Astronomy & Astrophysics manuscript no.

aanda ©ESO 2025

June 5, 2025

Interlinking internal and external magnetic fields of relativistically

rotating neutron stars
D. Ntotsikas and K.N. Gourgouliatos

Universe Sciences Laboratory, Department of Physics, University of Patras, Patras, Rio, 26504, Greece.
e-mail: [email protected], [email protected]

June 5, 2025
arXiv:2506.04198v1 [astro-ph.HE] 4 Jun 2025

ABSTRACT

Context. While both the internal and external magnetic field of neutron stars have been studied thoroughly via sophisticated methods
incorporating relativistic effects at the exterior and a magnetohydrostatic equilibrium at the interior, or even more complex regimes,
the fundamental issue of linking the internal to the external field in a self-consistent way remains yet unresolved. To achieve a
realistic depiction of the magnetic field, both the internal and external configurations need to be addressed into a single, simultaneous
calculation.
Aims. Our aim is to solve for the structure of the magnetic field of a neutron star within the stellar interior, and the relativistically
rotating magnetosphere, adhering to barotropic equilibria in the interior and the relativistic force-free condition at the exterior.
Methods. We solve the axisymmetric pulsar equation for the magnetosphere and the associated equilibrium equations for the neutron
star’s interior by employing an elliptic solver using the method of simultaneous relaxation for the magnetic field inside and outside
the star. Appropriate boundary conditions are implemented, at the interior of the star, the light cylinder and the external boundary of
the integration domain.
Results. We have found self-consistent solutions corresponding to a variety of combinations of internal and external fields. In all
cases the external field satisfies the force-free axisymmetric pulsar equation. The internal field satisfies a barotropic equilibrium and
extends to the center of the star. If a toroidal field is included at the interior of the star, then it has either the form of a twisted torus
confined within the flux surfaces that close inside the star, or it extends to the magnetosphere, but is contained to the field lines that
close within the light cylinder.
Conclusions. This work presents a global solution for the internal and the external field of an axisymmetric rotating neutron star. It
is shown that the twist of the internal field affects the external field, by increasing the number of open field lines and eventually the
spin-down rate of the star. This effect is far more drastic if the toroidal field, and consequently the poloidal current flowing within
the star, is allowed to populate the closed field lines of the magnetosphere, rather than if it remains confined in the star. We further
remark that the internal field structure depends on the presence of a twisted magnetosphere: if the twist current is not allowed to flow
in the magnetosphere it only occupied a narrow toroid at the interior of the star, whereas if the twist currents are allowed to flow in
the magnetosphere the internal toroidal field may occupy a significant volume of the stellar interior. Strong magnetospheric currents
may also impact the emission mechanisms, and lead to fluctuations in magnetar spin-down rates, moding and nulling of pulsars, a
correlation between angular shear and twist, and the general morphology of the pulsar magnetic field leading to various observational
manifestations. The magnetospheric toroidal fields may possibly dissipate, thus the system may switch from global twist to internal
twist and consequently exhibit transient behavior.
Key words. neutron stars – magnetic field – Barotropic Equilibrium

1. Introduction of its structure (Parfrey et al. 2012; Parfrey et al. 2013; Gour-
gouliatos & Hollerbach 2016; Gourgouliatos & Esposito 2018;
Gourgouliatos & Pons 2019), or possibly through electric dis-
The magnetic field of neutron stars provides the main route for
charges (Soglasnov et al. 2004; Chen & Beloborodov 2014) and
their observation, either via the generation of coherent radio
secular variations of the magnetic field (Pons et al. 2012).
emission as is the case in rotation powered pulsars (Philippov
et al. 2020), or through its decay (Pons et al. 2007) or particle This central role in the observable properties of neutron
acceleration and crust heating (Beloborodov 2009), which is rel- stars has motivated thorough studies of magnetic field structure
evant to thermally emitting X-ray neutron stars. Moreover, ex- both at the interior (Braithwaite & Nordlund 2006; Lander &
plosive and transient events in the form of flares (Mazets et al. Jones 2009; Ciolfi & Rezzolla 2013; Uryū et al. 2014; Armaza
1979; Hurley et al. 1999, 2005), bursts (Coti Zelati et al. 2018), et al. 2015; Suvorov & Glampedakis 2023), and at the exte-
giant pulses (Karuppusamy et al. 2010), changes in timing prop- rior; starting from axisymmetry (Contopoulos et al. 1999; Uz-
erties (Woods et al. 2002; Archibald et al. 2015; Pintore et al. densky 2003; Goodwin et al. 2004; Gruzinov 2005; Timokhin
2016; Scholz et al. 2017; Hu & Ng 2019; Levin et al. 2019), 2006), to three-dimensions (Spitkovsky 2006; Kalapotharakos
nulling, moding (Rankin et al. 2013; Lyne et al. 2010) and even & Contopoulos 2009), using techniques ranging from relax-
some forms of timing noise (Tsang & Gourgouliatos 2013) are ation to force-free electrodynamics and particle in cell (Philip-
related to greater or lesser extent to their magnetic field. This is pov & Spitkovsky 2014; Philippov et al. 2015; Kalapotharakos
mediated either through instabilities and major reconfigurations et al. 2018) and approaches based on machine learning (Ste-
Article number, page 1 of 11
 A&A proofs: manuscript no. aanda

fanou et al. 2022; Dimitropoulos et al. 2024). Despite the high where j is the electric current density, P is the pressure, Φ is
level of complexity of these studies, there have been limited at- the gravitational potential, and ρ is the stellar density. Under the
tempts to address simultaneously both the internal and the exter- assumption of barotropicity, pressure P is a function of density
nal field structure, and this has been done by simplifying the con- ρ, P = P(ρ). Taking the curl of equation (2) and using Ampère’s
figuration, i.e. by considering a non-relativistic magnetosphere law, we obtain:
(Glampedakis et al. 2014; Fujisawa & Kisaka 2014; Akgün et al.
2016; Akgün et al. 2018; Akgün et al. 2018) or a light-cylinder !
∇×B
sufficiently far from the surface of the star while focusing on the ∇× B× = 0. (3)
field structure at the interior and the near zone of the star (Uryū ρ
et al. 2014; Pili et al. 2015). Thus, the question of the impact Since the above expression is a curl of a vector, equal to 0,
of the internal field to the external and vice-versa remains unre- we can set the quantity inside curl equal to a gradient of a scalar
solved. function S (R, z):
In this work we address the issue of linking the magnetic
field of the interior of the star to the magnetosphere. We solve
simultaneously for the internal field so that it obeys a barotropic 1
∇S = B × (∇ × B) . (4)
MHD equilibrium, while the magnetospheric field is in a rel- ρ
ativistic force-free equilibrium. The system contains a toroidal
field, which is either of the form of a twisted torus confined From the requirement of axisymmetry, it is evident that:
within the star, or extends at the magnetosphere in the region
of the field lines that close within the light cylinder. Following [B × (∇ × B)]ϕ = 0 , (5)
this approach we provide a global solution for the magnetic field
structure of an axisymmetric relativistically rotating neutron star. thus any poloidal current must be parallel to the poloidal
The plan of the paper is the following. In section 2, we magnetic field, that leads to the following relation:
present the mathematical setup and we formulate the equations
that describe the system. In section 3, we present the solution
strategy we follow. In section 4, we present the simulations and ∇Ψ × ∇I = 0 ⇔ I = I(Ψ) , (6)
the results arising from the calculations. We discuss the prop- The poloidal component of equation (4) satisfies the follow-
erties of the solutions in section 5. We present some possible ing equation:
astrophysical applications in section 6. We conclude in section
7.
∂2 Ψ 1 ∂Ψ ∂2 Ψ
!
dI
− + 2 +I ∇Ψ + R2 ρ(r)∇S = 0 , (7)
2. Mathematical setup ∂R2 R ∂R ∂z dΨ

