Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Submitted to -
ABSTRACT
We derive kinematic properties of a sample of 122 r-process enriched stars from the halo of the Milky
Way based on GAIA DR2 and R-Process Alliance data. Over 4 derived orbital integrals of motion, we
apply a group of clustering analysis methods to the sample and receive 8 separate clusters containing
from 3 to 6 stars each. These clusters exhibit low discrepancy of r-process element abundances and are
likely to have originated in the same ultra-faint dwarf progenitors, accreted into the Milky Way halo.
This hypothesis is further supported by such findings as distinctively low scatter in relative Carbon
abundances within the clusters, suggesting common supernova enrichment sources, and lower scatter
than expected from random grouping in Barium abundances, indicating shared s-process enrichment
events. No conclusive evidence is found to suggest different origins of r-I and r-II stars.
very close to scaled solar abundance ratios, albeit the abundances here: [X/Fe] = [A(X) − A(Fe)]∗ − [A(X) − A(Fe)]Sun ,
where A(X) is the absolute abundance of element X.
2
although, it is important to note that theoretical predic- Holmbeck et. al. 2018). This “actinide boost” has
tions of the NSM frequencies vary, leading to significant an unknown origin, but its presence hints at the pos-
uncertainties in expected r-process abundances. Second, sibility of fundamental difference in origin between r-I
UFD galaxies could be the main source of the r-process and, at least, some r-II stars. In this case, the clusters
enhanced metal-poor stars found in the Milky Way halo, of r-II stars could differ from the clusters of r-I stars [ask
accreted into it (Frebel 2018). This opens an exciting Tim in what they could differ].
possibility to group those stars by their UFD progenitor, In this work, we study a large sample of r-I and r-
based on similarities in their kinematic properties. II stars found in literature by splitting them in small
Application of mathematical clustering algorithms in groups of 3-6 stars based on shared kinematic proper-
the space of orbital integrals of motion will result in ties. First we derive four orbital integrals of motion for
stars being grouped based on the closeness of their or- each of the stars, using their kinematic data available in
bits around the galaxy. If those stars had indeed origi- the literature combined with the radial velocities, dis-
nated in the same UFD galaxy, then they are expected to tances and proper motions obtained in the GAIA sur-
have small heavy r-process element abundance spread. vey (Gaia Collaboration 2016, 2018) [more citations,
As Roederer et. al. (2018) found for a sample of 35 r- as required by GAIA col.]. Then we apply 4 clustering
process enriched stars, the metallicity discrepancy also methods and receive a combined sample of 16 clusters
turns out to be smaller than expected by random chance. with 3-6 stars in each cluster. We study the distribu-
They also cross-matched the stellar metallicities with tion of metallicity and element abundances within each
the galactic luminosity-metallicity relation and found cluster and investigate whether they are likely to have
UFDs to be their most likely progenitors. The most originated in the same progenitor, as well as try to de-
likely alternative progenitor of the stars in a cluster is a termine the nature of that progenitor. We also study
single galactic cluster (GC) - however, GCs are known to the differences in clusters made of r-I and r-II stars to
demonstrate discrepancies in heavy element abundances test the hypothesis of the inherent difference between
such as [Eu/Fe] (Roederer 2011; Worley et. al. 2013), these groups or their subgroups.
hereby allowing us to discriminate between the two pos-
sible progenitors. 2. DATA AND KINEMATIC PARAMETERS
r-process enhanced stars can be split into multiple
For our initial r-process star sample, we select r-I and
groups based on their elemental abundances. The R-
r-II stars from two literature compilations: RPA (R-
Process Alliance (RPA) collaboration employs the fol-
Process Aliance) (Hansen et. al. 2018) and JINAbase
lowing classification (Hansen et. al. 2018), based on
(Abohalima & Frebel 2018). Then we cross-match this
Beers & Christlieb (2005):
data (the RA/DEC coordinates) with GAIA DR2 (Gaia
Collaboration 2016, 2018) in order to obtain geocen-
Table 1: r-process enhanced star classification. tric distances, radial velocities and two proper motions.
We then select all stars from the combined sample that
Sub-class Abundances have the estimated values for all 6 parameters. We
r-II [Eu/Fe] > +1.0, [Ba/Eu] < 0 further require that each of the stars have the follow-
r-I +0.3 ≤ [Eu/Fe] ≤ +1.0, [Ba/Eu] < 0 ing spectroscopically determined relative abundances:
limited-r [Sr/Ba] > +0.5, [Eu/Fe] < +0.3 [Fe/H], [C/Fe], [Sr/Ba], [Ba/Eu] and [Eu/Fe]. [Fe/H]
CEMP-r [C/Fe] ≥ +0.7, [Eu/Fe] > +0.3, [Ba/Eu] < 0 provides metallicity estimate, [C/Fe] indicates light el-
ement abundance, [Sr/Ba] and [Ba/Eu] provide com-
parisons of s-process abundances and light and heavy
r-process abundances respectively, and finally [Eu/Fe]
provides the estimate of r-process abundance. We also
Here, limited-r stars exhibit enhanced light r-process el- demand that only the values with the measured Eu-
ements, but inhibited heavy r-process elements, and al- ropeum abundance, rather than including the upper lim-
legedly have been enriched by the r-process producing its, are present. The resulting data, along with the
light elements only. r-I and r-II stars, in contrast, have G-magnitude and BrBp color (also taken from GAIA
been enriched by the main r-process producing mainly DR2) is shown on table 2, and the r-process enhance-
heavy elements, and r-II stars exhibit the extreme en- ment classes, relative chemical abundances and surface
richment. Moreover, some r-II star exhibit enhanced temperatures are shown on table 3. The majority of
abundances of the heaviest, actinide elements such as the chemical abundances are photometric, however the
Uranium and Thorium (Mashonkina et. al. 2014; abundances taken from (Roederer et. al. 2014) are
3
12 0.5
0.0 6500
10
−0.5
8
[Eu/H]
Frequency
−1.0
6000
6 −1.5
−2.0
4
Temperature (K)
−2.5 5500
2
0 2.0
0 1 2 3 4 5 6 7 8 5000
Distance (kpc) 1.6
[Eu/Fe]
Figure 1:
1.2
Distances of all 122 r-process enhanced stars in the sample, 4500
as featured in GAIA DR2. 0.8
Some of the orbital parameters derived from Staeckel stars with both prograde and retrograde motion around
potential approximation are shown on the bottom panel the Milky Way are present in comparable numbers. In
of fig. 3 (see section 3 for the description of the clus- the inner halo 35% stars have retrograde motion, while
tering procedure).The distance from the galactic plane in the outer halo the proportion is higher, specifically
Zmax and the apocentric distance rapo are consistent 45%; this aligns with the known studies of the two Milky
with the halo star orbits, versus thick disk stars, ex- Way halos, such as (Carollo et. al. 2007), although in
pected to have Zmax < 1 kpc and rapo < 3 kpc (Li & general the outer halo stars are expected to have a net
Zhao 2017). The majority of stars (∼ 83%) belong to retrograde motion.
