MNRAS 000, 1–13 (2022) Preprint 5 April 2022 Compiled using MNRAS LATEX style file v3.

Observational constraints on the deceleration parameter in a tilted

universe
Kerkyra Asvesta,1★ Lavrentios Kazantzidis,2 † Leandros Perivolaropoulos2 ‡ and Christos G. Tsagas1,3 §
1 Sectionof Astrophysics, Astronomy and Mechanics, Department of Physics, Aristotle University of Thessaloniki 54124, Greece
2 Department of Physics, University of Ioannina, GR-45110, Ioannina, Greece
3 Clare Hall, University of Cambridge, Herschel Road, Cambridge CB3 9AL, United Kingdom
arXiv:2202.00962v2 [astro-ph.CO] 2 Apr 2022

Accepted 2022 March 31. Received 2022 March 28; in original form 2022 February 14

ABSTRACT
We study a parametrization of the deceleration parameter in a tilted universe, namely a cosmological model equipped with
two families of observers. The first family follows the smooth Hubble flow, while the second is the real observers residing in a
typical galaxy inside a bulk flow and moving relative to the smooth Hubble expansion with finite peculiar velocity. We use the
compilation of Type Ia Supernovae (SnIa) data, as described in the Pantheon dataset, to find the quality of fit to the data and
study the redshift evolution of the deceleration parameter. In so doing, we consider two alternative scenarios, assuming that the
bulk-flow observers live in the ΛCDM and in the Einstein-de Sitter universe. We show that a tilted Einstein-de Sitter model can
reproduce the recent acceleration history of the universe, without the need of a cosmological constant or dark energy, by simply
taking into account linear effects of peculiar motions. By means of a Markov Chain Monte Carlo (MCMC) method, we also
constrain the magnitude and the uncertainties of the parameters of the two models. From our statistical analysis, we find that the
tilted Einstein-de Sitter model, equipped with one or two additional parameters that describe the assumed large-scale velocity
flows, performs similar to the standard ΛCDM paradigm in the context of model selection criteria (Akaike Information Criterion
and Bayesian Information Criterion).
Key words: cosmology: cosmological parameters – theory – dark energy – large-scale structure of Universe – supernovae:
general

1 INTRODUCTION mic Microwave Background (CMB) and Baryon Acoustic Oscilla-

tions (BAO) data, with the local Type Ia Supernovae (SnIa) mea-
The discovery of the accelerated expansion of the universe Riess
surements. The former use the inverse distance ladder method and
et al. (1998); Perlmutter et al. (1999) has restored the idea of the
lead to 𝐻0 = 67.4 ± 0.5 𝑘𝑚 𝑠𝑒𝑐−1 Mpc−1 Aghanim et al. (2020),
cosmological constant in the last 20 years Carroll (2001), rendering
while the latter give 𝐻0 = 73.2 ± 1.3 𝑘𝑚 𝑠𝑒𝑐−1 Mpc−1 by utilizing
the ΛCDM scenario as the broadly accepted concordance model of
the distance ladder method Riess et al. (2021b). This discrepancy
modern cosmology. In spite of its simplicity and ability to be in
is at 4𝜎 level and reaches the 5𝜎 threshold, when taking into ac-
tandem with a wide variety of up to date observational data Betoule
count the latest measurement of the Pantheon+ dataset, suggesting
et al. (2014a); Aubourg et al. (2015); Baxter et al. (2016); Alam et al.
𝐻0 = 73.04 ± 1.04 𝑘𝑚 𝑠𝑒𝑐−1 Mpc−1 Riess et al. (2021a). Another
(2017); Efstathiou & Lemos (2018); Scolnic et al. (2018a); Aghanim
similar inconsistency, though at a somewhat lower level, is the growth
et al. (2020), there are still theoretical and observational challenges
tension. The latter describes the inconsistency of the density rms mat-
that can not be explained in the context of the ΛCDM paradigm. The
ter fluctuations (in spheres of radius around 8ℎ−1 Mpc) and/or of the
theoretical challenges include the well known cosmological constant
matter density parameter (𝜎8 and Ω0𝑚 respectively), between the val-
Weinberg (1989) and coincidence P.J (1997); Velten et al. (2014)
ues reported by the Planck mission Aghanim et al. (2020) and those
problems. The observational challenges, which seem to have been
reported by dynamical probes, such as Weak Lensing Joudaki et al.
amplified during the last decade, refer to the discrepancies between
(2018); Abbott et al. (2018); Heymans et al. (2021) and the Red-
the reported values of some basic cosmological parameters using
shift Space Distortion data Macaulay et al. (2013); Nesseris et al.
different cosmological probes/methods.
(2017); Kazantzidis & Perivolaropoulos (2018); Perivolaropoulos
Probably the most important observational crack in the standard
& Kazantzidis (2019); Skara & Perivolaropoulos (2020). This mis-
scenario is the 𝐻0 tension, which refers to the inconsistency of the
match is currently at a 2 − 3𝜎 level.
Hubble parameter when comparing its value, obtained by the Cos-
These challenges led throughout the years to a wide variety
★ of alternative theories that have the potential to address some of
E-mail: [email protected]
† E-mail: [email protected] the issues discussed above (see Refs. Ishak (2019); Kazantzidis &
‡ E-mail: [email protected] Perivolaropoulos (2021); Di Valentino et al. (2021); Perivolaropou-
§ E-mail: [email protected] los & Skara (2021a); Saridakis et al. (2021) for recent reviews on the

© 2022 The Authors

2 K. Asvesta et al.
subject and references therein). However, a common assumption in sion of a recent global event Tsagas & Kadiltzoglou (2015); Tsagas
the majority of the proposed alternative scenarios is that any effects (2021b). Given that drift flows introduce apparent (Doppler-like)
that may emerge from the observed large-scale peculiar motions are dipolar anisotropies, the aforementioned observers should also “see”
neglected. As a matter of fact, the authors in Ref. Davis et al. (2011) an apparent dipole in the sky-distribution of the deceleration param-
have argued that neglecting coherent velocities in the SnIa sample eter Tsagas (2011). Moreover, the associated dipole axis should lie
can have a small but measurable impact on the estimation of cosmo- fairly close to that of CMB, assuming that both dipoles are (Doppler-
logical parameters. Peculiar motions are induced by the gravitational like) relative-motion effects. Interestingly, over the last decade or so,
attraction of the surrounding matter and lead to deviations from the a number of reports have claimed that such a dipolar anisotropy may
smooth Hubble law. As a result, peculiar velocities are used as probes actually exist in the local SnIa data Cooke & Lynden-Bell (2010);
to test different cosmological models. Antoniou & Perivolaropoulos (2010); Cai & Tuo (2012); Mariano &
Since the measurements of the individual peculiar velocities of Perivolaropoulos (2012, 2013); Colin et al. (2019); Krishnan et al.
galaxies still present large systematic uncertainties, the studies have (2021). Similar dipolar signatures have also been reported using
been conducted on peculiar-velocity statistics, averaging over many galaxy-cluster surveys. In the recent studies of Migkas et al. (2021a);
galaxies. By far the most common is the bulk flow (sometimes also Migkas et al. (2021b), for example, the authors reported a dipolar
called streaming motion). This is a sensitive probe of the density anisotropy in the distribution of the local Hubble constant (𝐻0 ), with
fluctuations on large enough scales, representing the coherent motion a statistical significance of more than 5𝜎, which could be attributed
of a large region (usually a sphere centered on the observer) moving to a bulk flow of 900 𝑘𝑚 𝑠𝑒𝑐−1 extending out to ∼ 500 Mpc. Similar
relative to the cosmological rest-frame (where the CMB temperature variations on 𝐻0 across the sky have been reported by Luongo et al.
dipole vanishes). Over the last couple of decades, several studies (2021) where QSOs and GRBs where used to trace the anisotropies.
have focused on the analysis of the peculiar-velocity field in search Additional hints for possible deviation from isotropy were recently
of an overall bulk flow. These studies use data from SnIa Riess et al. claimed by using 1.36 million quasars Secrest et al. (2021). The lat-
(1995); Colin et al. (2011); Dai et al. (2011); Turnbull et al. (2012); ter study revealed a large dipole well aligned with that of the CMB,
Feindt et al. (2013); Lavaux et al. (2013); Rathaus et al. (2013); Ma showing a 4.9𝜎 tension with respect to the one expected by standard
& Pan (2014); Boruah et al. (2020); Horstmann et al. (2021), employ ΛCDM. All these reports constitute a crucial test for the validity of
galaxies as tracers Feldman et al. (2010); Nusser & Davis (2011); the concordance ΛCDM paradigm.
Nusser et al. (2011); Ma & Scott (2013); Hong et al. (2014); Hoffman In the current study we focus on the implications of bulk pecu-
et al. (2015); Magoulas et al. (2016); Scrimgeour et al. (2016); Qin liar flows for the mean kinematics of the associated observers, by
et al. (2019); Qin (2021); Howlett et al. (2022), or appeal to the employing the latest publicly available SnIa (i.e. the Pantheon com-
kinematic Sunyaev-Zeldovich effect on the CMB, in order to measure pilation). More specifically, we provide the profile of the deceleration
bulk peculiar velocities of more than 1000 galaxy clusters Kashlinsky parameter in tilted almost-Friedmann universes, as measured by ob-
et al. (2009, 2010); Mak et al. (2011). While most studies agree on servers living inside (slightly) contracting bulk flows. This follows
the general direction of the bulk flow, there are strongly conflicting after adopting a functional form for the volume scalar, which mon-
claims as to the amplitude and its scale. The majority of the surveys itors the local contraction of the observers’ peculiar motion and in
report bulk flows on scales 20 − 100ℎ−1 Mpc and velocities around so doing determines the local deceleration parameter. Here, our re-
200 − 300 𝑘𝑚 𝑠𝑒𝑐−1 Ma & Pan (2014); Boruah et al. (2020); Qin sults apply to a perturbed, tilted Einstein-de Sitter universe, which
(2021). Also, although bulk peculiar flows should generally diminish was selected for its naturalness and simplicity.1 In such a model,
on progressively larger volumes, there have been some controversial the accelerated expansion does not require an ad hoc cosmological
claims of very large bulk flows that are commonly referred to as constant, or exotic dark energy, but it is achieved naturally after al-
dark flows. These can exceed 400ℎ−1 Mpc Kashlinsky et al. (2009) lowing for peculiar-velocity perturbations and by introducing tilted
and are inconsistent with the ΛCDM expectations. Alleviating this observers. Clearly, allowing for bulk peculiar motions is far less ar-
inconsistency may provide us with further insight into the nature of bitrary and has much sounder physical motivation than introducing
dark matter and dark energy. a fine-tuned cosmological constant, or appealing to an elusive dark-
Typically, most studies measure the amplitude and the direction energy component. In addition, since peculiar velocities are triggered
of the bulk flow in the rest-frame of the CMB radiation, which also by structure formation, the tilted scenario faces no coincidence prob-
defines the coordinate system of the smooth Hubble flow. However, lem either.
real observers do not simply follow the cosmological expansion, but This work ameliorates previous treatments (see Tsagas (2011);
drift relative to it, having their own peculiar motion. For example, Tsagas & Kadiltzoglou (2015); Tsagas (2021b)), which only qual-
from the dipole anisotropy seen in the CMB temperature map, it itatively constrained the parameters of the tilted model. The afore-
is estimated that the Local Group of galaxies (including our Milky mentioned earlier studies did not include a detailed data analysis, but
Way) has a peculiar velocity of about 600 𝑘𝑚 𝑠𝑒𝑐−1 relative to the used parameters such as the value of the deceleration parameter at
CMB rest-frame Kogut et al. (1993). 𝑧 = 0, the local expansion rate “experienced” by the real observers
The present analysis adopts the tilted cosmological scenario intro- and the “transition redshift” from global deceleration to (apparent)
duced in Refs. Tsagas (2010, 2011), which allows for two families of local acceleration.
observers that move with respect to each other. This model predicts
different values for the deceleration parameters measured in the two
frames, due to relative-motion effects alone. In particular, observers 1 The relative-motion effects analysed here are not exclusive to the Einstein-
residing inside locally contracting bulk flows may assign negative
de Sitter background, but apply to essentially all Friedmann-Robertson-
values to their local deceleration parameter, while the host universe is Walker (FRW) backgrounds, irrespective of their spatial curvature and equa-
still globally decelerating. The apparent acceleration experienced by tion of state, including all those with matter that satisfies the strong-energy
these observers is not real, but a local artefact of their peculiar motion. condition Tsagas (2021a). Also note that the effects are purely general
Nevertheless, the affected scales are large enough (typically between relativistic in nature and cannot be naturally reproduced by a Newtonian
few hundred and several hundred Mpc) to create the false impres- study Tsagas et al. (2021).

