Galaxy Images Research Articles

The Atacama Cosmology Telescope: arcminute-resolution maps of 18 000 square degrees of the microwave sky from ACT 2008–2018 data combined with Planck

This paper presents a maximum-likelihood algorithm for combining sky maps with disparate sky coverage, angular resolution and spatially varying anisotropic noise into a single map of the sky. We use this to merge hundreds of individual maps covering the 2008–2018 ACT observing seasons, resulting in by far the deepest ACT maps released so far. We also combine the maps with the full Planck maps, resulting in maps that have the best features of both Planck and ACT: Planck's nearly white noise on intermediate and large angular scales and ACT's high-resolution and sensitivity on small angular scales. The maps cover over 18 000 square degrees, nearly half the full sky, at 100, 150 and 220 GHz. They reveal 4 000 optically-confirmed clusters through the Sunyaev Zel'dovich effect (SZ) and 18 500 point source candidates at > 5σ, the largest single collection of SZ clusters and millimeter wave sources to date. The multi-frequency maps provide millimeter images of nearby galaxies and individual Milky Way nebulae, and even clear detections of several nearby stars. Other anticipated uses of these maps include, for example, thermal SZ and kinematic SZ cluster stacking, CMB cluster lensing and galactic dust science. The method itself has negligible bias. However, due to the preliminary nature of some of the component data sets, we caution that these maps should not be used for precision cosmological analysis. The maps are part of ACT DR5, and will be made available on \\href{https://lambda.gsfc.nasa.gov/}{LAMBDA} no later than three months after the journal publication of this article, along with an interactive sky atlas.

Towards a consistent framework of comparing galaxy mergers in observations and simulations

Aims.We aim to perform consistent comparisons between observations and simulations on the mass dependence of the galaxy major merger fraction at low redshift over an unprecedentedly wide range of stellar masses (∼109to 1012 M⊙).Methods.We first carry out forward modelling of ideal synthetic images of major mergers and non-mergers selected from the Next Generation Illustris Simulations (IllustrisTNG) to include major observational effects. We then train deep convolutional neural networks (CNNs) using realistic mock observations of galaxy samples from the simulations. Subsequently, we apply the trained CNNs to real the Kilo-Degree Survey (KiDS) images of galaxies selected from the Galaxy And Mass Assembly (GAMA) survey. Based on the major merger samples, which are detected in a consistent manner in the observations and simulations, we determine the dependence of major merger fraction on stellar mass atz ∼ 0.15 and make comparisons between the two.Results.The detected major merger fraction in the GAMA/KiDS observations has a fairly mild decreasing trend with increasing stellar mass over the mass range 109 M⊙ < M* < 1011.5 M⊙. There is good agreement in the mass dependence of the major merger fraction in the GAMA/KiDS observations and the IllustrisTNG simulations over 109.5 M⊙ < M* < 1010.5 M⊙. However, the observations and the simulations show some differences atM* > 1010.5 M⊙, possibly due to the supermassive blackhole feedback in its low-accretion state in the simulations which causes a sharp transition in the quenched fractions at this mass scale. The discrepancy could also be due to the relatively small volume of the simulations and/or differences in how stellar masses are measured in simulations and observations.

Satellites and central galaxies in SDSS: the influence of interactions on their properties

ABSTRACT We use SDSS-DR14 to construct a sample of galaxy systems consisting of a central object and two satellites. We adopt projected distance and radial velocity difference criteria and impose an isolation criterion to avoid membership in larger structures. We also classify the interaction between the members of each system through a visual inspection of galaxy images, finding ${\sim}80{{\ \rm per\ cent}}$ of the systems lack evidence of interactions whilst the remaining ${\sim}20{{\ \rm per\ cent}}$ involve some kind of interaction, as inferred from their observed distorted morphology. We have considered separately, samples of satellites and central galaxies, and each of these samples were tested against suitable control sets to analyse the results. We find that central galaxies showing signs of interactions present evidence of enhanced star formation activity and younger stellar populations. As a counterpart, satellite samples show these galaxies presenting older stellar populations with a lower star formation rate than the control sample. The observed trends correlate with the stellar mass content of the galaxies and with the projected distance between the members involved in the interaction. The most massive systems are less affected since they show no star formation excess, possibly due to their more evolved stage and less gas available to form new stars. Our results suggest that it is arguably a transfer of material during interactions, with satellites acting as donors to the central galaxy. As a consequence of the interactions, satellite stellar population ages rapidly and new bursts of star formation may frequently occur in the central galaxy.

