Professional Documents
Culture Documents
ABSTRACT
Abell 514 (A514) at z = 0.071 is an intriguing merging system exhibiting highly elongated (∼1 Mpc)
X-ray features and three large-scale (300 ∼ 500 kpc) bent radio jets. To dissect this system with its
multi-wavelength data, it is critical to robustly identify and quantify its dark matter (DM) substruc-
tures. We present a weak-lensing analysis of A514 using deep Magellan/Megacam observations. Com-
bining two optical band filter imaging data obtained under optimal seeing (∼0.′′ 6) and leveraging the
proximity of A514, we achieve a high source density of ∼46 arcmin−2 or ∼6940 Mpc−2 , which enables
high-resolution mass reconstruction. We unveil the complex DM substructures of A514, which are char-
acterized by the NW and SE subclusters separated by ∼0.7 Mpc, each exhibiting a bimodal mass distri-
+0.24
bution. The total mass of the NW subcluster is estimated to be MNW 14
200c = 1.08−0.22 × 10 M⊙ and is fur-
ther resolved into the eastern (MNW +1.4 13 NWW +2.3
200c = 2.6−1.1 × 10 M⊙ ) and western (M200c = 7.1−2.0 × 10 M⊙ )
E 13
SE +0.28 14
components. The mass of the SE subcluster is M200c = 1.55−0.26 × 10 M⊙ , which is also further re-
solved into the northern (MSE +1.8 13 SES +3.0 13
200c = 2.9−1.3 × 10 M⊙ ) and southern (M200c = 8.5−2.6 × 10 M⊙ ) com-
N
ponents. These four substructures coincide with the A514 brightest galaxies and are detected with
significances ranging from 3.4σ to 4.8σ. Comparison of the dark matter substructures with the X-ray
distribution suggests that A514 might have experienced an off-axis collision, and the NW and SE
subclusters are currently near their apocenters.
Keywords: galaxies: clusters: weak lensing, galaxies: clusters: individual: Abell 514
nor the galaxy distribution is a dependable proxy of the excellent opportunity to probe the complex substruc-
subhalo masses. tures in great detail. We note, however, that since the
Weak gravitational lensing (WL hereafter) is a pow- intrinsic lensing efficiency is low, the requirement for
erful tool for mapping the dark matter distribution of systematics control is high for low-redshift WL.
galaxy clusters. Since it probes the projected mass solely Throughout this paper, we assume a flat ΛCDM cos-
based on the gravitational lensing effects on the shapes mology with H0 = 70 km s−1 Mpc−1 and Ωm = 0.3. At
of background galaxies, WL does not necessitate as- the cluster redshift (z = 0.071), the angular size of
sumptions about the dynamical state of the lens. In 1′ corresponds to the physical size of ∼82 kpc. M200c
particular, merging clusters are believed to deviate sub- (M500c ) is defined as the mass enclosed by a sphere in-
stantially from the hydrostatic equilibrium, making this side which the average density equals to 200 (500) times
merit even more critical. Rich dark matter substruc- the critical density at the cluster redshift. All errors are
tures have been detected by WL in various merging clus- quoted at the 1-σ level unless otherwise noted.
ters (e.g., Okabe & Umetsu 2008; Jee & Tyson 2009; Jee
et al. 2012; Wittman et al. 2014; Martinet et al. 2016; 2. OBSERVATION AND DATA REDUCTION
Finner et al. 2017).
The A514 field was observed using the Magel-
In this study, we present the first WL analysis of the
lan/Megacam imager on the night of October 23, 2022,
low-redshift (z = 0.071) merging galaxy cluster Abell
in the g- and r- bands (PI: W. Lee). The Megacam im-
514 (hereafter A514). One prominent feature of A514
ager’s focal plane is composed of 36 CCDs, yielding a to-
is its extended X-ray emission stretching ∼1 Mpc from
tal field of view of 25′ ×25′ (McLeod et al. 2015). We ap-
northwest (NW) to southeast (SE), which connects two
plied dithering and field rotation among all 18 pointings
galaxy overdensities. With XMM-Newton observations,
for both g- and r- filters. This significantly reduces sev-
Weratschnig et al. (2008) claimed the detection of an
eral artifacts that negatively affect WL analysis around
ICM density discontinuity and suggested an ongoing
bright stars, including their diffraction spikes and satu-
NW-SE merger. Another intriguing characteristic of
ration trails. Consequently, the number of usable source
A514 is its peculiar radio features, including three head-
galaxies for WL increases. The total exposure for each
tail radio galaxies with bent morphologies (e.g., Burns
filter is 5400 s. The mean seeings of the g- and r- fil-
et al. 1994; Govoni et al. 2001). Remarkably, the ra-
ters are 0.′′ 61 and 0.′′ 60, respectively, which are ideal for
dio emission originating from the two radio lobes of
ground-based WL.
