Astronomy & Astrophysics manuscript no.

main_SZ2_DESI_v1 ©ESO 2024

November 8, 2024

Dark energy reconstructions combining BAO data with galaxy

clusters and intermediate redshift catalogs
Orlando Luongo1, 2, 3, 4, 5 and Marco Muccino1, 5, 6, 7

1
Università di Camerino, Divisione di Fisica, Via Madonna delle carceri 9, 62032 Camerino, Italy.
2
SUNY Polytechnic Institute, 13502 Utica, New York, USA.
3
INAF - Osservatorio Astronomico di Brera, Milano, Italy.
4
INFN, Sezione di Perugia, Perugia, 06123, Italy.
arXiv:2411.04901v1 [astro-ph.CO] 7 Nov 2024

5
Al-Farabi Kazakh National University, Al-Farabi av. 71, 050040 Almaty, Kazakhstan.
6
Institute of Nuclear Physics, Ibragimova, 1, 050032 Almaty, Kazakhstan.
7
ICRANet, Piazza della Repubblica 10, Pescara, 65122, Italy.
e-mail: [email protected], [email protected]

ABSTRACT

Context. Cosmological parameters and dark energy (DE) behavior are generally constrained assuming a priori models.
Aims. We work out a model-independent reconstruction to bound the key cosmological quantities and the DE evolution.
Methods. Through the model-independent Bézier interpolation method, we reconstruct the Hubble rate from the ob-
servational Hubble data and derive analytic expressions for the distances of galaxy clusters, type Ia supernovae, and
uncorrelated baryonic acoustic oscillation (BAO) data. In view of the discrepancy between Sloan Digital Sky Survey
(SDSS) and Dark Energy Spectroscopic Instrument (DESI) BAO data, they are kept separate in two distinct analyses.
Correlated BAO data are employed to break the baryonic–dark matter degeneracy. All these interpolations enable us
to single out and reconstruct the DE behavior with the redshift z in a totally model-independent way.
Results. In both analyses, with SDSS-BAO or DESI-BAO data sets, the constraints agree at 1–σ confidence level (CL)
with the flat ΛCDM model. The Hubble constant tension appears solved in favor of the Planck satellite value. The
reconstructed DE behavior exhibits deviations at small z (> 1–σ CL), but agrees (< 1–σ CL) with the cosmological
constant paradigm at larger z.
Conclusions. Our method hints for a slowly evolving DE, consistent with a cosmological constant at early times.
Key words. Dark energy; model-independent techniques; cosmological parameters; galaxies clusters.

1. Introduction Abdalla et al. 2022), as well as unresolved theoretical

challenges like the coincidence and fine-tuning problems
Our current understanding of the Universe does not ade- (Copeland et al. 2006), have motivated the exploration of
quately explain the physical reasons behind its accelerated alternative models Wolf et al. (2024a,b); Wolf & Ferreira
expansion, experimentally found by observations of SNe Ia (2023). New data, including DESI measurements of bary-
(see Riess et al. 1998; Perlmutter et al. 1999). To account onic acoustic oscillations (BAO), support numerous ap-
for this unexpected phenomenon, it is widely-accepted that proaches that may replicate or even surpass the predic-
an exotic fluid, exhibiting a negative equation of state, tive successes of the ΛCDM paradigm (Luongo & Muccino
might be included into Einstein’s energy-momentum ten- 2018; D’Agostino et al. 2022; Belfiglio et al. 2022).
sor.
Indeed, the presence of baryonic and cold dark matter To mitigate reliance on a specific cosmological
alone is insufficient to describe the Universe’s late-time ac- model, various model-independent methods have been
celeration (Peebles & Ratra 2003), leading to the hypothe- proposed (Capozziello et al. 2013; Dunsby & Luongo
sis of a time-dependent fluid, known as DE, that drives the 2016; Luongo & Muccino 2021b, 2023; Shafieloo 2007;
Universe to speed up. Shafieloo & Clarkson 2010; Haridasu et al. 2018). A pri-
Among various DE models, the ΛCDM paradigm mary challenge of these approaches is the difficulty of
posits that DE is in the form of a genuine cosmologi- reconciling data from low, intermediate, and early cosmic
cal constant (Carroll 2001). Thus, it is not hard to be- epochs, aiming to capture DE’s evolution across the full
lieve that this model, due to its minimal set of free expansion history. Additionally, a significant drawback
parameters, is the most statistically preferred for de- arises from assuming both cold dark matter and baryons
scribing large-scale cosmic dynamics, making it partic- as dust-like matter, which prevents separate measurements
ularly suitable for late-time cosmology. However, recent of their respective densities, creating a degeneracy within
observational cosmological tensions, such as the Hub- the matter sector.
ble constant tension and inconsistencies in the cluster- In this work, we propose a strategy that heals
ing amplitude S8 across low and high-redshift measure- the above issues by combining different probes –
ments (Bernal & Libanore 2023; Di Valentino et al. 2021; based on low, intermediate and early time data points
Article number, page 1 of 11
 A&A proofs: manuscript no. main_SZ2_DESI_v1

– and disentangles the matter sector in order to 1) The Bézier interpolation of the OHD catalog provides a
mold DE at different stages of its evolution. We model-independent expression for the Hubble rate H(z)
resort a model-independent approach based on the and an alternative estimate of the Hubble constant H0 .
so-called Bézier parametric interpolation (Amati et al. 2) This interpolation is used to derive analytic expressions
2019; Luongo & Muccino 2021b; Montiel et al. 2021; for the angular diameter distance DA (z) of GCs, the
Luongo & Muccino 2023; Muccino et al. 2023; Alfano et al. luminosity distance DL (z) of SNe Ia, and BAO observ-
2024a,b,c), which is used to: ables, which bear no a priori assumptions on Ωk that
can be extracted from the fits.
- reconstruct the Hubble rate H(z) fitting the observa- 3) The combination of SDSS-BAO or DESI-BAO data with
tional Hubble data (OHD) (see, e.g., Amati et al. 2019), correlated WiggleZ-BAO data breaks the baryonic–dark
- derive analytic expressions for the distances of galaxy matter degeneracy through the definition of rd .
clusters (GCs), SNe Ia, and BAO uncorrelated data, 4) The so-extracted cosmic bounds plus the H(z) recon-
- break the baryonic–dark matter degeneracy within the struction single out and reconstruct DE behavior in
comoving sound horizon rd (Efstathiou & Bond 1999), terms of z, in a quite fully model-independent way.
through the interpolation of the correlated BAO data.
The key feature of our recipe is therefore based on the
To check whether our method can give additional in- use of Bézier approximation that we describe below.
sights on the form of dark energy,

