اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدتوقعات التعلم العميق لمرحلة دمج المجرات وأهمية الواقعية الملاحظية
Deep learning predictions of galaxy merger stage and the importance of observational realism
749
0
0.0
(
0
)
اسأل ChatGPT حول البحث
تصبح التعلم الآلي أداة مشهورة لتحديد أشكال المجرات وتحديد الدمج. ومع ذلك، يعتمد هذا التقنية على استخدام مجموعة مناسبة من البيانات التدريبية للنجاح. بالإضافة إلى ذلك، بالإشارة إلى تحليلات الهيدروديناميكية، والملاحظات الاصطناعية، وشبكات الأوراق المدرجة (CNNs)، نحن نقيم كيفية الحقيقية الصور المجرى المصنوعة التي يجب أن تكون لتصنيف دقيق للدمج. بالطبع، نحن نقارن أداء الشبكات المدربة على نوعين من الصور المجرى، وهي الخرائط النجمية والصور المضافة الضوئية الشاملة للتراب، بحيث يتم تطبيق ثلاث مستويات من الواقعية الملاحظة: (1) بدون آثار الملاحظة (الصور المثالية)، (2) السماء الواقعية ونقطة الانتشار الواقعية (الصور النصف واقعية)، (3) إدراجها في صورة السماء الواقعية (الصور الكاملة الواقعية). ونجد أن الشبكات المدربة على الصور المثالية أو النصف واقعية لديها أداء سيء عند تطبيقها على الصور الواقعية للمسح. على المقارنة، تحقق الشبكات المدربة على الصور الواقعية كاملة أداء 87.1٪ في التصنيف. من المهم أن الواقعية في الصور التدريبية هي الأكثر أهمية بكثير من ما إذا كانت الصور تشمل التحويل الضوئي أو ببساطة استخدمت الخرائط النجمية (87.1٪ مقارنة 79.6٪ في الدقة). لذلك، يمكن تجنب التكلفة الحاسوبية والتخزين الكبيرة لتشغيل التحويل الضوئي مع ترابطة بسيطة في الأداء التصنيف. يتطلب تجنب الحساسية للألوان في الشبكات الفوتومترية تخفيضا بسيطا في الأداء مع البيانات الواقعية للمسح (86.0٪ مع r-only مقارنة 87.1٪ مع gri). يظهر هذا النتيجة أن على الرغم من أن اللون يمكن استغلاله بواسطة الشبكات الحساسة للألوان، فإنه لا يلزم الوصول إلى دقة عالية وبالتالي يمكن تجنبه إذا أردنا. نحن نقدم إصدار عام لجنة الواقعية الإحصائية الخاصة بنا، RealSim، كمقابلة لهذه المقالة.
Machine learning is becoming a popular tool to quantify galaxy morphologies and identify mergers. However, this technique relies on using an appropriate set of training data to be successful. By combining hydrodynamical simulations, synthetic observations and convolutional neural networks (CNNs), we quantitatively assess how realistic simulated galaxy images must be in order to reliably classify mergers. Specifically, we compare the performance of CNNs trained with two types of galaxy images, stellar maps and dust-inclusive radiatively transferred images, each with three levels of observational realism: (1) no observational effects (idealized images), (2) realistic sky and point spread function (semi-realistic images), (3) insertion into a real sky image (fully realistic images). We find that networks trained on either idealized or semi-real images have poor performance when applied to survey-realistic images. In contrast, networks trained on fully realistic images achieve 87.1% classification performance. Importantly, the level of realism in the training images is much more important than whether the images included radiative transfer, or simply used the stellar maps (87.1% compared to 79.6% accuracy, respectively). Therefore, one can avoid the large computational and storage cost of running radiative transfer with a relatively modest compromise in classification performance. Making photometry-based networks insensitive to colour incurs a very mild penalty to performance with survey-realistic data (86.0% with r-only compared to 87.1% with gri). This result demonstrates that while colour can be exploited by colour-sensitive networks, it is not necessary to achieve high accuracy and so can be avoided if desired. We provide the public release of our statistical observational realism suite, RealSim, as a companion to this paper.
قيم البحث
اقرأ أيضاً
We explore the galaxy-galaxy merger rate with the empirical model for galaxy formation, Emerge. On average, we find that between $2$ per cent and $20$ per cent of massive galaxies ($log_{10}(m_{*}/M_{odot}) geq 10.3$) will experience a major merger p
er Gyr. Our model predicts galaxy merger rates that do not scale as a power-law with redshift when selected by descendant stellar mass, and exhibit a clear stellar mass and mass-ratio dependence. Specifically, major mergers are more frequent at high masses and at low redshift. We show mergers are significant for the stellar mass growth of galaxies $log_{10}(m_{*}/M_{odot}) gtrsim 11.0$. For the most massive galaxies major mergers dominate the accreted mass fraction, contributing as much as $90$ per cent of the total accreted stellar mass. We reinforce that these phenomena are a direct result of the stellar-to-halo mass relation, which results in massive galaxies having a higher likelihood of experiencing major mergers than low mass galaxies. Our model produces a galaxy pair fraction consistent with recent observations, exhibiting a form best described by a power-law exponential function. Translating these pair fractions into merger rates results in an inaccurate prediction compared to the model intrinsic values when using published observation timescales. We find the pair fraction can be well mapped to the intrinsic merger rate by adopting an observation timescale that decreases linearly with redshift as $T_{mathrm{obs}} = -0.36(1+z)+2.39$ [Gyr], assuming all observed pairs merge by $z=0$.