the AGN in the SE region extends ∼0.7 Mpc towards
We applied initial bias and sky-flat correction, and
the southern outskirts with multiple bends (Lee et al.
cosmic rays were masked using ASTROSCRAPPY (McCully
2023a). Although there is no direct evidence to date,
et al. 2018). All flat-fielded frames underwent processing
it is highly probable that these peculiar radio features
by SExtractor (Bertin & Arnouts 1996) for the prepa-
are influenced by the ongoing merger. One of the goals
ration of the astrometric and photometric calibration
of the current WL study is to identify and quantify the
using SCAMP (Bertin 2006). The final deep mosaic im-
substructures in A514, which will provide critical input
ages where the weak-lensing signal is measured were cre-
to future numerical simulations. An outstanding ques-
ated with SWARP (Bertin et al. 2002) by stacking all the
tion in merging cluster physics is how AGN plasma gets
frames precisely with the refined World Coordinate Sys-
redistributed within the ICM due to merger-driven gas
tem (WCS) information.
motions (see Vazza & Botteon 2024, for review).
In general, mean-stacked images provide a better
WL analysis of low-redshift clusters has been consid-
S/N than the median-stacked images. However, a
ered a challenge relative to intermediate redshift (z ∼
plain mean-stacking scheme is vulnerable to outliers.
0.5) clusters because of their low lensing efficiency (i.e.,
For this reason, we used the SWARP keyword COM-
smaller distortion given the same mass and source red-
BINE TYPE=CLIPPED (Gruen et al. 2014) for outlier-
shift). However, this disadvantage is outweighed by the
clipped mean stacking. This option first generates a
large projected area thanks to their proximity, which
median-stacked image and then produces the final im-
provides a significantly higher number of background
age through inverse-variance weight-averaging, clipping
galaxies per physical area at the lens redshift. For
the outliers that deviate significantly from the median-
instance, HyeongHan et al. (2024) demonstrated that
stacked image. Additionally, this option generates a log
while at z = 0.02 the lensing efficiency is lower by an
file containing information about photometric outliers
order of magnitude than at z = 0.5, the net gain in S/N
in each frame. This information is essential for later
per physical area is approximately three times higher.
PSF modeling, where the appropriate weight needs to
Thus, the low-redshift merging cluster WL provides an
be applied. Readers are referred to Gruen et al. (2014)
Weak-Lensing Analysis of the Complex Merging Cluster Abell 514 3
and Jee et al. (2015) for more details. We found that relation can be linearized with the Jacobian matrix A:
the background rms is about ∼11% lower in the mean- !
stacked image than in the median-stacked image, both ∂β 1 − g1 −g2
A= = (1 − κ) , (4)
for g- and r- bands. Although the overall flux level ∂θ −g2 1 + g1
is higher in the r-band, the photometric S/N for faint
where g is the reduced shear, defined as
galaxies is slightly higher in the g-band. Therefore, we
conducted WL analysis using both g- and r-bands. γ
g= . (5)
We executed SExtractor in dual-image mode, where 1−κ
the detection image was generated by weight-averaging In Equation (5), γ is the shear, which is directly related
the g- and r- band mosaic images. The dual-image mode to the mass of the lens. In the weak-lensing regime (κ ≪
maintains consistent isophotal area across all filters, en- 1), the reduced shear g approaches the shear γ.
suring that object colors were measured from identical We can use the complex notation for the reduced
isophotal apertures. We defined objects as regions with shear, introducing its two components:
more than five connected pixels, each having a flux above
1.5 times the local background rms. g = g1 + ig2 . (6)
3. WEAK LENSING ANALYSIS Here, g1 distorts the image along the x-axis and y-axis
3.1. Basic Theory directions, while g2 distorts the image along the y =
x and y = −x directions. In the same way, we can
Measuring the local ensemble-averaged distortion of define the intrinsic (unlensed) ellipticity ϵ and observed
background galaxies caused by the lens allows us to (lensed) ellipticity e using the same complex notation:
investigate the dark matter distribution and mass of
the galaxy cluster. In this section, we provide a brief ϵ = ϵ1 + iϵ2 , e = e1 + ie2 . (7)
overview of the basic theory and formalism of WL. Read-
ers are referred to review papers (e.g., Narayan & Bartel- Then, with the presence of the local reduced shear g,
mann 1996; Bartelmann & Schneider 2001; Schneider the Equation (4) will transform the intrinsic ellipticity
2006) for details. ϵ to the observed ellipticity e as:
Light bundles from background galaxies bend under ϵ+g
the gravitational tidal field caused by a foreground clus- e= , (8)
1 + g∗ ϵ
ter. The coordinate mapping from the background
source plane β to the observed image plane θ can be where the asterisk (*) denotes the complex conjugate.
described by the lens equation: Assuming that the intrinsic orientation of background
galaxies is random, i.e. ⟨ϵ⟩ = 0, Equation (8) becomes:
β = θ − α(θ), (1)
g = ⟨e⟩ . (9)
where α is the scaled deflection angle, which involves
Therefore, we can use ⟨e⟩ as an unbiased estimator of
the convergence κ via:
the reduced shear for the weak distortions in an ideal
1
Z
θ − θ′ case where no systematic bias is present.