the inner halo region (0 < rapo < 15 kpc), with uncor- Two of the integrals of motion, I3 and Lp , are fairly
related distribution of apocentric distances with regards strongly correlated in a non-linear way. This dimin-
to metallicities. The stars in the outer halo (rapo > 15 ishes the effect from introduction of the last integral
kpc), on the other hand, exhibit low scatter, and they of motion, I3 , on the clustering process. Nonetheless,
are shifted towards lower metallicity values, as per accor- we use all four integrals to maximize the cluster separa-
dance with studies of inner and outer halo stars (Beers tion, which proves to lead to smaller relative abundance
et. al 2012). Pericentric distances are split into two spreads than with I3 omitted.
general regions, rperi < 4 kpc and rperi > 4 kpc, but
those regions do not seem to have any other system- 3. CLUSTERING PROCEDURE
atic differences. Orbits of a wide range of eccentricities
We employ 4 different clustering analysis algorithms
(0.15 < ε < 1) are present in the sample, with stars
in the space of integrals of motion. These algorithms
in the inner halo having lower eccentricities than in the
seek to find groups of objects with close values and
outer halo (median value of 0.55 versus 0.65) and higher
separate them from each other. We employ the fol-
eccentricity spread (standard deviation value of 0.25 ver-
lowing algorithms: K-means (Lloyd 1982), mean shift
sus 0.19).
(Fukunaga & Hostetler 1975), agglomerative cluster-
The orbital parameters are used to calculate the inte-
ing (Sibson 1973; Defays 1977) and affinity propaga-
grals of motion. The four reconstructed integrals are:
tion (Frey & Dueck 2007). We select the parameters
E for each of these algorithms so they produce approxi-
1. Specific orbital energy = m , defined as the full
energy of the star on the orbit divided by its mass. mately the same number of clusters and cluster sizes
(the code, including the parameter values, is available
2. Full angular momentum Lp . at https://github.com/DmitriiGudin/r pro clustering).
Then we find all the pairs of stars that have been as-
3. z-projection of the angular momentum Lz , where
signed to the same cluster in all 4 algorithms. These
z is the axis perpendicular to the galactic plane.
stars are put in the same final cluster. Finally, all the
Positive and negative values of Lz correspond to
clusters with less than 3 stars in them are removed from
prograde and retrograde motion of stars about the
consideration. This leaves out 8 clusters, each contain-
Milky Way center.
ing from 3 to 6 stars.
4. Approximated third integral of motion I3 de- The cluster selection process is demonstrated on fig.
fined through a combination of terms as shown [4]; we only show two of the integrals of motion, specifi-
in (Tohline & Voyages 2001). cally energy and orbital momentum. There is some cor-
relation between them for the sample, but also a very
The integrals calculated for our sample of 122 stars, significant spread. One of the interesting results is that
along with other derived kinematic values, are shown on the stars within each cluster for all methods seem to
table 4, and their distribution is displayed on the upper have much more similarity in their energies, than in the
panel of fig. 3. Note the small group of outliers by I3 in orbital momenta. This can be seen especially well for the
the region of I3 > 2 kpc2 km2 s−2 ; as we will see later, clusters denoted by yellow triangles and black crosses on
these stars are not associated with any known clusters, the figure for the affinity propagation clustering, which
and further observations may be needed to determine encompass half of the entire Lp range and each have
whether they have a similar formation history to the one star laying far to the left of the other two. It is
other r-process enhanced stars. The orbital momentum not clear what causes this behavior; it cannot be a bias
Lp distribution shares similar outliers in the region Lp > introduced by the clustering procedure, which uses the
1.5 kpc km s−1 , and most of these outliers correspond to normalised values of all integrals of motion. Instead,
the same subset of stars. Energy and projection Lz do it has to be a characteristic of star clusters themselves.
not seem to have any obvious outlier stars. Note that the This behavior was no reported by (Roederer et. al.
5
−0.8
2.5
2 4
−1.0
Energy (km2 s−2 × 105)
2
−1.4 1.0
0
0.5 1
−1.6
−1
0.0 0
−1.8
10
25 1.0
30
8
25 20 0.8
Zmax (kpc)
rperi (kpc)
rapo (kpc)
6 20 15 0.6
ε
4 15 10
0.4
10
2 5
0.2
5
0
0
−4 −3 −2 −4 −3 −2 −4 −3 −2 −4 −3 −2
[Fe/H] [Fe/H] [Fe/H] [Fe/H]
Figure 3:
Initial and derived from Staeckel potential kinematic parameters, as well as the integrals of motion, of the sample. The grey
circles denote non-clustered stars, and different markers of different colors correspond to stars within clusters (see section 3).