MNRAS 000, 1–13 (2022)

 The Deceleration of a Tilted Universe 3
In the context of this analysis we raise and address the following spatial divergence of (1) and keeping up to first-order terms leads
two questions: to Maartens (1998)
• Can the deceleration parameter of a tilted cosmological model Θ̃ = Θ + 𝜃˜ , (2)
achieve a reasonable fit to the Pantheon data set, without the assump-
tion of a cosmological constant or of dark energy? where Θ = D𝑎 𝑢 𝑎 , Θ̃ = D̃𝑎 𝑢˜ 𝑎 and 𝜃˜ = D̃𝑎 𝜐˜ 𝑎 are the associated
• What are the favored values of the tilted parameters and how expansion/contraction scalars.4 Although the first two scalars are al-
does the data-fit of the tilted scenario compares to that of the standard ways positive due to the universal expansion, the third can be either
ΛCDM paradigm, in the context of model selection criteria? positive or negative (i.e. 𝜃˜ ≷ 0, with | 𝜃|/Θ
˜  1 during the linear
phase), since the bulk flow may locally expand or contract.5 Follow-
The structure of our paper is as follows: In the next section we ing (2), observers inside contracting bulk flows (where 𝜃˜ < 0) will
present the basic theory describing the linear kinematics of relatively measure a smaller expansion rate than their Hubble-flow counterparts
moving observers and provide the deceleration parameter measured (i.e. Θ̃ / Θ). Inside a locally expanding bulk flow (where 𝜃˜ > 0), on
by the tilted (i.e. the real) observers. Section 3 contains the statistical the other hand, the tilted observers will “experience" faster expansion
analysis and the corresponding quality of fit, employing the Pantheon (with Θ̃ ' Θ). Nevertheless, in either case, the relative-motion effect
data set. Finally, in Section 4, we outline our results and discuss ˜
should be too small to leave a measurable imprint, since | 𝜃|/Θ 1
possible extensions and future work. at the linear level.
According to (2), the expansion rates measured in the Hubble and
the tilted frames differ simply because of relative-motion effects.
For the same reason, the two observer groups experience different
2 FORM OF THE PECULIAR DECELERATION
acceleration/deceleration rates as well. Indeed, differentiating Eq. (1)
PARAMETER
with respect to time and keeping up to first-order terms, leads to the
Following Tsagas (2010, 2011); Tsagas & Kadiltzoglou (2015); linear relation
Tsagas (2021b) and adopting natural units (with ℏ = 𝑐 = 8 𝜋 𝐺 ≡ 1),
Θ̃ 0 = Θ
¤ + 𝜃˜0 , (3)
we consider a tilted cosmological model equipped with two families
of relatively moving observers. The first are the idealised (fictitious) where Θ ¤ and Θ̃ 0 monitor the time evolution of Θ and Θ̃ in their
observers, which follow the smooth Hubble expansion of a dust- respective rest frames Tsagas (2010).6 At this point we should note
dominated FRW universe with worldlines tangent to the timelike ˜
that, although | 𝜃|/Θ  1 all along the linear phase, this is not
4-velocity 𝑢 𝑎 (normalised so that 𝑢 𝑎 𝑢 𝑎 = −1).2 The second family necessarily true for the ratio of their time derivatives.
are the real (the so-called tilted) observers, residing in a typical galaxy The linear differences between the volume scalars and between
like our Milky Way and drifting relative to the reference 𝑢 𝑎 -frame their time derivatives, seen in (2) and (3), imply that the deceleration
with finite peculiar velocity. The latter is non-relativistic, as indi- parameters measured in the CMB and the tilted frames differ as
cated by the observations Hong et al. (2014); Boruah et al. (2020); well. Indeed, expressed in terms of their volume scalars and their
Qin (2021). Assuming that 𝑢˜ 𝑎 and 𝜐˜ 𝑎 are respectively the 4-velocity derivatives, the deceleration parameters measured in the coordinate
and the peculiar velocity of the tilted observers (with 𝑢˜ 𝑎 𝑢˜ 𝑎 = −1, system of the smooth Hubble flow and in rest-frame of the bulk
𝑢 𝑎 𝜐˜ 𝑎 = 0 and 𝜐˜ 2 = 𝜐˜ 𝑎 𝜐˜ 𝑎  1), the aforementioned three velocity peculiar motion read
fields are related by the reduced Lorentz boost 
3Θ¤
𝑢˜ 𝑎 = 𝑢 𝑎 + 𝜐˜ 𝑎 . (1) 𝑞 =− 1+ 2 (4)
Θ
Note that the peculiar velocity vanishes in the FRW background by and
default, which makes 𝜐˜ 𝑎 a gauge-invariant linear perturbation Tsagas
3Θ̃ 0
 
et al. (2007); Ellis et al. (2012).3 𝑞˜ = − 1 + , (5)
In what follows, we will refer to the reference coordinate system of Θ̃2
the ideal observers as the Hubble-frame, or the CMB-frame, while respectively. Combining these definitions with Eqs. (2) and (3), leads
to that of the tilted observers as the bulk-flow frame or the tilted- to the following linear relation between the two deceleration param-
frame. We will also adopt the relativistic 1 + 3 covariant approach eters Tsagas (2010, 2011); Tsagas & Kadiltzoglou (2015); Tsagas
to cosmology (see Ehlers (1961); Ellis (1973) and also Tsagas et al.
(2007); Ellis et al. (2012) for more recent reviews), where spacetime
splits into a temporal direction and into 3-dimensional space-like
4 Throughout this manuscript “tildas” will denote variables and operators
slices (along and orthogonal to the observers 4-velocity).
evaluated in the tilted frame of the real observers. Thus, the differential oper-
The mean kinematics of the relatively moving observers are de-
ators D𝑎 = ℎ𝑎 𝑏 ∇𝑏 and D̃𝑎 = ℎ˜ 𝑎 𝑏 ∇𝑏 , with ∇𝑎 representing the spacetime
termined by the volume scalars of their motion, monitoring the ex-
covariant derivative operator, indicate spatial covariant derivatives in the
pansion/contraction of the corresponding velocity fields. Taking the Hubble and the tilted frames respectively.
5 By construction, the expansion scalar is related to the average Hubble

parameter (𝐻 ) by means of the simple relation Θ = 3𝐻 , where 𝐻 = 𝑎/𝑎 ¤

2 Hereafter, we will treat the 𝑢𝑎 -field as the reference frame of the uni- and 𝑎 = 𝑎 (𝑡) is the cosmological scale factor Tsagas et al. (2007); Ellis et al.
verse, with respect to which large-scale peculiar velocities can be defined and (2012). Here, for convenience, we will also express the Hubble parameter in
measured. terms of redshift (i.e. we will also adopt the notation 𝐻 = 𝐻 (𝑧)). Finally,
3 If 𝑔 when referring to the current Hubble constant, we will use the notation 𝐻0 .
𝑎𝑏 represents the metric of the whole spacetime, the symmetric
spacelike tensors ℎ𝑎𝑏 = 𝑔𝑎𝑏 + 𝑢𝑎 𝑢𝑏 and ℎ˜ 𝑎𝑏 = 𝑔𝑎𝑏 + 𝑢˜ 𝑎 𝑢˜ 𝑏 (with 6 Hereafter, overdots will indicate time differentiation in the Hubble/CMB-