α(θ) = d2 θ ′ κ(θ ′ ) . (2)
π R2 |θ − θ ′ |2
3.2. PSF Modeling
The convergence κ is a dimensionless quantity defined Observed galaxy shapes are affected by point spread
by the projected surface mass density (Σ) normalized by functions (PSFs), and this effect becomes larger for the
the critical surface density (Σc ): ground-based observations because of atmospheric tur-
Σ c2 Ds bulence. This phenomenon introduces a systematic bias
κ= , Σc = , (3) in the shear measurement, making Equation (9) invalid.
Σc 4πG Dl Dls
Therefore, precise and accurate PSF modeling is a crit-
where c is the speed of light, G is the gravitational ical step in WL signal detection. We adopted the prin-
constant, and Ds , Dl , and Dls denote the angular di- cipal component analysis (PCA) technique (Jee et al.
ameter distances from the observer to the source, from 2007; Jee & Tyson 2011) for our PSF modeling.
the observer to the lens, and from the lens to the source, In the mean-stacked image, the PSF is equivalent to
respectively. a linear combination of the PSFs from all contribut-
If the size of a source is much smaller than the angular ing individual frames, with the same weight used in the
scale over which the lens properties vary, the mapping mosaic image stacking stage. Since the PSF pattern
4 Ahn et al.
Figure 1. Megacam PSF modeling for g-band (left) and r-band (right) images. Red circles represent the ellipticity components
of the observed stars in the mosaic image. Black circles represent the ellipticity components of the residuals between the observed
stars and the modeled PSF. After correction, the residuals in both g- and r-bands are centered at the origin with reduced scatter,
ensuring the precise and accurate PSF modeling.
has discontinuities across the CCD boundaries (e.g., Jee earlier, these weights are identical to those applied in
et al. 2013), our previous WL studies modeled PSFs the creation of the mosaic image.
CCD by CCD and exposure by exposure (e.g., Finner Figure 1 shows the results from the PSF modeling in
et al. 2021; Cho et al. 2022). However, the A514 field g- and r- bands. The centroids of the residual ellip-
does not provide a sufficient number of stars suitable ticities (the difference between the model and observed
for modeling the PSF within each CCD. Fortunately, PSF pattern) are well-clustered around the origin. Both
we found that for a given exposure, the star ellipticity the reduction in scatter and the shift toward the origin
pattern flows smoothly across the CCD gaps when we (0,0) confirm that our PSF model on the mosaic image
put all 36 CCDs together, although the global pattern is precise and accurate.
changes exposure by exposure. Therefore, in this study,
we chose to model the PSF for the entire focal plane (all 3.3. Multiple-filter Shape Measurement
36 CCDs) exposure by exposure and stack the models We used a PSF-convolved elliptical Gaussian function
to create the final PSF model of the mosaic image. to fit the light profile of background galaxies. Although
We initially identified “good stars” in each expo- a Sérsic model provides a better description, we find
sure, which satisfy several criteria. First of all, that the resulting shape measurement is noisier since the
these stars should be isolated without being satu- fitting assigns higher weights to the noisy pixels in the
rated. And they should not be located in the peripheral region (Jee et al. 2013), reducing the number
vicinity of the CCD boundaries. Also, their cen- of usable galaxies.
troid measured by the peak positions (XPEAK IMAGE, The PSF-convolved elliptical Gaussian function
YPEAK IMAGE) should agree with the windowed cen- M (x, y) is:
troid (XWIN IMAGE, YWIN IMAGE) within 1 pixel. We
were able to select about ∼200 “good stars” from M (x, y) = G(x, y) ⊗ P (x, y), (10)
each exposure for both g- and r-band images. Then,
where P (x, y) is the model PSF at the source position
all surviving stars were cropped into 21 pixel × 21
and G(x, y) is the elliptical Gaussian function:
pixel postage-stamp images and used for PCA. To cre-
ate a PSF model in the mosaic image, we performed (∆x cos θ + ∆y sin θ)2
an inverse-variance weight-averaging of all contributing G(x, y) = Ibg + Ipk exp −
2σx2
PSF models from individual exposures. As mentioned (11)
(−∆x sin θ + ∆y cos θ)2
− ,
2σy2
Weak-Lensing Analysis of the Complex Merging Cluster Abell 514 5
where Ibg and Ipk denote the background level and max-
imum flux level, respectively. The ∆x and ∆y rep-
resent the distances from the centroid to each pixel
(x, y), respectively, σx2 and σy2 are the variances, and
θ is the position angle measured counterclockwise from
the x−axis. We fixed the background level and centroid
using the SExtractor measurements BACKGROUND and
(XWIN IMAGE, YWIN IMAGE), respectively. By reduc-
ing the free parameters in G(x, y) from seven to four,
we can further reduce the measurement error, thereby
increasing the source density of usable galaxies in our
analysis.
