Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userDeepShadows: Separating Low Surface Brightness Galaxies from Artifacts using Deep Learning
124
0
0.0
(
0
)
Added by
Dimitrios Tanoglidis
Publication date
2020
fields
Physics
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Searches for low-surface-brightness galaxies (LSBGs) in galaxy surveys are plagued by the presence of a large number of artifacts (e.g., objects blended in the diffuse light from stars and galaxies, Galactic cirrus, star-forming regions in the arms of spiral galaxies, etc.) that have to be rejected through time consuming visual inspection. In future surveys, which are expected to collect hundreds of petabytes of data and detect billions of objects, such an approach will not be feasible. We investigate the use of convolutional neural networks (CNNs) for the problem of separating LSBGs from artifacts in survey images. We take advantage of the fact that, for the first time, we have available a large number of labeled LSBGs and artifacts from the Dark Energy Survey, that we use to train, validate, and test a CNN model. That model, which we call DeepShadows, achieves a test accuracy of $92.0 %$, a significant improvement relative to feature-based machine learning models. We also study the ability to use transfer learning to adapt this model to classify objects from the deeper Hyper-Suprime-Cam survey, and we show that after the model is retrained on a very small sample from the new survey, it can reach an accuracy of $87.6%$. These results demonstrate that CNNs offer a very promising path in the quest to study the low-surface-brightness universe.
rate research
Read More
The existence of galaxies with a surface brightness $mu$ lower than the night sky has been known since three decades. Yet, their formation mechanism and emergence within a $rmLambda CDM$ universe has remained largely undetermined. For the first time, we investigated the origin of Low Surface Brightness (LSB) galaxies with M$_{star}$$sim$10$^{9.5-10}$M$_{odot}$, which we are able to reproduce within hydrodynamical cosmological simulations from the NIHAO suite. The simulated and observed LSBs share similar properties, having large HI reservoir, extended star formation histories and effective radii, low S{e}rsic index and slowly rising rotation curves. The formation mechanism of these objects is explored: simulated LSBs form as a result of co-planar co-rotating mergers and aligned accretion of gas at early times, while perpendicular mergers and mis-aligned gas accretion result in higher $mu$ galaxies by $z$=0. The larger the merger, the stronger the correlation between merger orbital configuration and final $mu$. While the halo spin parameter is consistently high in simulated LSB galaxies, the impact of halo concentration, feedback-driven gas outflows and merger time only plays a minor-to-no role in determining $mu$. Interestingly, the formation scenario of such `classical LSBs differs from the one of less massive, M$_{star}$$sim$10$^{7-9}$M$_{odot}$, Ultra-Diffuse Galaxies, the latter resulting from the effects of SNae driven gas outflows: a M$_{star}$ of $sim$10$^9$M$_{odot}$ thus represents the transition regime between a feedback-dominated to an angular momentum-dominated formation scenario in the LSB realm. Observational predictions are offered regarding spatially resolved star formation rates through LSB discs: these, together with upcoming surveys, can be used to verify the proposed emergence scenario of LSB galaxies.
We introduce a method for producing a galaxy sample unbiased by surface brightness and stellar mass, by selecting star-forming galaxies via the positions of core-collapse supernovae (CCSNe). Whilst matching $sim$2400 supernovae from the SDSS-II Supernova Survey to their host galaxies using IAC Stripe 82 legacy coadded imaging, we find $sim$150 previously unidentified low surface brightness galaxies (LSBGs). Using a sub-sample of $sim$900 CCSNe, we infer CCSN-rate and star-formation rate densities as a function of galaxy stellar mass, and the star-forming galaxy stellar mass function. Resultant star-forming galaxy number densities are found to increase following a power-law down to our low mass limit of $sim10^{6.4}$ M$_{odot}$ by a single Schechter function with a faint-end slope of $alpha = -1.41$. Number densities are consistent with those found by the EAGLE simulations invoking a $Lambda$-CDM cosmology. Overcoming surface brightness and stellar mass biases is important for assessment of the sub-structure problem. In order to estimate galaxy stellar masses, a new code for the calculation of galaxy photometric redshifts, zMedIC, is also presented, and shown to be particularly useful for small samples of galaxies.