ℎ𝑎𝑏 𝑢 𝑏 = 0 = ℎ˜ 𝑎𝑏 𝑢˜ 𝑏 and ℎ𝑎 𝑎 = ℎ˜ 𝑎 𝑎 = 3) project orthogonally to frame, while primes will denote the same in the tilted coordinate system
𝑢𝑎 and to the 𝑢˜𝑎 fields respectively, while they also act as the metric tensors of the real observers. In other words, Θ ¤ = 𝑢 𝑎 ∇𝑎 Θ, Θ̃0 = 𝑢˜ 𝑎 ∇𝑎 Θ̃ and
of the corresponding 3-spaces Tsagas et al. (2007); Ellis et al. (2012). ˜ 0 ˜
𝜃 = 𝑢˜ ∇𝑎 𝜃 Tsagas (2010, 2011).
𝑎

MNRAS 000, 1–13 (2022)

4 K. Asvesta et al.
(2021b) for the 𝑛-th harmonic Tsagas (2021b); Tsagas et al. (2021). Here, 𝜆 (𝑛)
𝜃˜ 0 1
  0
Ω 𝜃˜ is the physical scale of the peculiar-velocity perturbation (practically
𝑞˜ = 𝑞 − = 𝑞 + 1 + , (6) the size of the bulk flow), 𝜆 𝐻 = 1/𝐻 is the Hubble radius (quantifying
3𝐻 2 3 2 𝐻¤
the size of the observable universe, with 𝜆 𝐻 ≈ 4300 Mpc today) and
guaranteeing that 𝑞˜ ≠ 𝑞 due to the presence of peculiar motions. 𝜆 𝐾 = 𝑎/|𝐾 | is the curvature scale of the FRW background (with
Note that, in deriving the above, we have also used the background 𝐾 = ±1). Since (𝜆 𝐻 /𝜆 𝐾 ) 2 = |1 − Ω| and given that |1 − Ω| ≤ 10−2 ,
expressions Θ = 3𝐻 and 𝐻¤ = −𝐻 2 (1 + Ω/2), with Ω = 𝜅𝜌/3𝐻 2 due to the near spatial flatness of our cosmos Aghanim et al. (2020),
representing the density parameter of an FRW universe. the last term of (9) is negligible. The latter then simplifies to
The second term on the right-hand side of (6) constitutes a cor-
1 𝜆 𝐻 2 𝜃˜
 
rection term introduced by relative-motion effects, since it vanishes
𝑞˜ = 𝑞 + , (10)
in the absence of them. In order to analyse Eq. (6) further, we need 9 𝜆 𝐻
an expression for the time evolution of 𝜃˜ written in the bulk-flow
after dropping the mode index (𝑛) for economy. The above provides
frame. Put another way, we need the Raychaudhuri equation of the
a very simple relation between 𝑞˜ and 𝑞. The former is the local de-
bulk peculiar motion. In a perturbed, tilted FRW universe with dust,
celeration parameter measured in the rest-frame of a bulk flow with
the latter linearises to Tsagas & Kadiltzoglou (2013)
0 0
size 𝜆. The latter is measured in the coordinate system of the smooth
𝜃˜ = −𝐻 𝜃˜ + D̃𝑎 𝜐˜ 𝑎 . (7) Hubble expansion and coincides with the deceleration parameter of
The source term in the above, namely the spatial divergence of the ˜
the universe itself. Note that, even though | 𝜃|/𝐻  1 throughout the
0
time derivative of the peculiar velocity (D̃𝑎 𝜐˜ 𝑎 ), is not yet directly linear regime, the effect of the (purely relativistic) correction term
observable and requires further theoretical analysis. We will therefore on the right-hand side of Eq. (10) can be strong, depending on the
turn to linear relativistic cosmological perturbation theory. scale of the bulk flow in question. In fact, the relative-motion effect
Relative motions also affect the nature of the cosmic medium on 𝑞˜ increases as we move down to smaller scales, where 𝜆 𝐻 /𝜆  1.
“seen” by the associated observers. Following Maartens (1998), the Furthermore, the overall effect also depends on the sign of 𝜃. ˜ This
linear relations between the matter components measured in the two means that in slightly expanding bulk flows (where 0 < 𝜃/𝐻 ˜  1)
frames are 𝜌˜ = 𝜌, 𝑝˜ = 𝑝, 𝑞˜ 𝑎 = 𝑞 − (𝜌 + 𝑝) 𝜐˜ 𝑎 and 𝜋˜ 𝑎𝑏 = 𝜋 𝑎𝑏 . the deceleration parameter measured in the tilted frame is greater
Here, 𝜌 is the energy density, 𝑝 is the (isotropic) pressure, 𝑞 𝑎 is the than its Hubble-frame counterpart. In contrast, the deceleration pa-
energy flux and 𝜋 𝑎𝑏 is the viscosity of the matter. Accordingly, when rameter measured by observers inside slightly contracting bulk flows
the Hubble-flow observers see the cosmic medium as a pressureless (i.e. those with −1  𝜃/𝐻 ˜  0) is smaller than the deceleration
perfect fluid (with 𝑝 = 0 = 𝑞 𝑎 = 𝜋 𝑎𝑏 ), the tilted see it as imperfect, parameter of the universe.
with an effective energy flux 𝑞˜ 𝑎 = −𝜌 𝜐˜ 𝑎 due to relative-motion Following (10), the local value of 𝑞˜ is more sensitive to the scale-
effects alone. ratio 𝜆 𝐻 /𝜆. This means that on super-Hubble lengths (with 𝜆 𝐻 /𝜆 
In contrast to Newtonian gravity, the energy flux also contributes 1) the impact of the observer’s relative motion is negligible and the
to the relativistic gravitational field, via the local energy-momentum local deceleration parameter approaches its global value (i.e. 𝑞˜ → 𝑞).
tensor. In a sense, one may say that, in relativity, bulk flows grav- This also agrees with our expectation that the peculiar velocities
itate Tsaprazi & Tsagas (2020); Filippou & Tsagas (2021). When and their effects fade away as we move out to progressively longer
dealing with bulk peculiar flows, where there is a nonzero 𝑞˜ 𝑎 due to wavelengths.
the observers relative motion, this additional flux-contribution to the The situation changes drastically on sufficiently small scales,
local gravitational field acquires particular significance. where 𝜆 𝐻 /𝜆  1. There, even relatively slow bulk flows (with
The extra input of the bulk-flow flux to the Einstein field equations ˜
| 𝜃|/𝐻  1) can have a significant impact on the local deceleration
feeds into the relativistic conservation laws and eventually emerges parameter. In fact, the relative-motion effects dominate the right-
in the equations monitoring the linear evolution of the tilted cosmo- hand side of (10) when the correction term equals (in absolute value)
logical model. More specifically, in a tilted almost-FRW universe the deceleration parameter (𝑞) measured in the Hubble frame. This
with dust, expression (7) recasts into Tsagas (2021b); Tsagas et al. occurs at a characteristic length given by Tsagas (2021b)
(2021) √︄
˜
1 | 𝜃|
0
!
0 1 1 Δ̃ Z̃ 𝜆𝑇 = 𝜆𝐻 . (11)
2 9𝑞 𝐻
𝜃˜ = −2𝐻 𝜃˜ + D̃ 𝜃˜ − 2 + . (8)
3𝐻 𝑎 3𝐻 3𝐻
On scales smaller than 𝜆𝑇 , peculiar-velocity perturbations dominate
Note that Δ describes inhomogeneities in the spatial distribution of over the background Hubble expansion, dictating the linear bulk-
the matter density and Z represents spatial variations in the universal flow kinematics and determining the local value of the deceleration
expansion.7 For more details and further discussion, the reader is parameter.8 The latter can even turn negative when the bulk flow is
referred to Tsagas (2021b); Tsagas et al. (2021). contracting. In that case, 𝜆𝑇 also marks the “transition scale” where
Before substituting (8) into Eq. (6), it is worth noting the spatial the sign of 𝑞˜ changes from positive to negative Tsagas (2021a,b).
Laplacian term on the right-hand side of (8). The latter introduces a Using data from recent bulk-flow surveys, the typical values of 𝜆𝑇
scale-dependence, which becomes explicit after a simple harmonic were found to vary between few hundred and several hundred Mpc
decomposition. In so doing, one arrives at the following expression (see Tsagas (2021b) for a representative Table).
0
1 𝜆 𝐻 2 𝜃˜ (𝑛) 1 𝜆 𝐻 2 © Δ̃ (𝑛) Z̃(𝑛) ª
   
𝑞˜ (𝑛) = 𝑞 + + ­ + ®, (9)
9 𝜆 (𝑛) 𝐻 9 𝜆𝐾 𝐻 𝐻 8 It is worth pointing out the close analogy between 𝜆𝑇 and the familiar
« ¬
“Jeans length” (𝜆 𝐽 ). Recall that the latter determines the threshold inside
which pressure-gradient perturbations dominate over the background grav-
7 By construction Δ̃ = 𝑎D̃𝑎 Δ̃𝑎 , with Δ̃𝑎 = (𝑎/𝜌) D̃𝑎 𝜌. The former closely itational pull and thus dictate the linear evolution of density perturbations.
corresponds to the familiar density contrast 𝛿 = 𝛿𝜌/𝜌 Tsagas et al. (2007); In this respect, 𝜆𝑇 can be seen as the peculiar-motion analogue of the Jeans
Ellis et al. (2012). Also by construction, Z̃ = 𝑎D̃𝑎 Z̃𝑎 , with Z̃ = 𝑎D̃𝑎 Θ̃. length, or as the “peculiar Jeans length” Tsagas (2021b).

