spherically symmetric potential, the x, y, and z (see e.g. Keane & Kramer 2008; Ofek 2009) because they
components of the angular momentum and the total angular may have been heavily biased by a number of observational
momentum are conserved, out potential only admits two con- selection effects (Lorimer 2008).
served quantities, the total energy E , and the z-component of Regarding the kicks simulation, the kick velocity imparted
the angular momentum vector. Therefore, the motion of the to an NS at birth is one of the major problems in the the-
objects are in the plane and perpendicular to the vector of the ory of stellar evolution. However, the physical mechanism
total angular momentum. We assume that the rotation curve that causes this kick is presently unknown and a number
is composed of a spheriod, disk, averaged circular velocity, of physical models have been described and evaluated (see
and triaxial halo components in the Galaxy (see Figure 1), e.g. Kalogera 1996; Wang & Han 2010; Meng, Chen, &
and is calculated at any radius by v c2 (r) = r d
dr
. It is clearly Han 2009). But it is presumably the result of some asym-
seen from the same figure that each contribution prevails in metry in the core collapse or subsequent SNe explosion (see
different distances form the galacto-centric R. In particular, e.g. Pfahl, Rappaport, & Podsiadlowski, 2002; Podsiadlowski
at small distances when R ≥ 3 kpc, the contribution from the et al. 2004). However, the distribution of the kick amplitudes
spheriod dominates, while at distances of 3 < R < 19 kpc the is usually obtained from the analysis of radio pulsar proper
disk contribution is the dominant factor. On the other hand, at motions (Hobbs et al. 2005).
large enough galacto-centric distances, R > 19 kpc, we see The short-lived sudden kick in the phase space has a kick
that the contribution from the triaxial halo prevails, thus forc- direction uncorrelated with the orientation of the plane and
ing the rotation curve to remain at with increasing distance the system velocity that the object acquires because the ex-
from the centre (Zotos 2014). plosion is also uncorrelated with the rotation in the Galaxy.
PASA, 34, e024 (2017)
doi:10.1017/pasa.2017.17

Downloaded from https:/www.cambridge.org/core. Cambridge University Press, on 20 Jun 2017 at 15:23:44, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms
. https://doi.org/10.1017/pasa.2017.17
 4 Taani and Vallejo

Table 1. Selected representative orbits, θ and  in degrees, |v| in km s−1 .

Orbit label Initial position (x0 , y0 , x0 ) Initial position (r0 , θ0 , 0 ) Initial velocity (v x0 , v y0 , v z0 ) |v|

O1 (−1.11, −9.2, 0.008) (9.267, 0.049, −96.87) (172.4, −53.0, −121.2) 217.302
O2 (−2.14, −4.0, 0.1) (4.538, 1.26, −118.14) (15.8, −161.8, −137.4) 212.856
O3 (6.88, −6.78, 0.2) (9.661, 1.18, −44.58) (48.5, −44.7, −123.7) 140.185

