The combination of strong gravitational lensing and supernovae tests the cosmic

distance duality relation

Name11 and Name21
1
College of Sciences, Northeastern University, Shenyang 110819, China
(Dated: May 20, 2023)
The cosmic distance duality relation between the luminosity distance DL and the angular diameter
distance DA is one of the most fundamental and crucial formulae in cosmology, and Supernova Ia
are deemed to be standard candles in the universe. In this paper, we use the Gaussian process
to reconstruct the smoothed luminosity distance from the Pantheon type Ia supernovae samples.
Meanwhile, We also utilize the seven sets of strong gravitational lensing data from H0LiCOW,
combined with the fitting curve results of SNe Ia, to constrain the parameter η0 . We employ
parameterization and MCMC methods to perform variance tests in order to test the correctness of
the cosmic distance duality relation.

I. INTRODUCTION Supernova explosions are considered as a means of

testing the CDDR. Supernova explosions release a huge
In 1933, Etherington came up with the cosmic distance amount of energy in a short period, and this energy has
duality relation[1] (CDDR) connecting the observed lu- a relatively stable brightness feature. Therefore, they
minosity distance DL and angular diameter distance DA can be regarded as standard candles in the universe,
at the same redshift z, providing a reference standard for measuring distances.
By observing the brightness (such as the apparent
−1
DL DA magnitude) of supernovae and comparing them with
= 1. (1) the known distances of other supernovae, the distances
(1 + z)2
of unknown supernovae can be calculated. Although
It is based on two fundamental hypotheses, different types of supernovae have differences, overall,
they have sufficient similarity and repeatability to
1. The light travels always along null geodesics in the be used as important tools of cosmological distance
Riemannian geometry[2, 4]; measurement and provide a means of studying many
2. The number of photons is conserved[3]. cosmological issues such as dark energy.

The CDDR is always used for exploring the accelerating In this paper, we try to combine SGL and SNe Ia aim-
expansion of the late universe, which is important for ing to test the CDDR by using 1048 Pantheon SNe Ia
observational cosmology, especially in galaxy observing, samples together with results from seven strong gravi-
CMB observing and gravitational lensing, etc. While it tational lenses. In Sec. II, we introduce the time de-
is possible that one of the above two hypotheses would lay effect of strong gravitational lensing and the com-
be violated, which is the signal of new physics [4, 5], so monly used physical quantity, time delay distance. In
testing the CDDR through astronomical observations is Sec. III, based on the 1048 supernova data from the Pan-
worthy and necessary. theon survey, we introduce the concept of distance mod-
ulus through the standard candle feature of supernovae,
Several tests have been proposed in recent years, derive the relationship between luminosity distance and
through inducing a parameter η(z) relying on redshift, redshift, and reconstruct a luminosity-distance-redshift
which can deform CDDR into[6] curve using the principles of Gaussian processes. In Sec.
−1
IV, we discuss the use of the existing H0LiCOW data
DL DA from seven strong gravitational lenses, perform a vari-
= η(z). (2)
(1 + z)2 ance test on the time delay distance using Markov Chain
Monte Carlo methods(MCMC), and parameterize the pa-
Some people attempted to constrain the CDDR with rameter η0 with three different parameterization expres-
gravitational waves and strong gravitational lensing sions to test the correctness of the CDDR.
(SGL), through observed gravitational waves data and
observations of SNe Ia [7–12].

There are several different types of gravitational II. TIME DELAY DISTANCE IN SGL
lensing, distinguishing them by their effects. For weak
gravitational lensing, there is a geometric distortion The time delay is a vital observing consequence in
of the image. Multiple images will happen in strong SGL, which is the difference in time delay between the
gravitational lensing. perturbed and unperturbed light rays [13–17, 19], con-
necting three distances: the angular diameter distances
 2

between observer and lens, observer and source, and lens and β are two nuisance parameters, respectively. Usually,
and source[18], respectively. MB is approximately equal to []
For a single lens plane, the excess time delay of an im-
age at an angular position θ(θ1 , θ2 ) with corresponding H0
MB ≈ −19.3 ± 0.03 + 5 log10 , (7)
source position β(β1 , β2 ) relative to the case of no lensing 60 km s−1 Mpc−1
as
and the best-fitting value of MB is −19.18 when H0 =
l s
 