Strongly Lensed Supernovae in Well-Studied Galaxy Clusters with the Vera C. Rubin Observatory

Strong lensing by galaxy clusters can be used to significantly expand the survey reach, thus allowing observation of magnified high-redshift supernovae that otherwise would remain undetected. Strong lensing can also provide multiple images of the galaxies that lie behind the clusters. Detection of strongly lensed Type Ia supernovae (SNe Ia) is especially useful because of their standardizable brightness, as they can be used to improve either cluster lensing models or independent measurements of cosmological parameters. The cosmological parameter, the Hubble constant, is of particular interest given the discrepancy regarding its value from measurements with different approaches. Here, we explore the feasibility of the Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST) of detecting strongly lensed SNe in the field of five galaxy clusters (Abell 1689 and Hubble Frontier Fields clusters) that have well-studied lensing models. Considering the 88 systems composed of 268 individual multiple images in the five cluster fields, we find that the LSST will be sensitive to SNe Ia (SNe IIP) exploding in 41 (23) galaxy images. The range of redshift of these galaxies is between 1.01<z<3.05. During its 10 years of operation, LSST is expected to detect 0.2±0.1 SN Ia and 0.9±0.3 core collapse SNe. However, as LSST will observe many more massive galaxy clusters, it is likely that the expectations are higher. We stress the importance of having an additional observing program for photometric and spectroscopic follow-up of the strongly lensed SNe detected by LSST.

AstroVaDEr: astronomical variational deep embedder for unsupervised morphological classification of galaxies and synthetic image generation

ABSTRACT We present AstroVaDEr (Astronomical Variational Deep Embedder), a variational autoencoder designed to perform unsupervised clustering and synthetic image generation using astronomical imaging catalogues. The model is a convolutional neural network that learns to embed images into a low-dimensional latent space, and simultaneously optimizes a Gaussian Mixture Model (GMM) on the embedded vectors to cluster the training data. By utilizing variational inference, we are able to use the learned GMM as a statistical prior on the latent space to facilitate random sampling and generation of synthetic images. We demonstrate AstroVaDEr’s capabilities by training it on grey-scaled gri images from the Sloan Digital Sky Survey, using a sample of galaxies that are classified by Galaxy Zoo 2. An unsupervised clustering model is found that separates galaxies based on learned morphological features such as axial ratio, surface brightness profile, orientation, and the presence of companions. We use the learned mixture model to generate synthetic images of galaxies based on the morphological profiles of the Gaussian components. AstroVaDEr succeeds in producing a morphological classification scheme from unlabelled data, but unexpectedly places high importance on the presence of companion objects – demonstrating the importance of human interpretation. The network is scalable and flexible, allowing for larger data sets to be classified, or different kinds of imaging data. We also demonstrate the generative properties of the model, which allow for realistic synthetic images of galaxies to be sampled from the learned classification scheme. These can be used to create synthetic image catalogues or to perform image processing tasks such as deblending.