Therefore, we follow Hansen & Phinney (1997) and find that velocity vector with a modulus fulfilling a Maxwellian dis-
the distribution is consistent with Maxwellian distribution tribution with σv = 265 km s−1 (Hobbs et al. 2005; Story
of kick velocity. Finally, regarding the third element used in & Gonthier 2007) and σv =190 km s−1 (Hansen & Phinney
the calculation of the initial condition, we choose the initial 1997). The direction is given by selecting randomly a di-
circular rotation velocity of the NS before a kick. A plot of rection specified by choosing two angles, 0 <  < 360 and
averaged circular velocity for every parts of our galactic po- −90 < θ < 90 (Kiel & Hurly 2009).
tential model presented is in Figure 1. It rises linearly up to The equations above were solved numerically in three di-
a maximum value and then they become constant for larger rections using the -order Runge–Kutta method with adaptive
radii. control step sizes (Press et al. 1992). The initial step size is
A distinct approach to the analysis of location and veloc- dt = 10−4 Myr, and calculations of position and velocity in 3-
ity of old NSs is based on the Monte Carlo simulation of D then continue until 0.1 Myr with adjusted step sizes accord-
the evolution of a simulated sample with different initial pa- ing to the required accuracy. The data are recorded every 0.1
rameters, and then their orbits are numerically integrated. We Myr and the total energy Etot = 21 (v x2 + v y2 + v z2 ) + (R, z)
would utilise the method of the Poincaré sections to explore is checked. The position and velocity then are used for input
their motions. This has been studied in detail in Papers I and of the next 0.1 Myr. The procedure used here enabled us to
II). Once we have selected the above, we can chose a set control and achieve required levels of accuracy by using the
of initial conditions. We will consider isolated NSs (exclud- energy integral.
ing the MSPs and GCs), being of ages older than 109 yrs The Monte Carlo simulations were run and the results are
and with large radial expansions, 10 > R > 25 Kpc. The ra- given here. One hundred initial conditions were obtained.
dial distribution has an exponential scale length of λe−λ|z| , Different families are identified and we have selected O1, O2,
where λ = 1/0.07˜ kpc−1 . Here, we assumed a maximum O3. These orbits are listed in Table 1. A periodic orbit is found
birth height off the plane, zmax of 150 pc. when the initial and final coordinates coincide with an accu-
racy at least 10−15 . This is a representative set because they
show the typical behaviour of the whole distribution, with
3 NUMERICAL RESULTS very large Galactic radial expansion and they are rosette-like
type orbits. In addition, the orbits lie essentially in the plane of
3.1. Selection of initial conditions the long and intermediate axes and so reinforces the shape of
We deal in this section with the analysis of the dynamical the potential to some extent. The most general trajectory for
evolution of a representative set of NSs’ Orbits in the Galaxy an object in this potential, which allowed us to clarify prop-
calculated following the previously described distribution of erties on the structure and configuration of the invariant tori
velocities. We aim to obtain a set of different orbits repre- that we encounter in the vicinity of the periodic orbits. The
senting NSs at given distances from the axis. Our method exact initial conditions for the periodic orbit are calculated
follows Arzoumanian (2002). We start by choosing a ran- and listed in Table 1. The trajectories and 2-D projections
domly selected initial position, with boundaries located at corresponding to the integration of every initial condition up
the Galacto-centric distance in the r ≤ 25 kpc. to a timescale of 108 yr are presented in Figures 2–5.
The initial velocity components of the NSs are also ob- As we can see from Figure 2, the different initial heights
tained from a random sample, following the vector addition of have an influence on z-direction due to the perturbations and
the three different velocities described in the previous section: will naturally cause different trajectories, while if the objects
the Maxwellian distribution, the constant kick, and the circu- are located at different R, trajectories can be less different
lar motion velocities at the birth place. The direction of the and may prone to instability through the gravitational pertur-
initial velocity vector is chosen randomly within the geomet- bation.
ric shape of the potential. We start from a vector based on the It is noteworthy to mention here that if the sun received
circular velocity. Then we add a random kick, in the form of a kick velocity (50 km s−1 ) in the Galactic plane (Repetto,
a local perturbation, with modulus following the Maxwellian Davies, & Sigurdsson 2012), the typical orbit (considered as
distribution (Hansen & Phinney 1997) and direction to the a circular in the Galactic potential) would turn into a rosette
radial direction. Finally, we add the third vector. This is a orbit.
PASA, 34, e024 (2017)
doi:10.1017/pasa.2017.17

Downloaded from https:/www.cambridge.org/core. Cambridge University Press, on 20 Jun 2017 at 15:23:44, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms
. https://doi.org/10.1017/pasa.2017.17
 Dynamical Monte Carlo Simulations of 3-D Galactic Systems in Axisymmetric and Triaxial Potentials 5

Figure 2. Trajectories corresponding to the different initial conditions of Table 2. These conditions are given in
enlarged scale, in order to better view the corresponding morphology. The upper-left panel corresponds to orbit O1.
The upper-right panel corresponds to orbit O2. The trajectory is recognised as a precessing banana orbit (see Evans &
Bowden 2014). The bottom-left panel corresponds to orbit O3.