2018). However, we do observe the same behavior with values. Since clustering was done by kinematic integrals
regards to the energy range of the clusters as found in of motion and did not involve chemical abundances, we
their work, specifically that the stars in the clusters with expect similarity in chemical abundances within each re-
lower energy values overlap more, than with higher ones. sulting cluster to be an indication of the similarity in the
The resulting integrals of motion for the combined star formation and evolution history.
clusters are shown on fig. [5], and the resulting clus- The top panel of fig. 6 shows relative abundances of
ters with selected parameters are shown in the table stars within each cluster: [Eu/Fe], [C/Fe], [Sr/Ba] and
5. The figure displays some interesting cluster patterns. [Ba/Eu], versus metallicity [Fe/H]. The next three pan-
First, just as for the individual clustering methods, the els show cumulative distributions of dispersions of these
resulting clusters feature stars typically more clustered abundances for randomly drawn groups of 3, 4 and 6
by energy than by the other integrals, and especially Lz stars respectively from the initial sample of 122 stars, as
and I3 feature strong outliers for the clusters denoted as well as the positions of dispersions of the 8 selected clus-
purple stars and green marks. Another significant fea- ters on them. Low y-axis position on these plots implies
ture is that all clusters but one have stars with different that the respective chemical abundance does not vary
signs o Lz , which means that they combine stars with significantly between stars in the cluster, hinting at the
prograde with those with retrograde motion in the Milky stars having formed in a similar chemical environment,
Way. This was not found by (Roederer et. al. 2018), such as a shared UfD progenitor.
and it also diverges from the simulation predictions in Dedicated plot for [Fe/H] is not included in fig. 6,
(Gomez et. al 2010). The source of this behavior is since we do not detect any systematic reduction of
unknown and being investigated. metallicity spread within the clusters, compared to ran-
domly drawn group of stars. This differs from (Roederer
4. DISCUSSION et. al. 2018), where all clusters were found to feature
Due to significant scatter of kinematic and chemical the [Fe/H] spreads below the median for the randomly
parameters of stars within each cluster, we employ a sta- sampled groups, while in our case 3 of the 5 clusters had
tistical approach to estimating the “quality” of the clus- higher spreads than the median. In tab. 5 we see that,
ters. Specifically, for each variable we are investigating, while metallicities for clusters C1, C2 and C4 are all
we sample mock clusters of N stars randomly selected within 0.6 of each other, all of the other clusters feature
from the sample (where N = {3, 4, 6}) and measure the one outlier each, differing by, at least, 0.95 from one of
statistical significance of the sampled standard deviation the other stars. In general, metallicity is expected to be
6
−0.8
less stable than the other considered abundances, as the
motion around the Milky Way contaminates the origi-
−1.0
Energy (km2 s−2 × 105)
−1.2
−1.3
−1.4
−1.5
−1.6
−1.7
0.0 0.5 1.0 −1 0 1 2 0.0 0.5 1.0 1.5
Lp (kpc km s−1 × 103) Lz (kpc km s−1 × 103) I3 (kpc2 km2 s−2 × 106)
Figure 5: The resulting clusters of r-process stars. Different markers and colors correspond to stars in different
clusters. Small grey circles are the stars from the sample that have not been assigned a cluster.
sults, we see noticeable stronger spreads in abundances Until recently, the number of detected r-process en-
of these elements, than we do in that of Europium. hanced stars with high resolution r-process element
However, as shown in the last column of fig. 6, the abundance measurements remained very low, and espe-
relative Barium abundances are generally significantly cially r-II stars with precise [Eu/Fe] measurements were
clustered, with [Ba/Eu] passing the 1σ (68%) confidence few. As the large sky surveys become a new central
level. This, again, is in line with the assumption of the point in the modern astronomy, more and more interest-
stars having the same UfD progenitors, as those progen- ing r-process enhanced stars are found. While up until
itors are expected to have created an environment with the year 2018 the largest study of r-process enhanced
relatively uniform s-process abundances (Ji et. al 2019; stars with high resolution abundances, after all cuts in-
Simon 2019). One of the clusters, C1, has anomalously troduced, encompassed less than 40 stars, we managed
high [Ba/Eu] spread due to the star-outlier J00400685- to obtain a sample of 122 stars. By combining relaxed
4325183, with [Ba/Eu] differing by 1.18 from the closest selection process with a rigorous clustering parameters
star in the cluster. As can be seen from tab. 3, this is choosing process, we managed to obtain 8 quality clus-
one of the only two stars (the other star is HE 0102-0633) ters. Those clusters exhibited striking similarity of Eu-
in the entire sample with [Ba/Eu] of below -1.2, both of ropeum abundances for the stars in them, and coinci-
those stars featuring values below -1.9. The source of dentally the absolute carbon abundances showed close
this discrepancy is unclear. proximity as well. These results align with the previ-
Finally, we also note that the apocentric distances ous findings qualitatively, while being stronger quanti-
within clusters were extremely similar, all being non- tatively.
random with the 3σ (99.7%) confidence level. This is not Our stars featured a wide distribution of apocentric
surprising, as orbital parameters are strongly connected distances, from as low as slightly below 5 kpc, to over
to the integrals of motion over which the clustering was 30 kpc. Apocentric distances of stars within each cluster
performed. were close to each other, however the orbital momenta
These results strongly support the assertion that the were not, and often stars with prograde and retrograde
stars within the determined clusters, indeed, come from motion around the Milky Way coexisted within clus-
the same UfD progenitors, that were later accreted ters. Each cluster had either only r-I, or only r-II stars,
by the Milky Way. Their heavy r-process, s-process however this result is unstable and changes with slight
and light element abundances exhibit statistically low variation of clustering parameters.