The End of Galaxy Surveys

Abstract For nearly a century, imaging and spectroscopic surveys of galaxies have given us information about the contents of the universe. We attempt to define the logical end point of such surveys by defining not the next galaxy survey but rather the final galaxy survey at near-infrared wavelengths; this would be the galaxy survey that exhausts the information content useful for addressing extant questions. Such a survey would require incredible advances in a number of technologies, and the survey details will depend on the as yet poorly constrained properties of the earliest galaxies. Using an exposure time calculator, we define nominal surveys for extracting the useful information for three science cases: dark energy cosmology, galaxy evolution, and supernovae (SN). We define scaling relations that trade off sky background, telescope aperture, and focal plane size to allow for a survey of a given depth over a given area. For optimistic assumptions, a 280 m telescope with a marginally resolved focal plane of 20 deg2 operating at L2 could potentially exhaust the cosmological information content of galaxies in a 10 yr survey. For galaxy evolution (making use of gravitational lensing to magnify the earliest galaxies) and SN, the same telescope would suffice. We discuss the technological advances needed to complete the last galaxy survey. While the final galaxy survey remains well outside of our technical reach today, we present scaling relations that show how we can progress toward the goal of exhausting the information content encoded in the shapes, positions, and colors of galaxies.

The interacting nature of dwarf galaxies hosting superluminous supernovae

Context. Type I superluminous supernovae (SLSNe I) are rare, powerful explosions whose mechanism and progenitors remain elusive. Several studies have shown a preference for SLSNe I to occur in low-metallicity, actively star-forming dwarf galaxies. Aims. We investigate whether the host galaxies of SLSNe I show increased evidence for interaction. Galaxy interaction can trigger star formation and provide favourable conditions for these exceptional explosions to take place. Methods. Based on SLSN host galaxy images obtained with the Hubble Space Telescope (HST), we narrowed down a sample of 42 images obtained in the rest-frame ultraviolet over the redshift range between 0 < z < 2. The number of host galaxy companions was measured by counting the number of objects detected within a given projected radius from the host. As a comparison, we used two different Monte Carlo-based methods to estimate the expected average number of companion objects in the same HST images, as well as a sample of 32 dwarf galaxies that have hosted long gamma-ray bursts (GRBs). Results. About 50% of SLSN I host galaxies have at least one major companion (within a flux ratio of 1:4) within 5 kpc. The average number of major companions per SLSN I host galaxy is 0.70−0.14+0.19. Our two Monte Carlo comparison methods yield a lower number of companions for random objects of similar brightness in the same image or for the SLSN host after randomly redistributing the sources in the same image. The Anderson-Darling test shows that this difference is statistically significant (p-value < 10−3) independent of the redshift range. The same is true for the projected distance distribution of the companions. The SLSN I hosts are, thus, found in areas of their images, where the object number density is greater than average. The SLSN I hosts have more companions than GRB hosts (0.44−0.13+0.25 companions per host distributed over 25% of the hosts) but the difference is not statistically significant. The difference between their separations is, however, marginally significant with SLSN companions being closer, on average, than those of GRBs. Conclusions. The dwarf galaxies hosting SLSNe I are often part of interacting systems. This suggests that SLSNe I progenitors are formed after a recent burst of star formation. Low metallicity alone cannot explain this tendency.

Kinematic signatures of nuclear discs and bar-driven secular evolution in nearby galaxies of the MUSE TIMER project

The central regions of disc galaxies hold clues to the processes that dominate their formation and evolution. To exploit this, the TIMER project has obtained high signal-to-noise and spatial resolution integral-field spectroscopy data of the inner few kpc of 21 nearby massive barred galaxies, which allows studies of the stellar kinematics in their central regions with unprecedented spatial resolution. We confirm theoretical predictions of the effects of bars on stellar kinematics and identify box/peanuts through kinematic signatures in mildly and moderately inclined galaxies, finding a lower limit to the fraction of massive barred galaxies with box/peanuts at ∼62%. Further, we provide kinematic evidence of the connection between barlenses, box/peanuts, and bars. We establish the presence of nuclear discs in 19 galaxies and show that their kinematics are characterised by near-circular orbits with low pressure support and that they are fully consistent with the bar-driven secular evolution picture for their formation. In fact, we show that these nuclear discs have, in the region where they dominate, larger rotational support than the underlying main galaxy disc. In addition, we define a kinematic radius for the nuclear discs and show that it relates to bar radius, ellipticity and strength, and bar-to-total ratio. Comparing our results with photometric studies of galaxy bulges, we find that careful, state-of-the-art galaxy image decompositions are generally able to discern nuclear discs from classical bulges if the images employed have high enough physical spatial resolution. In fact, we show that nuclear discs are typically identified in such image decompositions as photometric bulges with (near-)exponential profiles. However, we find that the presence of composite bulges (galaxies hosting both a classical bulge and a nuclear disc) can often be unnoticed in studies based on photometry alone and suggest a more stringent threshold to the Sérsic index to identify galaxies with pure classical bulges.