- we seek non-flat Universe, adding the spatial curvature 2.1. Bézier interpolation of H(z)
into the Friedmann equations,
The NO = 34 OHD measurements of the Hubble rate
- we analyze the impact of our strategy on the cosmolog- (see Table 1) are obtained from the detection of cou-
ical tensions. ples of galaxies, assumed to form at same age, mostly
Results from our Monte Carlo–Markov chain (MCMC) and rapidly exhausted their gas reservoir, thence, evolv-
simulations, using the Metropolis-Hastings algorithm, in- ing passively. Once the difference in age, ∆t, and redshift,
dicate that this approach provides valuable insights. Pre- ∆z, of these pairs of galaxies are spectroscopically deter-
cisely, when comparing our results with expectations from mined, the Hubble parameter is estimated from the identity
the flat (non-flat) ΛCDM model, the constraints agree H(z) = −(1 + z)−1 ∆z/∆t (Jimenez & Loeb 2002a).
within 1–σ CL with the flat ΛCDM model, whereas the For the sake of clearness, age-dating galaxies is affected
Hubble constant tension is solved in favor of the Planck by large systematic errors, typically associated with star
satellite value. Accordingly, the reconstructed DE behav- formation history, stellar age, formation timescale, chemi-
ior exhibits deviations at small z (& 1–σ CL), but agrees cal composition, and so on. These uncertainties contribute
with the cosmological constant paradigm at larger z. Conse- with additional 20–30% errors (Moresco et al. 2022), lead-
quently, our method suggests a slowly evolving DE, however ing to measurements that are not particularly accurate. The
consistent with a cosmological constant at early times. great advantage relies on their determination though, i.e.,
The paper is structured as follows. The methods of our it is roughly model-independent, as much as the above hy-
treatment are reported in Sect. 2, where the Bézier inter- potheses on galaxy formation are fulfilled.
polation is explained in detail. The numerical findings are The function of z best-interpolating the OHD catalog is
thus reported in Sect. 4, where the MCMC analyses are a second order Bézier curve
summarized. The core of DE reconstructions is displayed α⋆ 
H(z) = 2 α0 (zm − z)2 + 2α1 z(zm − z) + α2 z 2 ,

and theoretically discussed in Sect. 5. Conclusions and per- (1)
zm
spectives are summarized in Sect. 6.
with normalization α⋆ = 100 km s−1 Mpc−1 and coefficients
αi , that is extrapolated up to zm = 2.33, which is the
2. Methods largest redshift for OHD, GC, Pantheon+, and BAO cat-
In this section, we describe the methodologies developed alogs. From Eq. (1), at z = 0 the dimensionless Hubble
throughout our manuscript in order to obtain model- constant can be defined by h0 ≡ H0 /α⋆ ≡ α0 .
independent cosmological bounds. To do so, With Gaussian distributed errors σHj , the coefficients
αi are found by maximizing the log-likelihood function
- we follow the methodology introduced in Alfano et al.
NO  2
(2024b) that makes use of OHD, GCs and BAO inter- 1X Hj − H(zj )
mediate redshift catalogs, ln LO ∝ − . (2)
2 j=1 σHj
- we include the Pantheon+ catalog of SNe Ia
(Scolnic et al. 2022) to refine the overall constraints,
and 2.2. Constraints on the curvature parameter from GC data
- we perform two separate MCMC analyses, depend-
ing whether either SSDS or DESI data are in- When CMB photons travel across intra-cluster high-energy
volved into computation, in view of the claimed evi- electrons in GCs, inverse Compton scattering occurs,
dence for evolving DE derived from DESI-BAO data causing the distortion of the CMB spectrum. This phe-
(DESI Collaboration 2024). nomenon is referred to as the Sunyaev-Zeldovich (SZ) effect
(Sunyaev & Zeldovich 1970, 1972; Carlstrom et al. 2002).
Then, we jointly fit OHD, GC, SNe Ia, and BAO cata- The SZ effect is redshift-independent and, combined
logs, based on the key steps, summarized below. with high signal-to-noise ratio X-ray surface brightness of
Article number, page 2 of 11
 Luongo & Muccino: Dark energy reconstructions combining BAO with GCs and intermediate redshift catalogs

z H References z DA
(km/s/Mpc) (Mpc)
0.07 69.0 ± 19.6 ± 12.4 Zhang et al. (2014) 0.023 103 ± 42
0.09 69.0 ± 12.0 ± 11.4 Jimenez & Loeb (2002b) 0.058 242 ± 61
0.12 68.6 ± 26.2 ± 11.4 Zhang et al. (2014) 0.072 165 ± 45
0.17 83.0 ± 8.0 ± 13.1 Simon et al. (2005) 0.074 369 ± 62
0.1791 75.0 ± 3.8 ± 0.5 Moresco et al. (2012) 0.084 749 ± 385
0.1993 75.0 ± 4.9 ± 0.6 Moresco et al. (2012) 0.088 448 ± 185
0.20 72.9 ± 29.6 ± 11.5 Zhang et al. (2014) 0.091 335 ± 70
0.27 77.0 ± 14.0 ± 12.1 Simon et al. (2005) 0.142 478 ± 126
0.28 88.8 ± 36.6 ± 13.2 Zhang et al. (2014) 0.176 809 ± 263
0.3519 83.0 ± 13.0 ± 4.8 Moresco et al. (2016) 0.182 451 ± 189
0.3802 83.0 ± 4.3 ± 12.9 Moresco et al. (2016) 0.183 604 ± 84
0.4 95.0 ± 17.0 ± 12.7 Simon et al. (2005) 0.202 387 ± 141
0.4004 77.0 ± 2.1 ± 10.0 Moresco et al. (2016) 0.202 806 ± 163
0.4247 87.1 ± 2.4 ± 11.0 Moresco et al. (2016) 0.217 1465 ± 407
0.4497 92.8 ± 4.5 ± 12.1 Moresco et al. (2016) 0.224 1118 ± 283
0.47 89.0 ± 23.0 ± 44.0 Ratsimbazafy et al. (2017) 0.252 946 ± 131
0.4783 80.9 ± 2.1 ± 8.8 Moresco et al. (2016) 0.282 1099 ± 308
0.48 97.0 ± 62.0 ± 12.7 Stern et al. (2010) 0.288 934 ± 331
0.5929 104.0 ± 11.6 ± 4.5 Moresco et al. (2012) 0.322 885 ± 207
0.6797 92.0 ± 6.4 ± 4.3 Moresco et al. (2012) 0.327 697 ± 183
0.75 98.8 ± 33.6 Borghi et al. (2022) 0.375 1231 ± 441
0.7812 105.0 ± 9.4 ± 6.1 Moresco et al. (2012) 0.451 1166 ± 262
0.80 113.1 ± 15.1 ± 20.2 Jiao et al. (2023) 0.541 1635 ± 391
0.8754 125.0 ± 15.3 ± 6.0 Moresco et al. (2012) 0.550 1073 ± 238
0.88 90.0 ± 40.0 ± 10.1 Stern et al. (2010) 0.784 2479 ± 1023
0.9 117.0 ± 23.0 ± 13.1 Simon et al. (2005) Table 2. GCs catalog from De Filippis et al. (2005) with red-
1.037 154.0 ± 13.6 ± 14.9 Moresco et al. (2012) shift (first column) and diameter angular distances (second col-
1.26 135.0 ± 65.0 Tomasetti et al. (2023) umn).
1.3 168.0 ± 17.0 ± 14.0 Simon et al. (2005)
1.363 160.0 ± 33.6 Moresco (2015)
1.43 177.0 ± 18.0 ± 14.8 Simon et al. (2005)
1.53 140.0 ± 14.0 ± 11.7 Simon et al. (2005) 2.3. Reinforcing the constraints with Pantheon+ data
1.75 202.0 ± 40.0 ± 16.9 Simon et al. (2005)
1.965 186.5 ± 50.4 Moresco (2015) The Pantheon+ is a catalog of NS = 1701 SNe Ia with a
Table 1. OHD catalog redshifts, values of H with statistical and
redshift coverage 0 < z ≤ 2.3, that comprises 18 different
systematic (or combinrd) errors, and references, respectively. samples (Scolnic et al. 2022).
The luminosity distance DL (in Mpc) of each SN Ia with
rest-frame B-band apparent magnitude m is given by

