58. ???? ? ????
Figure 1. Size vs stellar mass for the entire sample of old, pres-
sure supported systems with available central velocity dispersions.
Blue squares denote Local Group dSph galaxies, and green open
Forbes + 14
????
??? ??
¨¨ ??? ?? ??
¨¨ ??? ?? ??
59. ???? ? ????
Figure 1 ¨C Historical and current tensions between LCDM theory and observations of dwarf galaxies. We classify
these according to the level of tension/challenge they present to the cosmological LCDM scenario, after the critical effects of
baryonic physics have been considered. Left to right moves from ¡®no tension¡¯ to ¡®strong tension¡¯. The following sections
discuss each of the topics in this chart. We discuss the M?-Mhalo relation and the Too-big-to-fail problem in sections with those
respective names. We address the core-cusp problem and the diversity of rotation curves in the section: ¡®Dark matter
distribution within dwarf galaxies¡¯; the diversity of sizes in the section: ¡®Baryonic distribution within dwarf galaxies¡¯; and
Sales + 22
??????? ???? ??? ?? ??? ??
60. Missing satellites
By Marcel S. Pawlowski
Classical dwarfs
By Marcel S. Pawlowski
Classical dwarfs
SDSS (2005~)
DES (2015~)
DM-only Cosmological sim.
Moore et al. 1999
?? ?????(??)
Dark halo
Luminous halo
Sawala et al. 2016
Hydro Cosmological sim.
Buck et al. 2019
No tension anymore with accurate baryonic physics ?
????
???????
61. Too big to fail
Boylan-Kolchin et al. 2012
Massive satellites are not found around the MW
Wetzel et al. 2016
DM-only sim. Hydro sim.
72. ?? ??
?????(Simulation)? ??
mid-14c., simulacioun, "a false show, false profession," fro
m Old French simulation "pretence" and directly from Lati
n simulationem (nominative simulatio) "an imitating, feign
ing, false show, hypocrisy," noun of action from past-partici
ple stem of simulare "imitate," from stem of similis "like, re
sembling, of the same kind" (see similar). The meaning "a
model or mock-up for purposes of experiment or training" i
s from 1954.
(etymonline.com)
105. Vogelsberger et al. (2014)
?? ??
Eagle ????? Illustris ?????
Horizon-AGN ?????
The Eagle project The Illustris Collaboration
Yohan Dubois The HR5 collaboration
HR5 ?????
Horizon Run 5 (HR5)
114. ? Required conditions: a very wide dynamic range is necessary (>106)
- effective radius R1/2?1kpc: resolution ? 0.1kpc
- diversity across different environments: volume ? (100Mpc)3
DARWIN¡¯s strategy
1. Efficient utilization of exa-scale computing
- massively-parallel supercomputer
- hybrid parallelization (MPI, OMP, GPU)
2. Three-step zoom-in simulations
- aggressive application of the zoom-in technique
3. Implementing AI¡¯s prediction to low-resolution data
- deriving de facto high-resolution results
115. spatial res. star mass res. volume feature
DARWIN-1 500pc 105 Msun (~65Mpc)3 various galaxies in various environments
DARWIN-2 125pc, 62.25pc 2x103 Msun (5-20Mpc)3 formation/evolution of dwarf galaxies in various env.
DARWIN-3 4-8pc 1x103 Msun < (1Mpc)3 ISM and internal structures of dwarf galaxies
¦¤x=500pc ¦¤x=125pc, 62.25pc ¦¤x=4-8 pc
DARWIN-2 DARWIN-3
DARWIN-1
cluster
Galaxy zoo in HR5
group
filament
Isolated
3-step zoom-in simulations
~
65Mpc
116. DARWIN1 analogue (¦¤x= 500pc)
2cMpc
DARWIN2 analogue (¦¤x=125pc)
2cMpc
100ckpc 100ckpc
2ckpc
DARWIN3 analogue (¦¤x=16pc)
Resolution of DARWIN
DARWIN-1 Detailed internal properties of galaxies (roughly Mstar?3x107Msun) in various environments
DARWIN-2 Formation of dwarf galaxies (global: Mstar?105Msun, internal: Mstar?106Msun ) in various environments
DARWIN-3 ISM and formation of internal structures of dwarf galaxies
117. spatial res. strength weakness
DARWIN-1 500pc statistical analysis inability to describe dwarf galaxies
DARWIN-2 125pc, 62.25pc dwarf galaxies in various environments reduced reliability of statistical analysis
DARWIN-3 4-8pc most realistically describe the dwarf galaxies inability to perform statistical analysis
??? ?? ?????
??? ??? ???