Quantifying the structure of strong gravitational lens potentials with uncertainty-aware deep neural networks

ABSTRACTGravitational lensing is a powerful tool for constraining substructure in the mass distribution of galaxies, be it from the presence of dark matter sub-haloes or due to physical mechanisms affecting the baryons throughout galaxy evolution. Such substructure is hard to model and is either ignored by traditional, smooth modelling, approaches, or treated as well-localized massive perturbers. In this work, we propose a deep learning approach to quantify the statistical properties of such perturbations directly from images, where only the extended lensed source features within a mask are considered, without the need of any lens modelling. Our training data consist of mock lensed images assuming perturbing Gaussian Random Fields permeating the smooth overall lens potential, and, for the first time, using images of real galaxies as the lensed source. We employ a novel deep neural network that can handle arbitrary uncertainty intervals associated with the training data set labels as input, provides probability distributions as output, and adopts a composite loss function. The method succeeds not only in accurately estimating the actual parameter values, but also reduces the predicted confidence intervals by 10 per cent in an unsupervised manner, i.e. without having access to the actual ground truth values. Our results are invariant to the inherent degeneracy between mass perturbations in the lens and complex brightness profiles for the source. Hence, we can quantitatively and robustly quantify the smoothness of the mass density of thousands of lenses, including confidence intervals, and provide a consistent ranking for follow-up science.

Using Unreal Engine to Visualize a Cosmological Volume

In this work we present “Astera’’, a cosmological visualization tool that renders a mock universe in real time using Unreal Engine 4. The large scale structure of the cosmic web is hard to visualize in two dimensions, and a 3D real time projection of this distribution allows for an unprecedented view of the large scale universe, with visually accurate galaxies placed in a dynamic 3D world. The underlying data are based on empirical relations assigned using results from N-Body dark matter simulations, and are matched to galaxies with similar morphologies and sizes, images of which are extracted from the Sloan Digital Sky Survey. Within Unreal Engine 4, galaxy images are transformed into textures and dynamic materials (with appropriate transparency) that are applied to static mesh objects with appropriate sizes and locations. To ensure excellent performance, these static meshes are “instanced’’ to utilize the full capabilities of a graphics processing unit. Additional components include a dynamic system for representing accelerated-time active galactic nuclei. The end result is a visually realistic large scale universe that can be explored by a user in real time, with accurate large scale structure. Astera is not yet ready for public release, but we are exploring options to make different versions of the code available for both research and outreach applications.

Mitigating Shear-dependent Object Detection Biases with Metacalibration

Abstract Metacalibration is a new technique for measuring weak gravitational lensing shear that is unbiased for isolated galaxy images. In this work we test metacalibration with overlapping, or “blended” galaxy images. Using standard metacalibration, we find a few percent shear measurement bias for galaxy densities relevant for current surveys, and that this bias increases with increasing galaxy number density. We show that this bias is not due to blending itself, but rather to shear-dependent object detection. If object detection is shear-independent, no deblending of images is needed, in principle. We demonstrate that detection biases are accurately removed when including object detection in the metacalibration process, a technique we call metadetection. This process involves applying an artificial shear to images of small regions of sky and performing detection on the sheared images, as well as measurements that are used to calculate a shear response. We demonstrate that the method can accurately recover weak shear signals even in highly blended scenes. In the metacalibration process, the space between objects is sheared coherently, which does not perfectly match the real universe in which some, but not all, galaxy images are sheared coherently. We find that even for the worst case scenario, in which the space between objects is completely unsheared, the resulting shear bias is at most a few tenths of a percent for future surveys. We discuss additional technical challenges that must be met in order to implement metadetection for real surveys.