t(θ i , β) =
1 + zl DA DA 1 2
(θ i − β) − ψ⊥ (θ i ) . (3) 63.41 km s−1 Mpc−1 . For the Pantheon SNe Ia sample,
c DAls 2 the Hubble diagram called SALT2mu is used, which de-
termines α and β before fitting the cosmological param-
Where θ and β are image and source positions in SGLs, eters. Therefore, the Pantheon SNe Ia data only contain
respectively. Strong gravitational lensing can produce mB and its corresponding error, so we will set α = β = 0
multiple images, and the angular positions of these dif- when using the Pantheon SNe Ia data [31].
ferent images can be represented by the angle θ i . ψ⊥ (θ) The relation between DL (in unit of Mpc) and µB reads
is the effective gravitational potential of the lens related
to the source position θ. The light rays emitted at the µB = 5 log10 DL + 25. (8)
same time from a source will reach the observer at differ-
ent times as these paths have different path lengths and Combining Eqs. (7) and (8), we can obtain DL . Further-
pass through different gravitational potentials. There- more, the uncertainty of DL , denoted as ∆DL , becomes
fore, there is a time delay between the multiple images.
The time delay could be written as follows, ln 10 p
∆DL = DL (∆mB )2 + α2 (∆x)2 + β 2 (∆c)2 . (9)
  5
D∆t 1 1
∆tij = (θ i − β)2 − ψ⊥ (θ i ) − (θ j − β)2 + ψ⊥ (θ j ) . We use the Gaussian process method to obtain lumi-
c 2 2
(4) nosity distances and their errors with the redshift range
as 0 < z < 2.3. The Gaussian process is designed to use
Where θ i and θ j are the positions of images i and j, supervised learning to build a model (or function) from
respectively. the training data and then use the model to forecast the
The time delay is related to a quantity, called the “time new samples [32].
delay distance”, which can be used to estimate cosmolog- In this work, we choose the kernel function as the
ical parameters. The time delay distance (D∆t ) is related matérn (v = 9/2) form,
to the angular diameter distance between observer-lens,   
observer-source, and lens-source. It is defined as[20–22] 2 3|z − z̃| 3|z − z̃|
k(z, z̃) = σf − × 1+ (10)
l l
l s
D D
27|z − z̃|2 18|z − z̃|3 27|z − z̃|4

D∆t ≡ (1 + zl ) A ls A . (5)
DA + + + .
7l2 7l3 35l4
(11)
Applying the seven sets of SGL data from H0LiCOW[25],
we can use the MCMC process to constrain the param-
Then, we apply these new luminosity distances to strong
eter η(z) as well as use the Gauss process for curve-
gravitational lensing (SGL) systems observed by the
fitting, which has always been applying to check CDDR
Sloan Lens ACS Survey (SLACS), BOSS Emission-Line
in cosmology[26, 27]. This will be discussed in section IV
Lens Survey (BELLS), Lenses Structure and Dynam-
ics Survey (LSD), and Strong Legacy Survey SL2S sur-
III. SUPERNOVAE
veys, respectively [33]. Additionally, SGL is a system
that obtains angular diameter distances. The luminosity
distance and redshift relationship curve and the recon-
In some previous studies [], data reported in the form structed DL –z points from simulated SNe Ia data for
of distance moduli depend on the details of cosmological each SGL system are plotted in Fig. 1.
models, so it is inappropriate to use these distance moduli
data directly, as they depend on the cosmological models
assumed a priori. In our work, we instead use the original
IV. THE METHODOLOGY AND RESULTS
measurements of the observed B-band magnitude mB ,
the stretch factor x(z), and color parameter c(z) in the
distance modulus, Constraining the parameter η is still a question needed
to be done. Markov chain Monte Carlo process could be
µB (z; α, β, MB ) = mB (z) − MB + αx(z) − βc(z), (6) used in this section, with calculating the variance of the
time delay distance of SGL.
where MB is absolute B-band magnitude, x(z) and c(z) In previous studies, we can see that there are some
are the color and light-curve shape parameters [31], and α methods have been chosen for η parameterization. We
 3

three expressions can be obtained. The result of the first

expression is shown in Fig. 2.
Figure 2 is a limitation plot of the parameter η0 using
the MCMC method for three different parametrization
methods. This is obtained by utilizing the time delay
distances from seven strong gravitational lens data pro-
vided by H0LiCOW. Due to the correspondence between
time delay distance and luminosity distance, the corre-
sponding redshift values can be found in the DL −z curve
reconstructed from supernova data in Fig.1. The calcula-
tion shows that by using the parametrization η = 1 + η0
is restricted to be η0 = 0.0098+0.050
−0.047 at 2σ level and
η0 = 0.0098 ± 0.025 at 1σ level. In the subfigure
2(b), the limitations on parameter η0 at 1σ and 2σ lev-
els obtained by calculation are η0 = −0.023+0.058−0.077 and
FIG. 1. The reconstruction of corrected apparent luminosity
distance–redshift relation DL (z) from Pantheon data set. The η0 = −0.02+0.14
−0.13 in parameterization η = 1 + η0 z. In
+0.20 +0.44
luminosity distances DL and the errors ∆DL obtained from figure 2(c), it is η0 = 0.08−0.24 and η0 = 0.08−0.42 for 1σ
the SNe Ia data for each SGL system are shown with the green and 2σ when choose η = 1 + η0 z/(1 + z) in 2(c).
and blue open circles.