the intra-cluster gas, which is redshift-dependent, it is possi- DL = 10(m−M−25)/5 . (5)

ble to determine the triaxial structure of the GC and, thus,
the corresponding corrected angular diameter distance DA .
Table 2 lists the sample of NG = 25 of such determined The rest-frame B-band absolute magnitude M in Eq. (5)
GC distances DA (De Filippis et al. 2005). The systematic can be viewed as a nuisance parameter. Following
errors – mainly due to radio halos, X-ray absolute flux and Conley et al. (2011), the marginalization over M that max-
electron temperature calibrations hindering the SZ effect imizes the log-likelihood ln LS of SN Ia is M = b/e, leading
calibration – are typically ≈ 13% (Bonamente et al. 2006). to
Using Eq. (1), we obtain an interpolated angular diam-   e  b2 
eter distance defined as 1
ln LS ∝ − a + ln − , (6)
2 2π e
−1 Z z √ 
c (1 + z) α⋆ α0 Ωk dz ′
DA (z) = √ sinh , (3)
α⋆ α0 Ωk 0 H(z ′ ) with a ≡ ∆mT C−1 ∆m, b ≡ ∆mT C−1 1 and e ≡ 1T C−1 1.
In these definitions, C is the covariance matrix that includes
statistical and systematic errors1, and ∆m ≡ m−mth(z) is
that holds for any value of the curvature parameter Ωk . the vector of residuals with respect to the model magnitude
If the errors σDAj are Gaussian distributed, αi and Ωk vector mth (z), with elements defined as
are obtained by maximizing the log-likelihood
mth (z) = 5 log (1 + z)2 DA (z) + 25 ,
 
(7)
NG  2
1 X DAj − DA (zj )
ln LG ∝ − . (4)
2 j=1
σDAj where DA (z) is given by Eq. (3).
Article number, page 3 of 11
 A&A proofs: manuscript no. main_SZ2_DESI_v1

Survey z DM /rd DH /rd DV /rd A References

6dFGS 0.106 2.98 ± 0.13 Beutler et al. (2011)
SDSS MGS 0.15 4.51 ± 0.14 Alam et al. (2021)
SDSS DR12 0.38 10.27 ± 0.15 24.89 ± 0.58 Alam et al. (2021)
SDSS DR12 0.51 13.38 ± 0.18 22.43 ± 0.48 Alam et al. (2021)
SDSS DR16 LRG 0.70 17.65 ± 0.30 19.78 ± 0.46 Alam et al. (2021)
SDSS DR16 ELG 0.85 19.50 ± 1.00 19.60 ± 2.10 Alam et al. (2021)
SDSS DR16 QSO 1.48 30.21 ± 0.79 13.23 ± 0.47 Alam et al. (2021)
SDSS DR16 Lyα–Lyα 2.33 37.60 ± 1.90 8.93 ± 0.28 Alam et al. (2021)
SDSS DR16 Lyα–QSO 2.33 37.30 ± 1.70 9.08 ± 0.34 Alam et al. (2021)
DESI BGS 0.30 7.93 ± 0.15 DESI Collaboration (2024)
DESI LRG1 0.51 13.62 ± 0.25 20.98 ± 0.61 DESI Collaboration (2024)
DESI LRG2 0.71 16.85 ± 0.32 20.08 ± 0.60 DESI Collaboration (2024)
DESI LRG+ELG 0.93 21.71 ± 0.28 17.88 ± 0.35 DESI Collaboration (2024)
DESI ELG 1.32 27.79 ± 0.69 13.82 ± 0.42 DESI Collaboration (2024)
DESI QSO 1.49 26.07 ± 0.67 DESI Collaboration (2024)
DESI Lyα–QSO 2.33 39.71 ± 0.94 8.52 ± 0.17 DESI Collaboration (2024)
WiggleZ 0.44 0.474 ± 0.034 Blake et al. (2012)
WiggleZ 0.6 0.442 ± 0.020 Blake et al. (2012)
WiggleZ 0.73 0.424 ± 0.021 Blake et al. (2012)
Table 3. BAO catalogs with surveys (first column), redshifts (second column), measurements with errors (third–sixth columns), and
references (last column). For BAO from SDSS, see also https://www.sdss4.org/science/final-bao-and-rsd-measurements/.

3. Breaking the baryon–dark matter degeneracy BAO measurements are affected by systematics errors re-
with BAO lated to photometry or spectroscopy, survey geometries and
discrete volumes, etc. that are below 0.5% (Glanville et al.
BAO are density fluctuations of the baryonic matter, gener- 2021; DESI Collaboration 2024).
ated by acoustic density waves in the primordial Universe Resorting Eqs. (1), (3), (5) and (8), BAO uncorrelated
(Weinberg 2008). Their characteristic scale, embedded in observables X = {DM /rd , DH /rd , DV /rd } – the transverse
the galaxy distribution (Cuceu et al. 2019), corresponds to comoving distance, the Hubble rate distance, and the angle-
the maximum distance rd covered by the acoustic waves averaged distance ratios with rd , respectively – listed in Ta-
before their “froze in” due to the decoupling of baryons. ble 3, can interpolated by the quantities X = {X1 , X2 , X3 },
Based on non-parametric reconstructions, respectively, given by the following expressions:
Aizpuru et al. (2021) proposed for rd a very accurate
expression (1 + z)DA (z)
X1 (z) = , (9a)
rd
a2 (a3 +ων )2
a1 e c
rd = Mpc , (8) X2 (z) = , (9b)
a4 ωba5 + a6 ωm
a7
+ a8 (ωb ωm )a9 rd H(z)
1/3
X3 (z) = zX1 (z)X22 (z)