We used the MPFIT code (Markwardt 2009) to ob-
tain the best-fit model that minimizes the difference be-
tween the model M (x, y) and the observed galaxy profile
O(x, y). MPFIT performs non-linear least squares mini- Figure 2. Ellipticity measurement errors determined from
mization based on the Levenberg–Marquardt algorithm. MPFIT as a function of r−band magnitude. The measurement
error for each case is the average of the e1 and e2 measure-
With the best-fit parameters, it also returns their 1-
ment errors. Across all magnitude ranges, especially for faint
sigma errors computed from the Hessian matrix. The sources, combining two filters in the shape measurement de-
free parameters are Ipk , e1 , e2 , and b, where e1 and e2 creases the measurement error. We obtained ∼20% more
are the two ellipticity components, and b is the semi- sources through simultaneous fitting given the same error
minor axis. We defined the e1 and e2 as: cut (dotted line, δe = 0.3).
e1 = e cos(2θ),
(12) all magnitude ranges, and the difference is larger for
e2 = e sin(2θ), fainter sources. Consequently, extracting shape informa-
where ellipticity e is defined using semi-major axis a and tion from multiple filters reduces the measurement error,
semi-minor axis b as e = (a − b)/(a + b). thereby increasing the total number of usable galaxies
We cropped each source galaxy into a square postage given the same measurement error cut. In this study,
stamp image, which constructs the χ2 function. We we were able to obtain ∼20% more sources through si-
combined the two χ2 functions for g- and r−band im- multaneous fitting.
ages and obtained one best-fit model that minimizes the A potential concern is the difference in shape between
total χ2 function as follows: filters. Detailed morphologies of galaxies vary from ul-
traviolet to infrared (e.g., Kuchinski et al. 2000). How-
X [Og (x, y)i − Mg (x, y)i ]2 ever, we used two broadband optical filters with a rel-
f = χ2g + χ2r = 2
σg,i atively small separation in effective wavelength. As a
i
(13) result, the systematic difference in intrinsic ellipticity
[Or (x, y)i − Mr (x, y)i ]2
+ 2 , between the two filters is expected to be smaller than
σr,i the measurement error (e.g., Lee et al. 2018).
where O(x, y)i and M (x, y)i indicate the observed and The analytic galaxy profiles cannot be the perfect rep-
modeled galaxy postage stamp images, respectively, and resentation of the real galaxies. This model bias can lead
σi is the background rms noise. The summation over to systematic errors in the shear measurement. Also,
i represents the summation of all pixels belonging to non-linear relations between pixel and ellipticity con-
the square postage stamp image for each galaxy. The tribute to the systematic errors (noise bias). By run-
two models Mg (x, y) and Mr (x, y) share the same shape ning image simulations that match observational data,
parameters (e1 , e2 , b) except for the peak intensities. we derived a global multiplicative factor of m = 1.14
Figure 2 shows the measurement errors of ellipticity with a negligible (< ∼10−3 ) additive bias (Jee et al.
components as a function of r-band magnitude. We 2016). Readers are referred to Jee & Tyson (2011) and
compared three cases: two involving the fitting pro- Jee et al. (2013) for details.
cesses using only the g- or r-band image and the other
using both g- and r- bands simultaneously. Measure- 3.4. Source Selection
ment errors from simultaneous fitting were consistently Optimal selection of background sources could be
lower than those from single-band-only fitting across made by choosing objects whose photometric redshifts
6 Ahn et al.
Figure 3. Color-magnitude diagram of the A514 field. Red 6. The SExtractor flag should be less than 4 to
dots represent spectroscopically confirmed members from exclude potentially problematic detections.
the NASA/IPAC Extragalactic Database (NED) survey, and
Finally, through visual inspection, non-astrophysical
blue dots represent the background sources after applying
source criteria. Green dots represent photometric member sources, such as diffraction spikes or rings near bright
candidates selected through the red sequence (dashed-gray stars, were discarded. The final source catalog contains
line). Many of the PSF stars, denoted as cyan dots, exhibit a total 45,353 galaxies, which provides a mean source
the horizontal locus redder than the cluster red sequence. density of ∼46.3 arcmin−2 (∼6940 Mpc−2 at the A514
redshift). Since the contribution from each source de-
are greater than that of the cluster. However, this is not pends on its measurement error, we can define an effec-
the case for our data, and we selected WL sources based tive WL source density (Jee et al. 2014a) as follows:
on their color-magnitude-redshift relation. X 2
σSN
Figure 3 illustrates the color-magnitude diagram neff = 2 + (δe )2 , (14)
σSN i
(CMD) of the A514 field. In this diagram, the red-
sequence galaxies of A514 exhibit clear and tight color- where σSN is the dispersion of the source ellipticity dis-
magnitude relation. Typically, for weak-lensing analysis tribution per component (∼0.25), and δei is the ith
of intermediate redshift (z ∼ 0.5) clusters, background galaxy’s ellipticity measurement error per component.