3.2. Dynamical analysis quasi-periodic orbits that bound to the surface of 3-D tori in a
3-D Hamiltonian system. These quasi-periodic orbits are rep-
3.2.1. Poincaré sections
resented in the surface of section by 2-D tori (Manos, Skokos,
In Poincaré sections, periodic orbits appear in the surface of & Antonopoulos 2012).
section as a finite set of points. The amount of points depends Aiming to gain a better picture, Figure 4 shows the 2-D
from the multiplicity of the periodic orbit. This method has projection in x–z plane, meanwhile Figure 5 shows the 2-D
been extensively applied to 2-D Hamiltonians in a 2-D plane projection in x − v x . In the first figure, the intersection points
(Lichtenberg & Lieberman 1992; Wei et al. 2010a; Taani et al. distribute in some regular lines on the Poincaré hyper-surface
2012b). While in 3-D systems, we have to project the Poincaré section, while the x − v x projections are tend to have enclosed
surface to spaces with lower dimensions (Contopoulos 2004). curves. This behaviour could indicate a regular motion and
By setting y, v y0 equal to zero at t = 0 (remaining zero at all a quasi-periodic orbit. All the points along the orbits form
times) in the Hamiltonian equation, the 3-D Poincaré sections ensembles of invariant subset of phase space. The ensemble of
can be found in Figure 3. The points in these figures show the old NSs along given orbits for invariant blocks of the galaxy,
3-D section. We generate 2 500 consequents of the orbits in since the galaxies are seen as built not of stars, but of orbits
3-D (x, vx , z). We find that there are sometimes several or- (Guarinos 1992).
bits from the same family plotted very close to each other.
This happens when the sequence of orbits within the family
reverses its progression in the x–z plane towards a given di- 3.2.2. Lyapunov asymptotic exponents
rection and very close to the current space position, causing We can conclude from the inspection of these figures all orbits
an accumulation of orbits near this position with different seem to be regular orbits, because they are represented by a
local velocities. The stable periodic orbits are surrounded by tori in the surface of sections (Katsanikas & Patsis 2011).
PASA, 34, e024 (2017)
doi:10.1017/pasa.2017.17

Downloaded from https:/www.cambridge.org/core. Cambridge University Press, on 20 Jun 2017 at 15:23:44, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms
. https://doi.org/10.1017/pasa.2017.17
 6 Taani and Vallejo

Figure 3. The 3-D Poincaré section x > 0, at y = 0 of the 3-D trajectories corresponding to the selected initial conditions. The upper-left
panel corresponds to orbit O1. The upper-right panel corresponds to orbit O2. The bottom left panel corresponds to orbit O3.