MNRAS 000, 1–13 (2022)

 The Deceleration of a Tilted Universe 5
Focusing on contracting bulk flows (with 𝜃˜ < 0), Eqs. (10) and backgrounds, irrespective of their spatial curvature and equation of
(11) combine to give Tsagas et al. (2021) state (see Tsagas et al. (2021) and also footnote 1 here).
"  #
𝜆𝑇 2

𝑞˜ = 𝑞 1 − , (12)
𝜆
3 PANTHEON SNIA ANALYSIS
guaranteeing that 𝑞˜ < 0 on scales smaller than the transition length
In order to constrain the parameters 𝑚, 𝑝 and 𝑟 that appear in the tilted
(with 𝜆 < 𝜆𝑇 ). Accordingly, observers inside (slightly) contracting
deceleration parametrization (14), we use the latest publicly avail-
peculiar motions will assign negative values to their local decelera-
able SnIa sample, namely the Pantheon sample consisting of 1048
tion parameter within a region determined by the associated transition
spectroscopically confirmed SnIa originally compiled by Scolnic
length. On scales larger than 𝜆𝑇 , on the other hand, the same ob-
et al. (2018b).9
servers will assign positive values to 𝑞. ˜ Clearly, on sufficiently large
wavelengths, the correction term on the rght-hand side of (12) be-
comes negligible and the value of the local deceleration parameter
approaches that of its Hubble-flow counterpart (i.e. 𝑞˜ → 𝑞). 3.1 The Pantheon Supernovae Sample
It is therefore theoretically possible to achieve accelerated expan- The Pantheon sample corresponds to a compilation of different super-
sion in tilted almost-FRW universes without appealing to dark en- novae surveys detecting SnIa in both high and low redshifts regions.
ergy, or to a cosmological constant. There is also no need to modify In particular, it incorporates the CfA1-CfA4 Riess et al. (1999);
general relativity, to abandon the Friedmann models, or to introduce Jha et al. (2006); Hicken et al. (2009a,b, 2012) surveys, the Pan-
any new physics. Everything takes place within standard cosmology STARRS1 (PS1) Medium Deep Survey Scolnic et al. (2018b), the
and conventional physics. One only needs to allow for large-scale Sloan Digital Sky Survey (SDSS) Sako et al. (2018), the SuperNovae
peculiar motions, like those reported by an ever increasing number Legacy Survey (SNLS) Guy et al. (2010), ESSENCE Narayan et al.
of surveys. (2016), the Carnegie Supernova Project (CSP) Contreras et al. (2010)
Clearly, the accelerated expansion “experienced” by the bulk-flow as well as various Hubble Space Telescope (HST) samples, namely
observers is not real, but a local artefact of their peculiar motion the CANDELS/CLASH Graur et al. (2014); Rodney et al. (2014);
relative to the smooth Hubble expansion. Globally, the host universe Riess et al. (2018), GOODS Riess et al. (2004b, 2007) and SnIa
is still decelerating with 𝑞 > 0. Nevertheless, the affected scales form the Supernova Cosmology project (SCP) Suzuki et al. (2012)
are large enough (𝜆𝑇 typically ranges from few hundred to several subsamples. A summary of these surveys is illustrated in the follow-
hundred Mpc) to create the false impression of a recent global event. ing Table 1, where we also present the redshift range that each survey
Put another way, the unsuspecting observers have misinterpreted the covers as well as the total number of SnIa (𝑁 𝑆𝑛𝐼 𝑎 ). As a result the
local contraction of the bulk flow they happen to reside in, as global entire sample covers the redshift range 0.01 ≤ 𝑧 ≤ 2.26. The sky
acceleration of the surrounding universe. distribution (in galactic coordinates) of the SnIa is demonstrated in
Looking back at Eqs. (10) and (11), one can easily see the key Fig. 1 where we have colour-coded each SnIa with respect to the
role played by the local volume scalar (𝜃) ˜ in determining the values redshift, while the redshift distribution of the entire sample is shown
of both 𝑞˜ and 𝜆𝑇 . For simplicity, we have so far implicitly assumed in Fig. 2. Notice that the sky distribution of the Pantheon sample
that 𝜃˜ remains constant within the whole of the bulk-flow domain. In is quite inhomogeneous, since the majority pf the SnIa lies in the
practice, however, one expects the local volume scalar to vary with “southeast" quadrisphere.
scale. We therefore need to consider physically motivated profiles for The Pantheon sample contains significant improvements on the
𝜃˜ and then test their fit to the observations. 𝑃𝑆1 photometry to reduce the photometric calibration uncertainties
Given that peculiar velocities fade away with increasing scale, we to the milli-magnitude (mmag) level. The light curves of SnIa are
expect the value of 𝜃˜ to decrease as we move out to progressively parametrized by their brightness, observed colour and their decline
larger wavelength (i.e. 𝜃˜ → 0 as 𝜆 → ∞). Also, when dealing with rate. In Scolnic et al. (2018b), a recent version of light curve fitter is
contracting bulk motions, we expect faster contraction rates near used, namely the Spectral Adaptive Lightcurve Template - 2 (SALT2,
the outskirts of the flow and slower towards its center. Physically, see Ref. Betoule et al. (2014b) for more details) in order to optimize
this is a well motivated pattern, since it agrees with the typical the use of SnIa as standard candles to determine accurate distances
kinematic behaviour of any contracting self-gravitating system. The and to reduce systematic uncertainties due to K-corrections. Given
above described qualitative profile of 𝜃˜ is well parametrised by the the Tripp estimator Tripp (1998) together with the light curve fitter,
functional form the standardized observational SnIa distance modulus is given by
𝑚 𝜆2 𝜇𝑜𝑏𝑠 = 𝑚 𝑏 − 𝑀 + 𝛼 𝑥1 − 𝛽 𝑐 + Δ 𝑀 + Δ 𝐵 , (15)
𝜃˜ = 𝜃˜ (𝜆) = , (13)
𝑝 + 𝑟 𝜆3
where 𝑚 𝑏 corresponds to the observed peak magnitude at time of
where 𝑚, 𝑝 and 𝑟 are parameters decided by the data. Finally, given B-band maximum, 𝑀 is the absolute B-band magnitude of a fidu-
that the background Hubble parameter remains essentially constant cial SnIa, 𝑥1 is a light curve shape parameter, 𝑐 describes the SnIa
(i.e. 𝐻 ' 𝐻0 ) within the scales/redshifts of interest, Eqs. (13) and colour at maximum brightness, 𝛼 is a coefficient of the relation be-
(12) lead to the following functional form tween luminosity and stretch and 𝛽 is a coefficient of the relation
1

𝑚
 between luminosity and colour. Moreover, Δ 𝑀 and Δ 𝐵 are distance
𝑞˜ = 𝑞(𝜆)
˜ = 1− , (14) corrections based on the mass of the host galaxy and on predicted
2 𝑝 + 𝑟 𝜆3
for the deceleration parameter measured by the real observers. Note
that we have set 𝑞 = 1/2 in the above, which implies that expression 9 Actually, the latest SnIa sample corresponds to the Pantheon+ compilation
(14) holds on an Einstein-de Sitter background. Having said that, we Riess et al. (2021a). However, during the writing of the present work the data
remind the reader that the above result applies to essentially all FRW are not publicly available.

MNRAS 000, 1–13 (2022)

6 K. Asvesta et al.

Table 1. Summary of Sn1a subsamples of the Pantheon dataset. The column 𝑁𝑆𝑛𝐼 𝑎 includes the total number of SnIa of every sample and the two last columns
indicate the median value of the CMB redshift and its coverage for each sample respectively. The corresponding 𝑁𝑆𝑛𝐼 𝑎 number were identified using the
idsurvey column from the Ancillary_C11.FITRES file of the Pantheon data in the corresponding github repository.

Sample 𝑁𝑆𝑛𝐼 𝑎 Redshift Median Redshift Range

CfA 1-4 147 0.03 0.01 - 0.07

CSP 25 0.02 0.01 - 0.06
SDSS 335 0.20 0.03 - 0.40
SNLS 236 0.64 0.12 - 1.06
PS1 279 0.29 0.02 - 0.63
high-𝑧 26 1.26 0.73 - 2.26

all 1048 0.25 0.01 - 2.26

Figure 1. The sky distribution of the Pantheon SnIa sample in galactic coordinates classified by redshifts. The pseudo-colours indicate the supernovae redshifts
with respect to the CMB frame.

biases from simulations, respectively. As we can see, there are dif- 3.2 Pantheon Dataset Fit
ferent sources of systematic uncertainties that can yield to inaccurate
The SnIa are widely used as standard candles to probe the expan-
distance estimations. A distance-dependent bias that mostly affects
sion rate, by utilizing the theoretically predicted apparent magnitude
the SnIa at cosmological redshift is associated with the Malmquist
𝑚 𝑏 (𝑧). The latter reads
bias, where at redshifts near the survey magnitude threshold, brighter
SnIa are most likely to be observed, biasing the effective luminosity  
towards higher values. In Scolnic et al. (2018b), the authors used the 𝑑 𝐿 (𝑧)
𝑚 𝑏 (𝑧) = 𝑀 + 5 𝑙𝑜𝑔10 + 25 , (16)
BEAMS with Bias Corrections (BBC) method Kessler & Scolnic 1𝑀 𝑝𝑐
(2017) to account for errors due to intrinsic scatter and selection
effects (such as the the Malmquist bias), based on accurate SnIa where 𝑀 is the well known corrected intrinsic (absolute) magnitude
simulations. with respect to the colour and stretch. Also, 𝑑 𝐿 (𝑧) corresponds to
the luminosity distance, which in the context of a flat universe is
After all these corrections are applied, one can infer the corrected calculated by
apparent magnitudes 𝑚 𝑜𝑏𝑠 and perform a cosmological fit based on
a specific theoretical model. This approach is implemented in the 𝑑𝑧 0
∫ 𝑧
next subsection. 𝑑 𝐿 (𝑧) = 𝑐 (1 + 𝑧) 0 , (17)
0 𝐻 (𝑧 )

MNRAS 000, 1–13 (2022)

 The Deceleration of a Tilted Universe 7

Figure 2. The redshift distribution of the Pantheon SnIa subsamples, indicated in the legend, in the CMB frame. The blue colour indicates the number of SnIa
in the CfA 1-4 and CSP samples, while with purple the SnIa from the CANDELS/CLASH, GOODS and SCP subsamples are shown.

where 𝑧 denotes the SnIa redshift in the CMB rest frame and 𝑐 is the between the observed SnIa apparent magnitudes at redshift 𝑧 𝑖 with
speed of light. Typically, instead of the luminosity distance (𝑑 𝐿 ), the the theoretically predicted ones calculated from Eq. (18) and 𝐶𝑖−1 𝑗
Hubble free luminosity distance (𝐷 𝐿 (𝑧) ≡ 𝐻0 𝑑 𝐿 (𝑧)/𝑐) is used for is the inverse of the total covariance matrix. The total covariance
the theoretically predicted apparent magnitude, recasting Eq. (16) as matrix is constructed by taking the sum of a diagonal matrix 𝐷 stat
that includes the statistical uncertainties of the apparent magnitudes
𝑚 𝑜𝑏𝑠 (𝑧 𝑖 ) and a non diagonal matrix that is constructed using the
 