sources are often chosen from a population fainter and From this equation, we estimate the effective source den-
bluer than the cluster red sequence. However, given sity to be ∼40.6 arcmin−2 .
that A514 is located at z ∼ 0.07, a majority of galax- 3.5. Redshift estimation
ies sufficiently fainter than the cluster red sequence are
A weak-lensing signal is proportional to the angu-
likely behind the cluster regardless of their colors. We
lar diameter distance ratio β = Dls /Ds (Eqn 3). We
selected sources with r−band magnitudes in the range
estimated the source redshift by employing the COS-
22 < rauto < 27 and isophotal (g − r) colors within
MOS2020 photometric redshift catalog (Weaver et al.
−0.5 < (g − r)iso < 1.5. For the galaxies whose col-
2022). As the catalog does not provide magnitudes in
ors match those of the A514 red sequence, we increase
the Megacam photometric system, we used the Sub-
the magnitude upper limit to rauto = 24 to minimize
aru/Hyper Suprime-Cam g- and r-magnitudes as proxies
possible contamination from the faint end of the A514
for our Megacam g- and r-magnitudes, respectively. We
red sequence. The upper bound of the color is to avoid
applied the same color-magnitude criteria to the COS-
contamination from the Milky Way stars, which show
MOS2020 catalog, weighting the COSMOS2020 galaxies
distinct horizontal locus above the A514 red sequence
with the number density ratio in each magnitude bin be-
as shown in Figure 3. We estimated that only ∼4%
tween the COSMOS2020 and A514 fields (e.g., Finner
of the resulting source population is in the foreground
et al. 2017; Kim et al. 2019). This weighting scheme ac-
by applying the same source criteria to the COSMOS
counts for the difference in depth between the two fields.
photo-z catalog.
Assigning zero weights to galaxies with redshifts smaller
As a measure to exclude spurious sources whose shape
than that of A514, we calculated the effective β:
measurements are unreliable, the following additional
criteria were applied: Dls
⟨β⟩ = max 0, . (15)
Ds
Weak-Lensing Analysis of the Complex Merging Cluster Abell 514 7
We obtained ⟨β⟩ = 0.865, which corresponds to an ef- radio jet emissions (Lee et al. 2023a). The radio galaxy
fective source redshift of zeff = 0.607. The assumption A514SEb, which is hosted by the mass clump SEN , ex-
that all sources are located at a single redshift intro- hibits a relatively high line-of-sight velocity difference
duces bias because of the non-linearity in β. To address compared to the cluster redshift and the neighboring
the issue, we applied a first-order correction (Seitz & two brightest galaxies.
Schneider 1997; Hoekstra et al. 2000) as follows: Figure 5 presents overlays of the mass contours with
! the following: a) number density, b) r-band luminosity,
g′ β2 c) X-ray emission, and d) X-ray plus radio emissions.
g
=1+ 2 − 1 κ, (16)
The comparison with the galaxy number density and
⟨β⟩
luminosity maps shows that the galaxy distributions cor-
where g ′ and g are the observed and true reduced shear, relate well with the WL mass, although scrutiny sug-
respectively. We obtained β 2 = 0.781, which scales gests that the correlations might not be as strong as
the observed shear g ′ by a factor of (1 + 0.044κ). those observed in the BCG-mass peak comparison men-
tioned earlier. In the A514NW region, there are two
4. RESULTS number density peaks with ∼1′ offsets from their near-
4.1. Mass Reconstruction est mass peaks (Figure 5a). In contrast, the luminosity
map shows only one dominant peak centered on NWW
One of the primary goals of our weak-lensing anal-
(Figure 5b). In the A514SE region, the number density
ysis is the robust mass reconstruction of A514 and the
map shows one weak peak between SEN and SES . In
identification of its substructures. Since the convergence
the luminosity map, the distribution is highly elongated
κ is proportional to the projected surface mass density
in the N-S direction with its peak ∼1′ north of SEN .
