相关论文: Multi-modal Foundation Model for Cosmological Simu…
The versatility of self-attention mechanism earned transformers great success in almost all data modalities, with limitations on the quadratic complexity and difficulty of training. To apply transformers across different data modalities,…
Measuring environment for large numbers of distant galaxies is still an open problem, for which we need galaxy positions and redshifts. Photometric redshifts are more easily available for large numbers of galaxies, but at the price of…
In order to best improve constraints on cosmological parameters and on models of modified gravity using current and future galaxy surveys it is necessary maximally exploit the available data. As redshift-space distortions mean statistical…
Context. Strong lensing mass measurements require the knowledge of the redshift of both the lens and the source galaxy. Traditionally, spectroscopic redshifts are used for this purpose. Upcoming surveys, however, will lead to the discovery…
Modeling the mass distribution of galaxy-scale strong gravitational lenses is a task of increasing difficulty. The high-resolution and depth of imaging data now available render simple analytical forms ineffective at capturing lens…
We present a field-level emulator for large-scale structure, capturing the cosmology dependence and the time evolution of cosmic structure formation. The emulator maps linear displacement fields to their corresponding nonlinear…
Multi-modal image fusion integrates complementary information from different modalities into a unified representation. Current methods predominantly optimize statistical correlations between modalities, often capturing dataset-induced…
Multi-modal medical imaging enables comprehensive diagnostics, yet current foundation models process 2D (e.g. X-ray) and 3D (e.g. CT) data with separate, dimensionality-specific architectures. We present MultiMedVision, a unified framework…
Spatial transcriptomics (ST) enables gene expression mapping within anatomical context but remains costly and low-throughput. Hematoxylin and eosin (H\&E) staining offers rich morphology yet lacks molecular resolution. We present…
We employ high-resolution zoom-in cosmological simulations to analyze the emerging morphology of galaxies in dark matter halos at redshifts z > 2. We choose DM halos of similar masses of log (Mvir/Mo) ~11.65 +- 0.05 at the target redshifts…
The rapid expansion of multi-source satellite imagery drives innovation in Earth observation, opening unprecedented opportunities for Remote Sensing Foundation Models to harness diverse data. However, many existing models remain constrained…
With the launch and application of next-generation ground- and space-based telescopes, astronomy has entered the era of big data, necessitating more efficient and robust data analysis methods. Most traditional parameter estimation methods…
The Cosmic Evolution Survey (COSMOS) allows for the first time a highly significant census of environments and structures up to redshift one, as well as a full morphological description of the galaxy population. In this paper we present a…
We use the Millennium Simulation, a 10 billion particle simulation of the growth of cosmic structure, to construct a new model of galaxy clustering. We adopt a methodology that falls midway between the traditional semi-analytic approach and…
Given a dataset of expert trajectories, standard imitation learning approaches typically learn a direct mapping from observations (e.g., RGB images) to actions. However, such methods often overlook the rich interplay between different…
Foundation models are vital tools in various Computer Vision applications. They take as input a single RGB image and output a deep feature representation that is useful for various applications. However, in case we have multiple views of…
We combine magnitude and photometric redshift data on galaxies in the Hubble Deep Fields with morphological classifications in order to separate out the distributions for early type galaxies. The updated morphological galaxy number counts…
Large foundation models (FMs) are transforming Earth science by integrating heterogeneous multimodal data, such as multi-platform imagery, gridded reanalysis data, diverse geophysical and geochemical observations, and domain-specific text,…
We present a model for star formation and supernova feedback that describes the multi-phase structure of star forming gas on scales that are typically not resolved in cosmological simulations. Our approach includes radiative heating and…
In a hierarchical, dark matter-dominated Universe, stellar mass functions (SMFs), galaxy merger rates, star formation histories (SFHs), satellite abundances, and intracluster light, should all be intimately connected observables. However,…