where the density parameter for massive neutrino species . (9c)
is fixed to ων = 0.000645 (Aubourg et al. 2015), and the
density parameters for baryons only ωb = h20 Ωb and for Eqs. (9a)–(9c) reinforce the constraints on h0 and Ωk and
baryonic + dark matter ωm = h20 Ωm are the free parame- set bounds on ωb and ωm via Eq. (8) that, however, intro-
ters. The numerical coefficients ai have values duces a degeneracy between ωb and ωm (Efstathiou & Bond
1999) which is generally broken by fixing ωb with the value
a1 = 0.0034917, a2 = −19.972694, a3 = 0.000336186, got from the CMB (Planck Collaboration 2020) or Big
a4 = 0.0000305, a5 = 0.22752, a6 = 0.0000314257, Bang nucleosynthesis theory (Schöneberg 2024).
a7 = 0.5453798, a8 = 374.14994, a9 = 4.022356899. To break this degeneracy, we resort the correlated BAO
acoustic parameter A listed in Table 3 (Blake et al. 2012),
Table 3 lists the four kind of BAO measurements col- which described by the interpolation
lected from four different surveys: 1/3
√ (1 + z)2 DA 2

(z)
- 6dF Galaxy Survey (6dFGS), that mapped the nearby A(z) = g⋆ ωm , (10)
Universe over nearly half the sky (Beutler et al. 2011); c2 z 2 H(z)
- SDSS, providing galaxy and quasar spectroscopic sur- that does not depend upon rd and hence enables constraints
veys (Alam et al. 2021); on h0 , Ωk and only ωm (Alfano et al. 2024b).
- DESI, that collected galaxy and quasar optical spectra The log-likelihood function of each of the uncorrelated
to measure DE effect (DESI Collaboration 2024); BAO data, with corresponding errors σX , is given by
- WiggleZ Dark Energy Survey, furnishing correlated es-
timates of the acoustic parameter (Blake et al. 2012). NX  2
1X Xj − X (zj )
ln LX ∝ − , (11)
1
https://github.com/PantheonPlusSH0ES/DataRelease 2 j=1 σXj

Article number, page 4 of 11

 Luongo & Muccino: Dark energy reconstructions combining BAO with GCs and intermediate redshift catalogs

whereas the log-likelihood function for correlated BAO data Bézier interpolations of OHD, GC, SN Ia and BAO cata-
with covariance matrix CB (Blake et al. 2012) is logs.
Focusing on the cosmological parameters ωb , ωm , h0 ,
1 and Ωk (see Table 4), we can deduce what follows below.
ln LA ∝ − ∆AT C−1
B ∆A , (12)
2
- The MCMC1 analysis confirms and further refines the
with ∆A ≡ Aj − A(zj ). Combining Eqs. (11)–(12) leads to findings of Alfano et al. (2024b), which where based on
the total BAO log-likelihood function SDSS data points though not in their final version pre-
X sented by Alam et al. (2021).
ln LB = ln LX + ln LA . (13) - The MCMC2 bounds tend to agree with the MCMC1
X results, albeit with a) a smaller value of ωb and b) a
barely-consistent (at ≈ 1–σ CL) and positive Ωk .
4. Numerical results - In both the analyses, the inclusion of the Pantheon+
catalog improved the constraints on Ωk , which are more
Before proceeding with the numerical analysis, it is worth compatible (within 1–σ CL) with the flat scenario or
comparing the BAO data argued either from SDSS or from with small spatial curvature geometries.
DESI. - Both MCMC1 and MCMC2 results are in agreement
As pointed out by the DESI Collaboration (2024), the within 1–σ CL with the flat concordance model, though
region of the sky and the redshift ranges (see Table 3) with larger attached errors.
observed by DESI partially overlaps with those from the - For both analyses the consistency with the non-flat ex-
SDSS, therefore, a joint fit would require the knowledge of tension of the ΛCDM (Planck Collaboration 2020) is at
the covariance matrix. 2–σ CL, due to the estimate on h0 .
To this end, Ref. DESI Collaboration (2024) highlighted - The Hubble tension seems to be solved in favor of the
a large discrepancy (∼ 3–σ) between the DESI and SDSS Planck Collaboration (2020) value h0 = 0.6736±0.0054,
results, emphasized at redshift z ∼ 0.7. consistent at 1–σ CL with both MCMC1 and MCMC2
In addition to the above considerations, recent works estimates, whereas the value h0 = 0.7304 ± 0.0104 got
also evidenced possible anomalies in the DESI data set from SNe Ia (Riess et al. 2022) is only consistent within
(Colgáin et al. 2024; Luongo & Muccino 2024) and incon- 2–σ CL with MCMC1 and MCMC2 analyses.
clusive evidence in favor of a dynamical DE over the stan- - In general, the best-fit values got from the MCMC2
dard cosmological paradigm (see, e.g., Carloni et al. 2024; analysis (performed using the DESI-BAO data)
Giarè et al. 2024; Wang 2024, for an overview). seems to be closer to the flat ΛCDM best-fits
For these reasons, DESI and SDSS data (see Table 3) (Planck Collaboration 2020) than those from the
will not be jointly fit, but rather will be kept separated into MCMC1 procedure.
two MCMC analyses involving OHD, GCs, SNe Ia and the
following combinations of BAO, dubbed as follows:
- MCMC1, with BAO log-likelihood LB1 given by 5. Reconstruction of the dark energy behavior
Eq. (13), that combines the only data point from 6dFGS From the best-fit values of Table 4 we can now attempt the
and NA = 3 correlated data from WiggleZ with the reconstruction of the DE behavior.
NX = 15 uncorrelated measurements X from SDSS; We can use directly Ωk to model the curvature con-
- MCMC2, with BAO log-likelihood LB2 given by tribution and the combination of h0 and ωm to constrain
Eq. (13), in which the data points from 6dFGS and the matter density parameter. Next, we use the CMB
WiggleZ are combined with NX = 12 uncorrelated mea- temperature T0 = 2.7255 ± 0.0006K and the effective
surements X from DESI. extra relativistic degrees of freedom Neff = 2.99 ± 0.17
(Planck Collaboration 2020) to compute the radiation den-
We get the best-fit parameters of MCMC1 and MCMC2
analyses by maximizing the log-likelihood functions sity parameter Ωr = 9.15+0.26
−0.26 × 10
−5
. Putting all these
contributions together in a ΛCDM-like fashion, we define
MCMC1 : ln L1 = ln LO + ln LG + ln LS + ln LB1 , (14a) the following function of the redshift
MCMC1 : ln L2 = ln LO + ln LG + ln LS + ln LB2 . (14b) f (z) = α−2 3 2 4
(15)
0 ωm (1 + z) + Ωk (1 + z) + Ωr (1 + z) .
For both analyses, we impose a wide range of priors on the At this point, we can single out the contribution of the DE
parameters of our model-independent reconstructions: density by subtracting Eq. (15) from H2 (z), obtained by
squaring Eq. (1), namely
α0 ≡ h0 ∈ [0, 1] , Ωk ∈ [−2, 2] ,
α1 ∈ [0, 2] , ωb ∈ [0, 1] , 4
X
α2 ∈ [0, 3] , ωm ∈ [0, 1] . −2
Ωde (z) = (g⋆ α0 ) 2
H (z) − f (z) = βi (1 + z)i , (16)
i=0
Details on the MCMC1 and MCMC2 analyses and their
corresponding 1–σ and 2–σ contour plots (see Figs. A.1– where the coefficients βi in the last expressions and listed in
A.2, respectively) can be found in Appendix A. Table 5 depend upon combinations of the coefficients αi and
The best-fit values are listed in Table 4 and com- the cosmological parameters ωm , Ωk and Ωr , as reported in
pared with the constrains on flat and non-flat ΛCDM Appendix B.
model got from Planck TT, TE, EE+lowE+lensing data The DE reconstructed behaviors for both MCMC1 and
(Planck Collaboration 2020). In particular, in Fig. 2 the MCMC2 analyses, obtained inputing the best-fit values of
flat ΛCDM case is used as a benchmark for the best-fitting Table 4 in Eq. (16), are portrayed in Fig. 2 and compared
Article number, page 5 of 11
 A&A proofs: manuscript no. main_SZ2_DESI_v1