𝑐/𝐻0
𝑚 𝑏 (𝑧) = 𝑀 + 5 𝑙𝑜𝑔10 [𝐷 𝐿 (𝑧)] + 5 𝑙𝑜𝑔10 + 25 . (18) systematic uncertainties due to the bias correction method (see Ref.
𝑀 𝑝𝑐
Scolnic et al. (2018a) for more details). The diagonal matrix includes
From Eq. (18) we clearly see a degeneracy between the parameters
the total distance errors associated with every SnIa and takes the form
𝑀 an 𝐻0 , which in the context of a ΛCDM background 𝐻 (𝑧) is
considered constant. As a result, the two parameters are combined 𝜎2 0 ... 0
© 𝑚𝑜𝑏𝑠,1
for the definition of the parameter M, which is determined as 2 ª
­ 0 𝜎𝑚 𝑜𝑏𝑠,2
... 0 ®
𝐷 stat = ­ .
­ ®
­ .. .. .. .. ®

𝑐/𝐻0

­ . . . ®
®
M ≡ 𝑀 + 5 𝑙𝑜𝑔10 + 25 = 𝑀 − 5 𝑙𝑜𝑔10 (ℎ) + 42.38 , (19) 2
1𝑀 𝑝𝑐 « 0 0 ... 𝜎𝑚 𝑜𝑏𝑠, 𝑁 ¬

where 𝐻0 = 100ℎ 𝑘𝑚 𝑠𝑒𝑐−1 Mpc−1 . Even though this degenerate where 𝜎𝑚 2

𝑜𝑏𝑠
= 𝜎𝑀 2 + 𝜎2 2 2 2 2
𝑀 𝑎𝑠𝑠 + 𝜎𝜇−𝑧 + 𝜎𝑙𝑒𝑛𝑠 + 𝜎𝑖𝑛𝑡 + 𝜎𝐵𝑖𝑎𝑠 . The
combination is marginalized in most cases Conley et al. (2011); individual uncertainty contributions to the matrix are the photometric
Scolnic et al. (2018a), recent studies Zhao et al. (2019); Kazantzidis error, the mass step correction, the peculiar velocity and redshift
& Perivolaropoulos (2020); Sapone et al. (2021); Kazantzidis et al. uncertainty in quadrature, the stochastic gravitational redshift, the
(2021); Dainotti et al. (2021, 2022) argue that this process may intrinsic scatter and the distance bias correction respectively. From
lead to physical information loss, since a physical model with an the definition of the luminosity distance, it is evident that in order
abrupt transition on the absolute magnitude 𝑀 at a low redshift to apply the maximum likelihood method we need a specific form
𝑧 𝑡 has the potential to alleviate simultaneously the 𝐻0 and growth for 𝐻 (𝑧). From the deceleration parameter, we can construct the
tensions Alestas et al. (2021c); Camarena & Marra (2021); Marra corresponding evolution of 𝐻 (𝑧) via Gong & Wang (2007)
& Perivolaropoulos (2021); Alestas et al. (2021b); Perivolaropoulos ∫ 𝑧   
1 + 𝑞(𝑢)
& Skara (2021b); Alestas et al. (2021a, 2022); Perivolaropoulos 𝐻 (𝑧) = 𝐻0 exp 𝑑𝑢 . (21)
(2022). Hence, we decide to use this degenerate parameter in the 0 1+𝑢
minimization process. Hence, for a specific form of the deceleration parameter, we can easily
Armed with Eqs. (16)-(19), we can apply the maximum likelihood construct the theoretically predicted apparent magnitude 𝑚 𝑏 (𝑧) and
method Arjona et al. (2019), and construct the relevant 𝜒2 function as a result the corresponding 𝜒2 function, solving Eq. (21) for a
as follows specific 𝑞(𝑧) parametrization and substituting the derived formula of
𝜒2 = 𝑉𝑃𝑎𝑛𝑡
𝑖 −1 𝑗
(20) 𝐻 (𝑧) to the luminosity distance 𝑑 𝐿 (𝑧) of Eq. (17).
ℎ. 𝐶𝑖 𝑗 𝑉𝑃𝑎𝑛𝑡 ℎ. , However, our parametrization (14) depends on the physical size
𝑖
where 𝑉𝑃𝑎𝑛𝑡 ≡ 𝑚 𝑜𝑏𝑠 (𝑧 𝑖 ) − 𝑚 𝑏 (𝑧) corresponds to the difference of the bulk flow 𝜆 and can not be used on the current form. So,
ℎ.

MNRAS 000, 1–13 (2022)

8 K. Asvesta et al.

   
1 1 1 1
Figure 3. The evolution of 𝑞˜ (𝜆(𝑧)) = 2 1− 3 (at the left panel) and 𝑞˜ (𝜆(𝑧)) = 2 1− 3 (at the right panel) by fitting it to the full
𝛼+𝑏 𝜒¯ 𝐸 𝑑𝑆
𝛼+𝑏 𝜒¯ Λ𝐶𝐷 𝑀
Pantheon dataset. The red solid line is drawn by using the best fit parameters for each
 model (see Table 2). For comparison, the dashed blue line corresponds to
the best fit ΛCDM which has a deceleration parameter of the form 𝑞Λ𝐶𝐷 𝑀 (𝑧) = Ω0𝑚 (1 + 𝑧 3 ) − 2(1 − Ω0𝑚 ) / 2(Ω0𝑚 (1 + 𝑧) 3 + 1 − Ω0𝑚 ) .
 

Table 2. Table of the best fit parameters for the two tilted cosmological models T-Λ and T−𝐸 𝑑𝑆 as well as the standard ΛCDM scenario. Notice that with the
acronyms T-Λ and T−𝐸 𝑑𝑆 we refer to a tilted cosmological model with a ΛCDM or an EdS line-of-sight comoving distance respectively. The tilted cosmological
2
models have been fitted using Eqs. (16)-(21). The three models in comparison giving similar 𝜒𝑚𝑖𝑛 values, using the Pantheon data.

Model M 𝛼 𝑏 Ω0𝑚 2
𝜒min 2
𝜒red

𝚲CDM 23.809 ± 0.011 − − 0.299 ± 0.022 1026.67 0.981

T-Λ 23.815+0.014
−0.012
0.517+0.039
−0.038
3.9+3.6
−2.4
0.3 1026.69 0.982
T-Λ (𝛼 fixed) 23.808 ± 0.007 0.5 +2.6
5.20−1.9 0.3 1027.21 0.982
T-EdS 23.813+0.015
−0.014
0.512 ± 0.041 6.7+5.6
−3.8
1.0 1026.76 0.982
T-EdS (𝛼 fixed) 23.809 ± 0.007 0.5 +3.8
8.56−2.9 1.0 1027.05 0.982

in order to transform the deceleration parameter 𝑞˜ to a redshift de- the deceleration parameter measured in the tilted frame approaches
pendent function, we set 𝜆 ≡ 𝜒(𝑧),
¯ where 𝜒(𝑧)
¯ corresponds to the its value in the CMB frame. This behaviour is expected since, as
line-of-sight comoving distance. In the present analysis we choose we discussed in the Introduction, the peculiar velocities and their
two different cosmologies for the line-of-sight comoving distance. effects fade away on large wavelengths, assuming that the universe
The standard ΛCDM cosmology, where the line-of-sight comoving approaches an exact FRW model on large scales. This is illustrated
distance is given as in Fig. 3, where we show the deceleration parameter 𝑞(𝜆(𝑧))
˜ for
both the ΛCDM (right panel) and the EdS (left fpanel) line-of-sight
𝑐 𝑑𝑧 0
∫ 𝑧
𝜒¯ Λ𝐶𝐷 𝑀 (𝑧) = √︁ , (22) comoving distance as a function of the redhsift 𝑧, superimposed with
0 𝐻0 Ω0𝑚 (1 + 𝑧 0 ) 3 + (1 − Ω0𝑚 ) the standard ΛCDM scenario.
as well as the Einstein-de Sitter (EdS) form for the line-of-sight
comoving distance (Λ = 0) which is defined as
  3.3 Numerical Results
2𝑐 1
𝜒¯ 𝐸 𝑑𝑆 (𝑧) = 1− √ . (23) Now we are ready to apply the maximum likelihood method utilizing
𝐻0 1+𝑧 the aforementioned equations. We construct two separate codes, one
Moreover, we can further reduce the the number of parameters in written in Python and one written in Mathematica, and apply the
Eq. (14) rearranging the constants 𝑚, 𝑝 and 𝑟 as follows maximum likelihood method for the standard ΛCDM case as well
1

1
 as for the deceleration parameter of the tilted cosmological model
˜
𝑞(𝜆(𝑧)) = 1− , (24) assuming a ΛCDM and an EdS line-of-sight comoving distance 𝜒(𝑧). ¯
2 𝛼 + 𝑏 𝑑𝑟3 (𝑧) In this subsection, we present only the derived results of the Python
where 𝛼, 𝑏 are dimensionless parameters of the tilted cosmological code, however the two publicly available codes are fully consistent
models to be fixed by the data and 𝑑𝑟 (𝑧) ≡ 𝐻0 𝜒(𝑧)/𝑐.
¯ Recall that with each other. In particular, for the standard ΛCDM scenario we
we have set 𝜆 ≡ 𝜒(𝑧)
¯ and depending on the bulk flow model, 𝜒(𝑧)
¯ obtain M = 23.809 ± 0.011 and Ω0𝑚 = 0.299 ± 0.022 (see the first
is given either by Eq. (22) or by Eq. (23). Note also that in Eq. row of Table 2) in agreement with previous studies Scolnic et al.
(24), at early times (𝑧 ≫ 1), 𝑞(𝜆(𝑧))
˜ → 1/2, which means that (2018a); Zhao et al. (2019); Kazantzidis & Perivolaropoulos (2020);

MNRAS 000, 1–13 (2022)

 The Deceleration of a Tilted Universe 9

Figure 4. One-dimensional and two-dimensional posterior distributions on the parameters 𝛼, 𝑏 and M of the parametrization (24) using the EdS line-of-sight
comoving distance (23). The shaded area of the histograms shows the 68% error on the parameters. The contours represent the 68% and 95% confidence levels.
The histograms have been smoothed by applying the Gaussian Kernel Density Estimation (KDE) feature from the ChainConsumer package. The dashed lines
denote the best fit values of the parameters of the model (maximum likelihood method).