In order to crosscheck the results, we have computed the The selection of the initial deviation vectors is of impor-
Lyapunov asymptotic exponent for the selected orbits. The tance when computing the Lyapunov exponents during finite
ordinary, or asymptotic Lyapunov exponent can be defined integrations, leading to the so-called finite-time Lyapunov
as exponents (Vallejo, Aguirre, & Sanjuan 2003). But when
1 one uses long enough integration times, the axes evolve to-
λ(x, v) = lim ln
Dφ(x, t )v
(6)
t→∞ t wards the fastest growing direction and the computation of the
provided this limit exists (Ott & Yorke 2008). Here, φ(x, t ) growth rates return the asymptotic Lyapunov values (Vallejo,
denotes the solution of the flow equation, such that φ(x0 , 0) = Viana, & Sanjuan 2008).
x0 , and D is the spatial derivative in the direction of an in- We have used integration of up T = 105 time-units and the
finitesimal displacement v. resulting maximal Lyapunov values have been nearly zero in
A system trajectory is chaotic if it shows at least one posi- all cases, confirming the analysis done using the Poincaré
tive Lyapunov exponent, the movement is confined within section methods, it is seen that all orbits retain their regular
certain limited region, and the ω-limit set is not periodic characteristics along the stellar evolution, the regular orbit
neither composed of equilibrium points (Alligood, Sauer, & bound to the surface of 3-D torus (Kovár et al. 2013) in the
Yorke 1996). Conversely, if the maximum asymptotic Lya- phase space forms a narrow curve with zero width. This may
punov exponent is zero, this reflects the existence of a regu- be due to a different kick velocities due to SNe mass-loss
lar motion (that is, a quasi-periodic orbit). Finally, a negative and natal kicks to the newly formed NS (Podsiadlowski et al.
value will reflect the existence of one attractor, but this is not 2004). By examining the Poincaré section in each case, we
possible in a conservative system like the Hamiltonian we are found that the system looks integrable and its trajectories lie
analysing. mostly on tori. These tori are represented by 2-D tori in the
The Lyapunov exponents are computed by calculating the surface of section. This is known the KAM (Kolmogorov–
growth rate of the orthogonal semi-axes (equivalent to the Arnold–Moser) theorem (Kolmogorov 1954; Moser 1962;
initial deviation vectors) of one ellipse centred at the initial Arnold 1963).
position as the system evolves. By solving at the same time,
the flow equation and the fundamental equation of the flow
3.3. Orbits in the dark triaxial halo case
(that is, the distortion tensor evolution), we can follow the
evolution of the vectors, or axes, along the trajectory, and in To investigate the possible effects of spiral structure on the
turn, their growth rate. This method is described in Benettin orbital characteristics of trajectory and projections, we adopt
et al. (1980). the triaxial model of Law, Majewski, & Johnston (2009), that
PASA, 34, e024 (2017)
doi:10.1017/pasa.2017.17

Downloaded from https:/www.cambridge.org/core. Cambridge University Press, on 20 Jun 2017 at 15:23:44, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms
. https://doi.org/10.1017/pasa.2017.17
 Dynamical Monte Carlo Simulations of 3-D Galactic Systems in Axisymmetric and Triaxial Potentials 7

Figure 4. The projection on x–z plane of the 3-D Poincaré section. The upper left panel corresponds to orbit O1. The upper right panel
corresponds to orbit O2. The bottom left panel corresponds to orbit O3.

uses a triaxial halo as follows: role in the dynamics of the galaxies is presented in Vallejo &
z Sanjaun 2015 and references there.)
halo = v halo
2
ln(C1 x + C2 y + C3 xy + ( )2 + rhalo
2 2 2
) (7)
qz This potential is triaxial, rotated, and more realistic. As
where vhalo = 128 km s−1 , qz represents the flattening per- consequent, it reproduces the flat rotation curve for a Milky
pendicular to the Galactic plane. The various constants C1 , Way-type galaxy and it can be easily shaped to the axial ra-
C2 , and C3 are given by tios of the ellipsoidal isopotential surfaces (see Vallejo &
  Sanjuan 2015 for details). We include computations of the
cos2 φ sin2 φ three orbits correspondence to O1, O2, and O3 which can be
C1 = 2
+ (8)
q1 q22 found in realistic galactic-type potentials that incorporates
  spiral arms (see Figures 6 and 7). The main parameters to
cos2 φ sin2 φ
C2 = + (9) play with are the flattening (qz ) and the orientation (φ) of the
q22 q12
  dark halo, since the efficiency of bar formation depends very
1 1 strongly on the initial orientation of the galaxy disk (Lokas
C1 = 2 sin φ cos φ − (10)
q12 q22 et al. 2015). A lot of work on periodic orbits in the plane
The control parameters of this model are the orientation of of a rotating barred galaxy was carried out by Contopou-
the major axis of the triaxial halo φ and its flattening qz . (A los & Papayannopoulos (1980), Papayannopoulos & Petrou
detailed discussion on the effects of dark haloes and their (1983), Mulder & Hooimeyer (1984), Harsoula et al. (2011a,
PASA, 34, e024 (2017)
doi:10.1017/pasa.2017.17

Downloaded from https:/www.cambridge.org/core. Cambridge University Press, on 20 Jun 2017 at 15:23:44, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms
. https://doi.org/10.1017/pasa.2017.17
 8 Taani and Vallejo

Figure 5. The projection on x–v x plane of the 3-D Poincaré section. The upper-left panel corresponds to orbit O1. The upper-right panel
corresponds to orbit O2. The bottom left panel corresponds to orbit O3.