Morphological classification of galaxies using Conv-nets

Since the beginning of space exploration, the galaxy classification has been a vexing problem that has always muddled the astrophysicists. A number of techniques have proven their remarkable utility in the classification of galaxies, however, upon analysis, these methods revealed certain inefficiencies that cannot be overlooked. The traditional classification of galaxies in the universe contains a significant part of their history in the authority of government agencies where the classifications in the previous years were performed primarily by experts manually. Today’s astronomical research produces large amounts of data and manually labelling the galaxy images based on morphological features can be time-consuming and error-prone. The objective of this paper is to study and analyze the different types of machine learning methodologies used for classifying galaxies. An inference drawn from this study is that using deep learning algorithms in conjunction with some data augmentation techniques provide excellent classification results of galaxies. Considering the aforementioned fact, the authors have proposed a layered CNN based classification model along with certain data augmentation techniques to classify galaxies morphological. “The Galaxy Zoo” dataset has been used from Kaggle which is further handcrafted for ease of classification. The galaxies are classified into three classes: spiral, elliptical, and lenticular (somewhere in-between). It has been observed from the experimental work that the proposed model outperform than its earlier contemporaries and can be used effectively to classify the galaxies.

Stellar masses of giant clumps in CANDELS and simulated galaxies using machine learning

ABSTRACT A significant fraction of high redshift star-forming disc galaxies are known to host giant clumps, whose nature and role in galaxy evolution are yet to be understood. In this work, we first present a new method based on neural networks to detect clumps in galaxy images. We use this method to detect clumps in the rest-frame optical and UV images of a complete sample of ∼1500 star forming galaxies at 1 < z < 3 in the CANDELS survey as well as in images from the VELA zoom-in cosmological simulations. We show that observational effects have a dramatic impact on the derived clump properties leading to an overestimation of the clump mass up to a factor of 10, which highlights the importance of fair comparisons between observations and simulations and the limitations of current HST data to study the resolved structure of distant galaxies. After correcting for these effects with a mixture density network, we estimate that the clump stellar mass function follows a power law down to the completeness limit (107 solar masses) with the majority of the clumps being less massive than 109 solar masses. This is in better agreement with recent gravitational lensing based measurements. The simulations explored in this work overall reproduce the shape of the observed clump stellar mass function and clumpy fractions when confronted under the same conditions, although they tend to lie in the lower limit of the confidence intervals of the observations. This agreement suggests that most of the observed clumps are formed in situ.

Deep learning for a space-variant deconvolution in galaxy surveys

The deconvolution of large survey images with millions of galaxies requires developing a new generation of methods that can take a space-variant point spread function into account. These methods have also to be accurate and fast. We investigate how deep learning might be used to perform this task. We employed a U-net deep neural network architecture to learn parameters that were adapted for galaxy image processing in a supervised setting and studied two deconvolution strategies. The first approach is a post-processing of a mere Tikhonov deconvolution with closed-form solution, and the second approach is an iterative deconvolution framework based on the alternating direction method of multipliers (ADMM). Our numerical results based on GREAT3 simulations with realistic galaxy images and point spread functions show that our two approaches outperform standard techniques that are based on convex optimization, whether assessed in galaxy image reconstruction or shape recovery. The approach based on a Tikhonov deconvolution leads to the most accurate results, except for ellipticity errors at high signal-to-noise ratio. The ADMM approach performs slightly better in this case. Considering that the Tikhonov approach is also more computation-time efficient in processing a large number of galaxies, we recommend this approach in this scenario.