α0 ≡ h0 α1 α2 Ωk ωb ωm M (mag)
MCMC1 analysis
+0.023 (0.039) +0.053 (0.086) +0.11 (0.19) +0.14 (0.24) +0.0061 (0.0109) +0.022 (0.035) +0.045 (0.093)
0.682−0.027 (0.041) 1.169−0.055 (0.085) 2.33−0.11 (0.18) −0.09−0.13 (0.21) 0.0253−0.0070 (0.0100) 0.145−0.018 (0.029) −19.389−0.055 (0.104)
MCMC2 analysis
+0.024 (0.039) +0.051 (0.082) +0.10 (0.17) +0.13 (0.21) +0.0075 (0.0119) +0.021 (0.034) +0.051 (0.099)
0.677−0.024 (0.042) 1.170−0.055 (0.084) 2.39−0.12 (0.18) +0.08−0.12 (0.19) 0.0212−0.0051 (0.0081) 0.144−0.018 (0.028) −19.407−0.049 (0.099)
Planck Collaboration
0.6736+0.0054
−0.0054 – – – 0.02237+0.00015
−0.00015 0.1430+0.0011
−0.0011 –
0.636+0.021
−0.023 – – −0.011+0.013
−0.012 0.02249+0.00016
−0.00016 0.1410+0.0015
−0.0015 –
Table 4. MCMC best-fits and 1–σ (2–σ) errors, compared with flat and non-flat ΛCDM models (Planck Collaboration 2020).

250 250
– OHD+GC+BAO+SN – OHD+GC+BAO+SN
-- Flat LCDM HPlanck 2020L -- Flat LCDM HPlanck 2020L
HHzL @kmsMpcD

HHzL @kmsMpcD
200 æ 200 æ
æ æ
æ æ
æ æ
æ æ
150 æ
æ
150 æ
æ
æ æ
æ æ
æ æ æ æ
æ æ
100 æ
æ ææ
æ
æ
æ
æ
æ
100 æ
æ ææ
æ
æ
æ
æ
æ
æ ææ æ æ æ ææ æ æ
ææ æ æ ææ æ æ
æææ æ æææ æ

50 50

– OHD+GC+BAO+SN – OHD+GC+BAO+SN
3000 -- Flat LCDM HPlanck 2020L 3000 -- Flat LCDM HPlanck 2020L
DAHzL @MpcD

DAHzL @MpcD
æ æ

2000 2000
æ æ
æ æ
æ æ æ æ
ææ æ ææ æ
1000 æææ 1000 æææ
æ ææ æ æ ææ æ
æ æ
æ ææ æ ææ
ææ æ ææ æ
æ æ
ææ ææ

20000 – OHD+GC+BAO+SN 20000 – OHD+GC+BAO+SN

-- Flat LCDM HPlanck 2020L æ -- Flat LCDM HPlanck 2020L æ
DLHzL @MpcD

DLHzL @MpcD

15000 15000
æ æ æ æ
æ æ æ æ
æ æ æ æ æ æ
10000 æ æææææ
ææ 10000 æ ææææ
ææ
æ
æ æ
æ æ æ æ æ æ
ææ æ ææ æ
ææ
æ ææ
æ
æ æ æ æ æ æ
5000 æææææ
æ
ææææ
æ
æ
æææ
æææ
æ
ææ
æ
æ
æ
ææ
æ
æ
æ
ææ
æ
æ
ææ
æ
ææ
æ
æ 5000 ææææ
æ
ææ
ææ
æææ
æ
æ
æææ
æææ
æ
æ
ææ
ææ
æ
ææ ææ
æ
æ
æææ
æ
ææ
æ
æ
ææ
ææ
æ
ææ
æ
æ
ææ
ææ
æ
æ
ææ
æ ææ
ææ
æ
ææ
æ
æ
ææ
ææ
ææ
æ
æ
æ
æææ
æ
æ
ææ
æ
ææ
æ
æ
æ
æ
ææ
æ
ææ
æ
ææ
æ
ææ
æ æ
æ æææ
æ
æ
ææ
ææ
æ
æ
æ
æ
ææ
æ
ææ
æ
ææ
æ
ææ
æ æ
ææ
æ
ææ
ææ
æ
æ
æ
ææ
æ
æ
æ
ææ
æ
æ
æ
æ
ææ
æ æ ææ
æ
ææ
ææ
æ
æ
æ
ææ
æ
æ
ææ
æ
æ
ææ
ææ
æ æ æ
ææ
ææ
æ
æ
æ
ææ
æ
æ
æ
ææ
æ
ææ
æ
ææ
æ ææ
ææ
æ
æ
æ
ææ
æ
æ
æ
ææ
æ
ææ
æ
ææ
æ
ææ
ææ
æ
ææ
æ
æ
æ
ææ
æ
æ
æ
æ
æ
æ
æ
ææ
æ
æ
ææ
æ
ææ
ææ
ææ
ææ
æ
ææ
æ
æ
æ
ææ
æ
æ
æ
æ
æ
æ
æ
ææ
æ
æ
æææ
æ
æææ
æ
æ
æ
ææ
ææ
æ æ
æ
æ
ææ
ææ
æ
0 æ
æ
æ
ææ
æ
æ
ææ
æ
ææ
æ
æ
ææ
æ
0 æ
æ
æ
ææ
æ
æ
ææ
æ
ææ
æ
æ
ææ
æ