The Rastall gravity is a modification of Einsteins general relativity, in which the energy-momentum conservation is not satisfied and depends on the gradient of the Ricci curvature. It is in dispute whether the Rastall gravity is equivalent to the general relativity (GR). In this work, we constrain the theory using the rotation curves of Low Surface Brightness (LSB) spiral galaxies. Through fitting the rotation curves of LSB galaxies, we obtain the parameter $beta$ of the Rastall gravity. The $beta$ values of LSB galaxies satisfy Weak Energy Condition (WEC) and Strong Energy Condition(SEC). Combining the $beta$ values of type Ia supernovae and gravitational lensing of elliptical galaxies on the Rastall gravity, we conclude that the Rastall gravity is equivalent to the general relativity.
We present an in-depth study of surface brightness fluctuations (SBFs) in low-luminosity stellar systems. Using the MIST models, we compute theoretical predictions for absolute SBF magnitudes in the LSST, HST ACS/WFC, and proposed Roman Space Telescope filter systems. We compare our calculations to observed SBF-color relations of systems that span a wide range of age and metallicity. Consistent with previous studies, we find that single-age population models show excellent agreement with observations of low-mass galaxies with $0.5 lesssim g - i lesssim 0.9$. For bluer galaxies, the observed relation is better fit by models with composite stellar populations. To study SBF recovery from low-luminosity systems, we perform detailed image simulations in which we inject fully populated model galaxies into deep ground-based images from real observations. Our simulations show that LSST will provide data of sufficient quality and depth to measure SBF magnitudes with precisions of ${sim}0.2$-0.5 mag in ultra-faint $left(mathrm{10^4 leq M_star/M_odot leq 10^5}right)$ and low-mass classical (M$_starleq10^7$ M$_odot$) dwarf galaxies out to ${sim}4$ Mpc and ${sim}25$ Mpc, respectively, within the first few years of its deep-wide-fast survey. Many significant practical challenges and systematic uncertainties remain, including an irreducible sampling scatter in the SBFs of ultra-faint dwarfs due to their undersampled stellar mass functions. We nonetheless conclude that SBFs in the new generation of wide-field imaging surveys have the potential to play a critical role in the efficient confirmation and characterization of dwarf galaxies in the nearby universe.
We present a catalog of 23,790 extended low-surface-brightness galaxies (LSBGs) identified in $sim 5000 deg^2$ from the first three years of imaging data from the Dark Energy Survey (DES). Based on a single-component Sersic model fit, we define extended LSBGs as galaxies with $g$-band effective radii $R_{eff}(g) > 2.5$ and mean surface brightness $bar{mu}_{eff}(g) > 24.2 ,mag .arcsec^{-2}$. We find that the distribution of LSBGs is strongly bimodal in $(g-r)$ vs. $(g-i$) color space. We divide our sample into red ($g-i geq 0.60$) and blue ($g-i<0.60$) galaxies and study the properties of the two populations. Redder LSBGs are more clustered than their blue counterparts and are correlated with the distribution of nearby ($z < 0.10$) bright galaxies. Red LSBGs constitute $sim 33%$ of our LSBG sample, and $sim 30%$ of these are located within 1 deg of low-redshift galaxy groups and clusters (compared to $sim 8%$ of the blue LSBGs). For nine of the most prominent galaxy groups and clusters, we calculate the physical properties of associated LSBGs assuming a redshift derived from the host system. In these systems, we identify 41 objects that can be classified as ultra-diffuse galaxies, defined as LSBGs with projected physical effective radii $R_{eff} > 1.5 ,kpc$ and central surface brighthness $mu_0(g) > 24.0, mag ,arcsec^{-2}$. The wide-area sample of LSBGs in DES can be used to test the role of environment on models of LSBG formation and evolution.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