(Eqn 3), hereafter we use the terms “mass map” and
The smoothed northern X-ray peak aligns well with
“convergence map” interchangeably to refer to the re-
NWW , while the southern X-ray peak is ∼1′ north of
constructed convergence map. We used the FIATMAP
SEN (Figure 5c). Since the current X-ray observation is
code (Fischer & Tyson 1997; Wittman et al. 2006),
not sufficiently deep (three pointings of ∼15 ks observa-
which implements the Kaiser & Squires (1993) shear-
tion), the presence and location of the southern X-ray
to-convergence inversion in real space.
peak are uncertain. However, despite the low exposure,
Figure 4 presents the reconstructed mass distribu-
the current X-ray data indicate significant emission be-
tion of the A514 field. The overall mass distribution
tween A514NE and A514SW forming a “ridge”, which
can be characterized by the two bimodal mass distri-
serves as evidence for the post-merger status.
butions: A514NW and A514SE separated by ∼750 kpc
Figure 5d shows that, as mentioned above, three
(∼9′ ) with A514NW (A514SE) further resolved into
bright bent radio jets originate from the three brightest
NWE and NWW (SEN and SES ). Our bootstrapping
galaxies A514NWa, A514NWb, and A514SEb, which are
analysis shows that all four mass peaks have a signif-
hosted by the mass clumps NWW , NWE , and SEN , re-
icance greater than 3.4 σ (see §5.1). We found that
spectively. The surface brightness discontinuity claimed
the mass clump denoted by WL J0447.7-2020 is a back-
by Weratschnig et al. (2008) is located near the north-
ground cluster at z ∼ 0.6. The mass peak is in excellent
western boundary of the northern X-ray emission.
agreement with its BCG candidate. Since this cluster
has not been reported in the literature, this marks its
first detection based purely on WL signal. We present
additional information on this cluster in Appendix A. 4.2. Mass Estimation
We note that the positions and significances of the five In this section, we present mass estimation based on
mass peaks described above are also consistent with the fitting multiple Navarro–Frenk–White (NFW; Navarro
results obtained from other mass reconstruction algo- et al. 1997) halos. Although the three mass clumps
rithms such as the one in Kaiser & Squires (1993). (A514NW, A514SE, and WL J0447.7-2020) are widely
We label the six brightest cluster galaxies in A514 in separated (∼9.0′ between A514NW and A514SE, and
Figure 4. Since it is difficult to obtain accurate photom- ∼6.5′ between A514NW and WL J0447.7-2020), we em-
etry, we do not attempt to order them according to their ploy simultaneous fitting to minimize the influence from
brightness level; the suffix “a”-“c” is assigned arbitrarily. the neighboring structures in our three-halo fitting. We
All four WL peaks in A514 coincide with the brightest also present the results from five-halo fitting (NWE ,
cluster galaxies. Remarkably, among these four galax- NWW , SEN , SES , and WL J0447.7-2020). Because of
ies, three of them, A514SEb, A514NWa, and A514NWb, the degeneracy between the two parameters (concentra-
are the active galactic nuclei (AGNs) with strong bent tion c and scale radius rs ) in the NFW halo model, we
8 Ahn et al.
Figure 4. Convergence contours of the A514 field over a Magellan/Megacam color composite image. The contour levels
represent statistical significance (see §5.1). On a large scale, A514 is comprised of two subclusters: A514NW and A514SE.
A514NW (A514SE) is further resolved into NWE and NWW (SEN and SES ). All four mass peaks, whose significances range
from 3.4σ to 4.8σ, are centered on the brightest galaxies. The mass clump denoted as WL J0447.7-2020 is a background cluster
detected at the 6.1σ level.
used the mass-concentration (M − c) relation of Dutton where gjm is the jth component of the predicted reduced
& Macciò (2014). shear at the ith source galaxy position (xi , yi ) as a
For three-halo fitting, the mass center for each clump function of the halo mass vector M = [M1 , M2 , ..., Mn ],
within A514 was determined by computing the first mo- where n is the number of halos, and gjo is the jth com-
ment of the local convergence (white triangles in Figure ponent of the observed ellipticity at the same location.
5d). We excluded background sources within R < 100′′ We used the mathematical formulation of NFW shear
to minimize the influence of the substructures. (Wright & Brainerd 2000) to calculate the gjm at every
We performed the Markov Chain Monte Carlo source galaxy position. Note that the first-order correc-
(MCMC) analysis to obtain the mass posteriors with the tion in Equation (16) is applied to gjm . We assumed a
following log-likelihood function L (Kim et al. 2019): flat prior of 1012 M⊙ < M200c < 1016 M⊙ . We display
the resulting posteriors in Figure 6 (see also Table 1 for
the summary of the best-fit results and the used mass
centers).