Galaxy Zoo Builder: Four-component Photometric Decomposition of Spiral Galaxies Guided by Citizen Science

Abstract Multicomponent modeling of galaxies is a valuable tool in the effort to quantitatively understand galaxy evolution, yet the use of the technique is plagued by issues of convergence, model selection, and parameter degeneracies. These issues limit its application over large samples to the simplest models, with complex models being applied only to very small samples. We attempt to resolve this dilemma of “quantity or quality” by developing a novel framework, built inside the Zooniverse citizen-science platform, to enable the crowdsourcing of model creation for Sloan Digital Sky Survey galaxies. We have applied the method, including a final algorithmic optimization step, on a test sample of 198 galaxies, and examine the robustness of this new method. We also compare it to automated fitting pipelines, demonstrating that it is possible to consistently recover accurate models that either show good agreement with, or improve on, prior work. We conclude that citizen science is a promising technique for modeling images of complex galaxies, and release our catalog of models.

CO Multi-line Imaging of Nearby Galaxies (COMING). X. Physical conditions of molecular gas and the local SFR–mass relation

Abstract We investigate the molecular gas properties of galaxies across the main sequence of star-forming (SF) galaxies in the local Universe using 12CO(J = 1–0), hereafter 12CO, and 13CO(J = 1–0), hereafter 13CO, mapping data of 147 nearby galaxies obtained in the COMING project, a legacy project of the Nobeyama Radio Observatory. In order to improve the signal-to-noise ratios of both lines, we stack all the pixels where 12CO emission is detected after aligning the line center expected from the first-moment map of 12CO. As a result, 13CO emission is successfully detected in 80 galaxies with a signal-to-noise ratio larger than three. The error-weighted mean of the integrated-intensity ratio of 12CO to 13CO lines (R1213) of the 80 galaxies is 10.9, with a standard deviation of 7.0. We find that (1) R1213 positively correlates to specific star-formation rate (sSFR) with a correlation coefficient of 0.46, and (2) both the flux ratio of IRAS 60 μm to 100 μm (f60/f100) and the inclination-corrected linewidth of 12CO stacked spectra ($\sigma _{{\rm ^{12}CO},i}$) also correlate with sSFR for galaxies with the R1213 measurement. Our results support the scenario where R1213 variation is mainly caused by changes in molecular gas properties such as temperature and turbulence. The consequent variation of the CO-to-H2 conversion factor across the SF main sequence is not large enough to completely extinguish the known correlations between sSFR and Mmol/Mstar (μmol) or star-formation efficiency (SFE) reported in previous studies, while this variation would strengthen (weaken) the sSFR–SFE (sSFR–μmol) correlation.

A new way to constrain the densities of intragroup medium in groups of galaxies with convolutional neural networks

ABSTRACT Ram pressure (RP) can influence the evolution of cold gas content and star formation rates of galaxies. One of the key parameters for the strength of RP is the density of intragroup medium (ρigm), which is difficult to estimate if the X-ray emission from it is too weak to be observed. We propose a new way to constrain ρigm through an application of convolutional neural networks (CNNs) to simulated gas density and kinematic maps galaxies under strong RP. We train CNNs using 9 × 104 2D images of galaxies under various RP conditions, then validate performance with 104 new test images. This new method can be applied to real observational data from ongoing WALLABY and SKA surveys to quickly obtain estimates of ρigm. Simulated galaxy images have 1.0 kpc resolution, which is consistent with that expected from the future WALLABY survey. The trained CNN models predict the normalized IGM density, $\hat{\rho }_{\rm igm}$ where $0.0 \le \hat{\rho }_{\rm igm, n} \lt 10.0$, accurately with root mean squared error values of 0.72, 0.83, and 0.74 for the density, kinematic, and joined 2D maps, respectively. Trained models are unable to predict the relative velocity of galaxies with respect to the IGM (vrel) precisely, and struggle to generalize for different RP conditions. We apply our CNNs to the observed H i column density map of NGC 1566 in the Dorado group to estimate its IGM density.