Here we present the derivation of the equations that will be where we have assumed √ that the density is a function only of
solved for the equilibrium of the magnetic field at the interior the spherical radius r = R2 + z2 . From equation (7), it is evi-
and exterior of the neutron star. We use cylindrical coordinates dent that the gradient of S is parallel to that of Ψ, ∇Ψ||∇S , lead-
(R, ϕ, z) and assume axial symmetry throughout the system. The ing to the conclusion that S = S (Ψ). Eventually, the barotropic
length units are chosen so that the light cylinder RLC = 1 = c/Ω equilibrium equation is expressed as follows:
where c is the speed of light and Ω the angular frequency of the
star. ′ ′
∆∗ Ψ + II + R2 ρ(r)S = 0 , (8)
We express the magnetic field through two scalar functions
Ψ(R, z) and I(R, z); as follows: where:
∂2 1 ∂ ∂2
B = ∇Ψ × ∇ϕ + I∇ϕ , (1) ∆∗ = − + 2, (9)
∂R 2 R ∂R ∂z
where ∇ϕ = ϕ̂/R. The function Ψ(R, z) represents the is the Grad-Shafranov operator and a prime denotes differen-
poloidal flux, while I(R, z) is proportional to the poloidal electric tiation with respect to Ψ.
current passing through a circular disk parallel to the horizontal Regarding the mass density ρ we assume the mass density ρ
plane, whose radius is R and is centered on the axis of symme- of the star is approximated through the relation:
try at a distance z from the horizontal plane. From Gauss’s law,
the magnetic field has zero divergence which is satisfied by con- 2
rns − r2
struction, given the definition of equation (1). ρ(r) = ρ0 2
, (10)
Next we will use the above expressions in the appropriate rns
framework to describe the equilibrium at the stellar interior and
where rns = 0.1 is the radius of the star and ρ0 is the density
exterior.
at the center of the star. Equation (10) corresponds to the solution
of the mass density of a neutron star, for an n = 1 polytrope,
2.1. Internal field where P = ρ2 (Lattimer & Prakash 2001).
Despite the functional freedom of S (Ψ), as shown by
In an MHD equilibrium, the force equilibrium is given by the Glampedakis & Lasky (2016) there are some restrictions on the
following expression (Reisenegger 2009): applicable forms. Here, we assume that S is a linear function
of Ψ, thus its derivative appearing in the final partial differen-
1 1 tial equation is a constant and we set its value S ′ = 1. This form
(j × B) = ∇P + ∇Φ (2) is compatible with the requirements of the aforementioned work.
cρ ρ
Article number, page 2 of 11
 D. Ntotsikas and K.N. Gourgouliatos: Interlinking magnetic fields of neutron stars

The choice of a linear form for S (Ψ) restricts the solutions. How-
ever, it allows magnetic fields in the form of a dipole for I = 0
and assuming vacuum external boundary conditions, something
that is not possible for non-linear expressions of S in Ψ (Gour-
gouliatos et al. 2013). Moreover, given the large number of free
parameters appearing in the problem, we have chosen to adopt
the linear form as a first approach and we will explore the wider
parameter space in future work.

2.2. External field

Considering the relativistically corotating, plasma-filled magne-
tosphere, we assume that electromagnetic forces prevail over
gravitational, inertial and pressure gradient forces. Therefore, a
system in equilibrium must have zero Lorentz force, specifically:

1
ρq E + j × B = 0 , (11)
c

where ρq is the electric charge density. By assuming ideal

MHD, the expression of equation (1) for the magnetic field and
Ohm’s law for the electric field corresponding to a corotating Fig. 1. The various regions of the domain, labeled from (I) to (V). The
magnetosphere, where the rotation axis coincides with the dipole field lines are depicted in black, the last open field line of the magneto-
axis, so that: sphere in red, the stellar surface in green and the light cylinder in dashed
blue. The top right inlet provides a zoom on the star.

ΩR
E=− ϕ̂ × B (12) 3. Solution strategy
c
We study two main families of models. One, in which the
we obtain the axisymmetric pulsar equation (Scharlemann &
toroidal twist field is allowed to exist only within the star. We
Wagoner 1973):
will refer to this type of twist as an internal twist. In the sec-
ond family of models, closed magnetospheric fields lines and
field lines at the interior of the star that are connected to the
∂2 Ψ 1 ∂Ψ ∂2 Ψ ∂Ψ
!
2
(1 − R ) − + 2 − 2R = −I(Ψ)I ′ (Ψ) , (13) former can be twisted. We will refer to this twist as the global
∂R 2 R ∂R ∂z ∂R twist, since it extends both at the stellar interior and the mag-
netosphere. We remark that the open field lines have, in either
where the lengths have been suitably normalized to the light- model, a toroidal field, which is related to the smooth crossing
cylinder radius. of the light cylinder.
The solution domain is partitioned in five regions, as shown
Considering the relativistic Grad-Shafranov operator: in Figure 1. Region (I) is the part of the magnetosphere contain-
ing open magnetic field lines that cross the light cylinder. Region
∂2 1 ∂ ∂2 ∂
!
(II) is the area of the closed magnetic field lines in the magne-
∆∗ = (1 − R2 ) − + − 2R (14)
∂R2 R ∂R ∂z2 ∂R tosphere containing closed magnetic field lines. Regions (III),
(IV), and (V) are all the stellar interior, where region (III) con-
equation (13) can be written as: tains field lines that cross the stellar surface and connect to the
open magnetospheric field lines, region (IV) contains magnetic
field lines that connect to the closed field lines of the magneto-
dI(Ψ) sphere, and region (V) contains field lines that close within the
∆∗ Ψ = −I(Ψ) . (15) star; thus, they are not connected to the magnetosphere and form
dΨ
closed toroids.
In the standard pulsar solution (Contopoulos et al. 1999), The magnetospheric field in region (I) requires a poloidal
I(Ψ) = 0 in the area of closed field lines, while for the open return current Irt , essential for the smooth crossing of the light-
field it is I(Ψ) , 0 ensuring a smooth transition through the cylinder. The magnetospheric field in region (II) satisfies the rel-
light cylinder region. Magnetic fields of minimal complexity ex- ativistic force-free equation; however, unlike region (I), there is
hibit a condition of zero poloidal current inside regions of closed freedom in the form of the current as no field lines cross the
magnetic lines, based on the hypothesis that pulsar fields are light cylinder and thus there is no requirement for a return cur-
not inherently twisted. The implementation of a nonzero cur- rent. However, it is possible that these field lines are twisted.
rent I(Ψ) ≡ Itw , 0 in this area, along with the concept of The magnetic field at the interior, regions (III), (IV) and (V) sat-
a twisted magnetosphere, is prevalent in magnetar models and, isfies equation (8) corresponding to the barotropic equilibrium.
subsequently, in strongly magnetized systems that will be stud- We note that formally, the cross product of the poloidal mag-
ied here. We will refer to this current as the twist current Itw and netic field and the electric current in region (III) exert a force in
it will appear in the right-hand-side of equation (13). the ϕ direction which is required for the spin-down of the star
Article number, page 3 of 11
 A&A proofs: manuscript no. aanda