2011b), and Zotos (2014). All integrable triaxial potentials Table 2. Maximum asymptotic Lyapunov exponents
have a similar orbital structure (e.g. de Zeeuw 1985; Val- for the selected representative orbits, two different ori-
luri & Merritt 1998; Skokos 2001; Contopoulos 2004; Patsis entations of dark halo.
et al. 2009; Zotos & Carpintero 2013; Patsis et al. 2014).
Orbit label λ(φ = 0) λ(φ = 90)
The plots we delivered were done in spherical coordinates
by two different values of the dark halo orientation (φ = 0 O1 0.00015160387 0.00015288824
and φ = 90). Figures 6 and 7 (φ = 0, qz = 1.25, and λ rel- O2 4.5184949 4.3155085
O3 1.9746514 2.0018652
atively small value and close to zero for O1) show regular
orbits (tube orbits with short and long axes) for oblate and
prolate cases, respectively. The effect of spiral structure is
presented for all orbits and seems to be chaotic regions on
the Poincaré sections (φ = 90, qz = 1.5, and λ ≥ 1 for O2 (stationary) q1 is then aligned at stable equilibrium point with
and O3 in Figures 6 and 7). Thus, NSs in this potential follow the Galactic x-axis, and the motion is stable along its axis. The
harmonic and rotating motion in each of the x, y, z directions results in this case are comparable with non-triaxial, purely
independently. These control parameter values are given in logarithmic potentials. When φ = 90, q1 (see Figures 6 and
Table 2 for all orbits. In Figures 6 and 7 (O1), when φ = 0 7 for O2 and O3) is then aligned with the Galactic y-axis and
PASA, 34, e024 (2017)
doi:10.1017/pasa.2017.17

Downloaded from https:/www.cambridge.org/core. Cambridge University Press, on 20 Jun 2017 at 15:23:44, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms
. https://doi.org/10.1017/pasa.2017.17
 Dynamical Monte Carlo Simulations of 3-D Galactic Systems in Axisymmetric and Triaxial Potentials 9

Figure 6. Physical trajectories and the corresponding Poincaré sections y–v y , with plane x = 0 and v x > 0, for the selected representative
orbits, for a dark halo orientation of φ = 0.

it takes the role of q2 . We note that tube tori appear in the in Figure 6 is wider in the x direction than Figure 7 when
3-D projections of the spaces of section as soon as a pertur- seen in projection. It is noteworthy to mention that Figure 6
bation is introduced, even if it is a small one (Katsanikas & is symmetric in the vy –y plane, while Figure 7 is not due to
Patsis 2011). At (φ, qz ) = (0,1.25) the outer long-axis tube the Coriolis force (Gajda, Lokas, & Athanassoula 2016).

PASA, 34, e024 (2017)

doi:10.1017/pasa.2017.17

Downloaded from https:/www.cambridge.org/core. Cambridge University Press, on 20 Jun 2017 at 15:23:44, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms
. https://doi.org/10.1017/pasa.2017.17
 10 Taani and Vallejo

Figure 7. Physical trajectories and the corresponding Poincaré sections y–v y , with plane x = 0 and v x > 0, for the selected representative
orbits, for a dark halo orientation of φ = 90.

4 CONCLUSIONS itational potential models of normal non-barred galaxies. The

models consist of axisymmetric and a triaxial dark halo po-
We have found many interesting aspects for simulation of tentials with three components: bulge, disc, and dark halo.
several prototypical orbits of isolated old NSs and their con- These are relatively simple potentials that can show complex
nections with the spatial distribution phenomena, through the behaviours, which are found in more realistic galactic-type
long-term stellar dynamics. We employed 3-D Galactic grav- potentials. Then we found that orbits can rotate around the
PASA, 34, e024 (2017)
doi:10.1017/pasa.2017.17

