اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدExploring galaxy evolution with generative models
73
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Context. Generative models open up the possibility to interrogate scientific data in a more data-driven way. Aims: We propose a method that uses generative models to explore hypotheses in astrophysics and other areas. We use a neural network to show how we can independently manipulate physical attributes by encoding objects in latent space. Methods: By learning a latent space representation of the data, we can use this network to forward model and explore hypotheses in a data-driven way. We train a neural network to generate artificial data to test hypotheses for the underlying physical processes. Results: We demonstrate this process using a well-studied process in astrophysics, the quenching of star formation in galaxies as they move from low-to high-density environments. This approach can help explore astrophysical and other phenomena in a way that is different from current methods based on simulations and observations.
قيم البحث
اقرأ أيضاً
One of the key science drivers for the development of the SKA is to observe the neutral hydrogen, HI, in galaxies as a means to probe galaxy evolution across a range of environments over cosmic time. Over the past decade, much progress has been made
in theoretical simulations and observations of HI in galaxies. However, recent HI surveys on both single dish radio telescopes and interferometers, while providing detailed information on global HI properties, the dark matter distribution in galaxies, as well as insight into the relationship between star formation and the interstellar medium, have been limited to the local universe. Ongoing and upcoming HI surveys on SKA pathfinder instruments will extend these measurements beyond the local universe to intermediate redshifts with long observing programmes. We present here an overview of the HI science which will be possible with the increased capabilities of the SKA and which will build upon the expected increase in knowledge of HI in and around galaxies obtained with the SKA pathfinder surveys. With the SKA1 the greatest improvement over our current measurements is the capability to image galaxies at reasonable linear resolution and good column density sensitivity to much higher redshifts (0.2 < z < 1.7). So one will not only be able to increase the number of detections to study the evolution of the HI mass function, but also have the sensitivity and resolution to study inflows and outflows to and from galaxies and the kinematics of the gas within and around galaxies as a function of environment and cosmic time out to previously unexplored depths. The increased sensitivity of SKA2 will allow us to image Milky Way-size galaxies out to redshifts of z=1 and will provide the data required for a comprehensive picture of the HI content of galaxies back to z~2 when the cosmic star formation rate density was at its peak.
We combine inferred galaxy properties from a semi-analytic galaxy evolution model incorporating dark matter halo merger trees with new estimates of supernova and gamma ray burst rates as a function of metallicity from stellar population synthesis mod
els incorporating binary interactions. We use these to explore the stellar mass fraction of galaxies irradiated by energetic astrophysical transients and its evolution over cosmic time, and thus the fraction which is potentially habitable by life like our own. We find that 18 per cent of the stellar mass in the Universe is likely to have been irradiated within the last 260 Myr, with GRBs dominating that fraction. We do not see a strong dependence of irradiated stellar mass fraction on stellar mass or richness of the galaxy environment. We consider a representative merger tree as a Local Group analogue, and find that there are galaxies at all masses which have retained a high habitable fraction (>40 per cent) over the last 6 Gyr, but also that there are galaxies at all masses where the merger history and associated star formation have rendered galaxies effectively uninhabitable. This illustrates the need to consider detailed merger trees when evaluating the cosmic evolution of habitability.
Despite recent advances, the remaining bottlenecks in deep generative models are necessity of extensive training and difficulties with generalization from small number of training examples. We develop a new generative model called Generative Matching
Network which is inspired by the recently proposed matching networks for one-shot learning in discriminative tasks. By conditioning on the additional input dataset, our model can instantly learn new concepts that were not available in the training data but conform to a similar generative process. The proposed framework does not explicitly restrict diversity of the conditioning data and also does not require an extensive inference procedure for training or adaptation. Our experiments on the Omniglot dataset demonstrate that Generative Matching Networks significantly improve predictive performance on the fly as more additional data is available and outperform existing state of the art conditional generative models.
When using deep, multi-layered architectures to build generative models of data, it is difficult to train all layers at once. We propose a layer-wise training procedure admitting a performance guarantee compared to the global optimum. It is based on
an optimistic proxy of future performance, the best latent marginal. We interpret auto-encoders in this setting as generative models, by showing that they train a lower bound of this criterion. We test the new learning procedure against a state of the art method (stacked RBMs), and find it to improve performance. Both theory and experiments highlight the importance, when training deep architectures, of using an inference model (from data to hidden variables) richer than the generative model (from hidden variables to data).
We present a machine learning framework to simulate realistic galaxies for the Euclid Survey. The proposed method combines a control on galaxy shape parameters offered by analytic models with realistic surface brightness distributions learned from re
al Hubble Space Telescope observations by deep generative models. We simulate a galaxy field of $0.4,rm{deg}^2$ as it will be seen by the Euclid visible imager VIS and show that galaxy structural parameters are recovered with similar accuracy as for pure analytic Sersic profiles. Based on these simulations, we estimate that the Euclid Wide Survey will be able to resolve the internal morphological structure of galaxies down to a surface brightness of $22.5,rm{mag},rm{arcsec}^{-2}$, and $24.9,rm{mag},rm{arcsec}^{-2}$ for the Euclid Deep Survey. This corresponds to approximately $250$ million galaxies at the end of the mission and a $50,%$ complete sample for stellar masses above $10^{10.6},rm{M}_odot$ (resp. $10^{9.6},rm{M}_odot$) at a redshift $zsim0.5$ for the wide (resp. deep) survey. The approach presented in this work can contribute to improving the preparation of future high-precision cosmological imaging surveys by allowing simulations to incorporate more realistic galaxies.
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