Kazantzidis et al. (2021). The maximum likelihood method for this scenario are illustrated in Table 2. In Table 2, we also include the
case is applied using the corresponding expansion rate which is ratio 𝜒𝑟2𝑒𝑑 ≡ 𝜒min
2 /𝑑𝑜 𝑓 , where 𝑑𝑜 𝑓 corresponds to the degrees of
h i freedom and in order to achieve a good fit to the Pantheon data this
𝐻 2 (𝑧) = 𝐻02 Ω0𝑚 (1 + 𝑧) 3 + (1 − Ω0𝑚 ) . (25) ratio needs to be lower than unity.
Then, we use Eq. (25) to compose the theoretically predicted appar- For the determination of the best fit parameters, we minimize
ent magnitude 𝑚 𝑏 (𝑧) through the Hubble free luminosity distance Eq. (20) using the Python data fitting library lmfit Newville et al.
𝑖
𝐷 𝐿 (𝑧) and substitute it in the 𝑉𝑃𝑎𝑛𝑡 ℎ.
deriving as a result the relevant (2014). Also, Fig. 3 is produced with matplotlib Hunter (2007).
2
𝜒min which is minimized in the context of the maximum likelihood For the construction of the posterior probability distributions of the
method. Equivalently one can start directly from Eq. (21), substitut- parameters and the contours that are shown above, we apply an
ing the form of the deceleration parameter for the standard ΛCDM algorithm relying on the open-source Python package emcee, an
case which is implementantion of the Affine-Invariant MCMC Ensemble sampler
by Foreman-Mackey et al. (2013). All the plots are generated using
Ω0𝑚 (1 + 𝑧 3 ) − 2(1 − Ω0𝑚 )
 
𝑞 Λ𝐶𝐷 𝑀 (𝑧) =   , (26) ChainConsumer Hinton (2016b); Hinton (2016a), which analyzes
2(Ω0𝑚 (1 + 𝑧) 3 + 1 − Ω0𝑚 ) the chains and produces plots of the posterior inferred from the chain
and solving the integral obtain the corresponding evolution of the distributions. For all the models of this study we use 100 random
expansion rate (25). chains (walkers) and 2000 iterations (steps) for our MCMC analysis.
In order to distinguish between each case for the tilted cosmo- We choose flat priors for all parameters, allowing the parameters
logical models, we denote the 𝑞(𝜆(𝑧))
˜ parametrization assuming a to vary in ranges : 0.1 < 𝛼 < 0.9, 0 < 𝑏 < 35, 23 < M < 24.
ΛCDM background as Tilted-Λ (T-Λ) while the 𝑞(𝜆(𝑧))
˜ parametriza- Furthermore, we construct the contours which correspond to the
tion assuming an EdS background as Tilted-𝐸 𝑑𝑆 (T-𝐸 𝑑𝑆) where 1𝜎 − 2𝜎 confidence levels for the T-𝐸 𝑑𝑆 parametrization in Fig. 4,
˜
𝑞(𝜆(𝑧)) is given by Eq. (24). The results of the maximum likelihood while we show the same contours for the T-Λ parametrization case
method, i.e. the best fit values and the corresponding 𝜒min2 values in Fig. 5.
of the two tilted cosmological models as well as that of the ΛCDM From Table 2 it is clear that, not only the two tilted cosmological

MNRAS 000, 1–13 (2022)

10 K. Asvesta et al.

Figure 5. Same as Fig. 4, but for a ΛCDM line-of-sight comoving distance (22).

models perform equally well (provide a similar 𝜒red 2 value) with the (i.e. well within the 1𝜎 threshold), with the derived results where the
standard ΛCDM scenario, but they also have the additional advantage parameter 𝛼 is free to vary. The results are also presented in Table 2.
of not suffering from the fine tuning problem as the standard ΛCDM In order to identify the optimal model, we need to take into account
paradigm. Moreover, we can see that the form of the line-of-sight not only the quality of the provided fit 𝜒𝑟2𝑒𝑑 of Table 2, but we also
comoving distance 𝜒(𝑧)¯ does not affect the quality of fit to the data, need to consider the number of free parameters of each model used
since the errors of the parameter 𝑏 are quite large. Notice that, in the to obtain the particular 𝜒𝑟2𝑒𝑑 value. Even though, the best choice
case of the tilted ΛCDM cosmological model, the acceleration that between the different information criteria that have been presented
the observer inside the bulk flow measures is only a local effect and in the literature is not straightforward Liddle (2004) we use the
happens due to the impact of peculiar motions and not due to the most popular ones. The Akaike Information Criterion (AIC) Akaike
presence of the cosmological constant. From Fig. 3, it is also evident (1974); Liddle (2004); Nesseris & Garcia-Bellido (2013) defined as
that the transition from a decelerated to an accelerated phase occurs
around 𝑧 ≈ 0.6, i.e. close to previous studies which assume ΛCDM 2
𝐴𝐼𝐶 ≡ −2 ln Lmax + 2 𝑝 tot = 𝜒min + 2 𝑝 tot (27)
model Riess et al. (2004a); Turner & Riess (2002).
where 𝑝 tot corresponds to the total number of free parameters of
the considered model and Lmax corresponds to the maximum like-
This transition is determined by the parameter 𝑏 and is directly lihood of the model under consideration. Also, we implement the
connected to the scale of the bulk flow. On the contrary, the param- Bayesian Information Criterion (BIC) which was introduced by
eter 𝛼 determines the current value of the deceleration parameter as Schwarz (1978); Liddle (2004); Nesseris & Garcia-Bellido (2013)
measured by an observer at the center of the bulk flow, since for 𝑧 = 0 and is defined as
we obtain 𝑞(𝜆(𝑧
˜ = 0)) ≡ 𝑞˜ 0 = 21 (1 − 𝛼−1 ). For consistency with
𝐵𝐼𝐶 ≡ −2 ln Lmax + 𝑝 tot 𝑙𝑛(𝑁tot ) (28)
current measurements which report 𝑞˜ 0 ≈ −0.5 we can assume that
the parameter 𝛼 takes the generic value, 𝛼 = 1/2, thus reducing the Using the definitions (27) and (28) we construct the differ-
total number of parameters of the two tilted cosmological models. ences Δ𝐴𝐼𝐶 and Δ𝐵𝐼𝐶 of the models in question with respect to
Applying the maximum likelihood method, we can derive the best fit ΛCDM. According to the calibrated Jeffreys’ scales Jeffreys (1961),
and the corresponding 𝜒min2 values for this case that are very similar if 0 < |Δ𝐴𝐼𝐶 | ≤ 2, then the confronted models can be interpreted

MNRAS 000, 1–13 (2022)

 The Deceleration of a Tilted Universe 11

2 along with the corresponding 𝐴𝐼 𝐶, 𝐵𝐼 𝐶 values and the differences Δ𝐴𝐼 𝐶, Δ𝐵𝐼 𝐶 for the three cosmological models in
Table 3. The goodness-of-fit 𝜒min
question using the full Pantheon dataset.

Model 2
𝜒min 2
𝜒red 𝐴𝐼 𝐶 𝐵𝐼 𝐶 Δ𝐴𝐼 𝐶 Δ𝐵𝐼 𝐶

𝚲CDM 1026.67 0.981 1030.67 1040.58 − −

T-Λ 1026.69 0.982 1032.69 1047.55 2.02 6.97
T-Λ (𝛼 fixed) 1027.21 0.982 1031.21 1041.12 0.54 0.54
T-EdS 1026.76 0.982 1032.76 1047.62 2.09 7.04
T-EdS (𝛼 fixed) 1027.05 0.982 1031.05 1040.96 0.38 0.38