Downloaded from https:/www.cambridge.org/core. Cambridge University Press, on 20 Jun 2017 at 15:23:44, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms
. https://doi.org/10.1017/pasa.2017.17
 Dynamical Monte Carlo Simulations of 3-D Galactic Systems in Axisymmetric and Triaxial Potentials 11

axis of symmetry of the phase space on a surface of section, Arzoumanian, Z., Chernoff, D. F., & Cordes, J. M. 2002, ApJ, 568,
in the same direction in 3-D and 2-D loops for various val- 289
ues of the initial conditions for the family of axisymmetric Benettin, G., et al. 1980, Mecc, 15, 9
and triaxial models. We have used the Poincaré technique to Binney, J., Merrifield, S., & Shu, F. 1988, Nature, 333, 219
study the 3-D NS trajectories and their 2-D projections. It Caldwell, J., & Ostriker, J. 1981, ApJ, 251, 61
Caranicolas, N. D., & Zotos, E. E. 2009, BaltA, 18, 205
is shown that both loop and family of orbits arise as a natu-
Caranicolas, N. D., & Zotos, E. E. 2013, PASA, 30, 49
ral consequence of the dynamics of the stable periodic orbits Chen, B., et al. 2001, ApJ, 553, 184
(with regular motion). In addition, the morphology of orbits Contopoulos, G., & Papayannopoulos, T. 1980, A&A, 92, 33
in triaxial potentials can determine the structure of triaxial Contopoulos, G. 2004, Order and Chaos in Dynamical Astronomy
galaxy itself. There are 3-D invariant tori containing quasi- (New York: Springer)
periodic motions, these tori are represented by 2-D tori on Cordes, J. M., & Chernoff, D. F. 1998, ApJ, 505, 315
the surface of the section associated with phase space. In ad- de Zeeuw, T. 1985, MNRAS, 216, 273
dition, the Lyapunov asymptotic exponent is equal to zero Doroshenko, V., Ducci, L., Santangelo, A., & Sasaki, M. 2014,
in case of non-triaxial. However, it is clear that the chaotic A&A, 567, 7
orbit have a non-zero real exponent, since the finite-time Lya- Evans, N. W. 1994, MNRAS, 267, 333
punov exponents distributions reflect the underlying dynam- Evans, N. W., & Bowden, A. 2014, MNRAS, 443, 2
Evans, N. W., & Williams, A. A. 2014, MNRAS, 443, 791
ics (Vallejo et al. 2003). We also show that the phase space
Faucher-Giguère, C.-A., & Kaspi, V. M. 2006, ApJ, 643, 332
in triaxial galaxies with a rotated halo is rich in regular and Flynn, C., Sommer-Larsen, J., & Christensen, P. R. 1996, MNRAS,
chaotic regions, which is consistent with the analysis done by 281, 1027
the Poincaré cross-sections. The results of these motions are Frei, Z., Huang, X., & Paczyński, B. 1992 ApJ, 643, 332
strongly affected by the gravitational potential of the galactic Fryer, C., Benz, W., Herant, M., & Stirling, C. A. 1999, ApJ, 516,
disk associated with the effect of kick velocities on the orbital 892
parameters. This can provide us a better visualisation of the Haberl, F. 2007, Ap&SS, 308, 181
old NS dynamics. Halpern, J. P., & Gotthelf, E. V. 2010, ApJ, 709, 436
The conclusions of the present research are considered as Hansen, B. M. S., & Phinney, E. S. 1997, MNRAS, 291, 569
an initial effort and also as a promising step in the task of Harsoula, M., Kalapotharakos, C., & Contopoulos, G. 2011a, IJBC,