spreads, which, along with the similarity in their orbital Other initiatives in finding clusters of r-process en-
parameters, suggests shared history prior to arriving to hanced stars are currently in the works. One of the
the Milky Way halo, as supported by simulations (see, most interesting of them is the StarGO project (Yuan
for example, (Spitoni et. al 2016), especially for the et. al 2019; Tang et. al 2019) which seeks clusters
Barium spread within the metallicity range of -3.5 to by means of neural network training; it is currently be-
-1.5, and (Bullock & Johnston 2005; Johnston et. al ing used to find clusters of r-process and CEMP stars
2008) for a more general discussion). and will provide a second take on our results, as well as
those from (Roederer et. al. 2018). As the numbers
5. CONCLUSION of r-process stars with high resolution r-process element
8
1.2 0.0
0.50
0.50
0.25 −0.5
1.0 0.25
0.00
0.00
[Ba/Eu]
[Eu/Fe]
[Sr/Ba]
−1.0
[C/Fe]
0.8 −0.25
−0.25
−0.50 −1.5
0.6 −0.50
−0.75
−0.75 −2.0
−1.00
0.4
−3.5 −3.0 −2.5 −2.0 −1.5 −3.5 −3.0 −2.5 −2.0 −1.5 −3.5 −3.0 −2.5 −2.0 −1.5 −3.5 −3.0 −2.5 −2.0 −1.5
[Fe/H] [Fe/H] [Fe/H] [Fe/H]
abundance measurements grow, more and more clusters lution and theoretical models of the large scale cosmo-
will be found, which will allow us to put further theo- logical simulations, such as EAGLE (Joop et. al 2014),
retical constraints on the chemical evolution history in will provide a theoretical reference for interpreting our
galaxies. On the other hand, improvements in the reso- cluster abundance results.
REFERENCES
Burbidge, E. M., Burbidge, G. R., Fowler, W. A., & Hoyle, Hansen, T. T., Holmbeck, E. M., Beers, T. C., Placco, V.
F. 1957, Reviews of Modern Physics, 29, 547 M., Roeder, I. U., Frebel, A., Sakari, C. M. 2018, ApJ,
Cameron, A. G. W. 1957, PASP, 69, 201 858, 92
Arnould, M., Goriely, S., Takahashi, K. 2007, Physics Beers, T. C. & Christlieb, N. 2005, ARA&A, 43, 531
Reports, 450(4) Mashonkina, L., Christlieb, N., Eriksson, K. 2014, A&A,
Frebel, A. 2018, Annual Reviews, 68, 237 569, A43
Lattimer, J. M. & Schramm, D. N. 1974, ApJL, 192, L145 Holmbeck, E. M., Beers, T. C., Roederer, I. U., Placco, V.
Thielemann, K., F., Eichler, M., Panov, I. V. & Wehmeyer, M., Hansen, T. T., Sakari, C. M., Sneden, C., Liu, C.,
B. 2017, Annual Review of Nuclear and Particle Science, Lee, Y. S. 2018, 859(2), L24
67, 253 Gaia Collaboration, Prusti, T., de Bruijne, J. H. J., Brown,
Winteler, C., K. & R., Perego, A., et al. 2012, ApJL, 750, A. G. A., Vallenari, A., Babusiaux, C., Bailer-Jones, C.
L22 A. L., Bastian, U., Biermann, M., Evans, D. W., et. al.
Nishimura, N., Takiwaki, T. & Thielemann, F.-K. 2015, 2016, A&A, 595, A1
ApJ, 810, 109 Gaia Collaboration, Brown, A. G. A., Vallenari, A., Prusti,
T., de Bruijne, J. H. J., Babusiaux, C., Bailer-Jones, C.
Arcones, A., Janka, H.-T. & Scheck, L. 2007, A&A, 467,
A. L., Biermann, M., Evans, D. W., Eyer, L., et. al. 2018,
1227
A&A, 616, A1
Wanajo, S. 2013, ApJL, 770, L22
Abohalima, A., Frebel, A. 2018, AAS, 238(2), 36
Cowan, J. J., Pfeiffer, B., Kratz, K. L., Thielemann, F.K.,
Barklem, P. S., Christlieb, N., Beers, T. C., Hill, V.,
Sneden, C., Burles, S., Tytler, D. & Beers, T.C. 1999,
Bessell, M. S., Holmberg, J., Marsteller, B., Rossi, S.,
ApJL, 521, 194
Zickgraf, F.-J., Reimers, D. 2005, A&A, 439, 129
Burris, D. L. Pilachowski, C. A. Armandroff, T. E. Sneden,
Roederer, I. U., Preston, G. W., Thompson, I. B.,
C., Cowan, J. J. & Roe, H. 2000, ApJL, 544, 302
Shectman, S. A., Sneden, C. 2014, IOP, 784(2), 158
Cowan, J. J., Roederer, I. U., Sneden, C. & Lawler, J. E.
Christlieb, N., Gustafsson, B., Korn, A. J., Barklem, P. S.,
2011, ArXiv e-prints
Beers, T. C., Bessell, M. S., Karlsson, T.,
Kratz, K.-L., Farouqi, K., Pfeiffer, B., Truran, J. W.,
Mizuno-Wiedner, M. 2004, ApJ, 603, 708
Sneden, C. & Cowan, J. J. 2007, IOP, 662, 39
Jacobson, H. R., Keller, S., Frebel, A., Casey, A. R.,
Roederer, I. U., Mateo, M., Bailey, III, J. I., et al. 2016,
Asplund, M., Bessell, M. S., Da Costa, G. S., Lind, K.,
AJ, 151, 82
Marino, A. F., Norris, J. E., Peña, J. M., Schmidt, B. P.,
Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2017, Tisserand, P., Walsh, J. M., Yong, D., Yu, Q. 2015, ApJ,
Physical Review Letters, 119, 161101 807, 171
Kilpatrick, C. D., Foley, R. J., Kasen, D., et al. 2017, Siqueira Mello, C., Hill, V., Barbuy, B., Spite, M., Spite, F.,
Science, 358, 1583 Beers, T. C., Caffau, E., Bonifacio, P., Cayrel, R.,
Shappee, B. J., Simon, J. D., Drout, M. R., et al. 2017, François, P., Schatz, H., Wanajo, S. 2014, A&A, 565, A93
Science, 358, 1574 Honda, S., Aoki, W., Kajino, T., Ando, H., Beers, T. C.,
Hotokezaka, K., Benjamini, P., Piran, T. 2018, Izumiura, H., Sadakane, K., Takada-Hidai, M. 2004, ApJ,
International Journal of Modern Physics, 27, 13 607, 474
Roederer, I. U., Hattori, K., Valluri, M. 2018, AAS, 156(4), Cohen, J. G., Christlieb, N., McWilliam, A., Shectman, S.,
179 Thompson, I., Wasserburg, G. J., Ivans, I., Dehn, M.,
Roederer, I. U. 2011, IOP, 732, L17 Karlsson, T., Melendez, J. 2004, ApJ, 612, 1107
Worley, C. C., Hill, V., Sobeck, J. & Carretta, E. 2013, Lai, D. K., Bolte, M., Johnson, J. A., Lucatello, S., Heger,
A&A, 553, A47 A., Woosley, S. E. 2008, ApJ, 681, 1524
10
Allen, D. M., Ryan, S. G., Rossi, S., Beers, T. C., Bovy, J. 2015, AAS, 216(2), 29
Tsangarides, S. A. 2012, A&A, 548, A34 Tohline, J. E., Voyages, K. 2001, IOP, 555(1), 524
Cayrel, R., Depagne, E., Spite, M., Hill, V., Spite, F., Carollo, D., Beers, T. C., Lee, Y. S., Chiba, M., Norris, J.