We present results from recent Suzaku and Chandra X-ray, and MMT optical observations of the strongly merging double cluster A1750 out to its virial radius, both along and perpendicular to a putative large-scale structure filament. Some previous stud
ies of individual clusters have found evidence for ICM entropy profiles that flatten at large cluster radii, as compared with the self-similar prediction based on purely gravitational models of hierarchical cluster formation, and gas fractions that rise above the mean cosmic value. Weakening accretion shocks and the presence of unresolved cool gas clumps, both of which are expected to correlate with large scale structure filaments, have been invoked to explain these results. In the outskirts of A1750, we find entropy profiles that are consistent with self-similar expectations, and gas fractions that are consistent with the mean cosmic value, both along and perpendicular to the putative large scale filament. Thus, we find no evidence for gas clumping in the outskirts of A1750, in either direction. This may indicate that gas clumping is less common in lower temperature (kT~4keV), less massive systems, consistent with some (but not all) previous studies of low mass clusters and groups. Cluster mass may therefore play a more important role in gas clumping than dynamical state. Finally, we find evidence for diffuse, cool (<1 keV) gas at large cluster radii (R200) along the filament, which is consistent with the expected properties of the denser, hotter phase of the WHIM.
We estimate the evolution of the galaxy-galaxy merger fraction for $M_star>10^{10.5}M_odot$ galaxies over $0.25<z<1$ in the $sim$18.6 deg$^2$ deep CLAUDS+HSC-SSP surveys. We do this by training a Random Forest Classifier to identify merger candidates
from a host of parametric morphological features, and then visually follow-up likely merger candidates to reach a high-purity, high-completeness merger sample. Correcting for redshift-dependent detection bias, we find that the merger fraction at $z=0$ is 1.0$pm$0.2%, that the merger fraction evolves as $(1+z)^{2.3 pm 0.4}$, and that a typical massive galaxy has undergone $sim$0.3 major mergers since $z=1$. This pilot study illustrates the power of very deep ground-based imaging surveys combined with machine learning to detect and study mergers through the presence of faint, low surface brightness merger features out to at least $zsim1$.
Galaxy mergers are expected to have a significant role in the mass assembly of galaxies in the early Universe, but there are very few observational constraints on the merger history of galaxies at $z>2$. We present the first study of galaxy major mer
gers (mass ratios $>$ 1:4) in mass-selected samples out to $zapprox6$. Using all five fields of the HST/CANDELS survey and a probabilistic pair count methodology that incorporates the full photometric redshift posteriors and corrections for stellar mass completeness, we measure galaxy pair-counts for projected separations between 5 and 30 kpc in stellar mass selected samples at $9.7 < log_{10}(rm{M}_{*}/rm{M}_{odot}) < 10.3$ and $log_{10}(rm{M}_{*}/rm{M}_{odot}) > 10.3$. We find that the major merger pair fraction rises with redshift to $zapprox6$ proportional to $(1+z)^{m}$, with $m = 0.8pm0.2$ ($m = 1.8pm0.2$) for $log_{10}(rm{M}_{*} / rm{M}_{odot}) > 10.3$ ($9.7 < log_{10}(rm{M}_{*}/rm{M}_{odot}) < 10.3$). Investigating the pair fraction as a function of mass ratio between 1:20 and 1:1, we find no evidence for a strong evolution in the relative numbers of minor to major mergers out to $z<3$. Using evolving merger timescales we find that the merger rate per galaxy ($mathcal{R}$) rises rapidly from $0.07pm 0.01$ Gyr$^{-1}$ at $z < 1$ to $7.6pm 2.7$ Gyr$^{-1}$ at $z = 6$ for galaxies at $log_{10}(rm{M}_{*}/rm{M}_{odot}) > 10.3$. The corresponding co-moving major merger rate density remains roughly constant during this time, with rates of $Gamma approx 10^{-4}$ Gyr$^{-1}$ Mpc$^{-3}$. Based on the observed merger rates per galaxy, we infer specific mass accretion rates from major mergers that are comparable to the specific star-formation rates for the same mass galaxies at $z>3$ - observational evidence that mergers are as important a mechanism for building up mass at high redshift as in-situ star-formation.
We present predictions for the outcome of deep galaxy surveys with the $James$ $Webb$ $Space$ $Telescope$ ($JWST$) obtained from a physical model of galaxy formation in $Lambda$CDM. We use the latest version of the GALFORM model, embedded within a ne
w ($800$ Mpc)$^{3}$ dark matter only simulation with a halo mass resolution of $M_{rm halo}>2times10^{9}$ $h^{-1}$ M$_{odot}$. For computing full UV-to-mm galaxy spectral energy distributions, including the absorption and emission of radiation by dust, we use the spectrophotometric radiative transfer code GRASIL. The model is calibrated to reproduce a broad range of observational data at $zlesssim6$, and we show here that it can also predict evolution of the rest-frame far-UV luminosity function for $7lesssim zlesssim10$ which is in good agreement with observations. We make predictions for the evolution of the luminosity function from $z=16$ to $z=0$ in all broadband filters on the Near InfraRed Camera (NIRCam) and Mid InfraRed Instrument (MIRI) on $JWST$ and present the resulting galaxy number counts and redshift distributions. Our fiducial model predicts that $sim1$ galaxy per field of view will be observable at $zsim11$ for a $10^4$ s exposure with NIRCam. A variant model, which produces a higher redshift of reionization in better agreement with $Planck$ data, predicts number densities of observable galaxies $sim5times$ greater at this redshift. Similar observations with MIRI are predicted not to detect any galaxies at $zgtrsim6$. We also make predictions for the effect of different exposure times on the redshift distributions of galaxies observable with $JWST$, and for the angular sizes of galaxies in $JWST$ bands.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