understanding the underlying dynamics of the phase space by 21, 2221
Harsoula, M., Kalapotharakos, C., & Contopoulos, G. 2011b, MN-
mapping the regular regimes in 3-D axisymmetric potential.
RAS, 2411, 1111
Future work will go steps further in detection of the old NSs
Hobbs, G., Lorimer, D. R., Lyne, A. G., & Kramer, M. 2005, MN-
via accretion of the interstellar medium material which may RAS, 360, 974
make them shine, and their weak luminosity could be detected Jung, Ch., & Zotos, E. E. 2015, MeReC, 69, 45
as soft X-ray sources (0.5–2 keV). However, eROSITA all-sky Gajda, G., Lokas, E., & Athanassoula, E. 2016, ApJ, 830, 108
survey have an excellent capabilities of available survey for Goodman, J., & Schwarzschild, M. 1981, ApJ, 245, 1087
this kind of studies, and it should improve our knowledge of Guarinos, J. 1992, Evolution of Interstellar Matter and Dynamics
Galactic NSs phenomenally for the next decade. eROSITA of Galaxies (Cambridge: Cambridge University Press)
will be about 20 times more sensitive than the ROSAT all sky Kalamkar, M. N. 2013, PhD dissertation, University of Amsterdam
survey in the soft X-ray band (Merloni et al. 2012). Kalogera, V. 1996, ApJ, 471, 352
Kaplan, D. L. 2008, in AIP Conf. Ser., Vol. 983, 40 Years of Pul-
sars: Millisecond Pulsars, Magnetars and More (College Park:
ACKNOWLEDGEMENTS American Institute of Physics), 331
Katsanikas, M., Patsis, P. A., & Contopoulos, G. 2013, IJBC, 23,
We specifically would like to acknowledge Ying-Chun Wei for his 1330005
assistance with the simulation code. We are very grateful to Haris Keane, E. F., & Kramer, M. 2008, MNRAS, 391, 2009
Skokos, Panos Patsis, Matthaios Katsanikas, Georgios Contopou- Kolmogorov, A. 1954, Dokl. Akad. Nuak. SSSR, 98, 527
los, Nicola Sartore, Matthew Molloy, Diomar Cesar Lobâo, Barbara Kovár, J., Kopácek, O., Karas, V., & Kojima, Y. 2013, QGra, 30,
Pichardo, and María de los Angeles Perez Villegas for fruitful dis- 025010
cussions and useful remarks. A. Taani would like to thank Hamid Kulkarni, S. R., & van Kerkwijk, M. H. 1998, ApJ, 507, L49
Al-Naimiy and Awni Al-Khasawneh for their support. The authors Law, D. R., Majewski, S. R., & Johnston, K. V. 2009, ApJ, 703, L67
would like to thank the anonymous referees for the careful read- Lichtenberg, A. J., & Lieberman, M. A. 1992, Regular and Chaotic
ing of the manuscript and for all suggestions and comments which Dynamics (New York: Springer-Verlag)
allowed us to improve both the quality and the clarity of the paper. Lokas, E. L., Semczuk, M., Gajda, G., & D’Onghia, E. 2015, ApJ,
810, 100
Lorimer, D. R. 2008, LRR, 11, 8 http://relativity.livingreviews.org/
REFERENCES
Articles/lrr-2008-8/
Alligood, K., Sauer, T. D., & Yorke, J. A. 1996, Chaos: An Intro- Lyne, A. G., & Graham-Smith, F. 2007, Pulsar Astronomy, Cam-
duction to Dynamical Systems (New York: Springer-Verlag) bridge Astrophysics Series (3rd edn.; Cambridge: Cambridge
Alpar, M. A., et al. 1982, Nature, 300, 728 University Press)
Arnold, V. I. 1963, RuMaS, 18, 5 Manos, T., & Athanassoula, E. 2011, MNRAS, 415, 629
PASA, 34, e024 (2017)
doi:10.1017/pasa.2017.17