as consistent with each other, while if |Δ𝐴𝐼𝐶 | ≥ 4 it is an indication hood method for the two parametrizations. Our findings are summa-
that the model with the larger AIC value is disfavored by the data. rized in Table 2. It is important to note that, in the case of the tilted
Similarly, if 0 < |Δ𝐵𝐼𝐶 | ≤ 2 then the model with the larger BIC ΛCDM cosmological model, the local accelerated expansion that the
value is weakly disfavored by the data, while for 2 < |Δ𝐵𝐼𝐶 | ≤ 6 bulk flow observer measures, is mainly due to the peculiar motion
(|Δ𝐵𝐼𝐶 | > 6) the model with the larger BIC values is strongly (very relative to the Hubble flow and not due to the cosmological constant.
strongly) disfavored. The specific differences of the studied cosmo- Comparing them to the standard ΛCDM paradigm, we found that the
logical models are shown in Table 3. three different models perform equally well (provide a similar 𝜒red 2
According to the AIC, if 𝛼 is a free parameter, the two tilted value, as indicated in the last column of Table 2) and that the form of
cosmological models T-Λ and T-𝐸 𝑑𝑆 seem to be consistent with the comoving line-of-sight distance 𝜒(𝑧) ¯ does not affect significantly
ΛCDM. On the contrary, according to the BIC which penalizes more the derived quality of fit. However, the two tilted cosmological mod-
harshly any extra degrees of freedom, the two tilted cosmological els have an additional parameter compared to the standard scenario.
models seem to be strongly disfavored. However, if we fix 𝛼 to the Taking into account appropriate statistical criteria such as the Akaike
generic value 𝛼 = 1/2, the two cosmological models give Δ𝐴𝐼𝐶 and information criterion (AIC) and the Bayesian information criterion
Δ𝐵𝐼𝐶 significantly lower than unity, displaying that they are equally (BIC), we demonstrated that the three models are equally consistent
supported by the Pantheon sample as the standard ΛCDM model according to AIC (see the sixth column of Table 3). By contrast, BIC
does. clearly favours the ΛCDM scenario (see the last column of Table 3),
because it has one less degree of freedom.
Nevertheless, although the ΛCDM model is still the leading
cosmological paradigm, it faces a number of fundamental chal-
4 DISCUSSION AND CONCLUSIONS lenges Perivolaropoulos & Skara (2021a), which are not accounted
We have introduced and studied two novel parametrizations of the for by the aforementioned statistical criteria. Most importantly, the
deceleration parameter in the context of a tilted universe, with two ΛCDM scenario does not provide a physical explanation neither for
families of relatively moving observers. The first family are the ide- presence and the nature of the vacuum energy, nor for the fine tuning
alised observers following the smooth universal expansion, whereas of its value. In stark contrast, tilted cosmologies, even when applied to
the second are the tilted observers located inside a bulk flow that a simple Einstein-de Sitter background, can reproduce the observed
moves relative to the Hubble expansion with finite peculiar velocity. acceleration history of the universe naturally. In particular, applying
Due to their relative motion, the two observers assign different val- the Pantheon data to our tilted Einstein-de Sitter model, recovered
ues to their deceleration parameters. In fact, observers living inside both the early deceleration and the late acceleration phases of the
locally contracting bulk flows can measure negative deceleration pa- universe (see Fig. 3). This happened naturally, without appealing to
rameter, while the host universe is globally decelerating. Although exotic forms of matter, or introducing a cosmological constant and
the accelerating effect is a local artefact of the observers peculiar without any fine-tuning or coincidence problems. The tilted scenario,
motion, the affected scales can be large enough to create the false with its simple mathematical manifestation, can explain the recent
impression of recent global acceleration. accelerated expansion of the universe by accounting for the con-
The value of the locally measured deceleration parameter, as well sequences of the bulk peculiar motions, which dominate the linear
as the scale where its sign changes from positive to negative, namely kinematics of the local Universe.
˜ of the
the transition length, depend on the local contraction rate (𝜃) It is also worth noting that the statistical criteria that test the
bulk peculiar flow. The latter, however, lies well beyond our current observational viability of a theoretical model take into account only
observational capabilities. To address the problem, we introduced the number of free parameters and do not take into consideration the
a two-parameter function for the bulk-flow contraction rate (of the physical motivation behind each model. Here, BIC favoured the fine-
form 𝜃˜ = 𝜃˜ (𝜆) – see Eq. (13)), which is both mathematically simple tuned ΛCDM model over the physically motivated tilted Einstein-
and has sound physical motivation. Setting 𝜆 ≡ 𝜒(𝑧),¯ where 𝜒(𝑧)
¯ is de Sitter universe. The situation changed, however, when one of
the line-of-sight comoving distance, we then obtained an expression the two free parameters in Eq. (24) was fixed in advance. Setting
of the form 𝑞˜ = 𝑞(𝜆(𝑧))
˜ for the local deceleration parameter (see 𝛼 = 1/2 in particular ensured that 𝑞˜ 0 ' −1/2 in agreement with the
Eq. (24)), which could be directly constrained from the SnIa data. observations. Then, the tilted Einstein-de Sitter universe (T-𝐸 𝑑𝑆)
Employing the latest publicly available SnIa compilation, namely and the standard ΛCDM model achieved a similar quality of fit and
the Pantheon sample, we successively assumed a ΛCDM and an are equally supported by the Pantheon data according to BIC, as it
Einstein-de-Sitter bulk-flow model and applied the maximum likeli- can be seen by the Δ𝐵𝐼𝐶 difference presented in Table 3.

MNRAS 000, 1–13 (2022)

12 K. Asvesta et al.
Future work, which will extend and refine this study, should include Alestas G., Perivolaropoulos L., Tanidis K., 2022, preprint
alternative parametrisation profiles for Eq. (13), as well as the use (arXiv:2201.05846)
of different types of cosmological data. Those from the Pantheon+ Alves C. S., Leite A. C. O., Martins C. J. A. P., Matos J. G. B., Silva T. A.,
dataset Riess et al. (2021a), the SnIa compilation on Dhawan et al. 2019, Mon. Not. Roy. Astron. Soc., 488, 3607
(2021), or from SnIa and from Sn of type II Stahl et al. (2021), should Antoniou I., Perivolaropoulos L., 2010, JCAP, 12, 012
Arjona R., Cardona W., Nesseris S., 2019, Phys. Rev. D, 99, 043516
help to constrain further the parameters of the tilted model.
Aubourg E., et al., 2015, Phys. Rev. D, 92, 123516
From the observational view point, measuring the magnitude and
˜ which provides the spatial Baxter E., et al., 2016, Mon. Not. Roy. Astron. Soc., 461, 4099
the sign of the local volume scalar (𝜃),
Betoule M., et al., 2014a, Astron. Astrophys., 568, A22
divergence of the bulk-flow velocity (recall that 𝜃˜ = D̃𝑎 𝜐˜ 𝑎 ), will Betoule M., et al., 2014b, A&A, 568, A22
directly link both 𝑞˜ and 𝜆𝑇 to the observations. Further insight into Boruah S. S., Hudson M. J., Lavaux G., 2020, Mon. Not. Roy. Astron. Soc.,
the bulk-flow kinematics and their implications will come from fu- 498, 2703
ture estimates of the peculiar shear and the peculiar vorticity (𝜎 ˜ 𝑎𝑏 Cai R.-G., Tuo Z.-L., 2012, JCAP, 02, 004
and 𝜔˜ 𝑎𝑏 respectively). In this effort, measurements of the so-called Camarena D., Marra V., 2021, Mon. Not. Roy. Astron. Soc., 504, 5164
“redshift drift” (or “velocity shift”), namely of temporal variations Carroll S. M., 2001, Living Rev. Rel., 4, 1
in the redshifts of distant sources Liske et al. (2008); Alves et al. Colin J., Mohayaee R., Sarkar S., Shafieloo A., 2011, Mon. Not. Roy. Astron.
(2019), could also help. Soc., 414, 264
Refining the bulk-flow data may also allow to reconstruct their Colin J., Mohayaee R., Rameez M., Sarkar S., 2019, Astron. Astrophys., 631,
internal kinematics and thus link the parameters 𝛼 and 𝑏 employed L13
Conley A., et al., 2011, Astrophys. J. Suppl., 192, 1
in the (theoretically motivated) profile of 𝑞˜ (see Eq. (24)) directly
Contreras C., et al., 2010, AJ, 139, 519
to the observations. Last, but not least, is the systematic search for
Cooke R., Lynden-Bell D., 2010, Mon. Not. Roy. Astron. Soc., 401, 1409
anisotropies in the sky distribution of the deceleration parameter. Dai D.-C., Kinney W. H., Stojkovic D., 2011, JCAP, 04, 015
Recall that a key prediction of the tilted scenario is the existence of Dainotti M. G., De Simone B., Schiavone T., Montani G., Rinaldi E., Lambiase
an apparent (Doppler-like) dipolar anisotropy in the 𝑞-distribution,
˜ G., 2021, Astrophys. J., 912, 150
triggered by the observers peculiar motion. Dainotti M. G., De Simone B., Schiavone T., Montani G., Rinaldi E., Lambiase
We expect that the ongoing Kourkchi et al. (2020); Dhawan et al. G., Bogdan M., Ugale S., 2022, Galaxies, 10, 24
(2021) and the forthcoming Ivezić et al. (2019) wide-field surveys Davis T. M., et al., 2011, Astrophys. J., 741, 67
should achieve exquisite precision in their bulk-flow measurements Dhawan S., et al., 2021, Monthly Notices of the Royal Astronomical Society,
and in so doing provide us with accurate kinematic estimates of the 510, 2228
observed peculiar velocity fields. Di Valentino E., et al., 2021, Class. Quant. Grav., 38, 153001
Efstathiou G., Lemos P., 2018, Mon. Not. Roy. Astron. Soc., 476, 151
Ehlers J., 1961, Abh. Akad. Wiss. Lit. Mainz. Nat. Kl., 11, 793
Ellis G. F. R., 1973, Cargese lectures in Physics. Gordon and Breach.
Ellis G. F. R., Maartens R., MacCallum M. A. H., 2012, Relativistic Cosmol-
ACKNOWLEDGEMENTS ogy. Cambridge University Press, doi:10.1017/CBO9781139014403
Feindt U., et al., 2013, Astron. Astrophys., 560, A90
This work was supported by the Hellenic Foundation for Research and Feldman H. A., Watkins R., Hudson M. J., 2010, Mon. Not. Roy. Astron.
Innovation (HFRI - Progect No: 789). The authors also acknowledge Soc., 407, 2328
support by the IT Center of the Aristotle University of Thessaloniki Filippou K., Tsagas C. G., 2021, Astrophys. Space Sci., 366, 4
(AUTh) throughout this study. We would also like to thank Eoin Foreman-Mackey D., Hogg D. W., Lang D., Goodman J., 2013, Publ. Astron.
O Colgain for helpful comments and K.A. wishes to thank Niko- Soc. Pac., 125, 306
laos Karnesis for many helpful discussions concerning the MCMC Gong Y.-G., Wang A., 2007, Phys. Rev. D, 75, 043520
analysis. Graur O., et al., 2014, ApJ, 783, 28
Guy J., et al., 2010, A&A, 523, A7
Heymans C., et al., 2021, Astron. Astrophys., 646, A140
Hicken M., et al., 2009a, ApJ, 700, 331
DATA AVAILABILITY Hicken M., Wood-Vasey W. M., Blondin S., Challis P., Jha S., Kelly P. L.,
Rest A., Kirshner R. P., 2009b, ApJ, 700, 1097
The data used in the current analysis correspond to the Pantheon
Hicken M., et al., 2012, ApJS, 200, 12
compilation of SnIa that are provided in https://github.com/ Hinton S., 2016a, ChainConsumer, https://github.com/samreay/
dscolnic/Pantheon and described in detail in Ref. Scolnic et al. ChainConsumer
(2018a). The numerical files for the reproduction of the figures can Hinton S. R., 2016b, The Journal of Open Source Software, 1, 00045
be found in this github repository under the MIT license. Hoffman Y., Courtois H. M., Tully R. B., 2015, Mon. Not. Roy. Astron. Soc.,
449, 4494
Hong T., et al., 2014, Mon. Not. Roy. Astron. Soc., 445, 402
Horstmann N., Pietschke Y., Schwarz D. J., 2021, arXiv e-prints, p.
REFERENCES arXiv:2111.03055
Abbott T. M. C., et al., 2018, Phys. Rev. D, 98, 043526 Howlett C., Said K., Lucey J. R., Colless M., Qin F., Lai Y., Tully R. B., Davis
Aghanim N., et al., 2020, Astron. Astrophys., 641, A6 T. M., 2022, preprint (arXiv:2201.03112)
Akaike H., 1974, IEEE Transactions on Automatic Control, 19, 716 Hunter J. D., 2007, Comput. Sci. Eng., 9, 90
Alam S., et al., 2017, Mon. Not. Roy. Astron. Soc., 470, 2617 Ishak M., 2019, Living Rev. Rel., 22, 1
Alestas G., Camarena D., Di Valentino E., Kazantzidis L., Marra V., Nesseris Ivezić Ž., et al., 2019, ApJ, 873, 111
S., Perivolaropoulos L., 2021a, preprint (arXiv:2110.04336) Jeffreys H., 1961, Theory of Probability, 3rd edn. Oxford Classic Texts in the
Alestas G., Antoniou I., Perivolaropoulos L., 2021b, Universe, 7, 366 Physical Sciences. Oxford Univ. Press
Alestas G., Kazantzidis L., Perivolaropoulos L., 2021c, Phys. Rev. D, 103, Jha S., et al., 2006, AJ, 131, 527
083517 Joudaki S., et al., 2018, Mon. Not. Roy. Astron. Soc., 474, 4894