François, P., Plez, B., Beers, T. C., Primas, F., Andersen, E., Wilhelm, R., Sivarani, T., Marsteller, B., Munn, J.
J., Barbuy, B., Bonifacio, P., Molaro, P., Nordström, B. A., Bailer-Jones, C. A. L., Florentin, P. R., York, D. G.
2004, A&A, 416, 1117 2008, Nat, 451, 216
Preston, G. W., Thompson, I. B., Sneden, C., Stachowski, Li, C., Zhao, G. 2017, AAS, 850, 25
G., Shectman, S. A. 2006, AJ, 132, 1714 Beers, T. C., Carollo, D., Ivezić, Ž. An, D., Chiba, M.,
Hansen, T., Hansen, J. C., Christlieb, N., Beers, T. C., Norris, J. E., Freeman, K. C., Lee, Y. S., Munn, J. A.,
Yong, D., Bessell, S. M., Frebel, A., Perez, G. A., Placco, Fiorentin, P. R., Sivarani, T., Wilhelm, R., Yanny, B.,
V., Norris, E. J., Asplund, M. 2015, ApJ, 807 York, D. G. 2012, AJ, 746, 34
Behara, N. T., Bonifacio, P., Ludwig, H. G., Shordone, L., Lloyd, P. 1982, IEEE Transactions on Information Theory,
González Hernández, J. I., Caffau, E. 2010, A&A, 513, 28, 129
Fukunaga, K., Hostetler, L. 1975, IEEE Transactions on
A72
Information Theory, 21(1), 32
Li, H., Aoki, W., Zhao, G., Honda, S., Christlieb, N., Suda,
Sibson, R. 1973, The Computer Journal, 16(1), 30
T. 2015, PASJ, 67
Defays, D. 1977, The Computer Journal, 20(4), 364
Hayek, W., Wiesendahl, U., Christlieb, N., Eriksson, K.,
Frey, B. J., Dueck, D. 2007, Science, 315, 2007
Korn, A. J., Barklem, P. S., Hill, V., Beers, T. C.,
van der Maaten, L., Hilton, G. 2008, Journal of Machine
Farouqi, K., Pfeiffer, B., Kratz, K. L. 2009, A&A, 504,
Learning Research, 1, 48
511
Sharma, M., Theuns, T., Frenk, C. S., Cooke, R. J. 2017,
Cohen, J. G., Christlieb, N., Iam, T., Mcwilliam, A.,
MNRAS, 437(1), 984
Shectman, S., Reimers, D., Wisotzki, L., Kirby, E. 2013,
Gmez, F. A., Helmi, A., and Brown, A. G. A., Li, Y.-S.
IOP, 778, 56 2010, MNRAS, 408(2), 935
Cowan, J. J., Sneden, C., Burles, S., Ivans, I. I., Beers, T. Ji, A. P., Simon, J. D., Frebel, A., Venn, K. A., Hansen, T.
C., Truran, J. W., Lawler, J. E., Primas, F., Fuller, G. T. 2019, AAS, 870(2), 83
M., Pfeiffer, B., Kratz, K. L. 2002, ApJ, 572, 861 Simon, J. D. 2019, ARA&A, 57, 375
Howes, L. M., Casey, A. R., Asplund, M., Keller, S. C., Spitoni, E., Vincenzo, F., Matteucci, F., Romano, D. 2016,
Yong, D., Nataf, D. M., Poleski, R., Lind, K., Kobayashi, MNRAS, 458, 519
C., Owen, C. I., Ness, M., Bessell, M. S., da Costa, G. S., Bullock, J. S., Johnston, K. V. 2005, AJ, 635(2), 931
Schmidt, B. P., Tisserand, P., Udalski, A., Szymański, M. Johnston, K. V., Bullock, J. S., Sharma, S., Font, A.,
K., Soszyński, I., Pietrzyński, G., Ulaczyk. K., Robertson, B. E., Leitner, S. N. 2008, AJ, 689(2), 936
Wyrzykowski, L, Pietrukowicz, P., Skowron, J., Yuan, Z., Smith, M.C., Xue, X.-X., Li, J., Liu, C., Wang,
Kozlowski, S., Mróz, P. 2015, Nat, 527, 484 Y., Li, L., Chang, J. 2019, AAS, 881(2), 164
Ivans, I. I., Simmerer, J., Sneden, C., Lawler, J. E., Cowan, Tang, S.-Y., Pang, X., Yuan, Z., Chen, W.P., Hong, J.,