V. CONCLUSION
select three functions for parameterization, through add
a parameter η0 , which has been listed as follows High-redshift Type Ia supernovae are now being used
P1 : η(z) = 1 + η0 , (12) as an observational tool to probe the mechanism of late-
time accelerated expansion of the universe, which is based
P2 : η(z) = 1 + η0 z, (13) on two fundamental assumptions of the cosmic standard
P3 : η(z) = 1 + η0 z/(1 + z). (14) model. Verification of the cosmic distance-duality rela-
tion (CDDR) is therefore crucial to validating the mech-
By using these three different parameterizations and anism of cosmic acceleration and understanding the basic
substituting them into Eq.(2), we can confirm the properties of the universe. Using high-redshift Type Ia
correctness of CDDR, provided that we can constrain supernovae as an observation tool to test the CDDR can
η0 to be equal to zero in the three parameterization yield reliable results and further facilitate our exploration
procedures described above. of the unknown parts of the universe.
In this work, we reconstruct the luminosity distances
H0LiCOW provides seven sets of strong gravitational from the Pantheon SNe Ia samples by using the Gaussian
lensing (SGL) data, which include directly observed time process, apply the luminosity distances to strong gravita-
delay distances D∆t,obs , as well as the redshifts of the lens tional lensing system, and find it fitting the observation
and source. When performing variance testing using the well. By this means, we can combine the simulated lu-
Markov Chain Monte Carlo method, we can perform vari- minosity distances with the angular diameter distances
ance testing on the seven observed strong gravitational from other observation systems to test the validity of the
lensing data sets. According to Equation (5), the time CDDR.
delay distance is expressed in terms of angular diameter
distance. By using the parameterized CDDR and replac-
ing angular diameter distance with luminosity distance, ACKNOWLEDGMENTS
we can see that the time delay distance includes the pa-
rameter η.
After this paper, I would like to express my sincere
The variance test formula for strong gravitational lens-
gratitude and respect to all those who have supported
ing can be written as follows
me. First and foremost, I would like to thank my su-
7 th obs
!2 pervisor, Associate Professor Nan Li, for providing me
2
X D∆t − D∆t
χ = i
. (15) with valuable guidance and attentive care, enabling me
i=1
σi to continuously grow and make progress and complete
this work.
obs
Where D∆t i
is the observed time delay distance for lens Secondly, I would like to extend my sincere thanks to
th
system i, D∆t is the theoretical time delay distance, σi Xue Zhou, whose substantial efforts and help through-
is the corresponding error in the observable. out this project made our work possible and proceed
smoothly.
After considering three parameterization processes and I am deeply grateful to everyone who has been involved
using the MCMC method, the results of variance tests for in this research, including colleagues and institutions who
 4

(a) (b) (c)

FIG. 2. The likelihood functions from our analyses. The variance testing plot using MCMC method for the η = 1 + η0 ,
η = 1 + η0 z/(1 + z) and η = 1 + η0 z/(1 + z) parameterization corresponding to subfigures (a),(b) and (c) are shown in the
figure. We plot the results by using 1500 SNe Ia simulated. It can be seen that η0 is also well constrained around zero within
one and two standard deviations.

provided data and equipment that allowed me to have the opportunity to conduct this research.