arising from the penetration of the return current into the stellar To guarantee the smooth crossing of the field lines through
crust (Karageorgopoulos et al. 2019). In our approach, we will the light cylinder, we impose continuity on the function and its
neglect the impact of this current in the interior and find its equi- derivative as follows:
librium through equation (8) by setting I = 0. Regions (II), (IV),
and (V) may contain a poloidal current as a result of twisting
of the magnetic field, Itw . In summary, the subsequent relations Ψ(R = 1+ , z) = Ψ(R = 1− , z) (20)
describe the field in the five aforementioned regions:
and the regularization condition:
∆∗ Ψ = −Irt Irt′ (I) 1
∂R Ψ(R = 1− , z) = ∂R Ψ(R = 1+ , z) =
′
I(Ψ)I (Ψ) . (21)
∆∗ Ψ = −Itw Itw ′
(II) 2
′
∆∗ Ψ + R2 ρ(r)S = 0 (III) (16) The last equation also relates the value of the R-derivative of
∆ Ψ+
∗ ′
Itw Itw 2 ′
+ R ρ(r)S = 0 (IV), (V) Ψ to I. For the outer region and the open field lines of the mag-
netosphere, we implement the split monopole boundary condi-
The relativistic Grad-Shafranov operator appears explicitly tions to obtain radial behavior at the boundary of the integration
in regions (I) and (II) of the magnetosphere, whereas all of grid zmax , Rmax . Formally, boundary conditions of this type are
the interior of the star is characterized by the non-relativistic applied at infinite distance from the surface of the star, in which
Grad-Shafranov operator. This issue can be resolved even un- case the magnetic lines eventually acquire a radial shape; how-
der the assumption that all areas of the star’s magnetic field are ever, we observed that the application of split-monopole condi-
solved through the relativistic Grad-Shafranov operator. Given tions to the boundaries of our grid, compared to a run with twice
that the stellar surface radius is rns = 0.1RLC , the star’s rotation the size of the numerical domain, does not bring changes in the
exerts a minor influence on the internal structure of the mag- results, while it significantly increases the convergence speed of
netic field, as this would correspond to a maximum correction the computation. Relation (21) enables us to determine the form
of (rns /RLC )2 = 10−2 .We have confirmed that changes in the of I(Ψ) for magnetic flux function values less than Ψ0 , where Ψ0
magnetic flux function, when the above correction is performed, represents the magnetic flux function at the first closed magnetic
do not surpass 1%. Furthermore, the return current is in general field line.
much weaker than the twist current, especially if the latter has We note that a large fraction of the magnetospheric return
a remarkable effect on the magnetosphere, thus the assumption current enters the star through a current sheet flowing on the
of not including it in the calculation of the internal equilibrium separatrix. Such singularities cannot be handled by the finite dif-
does not alter the structure of the internal field. ference method used here; therefore, we approximate the current
Regarding the boundary conditions that will be imple- sheet using a Gaussian function with a width of 1 × 10−3 Ψ0 cen-
mented, we proceed as follows. We assume that the magnetic tered at Ψ = 0.99Ψ0 , ensuring that the entire return current flows
flux function, Ψ, is zero along the z-axis: through the open field lines. Unlike the solutions dealing exclu-
sively with the magnetosphere, in our approach, the field emerg-
ing from the star is not a pure dipole, thus the maximum flux
Ψ(R = 0, z) = 0 . (17)
function may have a different value depending on the twist im-
The value of Ψ at the star’s surface is not defined as that of a posed. To avoid such discrepancies we normalize appropriately
dipole, but is determined by solving the equation (16) within the the magnetic flux function so that the first closed field line for the
star. In this problem, the light cylinder is positioned at a distance α = 0 model corresponding to Ψ = 1.23 for direct comparison
of 10 stellar radii (RLC = 10rns ). The distance of the light cylin- with the standard solutions (Timokhin 2006) and adhere to this
der chosen correspond to an angular velocity of the neutron star normalization for the rest of the solutions. Here, α is a parame-
of Ω = 3000 rad/sec, indicating that we are formally examin- ter that determines the ratio of the electric current distribution to
ing a millisecond pulsar with a spin frequency of approximately the difference of the flux function Ψ minus its value at the first
500 Hz. closed magnetic field line.
The magnetic field is symmetric with respect to the north and With regard to the twist current, we will focus on two cases.
south hemisphere; hence, the subsequent boundary conditions The globally twisted model where the closed magnetospheric
are applied at the equator: field lines carry poloidal current (region II) connected to the star
(region IV):
∂z Ψ(R < RT , z = 0) = 0 (18) 

 Irt , (I)

I(Ψ) =  α(Ψ(R, Ψ

z) − ), (II),(IV),(V) (22)

where RT denotes the equatorial distance of the outermost closed  0
(III) .

field line of the magnetosphere. Subsequent to this intersection 0,

and along the equatorial plane, the equatorial current sheet of the
magnetosphere is characterized by the expression: Alternatively, in the internally twisted model, any poloidal cur-
rent corresponding to an internal poloidal field in the star closes
within the star, thus there is no poloidal current in the closed
Ψ(R ≥ RT , z = 0) = Ψ(RT , 0) (19) field lines (region II), nor in the field lines within the star that are
Although most works assume that RT lies exactly on the light connected to the closed magnetospheric field lines (region IV):
cylinder, this point has been argued to be located between 0.8RLC
and 0.9RLC (Contopoulos et al. 2024). When closed magnetic
Irt ,

field lines are twisted above a minimum amount, physically ac-


 (I)
I(Ψ) = 

0, (II), (III), (IV) (23)

ceptable solutions require RT to be placed towards the star (Ntot- 
sikas et al. 2024), which is accounted for in our models. α(Ψ(R, z) − Ψ ss f ), (V) .