J. J., Gallino, R., Bisterzo, S. 2006, ApJ, 645, 613 Goldman, B., Just, A., Shukirgaliyev, B., Lin, C.-C.
Mashonkina, L., Christlieb, N., Barklem, P. S., Hill, V., 2019, AAS, 877, 12
Beers, T. C., Velichko, A. 2010, AAP, 516, A46 Joop, S., Crain, R. A., Bower, R. G., Furlong, M., Schaller,
Beers, T. C., Christlieb, N. 2005, ARA&A, 43, 531 M., Theuns, T., Dalla Vecchia, C., Frenk, C. S.,
Kuzmin, G. G. 1956, Astr. Zh., 33, 27 McCarthy, I. G., Helly, J. C., Jenkins, A., Rosas-Guevara,
Kuzmin, G. G. 1962, Bull. Abastumani Ap. Obs., 27, 82 Y. M., White, S. D. M., Baes, Maarten, Booth, C. M.,
Zeeuw, T. de, Peletier, R., Franx, M. 1986, MNRAS, 18, 5 Camps, P., Navarro, J. F., Qu, Y., Rahmati, A., Sawala,
T., Thomas, P. A., Trayford, J. 2014, MNRAS, 446, 521
Batsleer, P., Dejonghe, H. 1994, AA, 287(1), 43
11
Table 2 continued
12
Table 2 (continued)
Table 2 continued
13
Table 2 (continued)
Table 2 continued
14
Table 2 (continued)
K
a
J00002259-1302275 r-I -2.90 -0.65 -0.82 -0.96 0.58 4879 HAN18
J00021668-2453494 r-I -1.81 -0.88 +0.49 -0.42 +0.52 4923 HAN18
b
HD 20 r-I -1.58 -0.38 -0.04 -0.53 +0.79 5445 BAR05
c
CS 30339-015 r-I -2.69 +0.69 +1.14 -1.08 +0.57 5580 ROE14b
d
CS 29497-004 r-II -2.68 +0.11 -0.31 -0.71 +1.71 5090 CHR04
e
SMSS J003055.81-482011.3 r-I -2.53 +0.05 -0.09 -0.31 +0.51 4720 JAC15
J00400685-4325183 r-I -2.55 -0.85 +0.04 -2.11 +0.55 4568 HAN18
J00405260-5122491 r-I -2.11 -0.04 +0.13 -0.90 +0.86 6154 HAN18
J00453930-7457294 r-I -2.00 +0.93 +0.46 -0.18 +0.55 4971 HAN18
f
HE 0057-4541 r-I -2.34 +0.49 +0.33 -0.61 +0.55 5144 SIQ14
g
HD 6268 r-I -2.62 -0.55 +0.12 -0.55 +0.54 4600 HON04
h
HE 0102-0633 r-II -2.95 ... -0.36 -1.94 +1.89 6350 COH04
HE 0105-6141 r-I -2.56 +0.23 +0.20 -0.38 +0.51 5234 SIQ14
J01202234-5425582 r-I -2.11 -0.09 +0.34 -0.14 +0.30 5370 HAN18
CS 31082-001 r-II -2.78 +0.18 -0.50 -0.52 +1.65 4922 BAR05
J01425422-5032488 r-I -2.09 +0.07 +0.26 -0.51 +0.38 5412 HAN18
J02165716-7547064 r-II -2.50 -0.33 +0.00 -0.87 +1.12 4842 HAN18
J02355867-6745520 r-I -1.55 +0.97 -0.39 -0.01 +0.50 4688 HAN18
HE 0240-0807 r-I -2.86 -0.37 -0.27 -0.58 +0.76 4740 SIQ14
J02462013-1518419 r-II -2.71 +0.04 -0.02 -0.85 +1.45 5755 HAN18
J02500719-5145148 r-I -2.20 +1.04 -0.53 -0.14 +0.62 4774 HAN18
i
CS 31078-018 r-II -2.83 +0.45 -0.33 -0.55 +1.14 5257 LAI08
HE 0300-0751 r-I -2.27 +0.06 +0.25 -0.78 +0.76 5280 BAR05
HE 0315+0000 r-I -2.73 +0.14 -0.21 -0.34 +0.64 5013 BAR05
j
CS 29526-110 r-II -2.19 +2.37 -1.85 -0.04 +2.18 6650 ALL12
HE 0337-5127 r-I -2.61 +0.12 +0.16 -0.66 +0.59 5247 BAR05
HE 0341-4024 r-I -1.82 +0.22 +0.13 -0.81 +0.68 6108 BAR05
CS 22186-002 r-I -2.72 +0.06 +0.17 -0.74 +0.65 5500 ROE14b
Table 3 continued
15
Table 3 (continued)
K
CS 22182-033 r-I -2.65 +0.31 +0.51 -0.87 +0.58 5810 ROE14b
k
CS 22186-025 r-I -3.00 -0.45 +0.01 -0.56 +0.53 4900 CAY04
HE 0430-4901 r-II -2.72 +0.05 -0.36 -0.69 +1.15 5296 BAR05
HE 0432-0923 r-II -3.19 +0.20 -0.09 -0.57 +1.24 5131 BAR05
HE 0442-1234 r-I -2.41 -0.65 -0.03 -0.70 +0.51 4604 BAR05
l
CS 22191-029 r-I -2.71 +0.25 +0.48 -0.70 +0.54 6000 PRE06
m
HE 0448-4806 r-II -2.26 +2.24 -0.68 -0.09 +1.87 5900 HAN15
SMSS J051008.62-372019.8 r-I -3.20 +0.56 -0.34 -0.20 +0.95 5170 JAC15
HE 0516-3820 r-I -2.48 +0.40 +0.23 -0.59 +0.62 5269 SIQ14
HE 0538-4515 r-I -1.52 +0.11 +0.41 -0.65 +0.50 5896 BAR05
n
SDSS J103649.93+121219.8 r-II -3.23 +1.57 -1.69 -0.09 +1.29 6000 BEH10
HE 1044-2509 r-I -2.88 +0.48 +0.15 -0.74 +0.92 5227 BAR05
o