30 30
æ
DVHzLrd

DVHzLrd

20 – OHD+GC+BAO+SN 20 – OHD+GC+BAO+SN
10
-- Flat LCDM HPlanck 2020L 10
-- Flat LCDM HPlanck 2020L
æ
æ
æ æ
0 0
– OHD+GC+BAO+SN – OHD+GC+BAO+SN
0.50 0.50
-- Flat LCDM HPlanck 2020L -- Flat LCDM HPlanck 2020L
AHzL

AHzL

0.25 0.25

0.00 0.00
25 – OHD+GC+BAO+SN 25 – OHD+GC+BAO+SN
DAHzLrd — DHHzLrd

DAHzLrd — DHHzLrd

-- Flat LCDM HPlanck 2020L -- Flat LCDM HPlanck 2020L

20 20

15 15
æ æ æ
æ æ
10 æ
æ æ
10 æ
æ
æ

5 5

0 0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
redshift z redshift z

Fig. 1. Plots of OHD, GC, SN Ia and BAO data sets with MCMC1 (left column) and MCMC2 (right) best-interpolating Bézier
curves (blue curves) and 1–σ confidence bands for H(z), DA (z), (1 + z)2 DA (z), X3 (z), and X1 (z)/(1 + z) (gray bands), and A(z)
and X2 (z) (green bands), compared to the ΛCDM paradigm from the Planck Collaboration (2020) (dashed red curves).

Article number, page 6 of 11

 Luongo & Muccino: Dark energy reconstructions combining BAO with GCs and intermediate redshift catalogs

β0 β1 β2 β3 β4
MCMC1 0.324 ± 0.068 0.284 ± 0.062 0.355 ± 0.091 −0.221 ± 0.028 0.0327 ± 0.0083
MCMC2 0.328 ± 0.066 0.262 ± 0.062 0.197 ± 0.086 −0.222 ± 0.030 0.0392 ± 0.0084
Table 5. DE reconstruction best-fit coefficients βi and 1–σ errors for MCMC1 and MCMC2 analyses.

2.0 2.0

1.5 1.5

1.0 1.0

WDE HzL
WDE HzL

ç
á á á á á
æ á
æ á
æ á
æ á ç
á á
æ ç á
æ ç á
æ ç á
æ á
æ ç æ á
ç æ á
ç æ á
ç æ á
ç æ á
ç æ á
ç æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ
ç áç
ç áçáçáç
æ ç
æ ç
æ ç
æ ç
ç æ á
æ á
æ á
æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ æ áçáçáçáçáçáçáçáçáç
ç ç ç ç ç ç áç
ç áçáçáçáçáçáçáçáçáçáçáçáçáçáçáç çç ççççççççççççççççççç áçáçáç çç ççççççççççççççççççç
áááç
áááç áááááááááá áááááááááá
áááááááááá
0.5 áááááááááá 0.5

0.0 0.0

-0.5 -0.5
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
redshift z redshift z

Fig. 2. Reconstructed DE behavior with z (blue curve) with attached confidence band (gray area), compared to the expected
behaviors for the ΛCDM (filled circles), the wCDM (empty circles) and the CPL (empty squares) models.

with the expectations for ΛCDM, ωCDM and CPL models Thus, the use of Bézier parametric curves enables one
(Planck Collaboration 2020). to obtain model-independent bounds on h0 , Ωk , ωb and ωm ,
The behavior from the MCMC1 analysis indicates a > having been used, in particular, to:
1–σ deviation from the cosmological constant paradigm at
z . 1.5, with a preference for a dynamical DE behavior (a) infer an interpolated Hubble rate H(z) and estimate h0
which differs (again, at > 1–σ CL) also from the simplest through OHD,
models, i.e., ωCDM and CPL. (b) evaluate an analytic expressions for the angular diam-
The MCMC2 analysis provides a reconstruction that, eter distance DA (z) of GCs, the luminosity distance
besides a small deviation at z . 0.3, is always consistent (1 + z)2 DA (z) of SNe Ia, and BAO uncorrelated ob-
(within 1–σ CL) with ΛCDM, ωCDM and CPL models. servables X with no a priori assumptions on Ωk ,
This indicates that the DE equation of state slowly evolves (c) break the degeneracy between ωb and ωm in the defini-
with z, mimicking the behavior of the cosmological constant tion of rd with the interpolation of the correlated acous-
at z & 0.3 and deviating from it at smaller redshifts. tic parameter A(z).

With respect to Alfano et al. (2024b), the inclusion

6. Final outlooks and perspectives of the Pantheon+ catalog significantly improved the con-
straints on Ωk , further narrowing down the magnitude of
In this paper, we proposed a strategy to separately measure a possible spatial curvature towards negligible values, and
baryons and cold dark matter, by using different probes contributed in solving the Hubble tension in favor of CMB-
that may be classified in terms of low, intermediate and consistent values, i.e., h0 = 0.682+0.023
−0.027 for MCMC1 and
early redshift points. Disentangling the matter sector we +0.024
h0 = 0.677−0.024 for MCMC2. Notably, the inclusion of SNe
infer DE at different stages of its evolution, resorting a
Ia disfavors the estimate h0 = 0.7304±0.0104 based on local
model-independent approach, based on the use of Bézier
SNe Ia anchored to Cepheid stars (Riess et al. 2022).
interpolation.
In general, the cosmic bounds got from MCMC1 and
In this respect, we extended the analysis performed in
MCMC2 (see Table 4 and Figs. A.1–A.2) are in great agree-
Alfano et al. (2024b) including the Pantheon+ catalog of
ment (within 1–σ CL) with the flat ΛCDM model. In par-
SNe Ia, involving de facto all low- and intermediate-redshift
ticular, albeit a slightly positive and small curvature, the
probes, above mentioned.
MCMC2 analysis closely matches the expectations of the
Further, in view of the recent BAO data release by the ΛCDM model (Planck Collaboration 2020), which is quite
DESI Collaboration (2024) – showcasing a ∼ 3–σ discrep- unexpected since DESI-BAO findings seem to question the
ancy at z ∼ 0.7 and a claimed evidence for dynamical DE standard paradigm (DESI Collaboration 2024).
with respect to the SDSS results – we performed two sepa- The corresponding results can be therefore summarized
rate MCMC analyses, namely for the underlying analyses, namely:

- MCMC1, involving SDSS data in conjunction with the - MCMC1, i.e., using SDSS-BAO data, highlighted a de-
other catalogs, and viation from the cosmological constant at z . 1.5 and
- MCMC2, replacing SDSS data set with DESI release. an overall agreement at larger redshifts.
Article number, page 7 of 11
 A&A proofs: manuscript no. main_SZ2_DESI_v1