X X [gjm (M, xi , yi ) − gjo (xi , yi )]2
L=− 2 + (δe )2 (17)
i j=1,2
σSN i
Weak-Lensing Analysis of the Complex Merging Cluster Abell 514 9
Figure 5. Comparison of the A514 mass distribution with the galaxy, X-ray, and radio distributions. a) Mass (black) overlaid
with the adaptively smoothed galaxy number density. b) Mass (black) overlaid with the galaxy luminosity in r-band. c) Mass
(black) overlaid with the adaptively smoothed XMM-Newton X-ray emission. d) Mass (white) overlaid with the Magellan
color composite + XMM-Newton X-ray (red) + uGMRT band-3 radio (green). The galaxies used in panels a) and b) are the
spectroscopic and photometric (red-sequence) members. In panel d), the luminosity peaks marked with orange pentagons are
estimated by flux-weighted averaging. The mass centers marked with white triangles are obtained from the first moments of
the local convergence values. White dashed contours represent the 68% confidence intervals of the mass centroids determined
from 1000 bootstrap realizations (§5.1).
Figure 7. Five-halo mass estimation in the same format used in Figure 6. The mass centers were set to the convergence peaks.
In some cases where the distance between the two neighboring substructure is relatively small, the degeneracy is relatively high
(the Pearson correlation coefficients are ρ = −0.535 for the NWE − NWW pair and ρ = −0.685 for the SEN − SES pair).
Figure 9. Left: Color-magnitude diagram of the background cluster WL J0447.7-2020. Objects located within a r = 80′′ circle
around the BCG of this cluster are marked as orange circles, exhibiting a distinct red sequence redder than that of A514 (red
dashed line). Blue dots represent the sources selected in §3.4. Right: Mass contours (yellow) and XMM-Newton X-ray emission
(orange) overlaid on the color image of WL J0447.7-2020. Cluster member candidates are marked as white circles.
We use simultaneous NFW halo fitting with the M −c that on a large scale, A514 is an off-axis post-merger
relation to determine the substructure masses. The system observed nearly at its apocenter. However, with
total mass of the NW subcluster is estimated to be the current data, it is difficult to investigate the merger
+0.24
MNW 14
200c = 1.08−0.22 × 10 M⊙ and the mass of the SE phase between the two subclumps within each subclus-
SE +0.28
subcluster is M200c = 1.55−0.26 × 1014 M⊙ . The masses ter.
for the four substructures are on the order of 1013 M⊙ , M. J. Jee acknowledges support for the current re-
and their significances range from 3.4σ to 4.8σ. search from the National Research Foundation (NRF)
Both the X-ray morphology and the large (∼0.7Mpc) of Korea under the programs 2022R1A2C1003130 and
separation between the NW and SE subclusters suggest RS-2023-00219959.
APPENDIX
mass estimation of the cluster itself but also could affect the mass estimation of A514 because of variations in shear
models depending on the cluster’s redshift. However, even if we vary the redshift of this cluster within the broad range
of 0.2 < z < 1.0, scatters of the masses for both A514NW and A514SE are less than ∼3%. Thus, we conclude that the
A514 mass bias due to the background cluster contamination is much smaller than its statistical error.
REFERENCES
Bartelmann, M., & Schneider, P. 2001, Physics Reports, HyeongHan, K., Jee, M. J., Cha, S., & Cho, H. 2024,
340, 291–472 Nature Astronomy, arXiv:2310.03073
Becker, M. R., & Kravtsov, A. V. 2011, ApJ, 740, 25 Jee, M. J., Blakeslee, J. P., Sirianni, M., et al. 2007, PASP,
Bertin, E. 2006, in Astronomical Society of the Pacific 119, 1403
Conference Series, Vol. 351, Astronomical Data Analysis Jee, M. J., Dawson, W. A., Stroe, A., et al. 2016, ApJ, 817,
Software and Systems XV, ed. C. Gabriel, C. Arviset, 179
D. Ponz, & S. Enrique, 112 Jee, M. J., Hoekstra, H., Mahdavi, A., & Babul, A. 2014a,
Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393 ApJ, 783, 78
Bertin, E., Mellier, Y., Radovich, M., et al. 2002, in Jee, M. J., Hughes, J. P., Menanteau, F., et al. 2014b, ApJ,
Astronomical Society of the Pacific Conference Series, 785, 20
Vol. 281, Astronomical Data Analysis Software and Jee, M. J., Mahdavi, A., Hoekstra, H., et al. 2012, ApJ,
Systems XI, ed. D. A. Bohlender, D. Durand, & T. H. 747, 96
Handley, 228 Jee, M. J., & Tyson, J. A. 2009, ApJ, 691, 1337
Brammer, G. B., van Dokkum, P. G., & Coppi, P. 2008, —. 2011, PASP, 123, 596
ApJ, 686, 1503 Jee, M. J., Tyson, J. A., Schneider, M. D., et al. 2013, ApJ,
Burns, J. O., Rhee, G., Owen, F. N., & Pinkney, J. 1994, 765, 74
ApJ, 423, 94 Jee, M. J., Stroe, A., Dawson, W., et al. 2015, ApJ, 802, 46
Cassano, R., Ettori, S., Giacintucci, S., et al. 2010, ApJL, Kaiser, N., & Squires, G. 1993, ApJ, 404, 441
721, L82 Kim, J., Jee, M. J., Hughes, J. P., et al. 2021, ApJ, 923, 101
Cho, H., James Jee, M., Smith, R., Finner, K., & Lee, W. Kim, M., Jee, M. J., Finner, K., et al. 2019, ApJ, 874, 143
2022, ApJ, 925, 68 Kuchinski, L. E., Freedman, W. L., Madore, B. F., et al.