LAMOST J1109+0754 r-II -3.41 -0.57 -0.02 -0.85 +1.16 4440 LI15b
HE 1127-1143 r-II -2.72 +0.50 -0.23 -0.49 +1.06 5224 BAR05
HE 1131+0141 r-I -2.49 +0.08 -0.23 -0.39 +0.87 5236 BAR05
p
HE 1219-0312 r-II -2.97 +0.00 -0.24 -0.74 +1.39 5060 HAY09
q
HE 1226-1149 r-II -2.91 +0.54 -0.40 -0.69 +1.54 5120 COH13
BS 16929-005 r-II -3.15 +1.72 +0.40 -0.79 +1.14 5250 ALL12
HE 1311-1412 r-I -2.88 -0.20 +0.15 -0.63 +0.52 4796 BAR05
CS 22877-015 r-I -2.12 +0.28 +0.10 -0.43 +0.85 6150 ROE14b
J13164824-2743351 r-I -1.61 -0.05 +0.25 -0.53 +0.54 4982 HAN18
BS 16543-097 r-I -2.48 +1.36 +0.65 -0.54 +0.97 5220 ALL12
J14232679-2834200 r-I -1.90 +0.43 +0.51 -0.68 +0.61 4912 HAN18
HD 126587 r-I -2.77 +0.31 +0.18 -0.62 +0.55 4960 HON04
J14301385-2317388 r-I -1.40 -0.70 +0.06 -0.28 +0.62 4104 HAN18
J14325334-4125494 r-II -2.79 -0.14 -0.39 -0.90 +1.61 4964 HAN18
HE 1430+0053 r-I -3.03 +0.25 -0.54 -0.85 +0.71 5201 BAR05
CS 30306-132 r-I -2.41 +0.46 -0.07 -0.62 +0.87 5110 HON04
J15260106-0911388 r-II -2.83 -0.82 +0.21 -1.01 +1.70 4655 HAN18
J15271353-2336177 r-I -2.15 +0.33 +0.38 -0.73 +0.70 6004 HAN18
CS 30312-059 r-I -3.13 +0.35 +0.14 -0.68 +0.56 5021 LAI08
J15582962-1224344 r-I -2.54 -0.14 +0.49 -0.85 +0.89 4885 HAN18
J16024498-1521016 r-I -1.80 -0.05 +0.27 -0.47 +0.55 4982 HAN18
SMSS J164921.94-041943.9 r-I -2.24 -0.25 -0.16 -0.32 +0.68 4936 JAC15
J17093199-6027271 r-I -2.48 -0.45 +0.43 -0.75 +0.60 4661 HAN18
J17124284-5211479 r-I -2.78 -0.42 +0.55 -1.06 +0.48 4300 HAN18
J17225742-7123000 r-II -2.42 -0.33 -0.39 -0.37 +1.07 4923 HAN18
r
BD+17:3248 r-I -2.06 -0.37 -0.03 -0.55 +0.87 5200 COW02
J17400682-6102129 r-I -2.24 -0.26 0.47 -0.62 0.31 4838 HAN18
J17435113-5359333 r-I -2.24 -0.35 +0.07 -0.80 +0.73 4880 HAN18
Table 3 continued
16
Table 3 (continued)
K
SMSS J175738.37-454823.5 r-II -2.46 -0.43 -0.26 -0.55 +1.02 4617 JAC15
J18024226-4404426 r-II -1.55 +0.35 -0.27 -0.10 +1.05 4519 HAN18
J18285086-3434203 r-I -2.46 -0.58 +0.51 -0.82 +0.53 4567 HAN18
J18294122-4504000 r-I -2.48 -0.46 +0.53 -0.76 +0.70 4901 HAN18
J18362318-6428124 r-I -2.57 +0.10 -0.81 -0.39 +0.57 4996 HAN18
J18363613-7136597 r-I -2.52 -0.75 +0.38 -0.93 +0.56 4980 HAN18
s
J183647.89-274333.1 r-I -2.48 -0.47 +0.05 -0.69 +0.82 4083 HOW15
J18562774-7251331 r-I -2.26 -0.48 +0.01 -0.33 +0.32 4835 HAN18
J19014952-4844359 r-I -1.87 +0.02 +0.61 -1.17 +0.93 4940 HAN18
J19161821-5544454 r-II -2.35 -0.80 +0.60 -1.20 +1.08 4492 HAN18
J19215077-4452545 r-I -2.56 -0.77 +0.26 -1.05 +0.74 4372 HAN18
J19232518-5833410 r-I -2.08 +0.27 +0.45 -0.65 +0.76 4861 HAN18
J19324858-5908019 r-I -1.93 -0.51 +0.25 -0.25 +0.90 4082 HAN18
CS 22891-200 r-I -4.06 +0.53 -0.43 -0.86 +0.11 4490 ROE14b
CS 22896-154 r-I -2.68 +0.31 +0.18 -0.39 +0.84 5250 CAY04
HD 186478 r-I -2.49 -0.13 +0.19 -0.83 +0.63 4720 HON04
SMSS J195931.70-643529.3 r-I -2.58 -0.45 +0.21 -0.75 +0.74 4639 JAC15
J20093393-3410273 r-II -2.10 -0.86 +0.36 -1.09 +1.32 4122 HAN18
J20303339-2519500 r-I -2.18 -0.65 +0.70 -0.70 +0.41 4261 HAN18
CS 22955-174 r-I -2.89 +0.25 +0.54 -0.58 +0.60 5650 PRE06
J20492765-5124440 r-I -2.47 -0.77 +0.38 -0.83 +0.58 4565 HAN18
J20514971-6158008 r-I -1.87 +0.10 +0.46 -0.61 +0.43 5284 HAN18
CS 22940-121 r-I -2.89 +0.25 +0.23 -0.44 +0.58 5400 PRE06
J21064294-6828266 r-II -2.76 +0.53 +0.38 -0.80 +1.32 5186 HAN18
J21091825-1310066 r-II -2.40 -0.28 +0.02 -1.13 +1.25 4967 HAN18
J21095804-0945400 r-I -2.73 -0.45 +0.72 -0.89 +0.77 4189 HAN18
J21162185-0213420 r-I -2.23 -0.98 +0.51 -0.43 +0.49 4163 HAN18