Trained AI
apply
AI model associating
low- and high- resolution
results
Predicting
high-resolution results based
on low-resolution results
o Training from overlapped region between DARWIN-1 and DARWIN-2
o Implementing AI¡¯s prediction in the low-resolution region outside the DARWIN-2
o Deriving de facto high-resolution results in DARWIN-1
o Preliminary study using TNG were recently published (Jung+24)
AI¡¯s prediction
Trained AI
118. cooling
stellar/BH feedback
star formation
chemical
enrichment
Tumlinson + 2017
Possible solutions in simulation side
o High-resolution to resolve internal structures
o Realistic chemistry and cooling modules
o Precise star formation recipe (sink SF model)
o Sophisticated stellar/BH Feedback
o Numerous samples comparable to observations
¨¨ Baryonic physics model in a self-consistent manner down to a several parsec scale
TIGRESS cooling model, stars seeded in sink scheme + Pop III + SN II + SN Ia + stellar winds
+ RAMSES-RTZ + PRISM + 4 pc resolution + large number of sample
Figure 1 ¨C Historical and current tensions between LCDM theory and observations of dwarf galaxies. We classify
these according to the level of tension/challenge they present to the cosmological LCDM scenario, after the critical effects of
baryonic physics have been considered. Left to right moves from ¡®no tension¡¯ to ¡®strong tension¡¯. The following sections
discuss each of the topics in this chart. We discuss the M?-Mhalo relation and the Too-big-to-fail problem in sections with those
respective names. We address the core-cusp problem and the diversity of rotation curves in the section: ¡®Dark matter
distribution within dwarf galaxies¡¯; the diversity of sizes in the section: ¡®Baryonic distribution within dwarf galaxies¡¯; and
satellite planes together with quiescent fractions grouped in the section: ¡®Satellite dwarf galaxies¡¯.
Alternately, on just the theoretical side, one can compare the predictions of different simulations regarding the relation
between galaxy stellar mass and dark-matter halo mass in the ultra-faint regime. Indeed, as discussed below, a careful look
into state-of-the-art numerical simulations that predict the correct number of MW-like galaxies and classical dwarf galaxies
suggests that their expected ultra-faint populations may differ, signaling an important theoretical uncertainty that persists. We
thus emphasize that our discussion of this relation between stellar mass and dark-matter halo mass is different from the others
in this review, because our comparison is only between different simulations, not (yet) between simulations and observations.
Fig. 2 shows the relation between stellar mass and dark-matter halo mass, where we collect the present-day relation
predicted from a sample of state-of-the-art cosmological simulations. Halo mass corresponds to the spherical radius within
which the average density is 200 times the critical density, the so-called virial radius. Where a different definition of halo mass
was presented in the published work, we convert those values using average mass-concentration relation from ref.70. On the left
panel, we include zoom-in simulations of MW-like or Local Group-like environments from various works: APOSTLE44,71
from the EAGLE project72, Latte and ELVIS suites45,65 from the FIRE-2 project73, Auriga74, NIHAO-UHD41, DC Justic
League75; or zooms of relatively large regions, like the Marvel Suite66. In all cases, we show only central (field) galaxies (not
satellites), which are located beyond a MW-mass halo within the zoom-in region and therefore have not been stripped of mass
like satellites have.
The numerical resolution of these simulations varies between a gas particle mass ? 103 M for the highest resolution case
(Marvel Suite), ? 5?103M for Auriga-L3 and FIRE-2, to ? 104 for APOSTLE and NIHAO-UHD. The physics modeled
and its particular implementation also vary from code to code, often with differences in predictions far more impacted by these
physics choices than by numerical resolution. A detailed and fair account of the physics included in each simulation is beyond
the scope of this review. But each simulation included in Fig. 2 is a good example of the current state of affairs in galaxy
formation modeling with demonstrated successes in the prediction of MW-like galaxies with realistic sizes, morphologies,
kinematics, metallicities, star-formation rates, among other properties.
There is substantial overlap on the space spanned by different simulations, which is encouraging given the different codes
LCDM tension with dwarf galaxies
High-precision baryonic physics
120. Research topics related to dwarf galaxies
o Correlation between the diversity of dwarf galaxies and large-scale structures
o Formation/evolution channels of mysterious dwarf galaxies: UFD, UCD, etc.
o Tracing the earliest star remnants in our Milky Way dwarf galaxies
o Relationship between primordial galaxies and supermassive black holes in the early universe
o Effects of ram pressure and hydrodynamic effects on dwarf galaxies
o Tracing the characteristics of stars and galaxies that first appear in the universe
o Internal evolution of gas instability and the impact of environmental factors
o Lyman-? properties of dwarf galaxies and emission/absorption to be compared with observations
o AI modelling to trace history of dwarf galaxies from observational characteristics of dwarf galaxies
¨¨ Various in-depth/detailed research topics from the DARWIN
¨¨ Maximizing synergy through collaboration with observation experts
Research topics using the DARWIN
121. DARWIN & LCDM tensions
500pc 125pc, 62.25pc 4 ¨C 8 pc
Statistical superiority
DARWIN-1 DARWIN-2
High precision
DARWIN-3
Core-cusp
Too big to fail
Diversity of rotation curves
Satellite planes
Missing satellites