[1] Etherington, I. M. H. 1933, PMag, 15, 761 Astronomy andAstrophysicsLibrary,(1992) ISSN 0941-
[2] Adler, S. L. 1971, AnPhy, 67, 599 7834.
[3] Liao, K., Avgoustidis, A., Li, Z. 2015a, PhRvD, 92, [16] S.H. Suyu et al., Dissecting the Gravitational lens
123539 B1608+656. II. Precision Measurements of the Hubble
[4] B. A. Bassett and M. Kunz, Astrophys. J. 607, 661 Constant, Spatial Curvature, and the Dark Energy Equa-
(2004). tion of State, ApJ 711 (2010) 201 [arXiv:0910.2773].
[5] B.A. Bassett and M. Kunz, Phys. Rev. D 69, 101305 [17] T.Treu, Strong Lensing by Galaxies, ARAA,
(2004). 48(2010)125[arXiv:1003.5567]
[6] Yang, Tao, R. F. L. Holanda, and Bin Hu, Astroparticle [18] Rana, Akshay, et al. ”Probing the cosmic distance duality
Physics 108 (2019): 57-62. relation using time delay lenses.” Journal of Cosmology
[7] R.F.L. Holanda, J.A.S. Lima, M.B. Ribeiro, Testing and Astroparticle Physics 2017.07 (2017): 010.
the distance-duality relation with galaxy clusters and [19] Kumar, D., Jain, D., Mahajan, S., Mukherjee, A., Rani,
type ia supernovae, Astrophys. J. 722 (L233) (2010). N. (2021). Constraining cosmological and galaxy param-
arXiv:1005.4458. [astro-ph.CO]. eters using strong gravitational lensing systems. Physical
[8] P. Wu, Z. Li, X. Liu, H. Yu, Cosmic distance-duality Review D, 103(6), 063511.
relation test using type ia supernovae and the baryon [20] Refsdal S., 1964, MNRAS, 128, 307
acoustic oscillation, Phys. Rev. D 92 (023520) (2011). [21] Schneider P., Ehlers J., Falco E. E., 1992, Gravitational
[9] X. Yang, H.R. Yu, T.J. Zhang, Constraining smooth- Lenses. Springer, doi:10.1007/978-3-662-03758-4
ness parameter and the DD relation of dyer-roeder [22] Suyu S. H., Halkola A., 2010, AA, 524, A94
equation with supernovae, JCAP 1306 (007) (2013). [23] Schneider, P., Ehlers, J., Falco, E.E. Gravitational
arXiv:1305.6989. [astro-ph.CO]. Lenses (Springer-Verlag Berlin Heidelberg, 1992)
[10] S. Jhingan, D. Jain, R. Nair, Observational cosmology [24] Schneider, P., Ehlers, J., Falco, E. E., Gravitational
and the cosmic distance- duality relation, J. Phys. Conf. Lenses, Springer-Verlag, Berlin, 1992.
Ser. 484 (012035) (2014). arXiv:1403.2070. [gr-qc]. [25] Wong, Kenneth C., et al. ”H0LiCOW XIII. A 2.4% mea-
[11] A. Rana, D. Jain, S. Mahajan, A. Mukherjee, R.F.L. surement of H0 from lensed quasars: 5.3 tension between
Holanda, Probing the cosmic distance duality relation early and late-Universe probes.” Monthly Notices of the
using time delay lenses, JCAP 1707 (07) (2017) 010. Royal Astronomical Society 498.1 (2020): 1420-1439.
arXiv:1705.04549. [astro-ph.CO]. [26] Lyu, M. Z., Li, Z. X., Xia, J. Q. (2020). Testing the
[12] K. Liao, Z. Li, S. Cao, M. Biesiada, X. Zheng, Z.H. Cosmic Distance Duality Relation with the Latest Strong
Zhu, The distance dual- ity relation from strong grav- Gravitational Lensing and Type Ia Supernovae. The As-
itational lensing, Astrophys. J. 822 (2) (2016) 74. trophysical Journal, 889(2), 114.
arXiv:1511.01318. [astro-ph.CO]. [27] Kumar, D., Jain, D., Mahajan, S., Mukherjee, A., Rani,
[13] S.Refsdal,On the possibility of determining Hubble’s pa- N. (2021). Constraining cosmological and galaxy parame-
rameter and the masses of galaxies from the gravitational ters using strong gravitational lensing systems. Research
lens effect, MNRAS 128 (1964) 307. in Astronomy and Astrophysics, 21(3), 045.
[14] P. Schneider, C.S. Kochanek, J. Wambsganss, Gravita- [28]
tional Lensing: Strong, Weak and Micro, Springer-Verlag [29] Bassett, B. A., Kunz, M. (2004). Cosmic distance-
Berlin Heidelberg (2006) eBook-ISBN 978-3-540-30310-7. duality as a probe of exotic physics and acceleration.
[15] P.Schneider, J. EhlersE.E.Falco, Gravitational Lenses, Physical Review D, 69(10), 101305.
 5

[30] Scolnic, D. M., Jones, D. O., Rest, A., Pan, Y. C., physical Journal, 888(1).
Chornock, R., Foley, R. J., ... Smith, K. W. (2018). [32] Seikel, M., Clarkson, C. (2013). Optimising Gaussian
The complete light-curve sample of spectroscopically processes for reconstructing dark energy dynamics from
confirmed SNe Ia from Pan-STARRS1 and cosmolog- supernovae.
ical constraints from the combined Pantheon sample. [33] Suyu, S. H., Bonvin, V., Courbin, F., Fassnacht, C. D.,
Physical Review D, 98(2), 022005. doi: 10.1103/Phys- Rusu, C. E., Sluse, D., ... Spiniello, C. (2017). H0LiCOW
RevD.98.022005. – I. H0 lenses in COSMOGRAIL’s Wellspring: program
[31] Lyu, M. Z., Li, Z. X., Xia, J. Q. (2020). Testing the overview. Monthly Notices of the Royal Astronomical So-
cosmic distance duality relation with the latest strong ciety, 468(3).
gravitational lensing and type Ia supernovae. The Astro-