J21224590-4641030 r-I -2.96 -0.18 +0.24 -0.89 +0.90 4912 HAN18
J21262525-2144243 r-I -2.81 -0.88 +0.39 -0.60 +0.58 4409 HAN18
G 214-1 r-I -2.29 -0.12 +0.19 -0.29 +0.56 5370 ROE14b
HE 2155+0136 r-I -2.07 -0.04 -0.11 -0.59 +0.68 5331 BAR05
BS 17569-049 r-I -2.85 -0.06 +0.26 -0.56 +0.68 4700 CAY04
HE 2206-2245 r-I -2.73 +0.17 +0.21 -0.76 +0.60 5100 BAR05
J22163596+0246171 r-I -2.37 -0.10 +0.44 -0.72 +0.45 5057 HAN18
HE 2216-1548 r-I -1.70 -0.42 -0.09 -0.33 +0.57 5154 BAR05
CS 22886-043 r-I -2.14 +0.25 +0.47 -0.50 +0.78 6000 PRE06
HE 2224+0143 r-II -2.58 +0.31 -0.20 -0.51 +1.04 5198 BAR05
CS 29491-069 r-II -2.60 +0.28 -0.03 -0.73 +1.05 5300 HAY09
HE 2242-1930 r-I -2.21 +0.05 +0.21 -0.69 +0.54 5281 BAR05
HE 2244-2116 r-I -2.35 +0.34 +0.10 -0.68 +0.58 5230 COH13
Table 3 continued
17
Table 3 (continued)
K
HE 2244-1503 r-I -2.87 +0.11 -0.50 -0.54 +0.94 5122 BAR05
J22492756-2238289 r-I -1.82 -0.08 +0.82 -0.91 +0.48 4795 HAN18
t
HE 2252-4225 r-II -2.94 -0.34 -0.36 -0.53 +1.12 4710 MAS14
HE 2252-4157 r-I -1.93 -0.19 -0.13 -0.28 +0.52 5090 BAR05
J23022289-6833233 r-I -2.64 +0.30 +0.44 -0.90 +0.61 5336 HAN18
HE 2301-4024 r-I -2.10 +0.26 -0.34 -0.53 +0.97 5743 BAR05
CS 29513-003 r-I -2.47 -0.04 +0.32 -0.57 +0.50 5480 ROE14b
u
HD 221170 r-I -2.16 -0.60 -0.05 -0.59 +0.78 4510 IVA06
v
HE 2327-5642 r-I -2.78 +0.09 -0.40 -0.67 +0.97 5050 MAS10
CS 30315-029 r-I -3.43 +0.00 -0.43 -0.52 +0.67 4570 SIQ14
J23362202-5607498 r-II -2.06 -0.05 +0.76 -0.90 +1.14 4531 HAN18
CS 29499-003 r-I -2.51 +0.26 +0.50 -0.96 +0.79 6080 ROE14b
CS 29499-058 r-I -2.81 +0.59 +0.58 -0.95 +0.64 6060 ROE14b
CS 22945-056 r-I -2.82 <+1.59 +0.30 -1.08 +0.67 6000 ROE14b
Table 3 continued
18
Table 3 (continued)
K
a (Hansen et. al. 2018)
b (Barklem et. al. 2005)
c (Roederer et. al. 2014)
d (Christlieb et. al. 2004)
e (Jacobson et. al. 2015)
f (Siqueira Mello et. al. 2014)
g (Honda et. al. 2004)
h (Cohen et. al. 2004)
i (Lai et. al. 2008)
j (Allen et. al. 2012)
k (Cayrel et. al. 2004)
l (Preston et. al. 2006)
m (Hansen et. al. 2015)
n (Behara et. al. 2010)
o (Li & Zhao 2017)
p (Hayek et. al. 2009)
q (Cohen et. al. 2013)
r (Cowan et. al. 2002)
s (Howes et. al. 2015)
t (Mashonkina et. al. 2014)
u (Ivans et. al. 2006)
v (Mashonkina el. al. 2010)
Table 4 continued
19
Table 4 (continued)
Table 4 continued
20
Table 4 (continued)
Table 4 continued
21
Table 4 (continued)
Cluster C2
HE2206-2245 r-I -2.73 0.60 0.17 0.21 -0.76
J18362318-6428124 r-I -2.57 0.57 0.10 -0.81 -0.39
J23022289-6833233 r-I -2.64 0.61 0.30 0.44 -0.90
HD126587 r-I -2.77 0.55 0.31 0.18 -0.62
CS22955-174 r-I -2.89 0.60 0.25 0.54 -0.58
CS22940-121 r-I -2.89 0.58 0.25 0.23 -0.44
Cluster C3
J02165716-7547064 r-II -2.50 1.12 -0.33 0.00 -0.87
J17225742-7123000 r-II -2.42 1.07 -0.33 -0.39 -0.37
LAMOSTJ1109+0754 r-II -3.41 1.16 -0.57 -0.02 -0.85
Cluster C4
J17435113-5359333 r-I -2.24 0.73 -0.35 0.07 -0.80
SMSS J164921.94-041943.9 r-I -2.24 0.68 -0.25 -0.16 -0.32
J18294122-4504000 r-I -2.48 0.70 -0.46 0.53 -0.76
Cluster C5
CS22186-002 r-I -2.72 0.65 0.06 0.17 -0.74
CS30315-029 r-I -3.43 0.67 0.00 -0.43 -0.52
HE0300-0751 r-I -2.27 0.76 0.06 0.25 -0.78
Cluster C6
HE1127-1143 r-II -2.72 1.06 0.50 -0.23 -0.49
CS31078-018 r-II -2.83 1.14 0.45 -0.33 -0.55
J18024226-4404426 r-II -1.55 1.05 0.35 -0.27 -0.10
Cluster C7
HE0337-5127 r-I -2.61 0.59 0.12 0.16 -0.66
HE0105-6141 r-I -2.56 0.51 0.23 0.20 -0.38
HE0538-4515 r-I -1.52 0.50 0.11 0.41 -0.65
Table 5 continued
23
Table 5 (continued)