Article number, page 4 of 11

 D. Ntotsikas and K.N. Gourgouliatos: Interlinking magnetic fields of neutron stars

In all cases, Irt is the appropriate return current determined by the Table 1. Simulation results for global twist.
smooth light crossing condition. The poloidal twist current in the
star is a linear expression given by the flux function Ψ reduced Global Twist
∆ϕ
by either its value at last closed field line (Ψ0 ), or its value at the α Ψnorm
0
Ltwisted
Luntwisted 2 RT Ψnorm
max Ψnorm
ss f
toroidal loop that touches the stellar surface, for which Ψ ss f = 0.0 1.23 1.0 0.0 1.00 11.39 10.61
Ψ(rns , 0), i.e. the value of the flux function on the equator of the 0.5 1.25 1.03 0.13 1.00 11.40 10.62
star. 1.0 1.29 1.09 0.29 1.00 11.43 10.65
Matching of the stellar and magnetospheric solution is made 1.5 1.37 1.24 0.38 1.00 11.47 10.69
by demanding continuity on Ψ and I. The solution in the stel- 2.0 1.49 1.47 0.44 1.00 11.53 10.76
lar interior satisfies ∆∗ Ψ + Itw Itw′
+ R2 ρS ′ = 0, while the cor- 2.5 1.63 1.76 0.50 1.00 11.60 10.83
responding one in the magnetosphere is ∆∗ Ψ + Itw Itw ′
= 0. The 3.0 1.81 2.17 0.58 0.98 11.68 10.92
two equations differ on the form of the Grad-Shafranov opera- 3.5 2.02 2.7 0.61 0.95 11.77 11.02
tor, being in the one case the non-relativistic and in the other the 4.0 2.23 3.29 0.67 0.95 11.85 11.11
relativistic one and on the term involving the density. We have 4.5 2.46 4.0 0.71 0.95 11.98 11.24
already commented on the impact of the non-relativistic term 5.0 2.73 4.93 0.79 0.90 12.10 11.36
of the Grad-Shafranov operator, leading to differences that scale 6.0 3.13 6.46 0.86 0.80 12.35 11.65
with the square of the ratio of the neutron star radius to the light-
7.0 3.77 9.4 0.90 0.70 12.78 12.10
cylinder, in which case is 0.01. The difference due to the density
8.0 4.10 11.10 0.94 0.60 13.10 12.44
term, is minimized as the density formally drops to zero on the
9.0 4.44 13.03 0.95 0.40 13.31 12.58
surface of the star, see equation (10), thus the same equation
is satisfied in either side of the stellar boundary. We note how- 10.0 4.90 15.90 1.01 0.20 13.51 12.71
ever, that due to discretizations and the cylindrical coordinates Notes. Simulation results for a range of values of the parameter α for
we have adopted, the density on the surface has some disconti- the global twist model. The first column is α, subsequent columns are
nuities, which, nevertheless do not lead to sharp transitions on Ψnorm
0 , where Ψnorm
0 = Ψ0 /(7.79 × 10−7 ) , the magnetic flux of the first
the field structure. closed field-line, Ltwisted /Luntwisted the ratio of the spin-down luminosity
In summary, two families of models of magnetic field config- of a magnetosphere with twist in the closed field-line region Ltwisted to
urations are developed. The first one, (equation 22), corresponds a standard pulsar solution Luntwisted , ∆ϕ is a measure of the twist of the
to the following state: regions (II), (IV), and (V) contain a cur- closed field-lines, RT the inner radius of the current sheet, Ψnorm max is the
rent associated to the internal twist of the magnetic field lines maximum value of Ψ attained everywhere in the domain Ψnorm ss f is the
that also leaks out in the closed magnetospheric field lines, while maximum value of of Ψ attained on the surface of the star.
open magnetospheric lines carry the essential return current from
the outer magnetosphere to the star’s interior. In the second one,
(equation 23) there is only a poloidal twist current in the field gouliatos & Lynden-Bell (2019) also used in Contopoulos et al.
lines that close within the star, region (V) without crossing the (2023), while the part of the interior code has been developed
surface of the magnetosphere, while the open field lines, region from the first principles for this particular application and has
(I) carry the essential return current. The choice of the linear ex- been benchmarked against Gourgouliatos et al. (2013).
pression for the twist current is restrictive given the wide param- Figure 2 shows magnetic field solutions for the global twist
eter space, however we have chosen to limit ourselves to a sim- model, for which Itw is applied in the area of the closed mag-
ple functional form and explore as a main parameter the ratio of netic field lines of the magnetosphere, regions (II), (IV) and (V)
the poloidal current to the flux function. The chosen expressions with coefficients ranging from α = 0.0 to α = 8.0. As the pa-
I = α(Ψ(R, z) − Ψ0 ) or I = α(Ψ(R, z) − Ψ ss f ) guarantee that the rameter α increases, we find that the current of the closed field
toroidal field vanishes along this boundary, thus there is no mag- lines becomes stronger. Additionally, the first closed field line
netic field discontinuity because of the twist nor an additional corresponds to a higher value of Ψ, and the equatorial current
current sheet that would have broken the north-south symmetry. sheet must end closer to the star. We present these effects for
A current sheet discontinuity is the one that already exists due to various values of α in Table 1, summarizing the main simulation
the presence of the return current that closes along the separatrix results. Regarding the magnetic field in the interior, we find that
between the closed and the open magnetic field lines. the region of the closed field lines shrinks once more twisted is
imposed. We have increased the value of α up to 10, as beyond
this value the twisted region is rather small and under-resolved.
4. Simulations and results Moreover, the innermost point of the equatorial current sheets is
We solve equation (16) for different values of the parameter α, twice the radius of the star for the maximum value of α imple-
which determines the current ratio over the magnetic field, ap- mented.
plying the simultaneous relaxation method and considering the We have also solved for the internally twisted model, Table
two cases of global and internal twist. Our simulations are con- 2 and Figure 3. As the field is only twisted in the interior, we
ducted using a FORTRAN 90 framework. As the system is ax- notice that the twisted region tends to shrink with increasing α
isymmetric, the functions will depend only on R and z. We di- as in the globally twisted case. No displacement of the innermost
vide the numerical domain into R and z with a typical resolution point of the current sheet is required here, as the twist is only at
of 800 × 800; nevertheless, we have run in higher resolutions, the interior, thus it affects mainly the magnetic flux emanating
1600 × 1600 to verify the numerical convergence of our results from the star and its multipolar structure, with higher multipoles
and a change of less than 0.5% was found. The derivatives are appearing for stronger toroidal fields. These effects impact both
computed numerically by finite difference; a central difference the flux value of the first closed field line and the structure of the
scheme is used for the first derivatives, and a three-point stencil of the magnetic field in the magnetosphere.
is utilized for the second derivatives. The part of the magneto- We note that for both families of solutions, increasing α leads
sphere code is based on a refined version of the code of Gour- to a larger amount of flux emerging from the star Ψ ss f and a
Article number, page 5 of 11
 A&A proofs: manuscript no. aanda

a b

c d
Fig. 2. The magnetic field of a relativistically rotating neutron star and its magnetosphere for the global twist model. The field lines are shown
in black, the stellar surface in green, the light cylinder is depicted with the dashed blue line at R = 1, along with the current sheet. The red line
represents the first closed field line of the untwisted case (panel a) and is shown in subsequent panels for comparison. A poloidal current is applied
in the area of the closed magnetic field lines with coefficient α = 0.0, 3.0, 6.0, 8.0. The color bar indicates the value of II ′ across several parts
of the star and the magnetosphere. The maximum value of II ′ appearing in the twisted field lines is saturated in color to allow the depiction of its
structure in the rest of the system. The top-right inlet is shows a zoomed-in area of the star.

