We use explainable neural networks to connect the evolutionary history of
dark matter halos with their density profiles. The network captures independent
factors of variation in the density profiles within a low-dimensional
representation, which we physically interpret using mutual information. Without
any prior knowledge of the halos' evolution, the network recovers the known
relation between the early time assembly and the inner profile, and discovers
that the profile beyond the virial radius is described by a single parameter
capturing the most recent mass accretion rate. The results illustrate the
potential for machine-assisted scientific discovery in complicated
astrophysical datasets.

利用可解释的神经网络将暗物质晕的演化历史与其密度配置文件相连接，并发现复杂的天体物理数据集中的潜在科学发现的潜力。

用神经网络解释暗物质晕密度分布

Explaining dark matter halo density profiles with neural networks

The density profiles of dark matter halos are typically modeled using
empirical formulae fitted to the density profiles of relaxed halo populations.
We present a neural network model that is trained to learn the mapping from the
raw density field containing each halo to the dark matter density profile. We
show that the model recovers the widely-used Navarro-Frenk-White (NFW) profile
out to the virial radius, and can additionally describe the variability in the
outer profile of the halos. The neural network architecture consists of a
supervised encoder-decoder framework, which first compresses the density inputs
into a low-dimensional latent representation, and then outputs $\rho(r)$ for
any desired value of radius $r$. The latent representation contains all the
information used by the model to predict the density profiles. This allows us
to interpret the latent representation by quantifying the mutual information
between the representation and the halos' ground-truth density profiles. A
two-dimensional representation is sufficient to accurately model the density
profiles up to the virial radius; however, a three-dimensional representation
is required to describe the outer profiles beyond the virial radius. The
additional dimension in the representation contains information about the
infalling material in the outer profiles of dark matter halos, thus discovering
the splashback boundary of halos without prior knowledge of the halos'
dynamical history.

该论文介绍了使用神经网络建模预测暗物质晕的密度分布，并利用低维的潜变量描述导致晕边界的物质的两种不同表现形式，其中包括 Navarro-Frenk-White 模型和 splashback boundary。

利用神经网络发现暗物质晕密度分布的基本组成部分

Discovering the building blocks of dark matter halo density profiles with neural networks

For nearly 40 years, dark matter has been widely assumed to be cold and
collisionless. Cold dark matter models make fundamental predictions for the
behavior of dark matter on small (<10 kpc) scales. These predictions include
cuspy density profiles at the centers of dark matter halos and a halo mass
function that increases as dN/dM ~ M^-1.9 down to very small masses. We suggest
two observational programs relying on extremely large telescopes to critically
test these predictions, and thus shed new light on the nature of dark matter.
(1) Combining adaptive optics-enabled imaging with deep spectroscopy to measure
the three-dimensional motions of stars within a sample of Local Group dwarf
galaxies that are the cleanest dark matter laboratories known in the nearby
universe. From these observations the inner slope of the dark matter density
profile can be determined with an accuracy of 0.20 dex, enabling a central cusp
to be distinguished from a core at 5 sigma significance. (2)
Diffraction-limited AO imaging and integral field spectroscopy of
gravitationally lensed galaxies and quasars to quantify the abundance of dark
substructures in the halos of the lens galaxies and along the line of sight.
Observations of 50 lensed arcs and 50 multiply-imaged quasars will be
sufficient to measure the halo mass function over the range 10^7 < M < 10^10
Msun at cosmological scales, independent of the baryonic and stellar
composition of those structures. These two observational probes provide
complementary information about the small scale structure, with a joint
self-consistent analysis mitigating limitations of either probe. This program
will produce the strongest existing constraints on the properties of dark
matter on small scales, allowing conclusive tests of alternative warm, fuzzy,
and self-interacting dark matter models.

这篇论文提出两种利用大型望远镜进行观测的方法，通过研究矮星系内恒星的运动轨迹以及通过引力透镜效应观测具有暗物质结构的系外星系和类星体，以检验冷暗物质模型对小尺度天体的占据和分布的预测，并为暗物质的性质提供新的限制。

使用极大望远镜测试暗物质的性质

Testing the Nature of Dark Matter with Extremely Large Telescopes

We use high-resolution N-body simulations to study the equilibrium density
profiles of dark matter halos in hierarchically clustering universes. We find
that all such profiles have the same shape, independent of halo mass, of
initial density fluctuation spectrum, and of the values of the cosmological
parameters. Spherically averaged equilibrium profiles are well fit over two
decades in radius by a simple formula originally proposed to describe the
structure of galaxy clusters in a cold dark matter universe. In any particular
cosmology the two scale parameters of the fit, the halo mass and its
characteristic density, are strongly correlated. Low-mass halos are
significantly denser than more massive systems, a correlation which reflects
the higher collapse redshift of small halos. The characteristic density of an
equilibrium halo is proportional to the density of the universe at the time it
was assembled. A suitable definition of this assembly time allows the same
proportionality constant to be used for all the cosmologies that we have
tested. We compare our results to previous work on halo density profiles and
show that there is good agreement. We also provide a step-by-step analytic
procedure, based on the Press-Schechter formalism, which allows accurate
equilibrium profiles to be calculated as a function of mass in any hierarchical
model.

通过高分辨率 N 体模拟，研究暗物质暗星系的平衡密度分布；找到了适用于代表这些分布的简单公式，并且发现这些公式适用于各种不同的宇宙模型。