MNRAS 000, 1–13 (2022)

 The Deceleration of a Tilted Universe 13
Kashlinsky A., Atrio-Barandela F., Kocevski D., Ebeling H., 2009, Astrophys. Riess A. G., Casertano S., Yuan W., Bowers J. B., Macri L., Zinn J. C., Scolnic
J. Lett., 686, L49 D., 2021b, Astrophys. J. Lett., 908, L6
Kashlinsky A., Atrio-Barandela F., Ebeling H., Edge A., Kocevski D., 2010, Rodney S. A., et al., 2014, AJ, 148, 13
Astrophys. J. Lett., 712, L81 Sako M., et al., 2018, PASP, 130, 064002
Kazantzidis L., Perivolaropoulos L., 2018, Phys. Rev. D, 97, 103503 Sapone D., Nesseris S., Bengaly C. A. P., 2021, Phys. Dark Univ., 32, 100814
Kazantzidis L., Perivolaropoulos L., 2020, Phys. Rev. D, 102, 023520 Saridakis E. N., et al., 2021, preprint (arXiv:2105.12582)
Kazantzidis L., Perivolaropoulos L., 2021, 𝜎8 Tension. Is Gravity Getting Schwarz G., 1978, The Annals of Statistics, 6, 461
Weaker at Low z? Observational Evidence and Theoretical Implications. Scolnic D. M., et al., 2018a, Astrophys. J., 859, 101
Springer International Publishing, pp 507–537 (arXiv:1907.03176), Scolnic D. M., et al., 2018b, ApJ, 859, 101
doi:10.1007/978-3-030-83715-0_33 Scrimgeour M. I., et al., 2016, Mon. Not. Roy. Astron. Soc., 455, 386
Kazantzidis L., Koo H., Nesseris S., Perivolaropoulos L., Shafieloo A., 2021, Secrest N. J., von Hausegger S., Rameez M., Mohayaee R., Sarkar S., Colin
Mon. Not. Roy. Astron. Soc., 501, 3421 J., 2021, Astrophys. J. Lett., 908, L51
Kessler R., Scolnic D., 2017, ApJ, 836, 56 Skara F., Perivolaropoulos L., 2020, Phys. Rev. D, 101, 063521
Kogut A., et al., 1993, Astrophys. J., 419, 1 Stahl B. E., de Jaeger T., Boruah S. S., Zheng W., Filippenko A. V., Hudson
Kourkchi E., Tully R. B., Anand G. S., Courtois H. M., Dupuy A., Neill J. D., M. J., 2021, Mon. Not. Roy. Astron. Soc., 505, 2349
Rizzi L., Seibert M., 2020, Astrophys. J., 896, 3 Suzuki N., et al., 2012, ApJ, 746, 85
Krishnan C., Mohayaee R., Ó Colgáin E., Sheikh-Jabbari M. M., Yin L., Tripp R., 1998, A&A, 331, 815
2021, preprint (arXiv:2106.02532) Tsagas C. G., 2010, Mon. Not. Roy. Astron. Soc., 405, 503
Lavaux G., Afshordi N., Hudson M. J., 2013, Mon. Not. Roy. Astron. Soc., Tsagas C. G., 2011, Phys. Rev. D, 84, 063503
430, 1617 Tsagas C. G., 2021a, preprint (arXiv:2112.04313)
Liddle A. R., 2004, Mon. Not. Roy. Astron. Soc., 351, L49 Tsagas C. G., 2021b, Eur. Phys. J. C, 81, 753
Liske J., et al., 2008, Mon. Not. Roy. Astron. Soc., 386, 1192 Tsagas C. G., Kadiltzoglou M. I., 2013, Phys. Rev. D, 88, 083501
Luongo O., Muccino M., Colgáin E. O., Sheikh-Jabbari M. M., Yin L., 2021, Tsagas C. G., Kadiltzoglou M. I., 2015, Phys. Rev. D, 92, 043515
preprint (arXiv:2108.13228) Tsagas C. G., Challinor A., Maartens R., 2007, Physics Reports, 465, 61
Ma Y.-Z., Pan J., 2014, Mon. Not. Roy. Astron. Soc., 437, 1996 Tsagas C. G., Kadiltzoglou M. I., Asvesta K., 2021, Astrophys. Space Sci.,
Ma Y.-Z., Scott D., 2013, Mon. Not. Roy. Astron. Soc., 428, 2017 366, 90
Maartens R., 1998, Phys. Rev. D, 58, 124006 Tsaprazi E., Tsagas C. G., 2020, European Physical Journal C, 80, 757
Macaulay E., Wehus I. K., Eriksen H. K., 2013, Phys. Rev. Lett., 111, 161301 Turnbull S. J., Hudson M. J., Feldman H. A., Hicken M., Kirshner R. P.,
Magoulas C., Springob C., Colless M., Mould J., Lucey J., Erdoğdu P., Jones Watkins R., 2012, Mon. Not. Roy. Astron. Soc., 420, 447
D. H., 2016, doi:10.1017/S1743921316010115, 308, 336 Turner M. S., Riess A. G., 2002, Astrophys. J., 569, 18
Mak D. S. Y., Pierpaoli E., Osborne S. J., 2011, Astrophys. J., 736, 116 Velten H. E. S., vom Marttens R. F., Zimdahl W., 2014, Eur. Phys. J., C74,
Mariano A., Perivolaropoulos L., 2012, Phys. Rev. D, 86, 083517 3160
Mariano A., Perivolaropoulos L., 2013, Phys. Rev. D, 87, 043511 Weinberg S., 1989, Rev. Mod. Phys., 61, 1
Marra V., Perivolaropoulos L., 2021, Phys. Rev. D, 104, L021303 Zhao D., Zhou Y., Chang Z., 2019, Mon. Not. Roy. Astron. Soc., 486, 5679
Migkas K., Pacaud F., Schellenberger G., Erler J., Nguyen-Dang N. T.,
Reiprich T. H., Ramos-Ceja M. E., Lovisari L., 2021a, Astron. Astro- This paper has been typeset from a TEX/LATEX file prepared by the author.
phys., 649, A151
Migkas K., Pacaud F., Schellenberger G., Erler J., Nguyen-Dang N. T.,
Reiprich T. H., Ramos-Ceja M. E., Lovisari L., 2021b, Astron. Astro-
phys., 649, A151
Narayan G., et al., 2016, ApJS, 224, 3
Nesseris S., Garcia-Bellido J., 2013, JCAP, 08, 036
Nesseris S., Pantazis G., Perivolaropoulos L., 2017, Phys. Rev. D, 96, 023542
Newville M., Stensitzki T., Allen D. B., Ingargiola A., 2014,
LMFIT: Non-Linear Least-Square Minimization and Curve-Fitting
for Python, doi:10.5281/zenodo.11813, https://doi.org/10.5281/
zenodo.11813
Nusser A., Davis M., 2011, Astrophys. J., 736, 93
Nusser A., Branchini E., Davis M., 2011, Astrophys. J., 735, 77
P.J S., 1997, Critical Problems in Physics. Princeton University Press
Perivolaropoulos L., 2022, preprint (arXiv:2201.08997)
Perivolaropoulos L., Kazantzidis L., 2019, Int. J. Mod. Phys. D, 28, 1942001
Perivolaropoulos L., Skara F., 2021a, preprint (arXiv:2105.05208)
Perivolaropoulos L., Skara F., 2021b, Phys. Rev. D, 104, 123511
Perlmutter S., et al., 1999, Astrophys. J., 517, 565
Qin F., 2021, Res. Astron. Astrophys., 21, 242
Qin F., Howlett C., Staveley-Smith L., Hong T., 2019, Mon. Not. Roy. Astron.
Soc., 482, 1920
Rathaus B., Kovetz E. D., Itzhaki N., 2013, Mon. Not. Roy. Astron. Soc., 431,
3678
Riess A. G., Press W. H., Kirshner R. P., 1995, Astrophys. J. Lett., 445, L91
Riess A. G., et al., 1998, Astron. J., 116, 1009
Riess A. G., et al., 1999, AJ, 117, 707
Riess A. G., et al., 2004a, Astrophys. J., 607, 665
Riess A. G., et al., 2004b, ApJ, 607, 665
Riess A. G., et al., 2007, ApJ, 659, 98
Riess A. G., et al., 2018, ApJ, 853, 126
Riess A. G., et al., 2021a, preprint (arXiv:2112.04510)

MNRAS 000, 1–13 (2022)