- MCMC2, i.e., involving DESI-BAO data, exhibits a Capozziello, S., De Laurentis, M., Luongo, O., & Ruggeri, A. 2013,
small deviation at z . 0.3 and either an overall con- Galaxies, 1, 216
sistency (within 1–σ CL) with the ΛCDM paradigm or Carloni, Y., Luongo, O., & Muccino, M. 2024, arXiv e-prints,
arXiv:2404.12068
in line with a slowly evolving DE equation of state. Carlstrom, J. E., Holder, G. P., & Reese, E. D. 2002, Annual Rev. of
Astron. Astrophys., 40, 643
The latter appeared quite unexpectedly but in line with Carroll, S. M. 2001, Living Reviews in Relativity, 4, 1
the above bounds on h0 , Ωk , ωb and ωm . Colgáin, E. Ó., Dainotti, M. G., Capozziello, S., et al. 2024, arXiv
Thus, the above improved cosmic bounds not only en- e-prints, arXiv:2404.08633
Conley, A., Guy, J., Sullivan, M., et al. 2011, ApJ Suppl. Ser., 192, 1
abled a better reconstruction of H(z), but also the first Copeland, E. J., Sami, M., & Tsujikawa, S. 2006, International Journal
attempt of a fully model-independent reconstruction of the of Modern Physics D, 15, 1753
DE evolution with z, based on the Bézier interpolation tech- Cuceu, A., Farr, J., Lemos, P., & Font-Ribera, A. 2019, JCAP, 2019,
nique that can be generalized in future works in terms of 044
D’Agostino, R., Luongo, O., & Muccino, M. 2022, Classical and Quan-
splines or alternatives to numerical reconstructions. Addi- tum Gravity, 39, 195014
tionally, a further refinement in our model-independent pro- De Filippis, E., Sereno, M., Bautz, M. W., & Longo, G. 2005, ApJ,
cedure is mandatory, especially for the reconstruction and 625, 108
the modeling of DE equation of state and, therefore, for the DESI Collaboration. 2024, arXiv e-prints, arXiv:2404.03002
understanding of its nature. In this sense, besides improv- Di Valentino, E., Mena, O., Pan, S., et al. 2021, Classical and Quan-
tum Gravity, 38, 153001
ing all the hereby catalogs, it would be crucial for future Dunsby, P. K. S. & Luongo, O. 2016, International Journal of Geo-
works to: metric Methods in Modern Physics, 13, 1630002
Efstathiou, G. & Bond, J. R. 1999, MNRAS, 304, 75
– resolve the discrepancy between DESI and SDSS data Giarè, W., Najafi, M., Pan, S., Di Valentino, E., & Firouzjaee, J. T.
sets and get joint cosmic bounds; 2024, JCAP, 2024, 035
– work out model- and probe-independent calibration Glanville, A., Howlett, C., & Davis, T. M. 2021, MNRAS, 503, 3510
Haridasu, B. S., Luković, V. V., Moresco, M., & Vittorio, N. 2018,
methods for gamma-ray bursts (Luongo & Muccino JCAP, 2018, 015
2021a) and quasars (Risaliti & Lusso 2019), investigat- Hastings, W. K. 1970, Biometrika, 57, 97
ing and strengthening the constraints up to z ∼ 9. Jiao, K., Borghi, N., Moresco, M., & Zhang, T.-J. 2023, ApJ Suppl.
Ser., 265, 48
Finally, the same procedure will be developed to addi- Jimenez, R. & Loeb, A. 2002a, ApJ, 573, 37
tional DE scenarios to check the goodness of alternative Jimenez, R. & Loeb, A. 2002b, ApJ, 573, 37
models describing the cosmic speed up. Luongo, O. & Muccino, M. 2018, Phys. Rev. D, 98, 103520
Luongo, O. & Muccino, M. 2021a, Galaxies, 9, 77
Luongo, O. & Muccino, M. 2021b, MNRAS, 503, 4581
Luongo, O. & Muccino, M. 2023, MNRAS, 518, 2247
Acknowledgements Luongo, O. & Muccino, M. 2024, A&A, 690, A40
Metropolis, N., Rosenbluth, A. W., Rosenbluth, M. N., Teller, A. H.,
OL acknowledges Gianluca Castignani for fruitful debates & Teller, E. 1953, J. Chem. Phys., 21, 1087
on the topic of this work during his stay at the University Montiel, A., Cabrera, J. I., & Hidalgo, J. C. 2021, MNRAS, 501, 3515
of Camerino. MM is grateful for the hospitality of the Uni- Moresco, M. 2015, MNRAS, 450, L16
versity of Camerino during the period in which this work Moresco, M., Amati, L., Amendola, L., et al. 2022, Living Reviews in
Relativity, 25, 6
has been written and acknowledges Grant No. BR21881941 Moresco, M., Cimatti, A., Jimenez, R., et al. 2012, JCAP, 2012, 006
from the Science Committee of the Ministry of Science Moresco, M., Pozzetti, L., Cimatti, A., et al. 2016, JCAP, 2016, 014
and Higher Education of the Republic of Kazakhstan. The Muccino, M., Luongo, O., & Jain, D. 2023, MNRAS, 523, 4938
authors are thankful to Anna Chiara Alfano for interest- Peebles, P. J. & Ratra, B. 2003, Reviews of Modern Physics, 75, 559
ing discussions related to the subject of model-independent Perlmutter, S., Aldering, G., Goldhaber, G., et al. 1999, ApJ, 517, 565
Planck Collaboration. 2020, A&A, 641, A6
techniques. Ratsimbazafy, A. L., Loubser, S. I., Crawford, S. M., et al. 2017,
MNRAS, 467, 3239
Riess, A. G., Filippenko, A. V., Challis, P., et al. 1998, AJ, 116, 1009
References Riess, A. G., Yuan, W., Macri, L. M., et al. 2022, ApJ Lett., 934, L7
Risaliti, G. & Lusso, E. 2019, Nature Astronomy, 3, 272
Abdalla, E., Abellán, G. F., Aboubrahim, A., et al. 2022, Journal of Schöneberg, N. 2024, JCAP, 2024, 006
High Energy Astrophysics, 34, 49 Scolnic, D., Brout, D., Carr, A., et al. 2022, ApJ, 938, 113
Aizpuru, A., Arjona, R., & Nesseris, S. 2021, Phys. Rev. D, 104, Shafieloo, A. 2007, MNRAS, 380, 1573
043521 Shafieloo, A. & Clarkson, C. 2010, Phys. Rev. D, 81, 083537
Alam, S., Aubert, M., Avila, S., et al. 2021, Phys. Rev. D, 103, 083533 Simon, J., Verde, L., & Jimenez, R. 2005, Phys. Rev. D, 71, 123001
Alfano, A. C., Capozziello, S., Luongo, O., & Muccino, M. 2024a, Stern, D., Jimenez, R., Verde, L., Kamionkowski, M., & Stanford,
Journal of High Energy Astrophysics, 42, 178
S. A. 2010, JCAP, 2010, 008
Alfano, A. C., Luongo, O., & Muccino, M. 2024b, A&A, 686, A30
Sunyaev, R. A. & Zeldovich, Y. B. 1970, Comments on Astrophysics
Alfano, A. C., Luongo, O., & Muccino, M. 2024c, arXiv e-prints,
and Space Physics, 2, 66
arXiv:2408.02536
Sunyaev, R. A. & Zeldovich, Y. B. 1972, Comments on Astrophysics
Amati, L., D’Agostino, R., Luongo, O., Muccino, M., & Tantalo, M.
and Space Physics, 4, 173
2019, MNRAS, 486, L46
Arjona, R., Cardona, W., & Nesseris, S. 2019, Phys. Rev. D, 99, Tomasetti, E., Moresco, M., Borghi, N., et al. 2023, A&A, 679, A96
043516 Wang, D. 2024, arXiv e-prints, arXiv:2404.13833
Aubourg, É., Bailey, S., Bautista, J. E., et al. 2015, Phys. Rev. D, 92, Weinberg, S. 2008, Cosmology
123516 Wolf, W. J. & Ferreira, P. G. 2023, Phys. Rev. D, 108, 103519
Belfiglio, A., Giambò, R., & Luongo, O. 2022, arXiv e-prints, Wolf, W. J., Ferreira, P. G., & García-García, C. 2024a, arXiv e-prints,
arXiv:2206.14158 arXiv:2409.17019
Bernal, J. L. & Libanore, S. 2023, Cosmic Tensions – Lecture Notes Wolf, W. J., García-García, C., Bartlett, D. J., & Ferreira, P. G. 2024b,
Beutler, F., Blake, C., Colless, M., et al. 2011, MNRAS, 416, 3017 Phys. Rev. D, 110, 083528
Blake, C., Brough, S., Colless, M., et al. 2012, MNRAS, 425, 405 Zhang, C., Zhang, H., Yuan, S., et al. 2014, Research in Astronomy
Bonamente, M., Joy, M. K., LaRoque, S. J., et al. 2006, ApJ, 647, 25 and Astrophysics, 14, 1221
Borghi, N., Moresco, M., & Cimatti, A. 2022, ApJ Lett., 928, L4