Downloaded from https:/www.cambridge.org/core. Cambridge University Press, on 20 Jun 2017 at 15:23:44, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms
. https://doi.org/10.1017/pasa.2017.17
 12 Taani and Vallejo

Manos, T., & Machado, R. E. G. 2014, MNRAS, 438, 2201 Repetto, S. 2012, Master’s thesis, Lund Observatory
Manos, T., Skokos, Ch., & Antonopoulos, Ch. 2012, IJBC, 22, 9 Repetto, S., Davies, M., & Sigurdsson, S. 2012, MNRAS, 425, 2799
Meng, X., Chen, X., & Han, Z. 2009, MNRAS, 395, 2103 Popov, S. B. 2008, PPN, 39, 1136
Merloni, A., et al. 2012, eprint, arXiv:1209.3114 Sartore, N., Ripamonti, E., Treves, A., & Turolla, R. 2010, A&A,
Miyamoto, M., & Nagai, R. 1975, PASJ, 27, 533 510, A23
Moser, J. K. 1962, Nachr. Akad. Wiss. Götingen, Math. Phys. K1. Shvartsman, V. G. 1971, SvA, 14, 662
IIa, 1 Skokos, Ch. 2010, LNP, 790, 63
Mulder, W. A., & Hooimeyer, J. R. A. 1984, A&A, 134, 158 Skokos, Ch., Antonopoulos, Ch., & Bountis, T. C. 2004, JPhA, 37,
Neuhäuser, R., & Trümper, J. E. 1999, A&A, 343, 151 6269
Ofek, E. O. 2009, PASP, 121, 814 Sun, X. H., & Han, J. L. 2004, MNRAS, 350, 232
Ostriker, J. P., Rees, M. J., & Silk, J. 1970, ApJ, 6, 179 Story, S. A., & Gonthier, P. L. 2007, ApJ, 671, 713
Ott, W., & Yorke, J. A. 2008, PhRvE, 78, 056203 Taani, A. 2016, RAA, 16, 1
Paczyński, B. 1990, ApJ, 348, 485 Taani, A., et al. 2012a, AN, 333, 53
Papayannopoulos, T., & Petrou, M. 1983, A&A, 119, 21 Taani, A., et al. 2012b, Ap&SS, 341, 601
Patsis, P. A., Kaufmann, D. A., Gottesman, S. T., & Boonyasait, V. Tabor, M. 1989, Chaos and Integrability in Nonlinear Dynamics
2009, MNRAS, 394, 142 (New York: Wiley)
Patsis, P. A., & Kalapotharakos, C. 2011, MSAIS, 18, 83 Taylor, J. H., & Cordes, J. M. 1997, ApJ, 411, 674
Patsis, P. A., & Katsanikas, M. 2014, MNRAS, 445, 3525 Treves, A., Turolla, R., Zane, S., & Colpi, M. 2000, PASP, 112, 297
Patsis, P. A., et al. 2002, MNRAS, 355, 1049 Vallejo, J. C., Aguirre, J., & Sanjuan, M. A. F. 2003, PhLA, 311, 26
Pavlov, G. G., Sanwal, D., & Teter, M. A. 2004, in IAU Symp. 218, Vallejo, J. C., Viana, R., & Sanjuan, M. A. F. 2008, PhLE, 78, 66204
Young Neutron Stars and their Environments, ed. F. Camilo & Vallejo, J. C., & Sanjaun, M. A. F. 2015, MNRAS, 447, 3797
B. M. Gaensler (San Francisco: ASP), 239 Wang, B., & Han, Z. 2010, Ap&SS, 329, 293
Perets, H. B., et al. 2009, ApJ, 697, 2097 Wei, Y. C., et al. 2010a, ChPhL, 27, 9801
Pfahl, E., Rappaport, S., & Podsiadlowski, P. 2002, ApJ, 573, 283 Wei, Y. C., et al. 2010b, ScChA, 53, 1939
Podsiadlowski, Ph., et al. 2004, ApJ, 612, 1044 Yakovlev, D. G., & Pethick, C. J. 2004, ARA&A, 42, 169
Press, W. H., et al. 1992, Numerical Recipes (2nd edn.; Cambridge: Zotos, E. E. 2014, BaltA, 23, 37
Cambridge University Press) Zotos, E. E., & Carpintero, D. D. 2013, CeMDA, 116, 417

PASA, 34, e024 (2017)

doi:10.1017/pasa.2017.17

Downloaded from https:/www.cambridge.org/core. Cambridge University Press, on 20 Jun 2017 at 15:23:44, subject to the Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms
. https://doi.org/10.1017/pasa.2017.17