larger maximum value of flux at the interior of the star Ψmax , face is dominated by a dipolar term, higher multipoles co-exist
while all other quantities involved in the equation (density pro- leading to a different surface magnetic field profile than a pure
file and S ) are kept the same. We note however that the change dipole.
on either the maximum flux and the total flux emerging from the
surface of the star is small and does not scale with the increase Comparing the two families of solutions, while the trend is
of Ψ0 . Thus the primary cause of increasing the fraction of open qualitatively the same, we find that the quantitative differences
field lines is the change of the magnetosphere. Furthermore, we between the two cases are rather pronounced, namely for the
note that while the form of the flux function on the stellar sur- same values of α the fraction of the open field lines is substan-
tially higher once twist currents flow through the magnetosphere.
Article number, page 6 of 11
 D. Ntotsikas and K.N. Gourgouliatos: Interlinking magnetic fields of neutron stars

3a b
Fig. 3. The magnetic field of the internal twist model. Panels a and b correspond to α = 5.0, 10.0 respectively. The various field lines, color bars
are shown as in Figure 2.

Table 2. Simulation results for internal twist.

Internal Twist
α Ψnorm
0
Ltwisted
Luntwisted Ψnorm
max Ψnorm
ss f
0.0 1.23 1.0 11.39 10.61
0.5 1.25 1.03 11.39 10.61
1.0 1.29 1.09 11.40 10.62
1.5 1.30 1.12 11.40 10.62
2.0 1.35 1.20 11.41 10.63
2.5 1.41 1.31 11.42 1.064
3.0 1.48 1.45 11.42 10.64
3.5 1.60 1.70 11.44 10.65
4.0 1.68 1.86 11.45 10.66
4.5 1.77 2.07 11.46 10.66
5.0 1.89 2.36 11.47 10.67
6.0 1.94 2.49 11.50 10.71
7.0 1.99 2.62 11.53 10.74
8.0 2.09 2.89 11.58 10.78
9.0 2.29 3.47 11.61 10.80
10.0 2.45 3.97 11.68 10.90
Notes. Simulation results for a range of values of the parameter α for
the internal twist model, where the current is confined to the region (V)
inside the star. The first column is α, subsequent columns are Ψnorm 0 , the
magnetic flux of the first closed field-line, Ltwisted /Luntwisted the ratio of Fig. 4. Magnetic flux function of the first closed magnetic field line as a
the spin-down luminosity of a magnetosphere with twist in the closed function of α.
field-line region Ltwisted to a standard pulsar solution with twist only in
the open field-lines Luntwisted , Ψnorm
max and Ψ ss f as shown in Table 1 We
norm

note that ∆ϕ = 0 and RT = 1, thus they are not shown here.

5. Discussion

Next we are going to address the consequences of increasing

This has major impact on various physical properties of the neu- the twist to the main physical properties of neutron stars. We
tron star ranging from the spin-down power to the twist of the will focus on the spin-down luminosity, the overall twist in the
magnetosphere and the polar cap opening angle. magnetosphere, and the polar cap opening angle.
Article number, page 7 of 11
 A&A proofs: manuscript no. aanda

Fig. 5. The spin-down luminosity of the various solutions scaled to the Fig. 6. The twist of the magnetospheric part of the field line for which
spin-down power of the untwisted solution, as function of α. Ψ = 1.1Ψ0 , as a function of the parameter α.

5.1. Spin-down Luminosity where BR is the radial component of the magnetic field, Bz the
axial and Bϕ the azimuthal. Thus, we obtain the following:
The general trend is that an increase in the twist current leads
to a larger fraction of open field lines. For small values of α ≤ Z
Bϕ
1, both models give the same result, Figure 4, however, once α ∆ϕ = dR . (26)
exceeds this value, the global twist model has a much higher RBR
value of Ψ0 . This is also clearly visible in Figures 2 and 3 where
the field line first closed of the untwisted model shown in red is As the twist may vary in different field lines, i.e. the fist closed
much more drastically displaced from the equator in the globally field lines have zero twist, as defined there by Bϕ = 0, while the
twisted model than in the internally twisted one. field line emerging from the equator of the star has the maxi-
The overall spin-down luminosity, which corresponds to the mum value of I, however its extend is formally zero. Therefore,
loss of electromagnetic energy from the star’s magnetic field, is we perform this integration is evaluated on the field line that cor-
expressed by the following relation: responds to Ψ = 1.1Ψnorm 0 . We integrate from the surface of the
star to the equator; therefore, the results presented in Table 1
represent half the twist values that would be derived by measur-
Z Ψnorm
0 ing the whole extent of the magnetic field line, emerging from
L=2 I(Ψ)dΨ. (24) the northern and closing at the southern hemisphere. In agree-
0 ment with earlier solutions (Lynden-Bell & Boily 1994; Gour-
Using equation (24), we evaluate the star’s spin-down power. In gouliatos & Lynden-Bell 2008; Pavan et al. 2009; Akgün et al.
models where the twist current flows in the magnetosphere, the 2018), the twist tends to approach a maximum value ∼ π/2 rather
spin-down power can be enhanced by a factor of approximately than increasing indefinitely. Therefore, the system can formally
16 for α = 10, Figure 5, whereas if the twist current closes within approach a split monopole state after a finite amount of twist. We
the star, for the same value of α the spin-down luminosity is have decided to investigate the properties in terms of the param-
enhanced by a factor of 4. The rapid increase of the spin-down eter α up to a maximum value of α = 10 in our calculation.
luminosity in the globally twisted model is related to the fact Figure 6 shows the twist of the closed magnetic field line
that the current sheet inner edge is closer to the stellar surface; region. It is evident that with increasing α, the twist in region (II)
therefore there is a drastic increase of the open magnetic field becomes higher; however, this region shrinks and is concentrated
lines. closer to the star, and the twist at the exterior will eventually
vanish.