Article number, page 8 of 11

 Luongo & Muccino: Dark energy reconstructions combining BAO with GCs and intermediate redshift catalogs

Appendix A: MCMC1 and MCMC2 details partial derivative matrixes Jij = ∂βi /∂xj with variables
xj = {α0 , α1 , α2 , Ωk , ωm },
We used a modified version of the code from Arjona et al.
(2019), which is based on the Metropolis-Hastings algo- 
X 4 X 4
rithm (Metropolis et al. 1953; Hastings 1970). 2 M
σ = Jij Cjk Jik , i = 0, ..., 3 ,


 βi


For each of the MCMC analyses run in this paper: j=0 k=0
4 X 4 (B.2)
 X
- we worked out a preliminary MCMC analysis to obtain σ =2 M 2
Jij Cjk Jik + σΩr , i = 4,

 βi

a test covariance matrix, and then j=0 k=0
- we performed the actual MCMC analysis to produce a
single chain. where σΩr is the error on the radiation density parameter,
which is not correlated with the variables xj .
For both MCMC1 and MCMC2 chains the initial 100 steps
have been removed as burn-in, leaving chains with overall
lengths of N ∼ 1.3 × 104 . To asses their convergence we
computed the corresponding autocorrelation functions at
lag k
PN −k


t=1 Lt − L̄ Lt+k − L̄
ρk (L) = PN
2 , (A.1)
t=1 L t − L̄

where Lt is the value of the log-likelihood at the step t

and L̄ is the mean of the chain. From Eq. (A.1), we define
the autocorrelation length l as the lag beyond which the
ACF drops below the threshold ρl (L) = 0.01. MCMC1 and
MCM2 analyses provide small values l = {12, 13}, respec-
tively, indicating that most of the samples in the chains are
independent, as indicated by the effective sample size

N
ESS = Pl , (A.2)
1+2 k=1 ρk (L)

giving high values ESS = {2178, 1967}, respectively.

The best-fit parameters got from MCMC1 and MCMC2
analyses are portrayed in the 1–σ and 2–σ contour plots of
Figs. A.1–A.2.

Appendix B: Coefficients of the dark energy

reconstruction
We here show the explicit expressions of the coefficients
βi used throughout the DE reconstruction reported in
Eq. (16):
2
α2 + α0 (1 + zm ) 2 − 2α1 (1 + zm )

β0 = , (B.1a)
α20 zm
4
√
4 β0 [α2 + α0 (1 + zm ) − α1 (2 + zm )]
β1 =− 2
, (B.1b)
α0 zm
6 (α0 − 2α1 + α2 ) 2 12 (α0 − α1 ) (α0 − 2α1 + α2 )
β2 = + +
α20 zm
4 α20 zm3

2 3α20 + (α2 − 6α1 ) α0 + 2α21

− Ωk , (B.1c)
α20 zm
2

β1 (α0 − 2α1 + α2 ) ωm
β3 =√ 2
− 2 , (B.1d)
β0 α0 zm α0
(α0 − 2α1 + α2 )2
β4 = − Ωr . (B.1e)
α20 zm
4

The errors σβi on the coefficients βi are computed using

M
the covariance matrix Cjk of each MCMC analysis and the
Article number, page 9 of 11
 A&A proofs: manuscript no. main_SZ2_DESI_v1

1.25
Α1

1.15

2.50
Α2

2.25

0.0
Wk

-0.2

0.04

0.03
Ωb

0.02

0.20

0.16
Ωm

0.12

0.65 0.70 1.15 1.25 2.25 2.50 -0.2 0.0 0.02 0.03 0.04 0.12 0.16 0.20
Α0 Α1 Α2 Wk Ωb Ωm

Fig. A.1. MCMC1 contour plots. Darker (lighter) areas display 1–σ (2–σ) confidence regions.

Article number, page 10 of 11

 Luongo & Muccino: Dark energy reconstructions combining BAO with GCs and intermediate redshift catalogs

1.25
Α1

1.15

2.55
Α2

2.40

2.25

0.2
Wk

0.0

0.03
Ωb

0.02

0.20

0.16
Ωm

0.12

0.65 0.70 1.15 1.25 2.25 2.40 2.55 0.0 0.2 0.02 0.03 0.12 0.16 0.20
Α0 Α1 Α2 Wk Ωb Ωm

Fig. A.2. MCMC2 contour plots. Darker (lighter) areas dispaly 1–σ (2–σ) confidence regions.

Article number, page 11 of 11