Deep learning for strong lensing search: tests of the convolutional neural networks and new candidates from KiDS DR3

ABSTRACT Convolutional neural networks have been successfully applied in searching for strong lensing systems, leading to discoveries of new candidates from large surveys. On the other hand, systematic investigations about their robustness are still lacking. In this paper, we first construct a neutral network, and apply it to r-band images of luminous red galaxies (LRGs) of the Kilo Degree Survey (KiDS) Data Release 3 to search for strong lensing systems. We build two sets of training samples, one fully from simulations, and the other one using the LRG stamps from KiDS observations as the foreground lens images. With the former training sample, we find 48 high probability candidates after human inspection, and among them, 27 are newly identified. Using the latter training set, about 67 per cent of the aforementioned 48 candidates are also found, and there are 11 more new strong lensing candidates identified. We then carry out tests on the robustness of the network performance with respect to the variation of PSF. With the testing samples constructed using PSF in the range of 0.4–2 times of the median PSF of the training sample, we find that our network performs rather stable, and the degradation is small. We also investigate how the volume of the training set can affect our network performance by varying it from 0.1 to 0.8 million. The output results are rather stable showing that within the considered range, our network performance is not very sensitive to the volume size.

Lens modelling of the strongly lensed Type Ia supernova iPTF16geu

ABSTRACT In 2016, the first strongly lensed Type Ia supernova (SN Ia), iPTF16geu, at redshift z = 0.409 with four resolved images arranged symmetrically around the lens galaxy at z = 0.2163, was discovered. Here, refined observations of iPTF16geu, including the time delay between images, are used to decrease uncertainties in the lens model, including the the slope of the projected surface density of the lens galaxy, Σ ∝ r1 − η, and to constrain the universal expansion rate H0. Imaging with Hubble Space Telescope provides an upper limit on the slope η, in slight tension with the steeper density profiles indicated by imaging with Keck after iPTF16geu had faded, potentially due to dust extinction not corrected for in host galaxy imaging. Since smaller η implies larger magnifications, we take advantage of the standard candle nature of SNe Ia constraining the image magnifications, to obtain an independent constraint of the slope. We find that a smooth lens density fails to explain the iPTF16geu fluxes, regardless of the slope, and additional substructure lensing is needed. The total probability for the smooth halo model combined with star microlensing to explain the iPTF16geu image fluxes is maximized at 12 per cent for η ∼ 1.8, in excellent agreement with Keck high-spatial-resolution data, and flatter than an isothermal halo. It also agrees perfectly with independent constraints on the slope from lens velocity dispersion measurements. Combining with the observed time delays between the images, we infer a lower bound on the Hubble constant, $H_0 \gtrsim 40\, {\rm km \ s^{-1} Mpc^{-1}}$, at 68.3 per cent confidence level.

Galaxy Morphology Classification Based On An Improved Deep Convolutional Neural Network

With each passing day telescopes around and above the Earth capture more and more images of distant galaxies. As better and bigger telescopes continue to collect these images, the datasets begin to explode in size. In order to better understand how the different shapes (or morphologies) of galaxies relate to the physics that create them, such images need to be sorted and classified. Combining with Kaggle public datasets called Galaxy Zoo, this paper devised a deep convolutional neural network for galaxy morphology image classification. The network contains eight convolutional layers, five max_pooling layers, one flatten layer, a fully connected layer contains 150 neurons, as well as the final layer which outputs galaxy category of probability distribution of each image. The experiment results show that our neural network on the validation set of cosine similarity reached to -0.8652 (the closer to -1 means that the closer the predicted output and the expected output), implying that our model for galaxy morphology classification is very effective.