5.2. Magnetospheric twist

5.3. Polar Cap
Next, we evaluate the twist of a magnetic field line by integration
of the following relation: We calculate the semi-opening angle of the polar cap by tracing
the foot points of the first closed field line corresponding to Ψ0 .
The location (RΨ0 , zΨ0 ) of its foot points satisfies the following
dR dz Rdϕ conditions: Ψ(RΨ0 , zΨ0 ) = Ψ0 and R2Ψ0 + z2Ψ0 = rns 2
. Then, the
= = (25) semi-opening angle is given by the following expression:
BR Bz Bϕ
Article number, page 8 of 11
 D. Ntotsikas and K.N. Gourgouliatos: Interlinking magnetic fields of neutron stars

flow in the magnetosphere, for α = 10 the spin-down luminosity

is about 16 times higher than in the untwisted case, whereas if
the twist current is enclosed in the star it is about 4 times higher
compared to the untwisted model. This effect will differentiate
the two families of models. This is related to the fact that the
equatorial current sheet needs to retract closer to the star if the
magnetosphere is twisted, whereas the current sheet can start at
the light cylinder in the absence of external twist. This calcu-
lation provides further evidence that the twist of the magneto-
sphere enhances the spin-down luminosity of the star (Thompson
et al. 2002; Lyutikov 2006; Viganò et al. 2011), but also quite re-
markably that internal twist also leads to spin-down luminosity,
even if it only indirectly affects the magnetosphere by altering
the lower boundary conditions.
We note that our calculations are constrained by the fact that
the light cylinder is only ten times larger than the stellar radius.
A smaller star or a more distant light cylinder would allow for a
finer exploration of the innermost point of the current sheet and
would not stop at this particular value. In the highly twisted case,
where the currents flow in the magnetosphere, we further notice
that the polar cap becomes larger and the magnetic field is rem-
iniscent of a split monopole even within the light-cylinder, thus
the field lines are much less curved in the poloidal direction, yet
there is still curvature due to the toroidal field. This stretching
Fig. 7. The polar cap’s semi-opening angle as a function of α. of the magnetic field disfavors the generation of curvature radi-
ation, as a smaller part of the field is curved, as opposed to the
previous case.
A critical question is what determines whether the internal
RΨ0 twist will also populate the magnetosphere, or if it is only going
θ pc = arctan , (27)
zΨ0 to be contained in the star, in principle the capacity of the mag-
netosphere to support twist currents. Twist currents can become
We plot the semi-opening angle of the polar cap in Figure 7. much stronger than the spin-down current which is related to
We find that as α increases, the semi-opening angle of the po- the ratio of the neutron star radius to the light cylinder. They re-
lar cap expands both for the internal and global twist. However, quire a larger population of charges in the magnetosphere which
there is a difference in magnitude. If the twist current is allowed is related to the multiplicity of the charges and may untwist
in the magnetosphere, the polar cap increases very rapidly, and within timescales of years (Beloborodov & Thompson 2007),
it even becomes a split monopole after some finite twist. In con- thus providing a transient, but not necessarily explosive behav-
trast, if the twist current lies entirely within the star and does not ior. As this question is still unresolved, we speculate that magne-
populate the closed magnetospheric field lines, changes on the tospheres with a higher charged particle density are more likely
polar cap are minor, increasing from 17.5◦ to 20◦ , due to the fact to host such currents, whereas magnetospheres with smaller par-
that the magnetic field on the surface is no longer dipolar, but it ticle densities are possibly reminiscent of vacuum or minimally
contains higher multipoles of north-south symmetry. twisted magnetospheres. This will have the following conse-
quences: neutron stars with twisted magnetospheres will have
a higher spin-down rate, the field lines will have a larger cur-
6. Applications
vature radius, and, as postulated, they will have a higher charge
The solutions found here illustrate the impact of twist on the density. This could provide an interpretation of transient magne-
global magnetic field structure of a neutron star, which spans tar behavior, as the spin-down rate is higher, the generation of
from the interior to the exterior. Our models explore the major curvature radiation related to radio emission is less likely due
qualitative difference of the presence of twist in the magneto- to the structure of the field lines, and the presence of charged
sphere or exclusively at the interior of the star, and employ an regions and currents within the magnetosphere may eclipse any
incremental exploration of the parameter space regarding the in- radio emission produced near the star (Levin et al. 2019; Lower
jection of twist of the field lines. These differences in the mag- et al. 2023). We note that such variations on the spin-down power
netic field structure both at the interior and the exterior are likely have been noted in simulations of twisted magnetospheres (Par-
to impact the observational behavior of neutron stars. frey et al. 2012; Ntotsikas et al. 2024) and also in the timing
Indeed, we confirm that the inclusion of a toroidal magnetic behavior of pulsars correlated with emission across the electro-
field strongly affects the spin-down rate. Even if a comparable magnetic spectrum and changes in pulse profile (Urama et al.
amount of poloidal magnetic flux crosses the stellar surface, a 2006; Camilo et al. 2007, 2008; Younes et al. 2020b; Lower et al.
twisted field will have a more pronounced spin-down rate. We 2025; Fisher et al. 2024), also accompanied by spectral evolution
notice major differences in the spin-down rate depending on (Younes et al. 2025).
whether the twist current is confined within the star or whether These models also introduce current sheets that extend
it populates the magnetosphere. For smaller values of α up to within the light cylinder, especially for highly twisted systems.
α ≈ 1, the spin-down power scales the same for both families As current sheets are prone to instabilities, they may undergo
of models. For values higher than that, we find that there is a tearing modes, which lead to dissipation. This effect may be
drastic difference. In particular, if we allow the twist current to leading to the release of energy and particle acceleration. Al-
Article number, page 9 of 11
 A&A proofs: manuscript no. aanda

