Unlocking the Milky Way's Secrets: How New Tech Bridges Theory and Observation
"A Powerful Tool for Comparing Galactic Models and Observational Data"
The past decade has witnessed remarkable advancements in both the theoretical understanding and observational mapping of the Milky Way. Extensive photometric surveys like 2MASS and SDSS have charted significant portions of our galaxy, while spectroscopic surveys such as GCS, SEGUE, and RAVE have provided radial velocity and spectral data for a multitude of stars. Furthermore, ambitious projects like LSST, GAIA, and SkyMapper promise an unprecedented wealth of new observations.
This surge in data, characterized by both the sheer number of stars cataloged and their distribution across the sky, has created a pressing need for efficient tools that can bridge the gap between theoretical models of our galaxy and the ever-expanding observational landscape. To facilitate this crucial comparison, theoretical models must be translated into the observational domain and then reconciled with inherent observational uncertainties. The Galaxia code is specifically designed to address this challenge.
Galaxia uses efficient and fast algorithms for creating a synthetic survey of the Milky Way. Both analytic and N-body models can be sampled by Galaxia. For N-body models, a scheme is presented that disperses the stars spawned by an N-body particle, in such a way that the phase space density of the spawned stars is consistent with that of the N-body particles.
Simulations and Milky Way Data
NASA’s June 2025 visualization describes how 100,000 computer simulations challenged the longstanding expectation that the Milky Way and Andromeda would collide in less than 5 billion years; the snippet says that future may change. Separately, an analysis using Gaia Space Observatory data examined the motions of 55,038,539 stars, selected in different directions from the Sun and out to distances of 3–6 kpc. These examples show simulations revisiting galactic forecasts alongside statistical analysis of large stellar datasets.
Cosmological Models and Galaxy Simulations
Numerical simulations based on the standard cosmological model have been used for decades to study cosmic structures and galaxy formation, and the model is described as a successful framework for explaining and predicting large-scale structure. A separate 2025 paper proposes an equilibrium Milky Way-like model with open, mostly two- and three-arm spiral patterns but no bar. Its authors suggest this as a more realistic alternative to the lower-disk-mass model used by the agora collaboration, which forms only multi-arm spiral patterns. The proposed model is intended to enable simulations that better represent those structures.
Tracing the Galaxy’s Long Formation
FIRE-based Latte simulations have tracked Milky Way-mass halos across the universe’s 13.8-billion-year history, modeling dark-matter halos, giant molecular clouds, and satellite galaxies. A 2026 account describes the Milky Way as assembling, merging, and rebuilding over nearly the entire history of the universe, rather than being born at one moment; it gives its age as roughly 13 billion years. Another report says a simulation of the Milky Way’s first several billion years took roughly three years to calculate on powerful supercomputers.
Galaxia: Bridging Theory and Observation
Galaxia employs the von Neumann rejection technique to sample analytic models effectively. This method is well-suited for continuous sampling across multidimensional spaces. Recognizing that a naive approach can be computationally expensive, the code divides the galaxy into a set of roughly equal mass nodes and applies rejection sampling to each. This division allows for optimization; for instance, depending on the distance of the node, a suitable lower limit can be set on the mass of the star to be generated, boosting processing speed.
- Efficiency: Speeds of about 0.16 million stars per second can be reached on a single 2.44 GHz CPU, higher for more shallow surveys.
- Comprehensive Surveys: A V < 20 magnitude limited survey of the NGP, covering 10,000 square degrees and consisting of about 35 million stars, can be generated in 220 seconds.
- Large Scale Applications: A V < 20 all sky GAIA like survey would require about 6 hours on a single CPU.
Comparing Simulated and Observed Stellar Streams
The study “Streams on FIRE: Populations of Detectable Stellar Streams in the Milky Way and FIRE” compares streams identified around 13 Milky Way analogs in the FIRE-2 simulations with streams observed by the Southern Stellar Stream Spectroscopic Survey (S⁵). The comparison examines simulated and observed stellar-stream populations. The supplied source summary does not state the study’s specific results.
Resolution and Modeling Challenges
The Rhea simulations source says that detailed comparison with observations requires modeling the Milky Way closely, while also noting that observations do not fully define details of Galactic structure. A separate account reports a simulation representing more than 100 billion individual stars over 10,000 years, while another describes combining physical simulations with a deep-learning surrogate model to capture galaxy-wide dynamics and small-scale stellar phenomena. A source on the computational challenge says a conventional simulation resolving the Milky Way down to individual stars would take 315 hours for every million years of simulated time, making a billion years take more than 36 years of real time.
Two Types of Atmospheric Halo
The listed sources concern optical halos rather than the Milky Way. One reports that the 22° halo is observed most frequently, while the 46° halo occurs less often. Another describes an unusual lunar-halo observation in which a 22° refraction halo, a circumscribed halo, and a reflection halo called the paraselenic circle were present. Together, these accounts contrast a commonly observed halo with a particular observation containing three halo forms.
Galactic Archeology with Velocity Information
Galaxia can be used to identify structures in the stellar halo. In the currently favored ACDM paradigm of structure formation the stellar halo is thought to have been built up by accretion of satellite galaxies.
A Star-by-Star Simulation Milestone
A November 2025 report says researchers created the first Milky Way simulation with more than 100 billion individual stars. It traced 10,000 years of galactic evolution, which the report describes as achieving accuracy beyond what a machine had previously achieved. The milestone reflects the scale of computing work now being applied to modeling our galaxy.
In order to objectively identify the structures we use the multi-dimensional group finding algorithm EnLink. The advantage of using EnLink being that it can work in spaces of arbitrary dimensions. Moreover, it has an in built significance estimator such that only genuine groups that stand out above the Poisson noise are considered.
As one adds radial velocity and proper motion information, we discover more groups and sample a larger number of unique accretion events. Specifically in the distance range 20–70 kpc, when velocity information is added, there is a potential for discovery of a large number of new structures.