Clowe, D., Bradač, M., Gonzalez, A. H., et al. 2006, ApJL, 2000, ApJS, 131, 441
648, L109 Lee, B., Chary, R.-R., & Wright, E. L. 2018, ApJ, 866, 157
Clowe, D., De Lucia, G., & King, L. 2004, MNRAS, 350, Lee, W., ZuHone, J., James Jee, M., et al. 2023a, ApJL,
1038 957, L4
Dutton, A. A., & Macciò, A. V. 2014, MNRAS, 441, 3359 Lee, W., Cha, S., Jee, M. J., et al. 2023b, ApJ, 945, 71
Feroz, F., & Hobson, M. P. 2012, MNRAS, 420, 596 Limousin, M., Morandi, A., Sereno, M., et al. 2013, SSRv,
Finner, K., Jee, M. J., Golovich, N., et al. 2017, ApJ, 851, 177, 155
46 Markevitch, M., Gonzalez, A. H., David, L., et al. 2002,
Finner, K., HyeongHan, K., Jee, M. J., et al. 2021, ApJ, ApJL, 567, L27
918, 72 Markwardt, C. B. 2009, in Astronomical Society of the
Fischer, P., & Tyson, J. A. 1997, AJ, 114, 14 Pacific Conference Series, Vol. 411, Astronomical Data
Golovich, N., van Weeren, R. J., Dawson, W. A., Jee, Analysis Software and Systems XVIII, ed. D. A.
M. J., & Wittman, D. 2017, ApJ, 838, 110 Bohlender, D. Durand, & P. Dowler, 251
Govoni, F., Taylor, G. B., Dallacasa, D., Feretti, L., & Martinet, N., Clowe, D., Durret, F., et al. 2016, A&A, 590,
Giovannini, G. 2001, A&A, 379, 807 A69
Gruen, D., Seitz, S., & Bernstein, G. M. 2014, PASP, 126, Mastropietro, C., & Burkert, A. 2008, MNRAS, 389, 967
158 McCully, C., Crawford, S., Kovacs, G., et al. 2018,
Herbonnet, R., von der Linden, A., Allen, S. W., et al. astropy/astroscrappy: v1.0.5 Zenodo Release, Zenodo,
2019, MNRAS, 490, 4889 doi:10.5281/zenodo.1482019
Hoekstra, H. 2003, MNRAS, 339, 1155 McLeod, B., Geary, J., Conroy, M., et al. 2015, PASP, 127,
Hoekstra, H., Franx, M., & Kuijken, K. 2000, ApJ, 532, 88 366
Weak-Lensing Analysis of the Complex Merging Cluster Abell 514 15
Meneghetti, M., Rasia, E., Merten, J., et al. 2010, A&A, van Weeren, R. J., Röttgering, H. J. A., Brüggen, M., &
514, A93 Hoeft, M. 2010, Science, 330, 347
Vazza, F., & Botteon, A. 2024, arXiv e-prints,
Molnar, S. M., & Broadhurst, T. 2015, ApJ, 800, 37
arXiv:2403.16068
Narayan, R., & Bartelmann, M. 1996, arXiv e-prints, astro Weaver, J. R., Kauffmann, O. B., Ilbert, O., et al. 2022,
Navarro, J. F., Frenk, C. S., & White, S. D. M. 1997, ApJ, ApJS, 258, 11
Weratschnig, J., Gitti, M., Schindler, S., & Dolag, K. 2008,
490, 493
A&A, 490, 537
Okabe, N., & Umetsu, K. 2008, PASJ, 60, 345 Wittman, D., Dawson, W., & Benson, B. 2014, MNRAS,
Schneider, P. 2006, in Saas-Fee Advanced Course 33: 437, 3578
Wittman, D., Dell’Antonio, I. P., Hughes, J. P., et al. 2006,
Gravitational Lensing: Strong, Weak and Micro, ed.
ApJ, 643, 128
G. Meylan, P. Jetzer, P. North, P. Schneider, C. S. Wright, C. O., & Brainerd, T. G. 2000, ApJ, 534, 34
Kochanek, & J. Wambsganss, 269–451 ZuHone, J. A., Markevitch, M., Weinberger, R., Nulsen, P.,
Seitz, C., & Schneider, P. 1997, A&A, 318, 687 & Ehlert, K. 2021, ApJ, 914, 73