though such effects are likely to occur in pulsars (Comisso et al. predetermined current is allowed to flow are not acceptable. The
2017; Cerutti & Giacinti 2021; Bransgrove et al. 2023), this presence of the light-cylinder adds an extra length-scale which,
model brings them within the light cylinder and very close to in combination with the twist-current, leads to contraction of
the star. The creation of current sheets is intimately related to the closed field line region and the presence of a current sheet
the twisting of the external field, thus, torque variations due to closer to the star. Furthermore, the open field lines, especially
switches from twisted and untwisted states will be accompanied near the light-cylinder and beyond have a structure of a split
by a rapid release of energy through explosive events (Archibald monopole: such a field is significantly different from the dipole
et al. 2020). Moreover, the twist, internal or global, affects the background assumed there. Notably, split monopole structures
size of the polar cap, with larger polar caps associated with can appear in spherical geometry only after a relatively small
higher twist (Tong 2019), as has been suggested to be the case in amount of twist (Lynden-Bell & Boily 1994; Gourgouliatos &
sources with flaring episodes (Younes et al. 2020a). Lynden-Bell 2008; Tong 2019) Apart from the expected differ-
In our simulations, the chosen ratio of the light-cylinder ra- ences in the magnetosphere, where the actual equilibrium equa-
dius over the neutron star radius corresponds to a millisecond tion is different, there are further, remarkable differences at the
pulsar whose frequency is approximately 500 Hz. Although this interior. In particular, the internal toroidal field is allowed in the
is much faster than any known magnetar, it may be directly com- equatorial and mid-latitude region, but not near the poles as in
parable to millisecond magnetars (Dall’Osso & Stella 2022). the non-rotating solution of Glampedakis et al. (2014). Conse-
These sources are newborn magnetars, expected to spin-down quently, the region where the twist current flows, at the interior
rapidly (Çıkıntoğlu et al. 2020). In addition to these sources, an this time, can maximally be, that of the closed field lines, a re-
interesting extension would be the scaling of the model to less gion determined by the relativistic solution. Overall, the solu-
rapidly rotating sources. In such systems, the light cylinder lies tion with a relativistically rotating magnetosphere is more con-
much farther away than the 10 stellar radii we have assumed strained than the non-rotation inside-out magnetosphere and pro-
here, and it is possible that a smaller fraction of the closed field vides direct estimation of astrophysically relevant quantities for
lines will be twisted. However, the inclusion of twist may impact neutron stars including their relativistic magnetosphere.
the spin-down properties of lower rotating sources, if it extends
all the way to the light cylinder. However, in models where the 7. Conclusions
twist current is contained within the star, part of the change in the
spin-down power is due to the fact that the field has a multipolar This research investigates equilibrium configurations for twisted
structure on the surface rather than a dipole. This will not affect magnetic fields of neutron stars considering both the star and
significantly the field structure near the light cylinder, as higher the magnetospheres. Previous studies, in general, either exam-
multipoles decrease rapidly. Nevertheless, even if these effects ined the external magnetosphere without accounting for the stel-
are mild, a toroidal field affects the overall amount of magnetic lar interior or presumed a vacuum exterior. Our work highlights
flux emerging from the star, thus leading to a higher spin-down a comprehensive method, in which the magnetic field is consis-
rate. tently solved for, including the stellar and magnetospheric cur-
A further remarkable difference between the models in rents.
which the twist is contained in the star versus the ones where The study indicates that the inclusion of twist impacts the
the magnetosphere is twisted is the volume occupied by the structure of the neutron stars magnetic field . Its effect is mod-
toroidal field within the star. In the former case, the toroidal field erate if the twist is only allowed in the interior of the star and
is confined in a very small region, within closed loops of mag- rather drastic if the twist is also in the closed magnetospheric
netic flux; on the contrary, if the twist is allowed to populate the field lines. We have quantified its impact on the spin-down rate,
magnetosphere, it reaches much higher latitudes up to approx- magnetospheric twist, and polar cap opening angle, offering in-
imately 45◦ , for the models presented here. This expansion of sights into phenomena such as transitions between pulsar and
the toroidal field region can have important implications for the magnetar states or intermittent sources. This work has provided
magnetic field energy, as the toroidal field can contribute more the framework for the extraction of observable parameters and
intensely to the total energy budget, even under equilibrium con- the eventual quantification of these effects.
ditions, whereas in cases the twisted field was contained in the Future research should progress to solutions beyond that of a
star this ratio was rather low (Lander & Jones 2009), unless a barotropic equilibrium and simple linear forms for S (Ψ) and the
special form of the poloidal current was postulated (Ciolfi & twist current Itw . Moreover, it is essential to adopt more realistic
Rezzolla 2013). It may also affect the ellipticity of the star, as models in which both the crust and the core are accounted for, as
the toroidal field tends to increase it, and in the fully confined the relevant equilibrium equations are rather different. Further-
models it can only contribute locally, whereas in this family of more, progress from axisymmetric models to three-dimensional
models, the toroidal field can be much stronger and extend to a simulations may provide a more precise understanding of mag-
larger region within the star (Haskell et al. 2008). netospheric processes.
The question of a hydromagnetic equilibrium of a neutron An in-depth examination of the communication between the
star with rotation was also posed in the work of Glampedakis star’s inside and exterior, particularly over its surface, is cru-
et al. (2014). In that work the problem was studied in the non- cial. These endeavors will be essential for addressing fundamen-
rotating system, while touching upon the question of rotation, tal inquiries regarding the evolution of magnetic fields in neu-
without, however, fully addressing the light-cylinder and the rel- tron stars and comprehending the relationship between magne-
ativistic magnetosphere. Our results provide a direct compari- tospheric distortions and observable phenomena, such as timing
son between the two regimes. The obvious difference is in the variations in magnetars and even normal sources.
very structure of the magnetosphere, where the additional phys-
ical constraints of the relativistic force-free magnetosphere have 8. Data availability statement
been added. In particular, the current flowing along the open field
lines is determined by the requirement that the magnetic field The data sets generated during the current study are available
crosses smoothly the light cylinder, therefore, solutions where a from the corresponding author upon reasonable request.
Article number, page 10 of 11
 D. Ntotsikas and K.N. Gourgouliatos: Interlinking magnetic fields of neutron stars

Acknowledgements. We are grateful to an anonymous referee whose construc- Lyutikov, M. 2006, MNRAS, 367, 1594
tive comments improved this paper. We are grateful to I. Contopoulos and S. Mazets, E. P., Golentskii, S. V., Ilinskii, V. N., Aptekar, R. L., & Guryan, I. A.
Lander for discussions during the preparation of this manuscript. This work was 1979, Nature, 282, 587
supported by computational time granted by the National Infrastructures for Re- Ntotsikas, D., Gourgouliatos, K., Contopoulos, I., & Lander, S. 2024, MNRAS,
search and Technology S.A. (GRNET S.A.) in the National HPC facility - ARIS 527, 6691
- under project ID pr017008/simnstar2. Parfrey, K., Beloborodov, A. M., & Hui, L. 2012, ApJ, 754, L12
Parfrey, K., Beloborodov, A. M., & Hui, L. 2013, ApJ, 774, 92
Pavan, L., Turolla, R., Zane, S., & Nobili, L. 2009, MNRAS, 395, 753
Philippov, A., Timokhin, A., & Spitkovsky, A. 2020, Phys. Rev. Lett., 124,
References 245101
Philippov, A. A. & Spitkovsky, A. 2014, ApJ, 785, L33
Akgün, T., Cerdá-Durán, P., Miralles, J. A., & Pons, J. A. 2018, MNRAS, 481, Philippov, A. A., Spitkovsky, A., & Cerutti, B. 2015, ApJS, 801, L19
5331 Pili, A. G., Bucciantini, N., & Del Zanna, L. 2015, MNRAS, 447, 2821
Akgün, T., Cerdá-Durán, P., Miralles, J. A., & Pons, J. A. 2018, MNRAS, 474, Pintore, F., Bernardini, F., Mereghetti, S., et al. 2016, MNRAS, 458, 2088
625 Pons, J. A., Link, B., Miralles, J. A., & Geppert, U. 2007, Phys. Rev. Lett., 98,
Akgün, T., Miralles, J. A., Pons, J. A., & Cerdá-Durán, P. 2016, MNRAS, 462, 071101
1894 Pons, J. A., Viganò, D., & Geppert, U. 2012, A&A, 547, A9
Archibald, R. F., Kaspi, V. M., Ng, C. Y., et al. 2015, ApJ, 800, 33 Rankin, J. M., Wright, G. A. E., & Brown, A. M. 2013, MNRAS, 433, 445
Archibald, R. F., Scholz, P., Kaspi, V. M., Tendulkar, S. P., & Beardmore, A. P. Reisenegger, A. 2009, A&A, 499, 557
2020, ApJ, 889, 160 Scharlemann, E. T. & Wagoner, R. V. 1973, ApJ, 182, 951
Armaza, C., Reisenegger, A., & Valdivia, J. A. 2015, ApJ, 802, 121 Scholz, P., Camilo, F., Sarkissian, J., et al. 2017, ApJ, 841, 126
Beloborodov, A. M. 2009, ApJ, 703, 1044 Soglasnov, V. A., Popov, M. V., Bartel, N., et al. 2004, ApJ, 616, 439
Beloborodov, A. M. & Thompson, C. 2007, ApJ, 657, 967 Spitkovsky, A. 2006, ApJ, 648, L51
Braithwaite, J. & Nordlund, Å. 2006, A&A, 450, 1077 Stefanou, P., Pons, J. A., & Cerdá-Durán, P. 2022, MNRAS, 518, 6390
Bransgrove, A., Beloborodov, A. M., & Levin, Y. 2023, ApJ, 958, L9 Suvorov, A. G. & Glampedakis, K. 2023, Phys. Rev. D, 108, 084006
Camilo, F., Cognard, I., Ransom, S. M., et al. 2007, ApJ, 663, 497 Thompson, C., Lyutikov, M., & Kulkarni, S. R. 2002, ApJ, 574, 332
Camilo, F., Reynolds, J., Johnston, S., Halpern, J. P., & Ransom, S. M. 2008,
Timokhin, A. N. 2006, MNRAS, 368, 1055
ApJ, 679, 681
Tong, H. 2019, MNRAS, 489, 3769
Çıkıntoğlu, S., Şaşmaz Muş, S., & Ekşi, K. Y. 2020, MNRAS, 496, 2183
Tsang, D. & Gourgouliatos, K. N. 2013, ApJ, 773, L17
Cerutti, B. & Giacinti, G. 2021, A&A, 656, A91
Chen, A. Y. & Beloborodov, A. M. 2014, ApJ, 795, L22 Urama, J. O., Link, B., & Weisberg, J. M. 2006, MNRAS, 370, L76
Ciolfi, R. & Rezzolla, L. 2013, MNRAS, 435, L43 Uryū, K., Gourgoulhon, E., Markakis, C. M., et al. 2014, Phys. Rev. D, 90,
Comisso, L., Lingam, M., Huang, Y. M., & Bhattacharjee, A. 2017, ApJ, 850, 101501
142 Uzdensky, D. A. 2003, ApJ, 598, 446
Contopoulos, I., Kazanas, D., & Fendt, C. 1999, ApJ, 511, 351 Viganò, D., Pons, J. A., & Miralles, J. A. 2011, A&A, 533, A125
Contopoulos, I., Ntotsikas, D., & Gourgouliatos, K. 2024, MNRAS: Letters, 527, Woods, P. M., Kouveliotou, C., Göǧüş, E., et al. 2002, ApJ, 576, 381
L127 Younes, G., Baring, M. G., Kouveliotou, C., et al. 2020a, ApJ, 889, L27
Contopoulos, I., Ntotsikas, D., & Gourgouliatos, K. N. 2023, MNRAS: Letters, Younes, G., Güver, T., Kouveliotou, C., et al. 2020b, ApJ, 904, L21
527, L127 Younes, G., Lander, S. K., Baring, M. G., et al. 2025, arXiv e-prints,
Coti Zelati, F., Rea, N., Pons, J. A., Campana, S., & Esposito, P. 2018, MNRAS, arXiv:2502.20079
474, 961
Dall’Osso, S. & Stella, L. 2022, in Astrophysics and Space Science Library, Vol.
465, Astrophysics and Space Science Library, ed. S. Bhattacharyya, A. Pa-
pitto, & D. Bhattacharya, 245–280
Dimitropoulos, I., Contopoulos, I., Mpisketzis, V., & Chaniadakis, E. 2024, MN-
RAS, 528, 3141
Fisher, R., Butterworth, E. M., Rajwade, K. M., et al. 2024, MNRAS, 528, 3833
Fujisawa, K. & Kisaka, S. 2014, MNRAS, 445, 2777
Glampedakis, K., Lander, S., & Andersson, N. 2014, MNRAS, 437, 2
Glampedakis, K. & Lasky, P. D. 2016, MNRAS, 463, 2542
Goodwin, S. P., Mestel, J., Mestel, L., & Wright, G. A. 2004, MNRAS, 349, 213
Gourgouliatos, K. & Lynden-Bell, D. 2008, MNRAS, 391, 268
Gourgouliatos, K. N., Cumming, A., Reisenegger, A., et al. 2013, MNRAS, 434,
2480
Gourgouliatos, K. N. & Esposito, P. 2018, in Astrophysics and Space Science
Library, Vol. 457, Astrophysics and Space Science Library, ed. L. Rezzolla,
P. Pizzochero, D. I. Jones, N. Rea, & I. Vidaña, 57
Gourgouliatos, K. N. & Hollerbach, R. 2016, MNRAS, 463, 3381
Gourgouliatos, K. N. & Lynden-Bell, D. 2019, MNRAS, 482, 1942
Gourgouliatos, K. N. & Pons, J. A. 2019, Physical Review Research, 1, 032049
Gruzinov, A. 2005, Physical Review Letters, 94, 021101
Haskell, B., Samuelsson, L., Glampedakis, K., & Andersson, N. 2008, MNRAS,
385, 531
Hu, C. P. & Ng, C. Y. 2019, Astronomische Nachrichten, 340, 340
Hurley, K., Boggs, S. E., Smith, D. M., et al. 2005, Nature, 434, 1098
Hurley, K., Cline, T., Mazets, E., et al. 1999, Nature, 397, 41
Kalapotharakos, C., Brambilla, G., Timokhin, A., Harding, A. K., & Kazanas,
D. 2018, ApJ, 857, 44
Kalapotharakos, C. & Contopoulos, I. 2009, A&A, 496, 495
Karageorgopoulos, V., Gourgouliatos, K. N., & Contopoulos, I. 2019, MNRAS,
487, 3333
Karuppusamy, R., Stappers, B. W., & van Straten, W. 2010, A&A, 515, A36
Lander, S. K. & Jones, D. I. 2009, MNRAS, 395, 2162
Lattimer, J. M. & Prakash, M. 2001, ApJ, 550, 426
Levin, L., Lyne, A. G., Desvignes, G., et al. 2019, MNRAS, 488, 5251
Lower, M. E., Karastergiou, A., Johnston, S., et al. 2025, MNRAS, 538, 3104
Lower, M. E., Younes, G., Scholz, P., et al. 2023, ApJ, 945, 153
Lynden-Bell, D. & Boily, C. 1994, MNRAS, 267, 146
Lyne, A., Hobbs, G., Kramer, M., Stairs, I., & Stappers, B. 2010, Science, 329,
408

Article number, page 11 of 11