Improving Photometric Redshift Estimates with Training Sample Augmentation

The Vera C. Rubin Observatory will advance many areas of astronomy over the next decade with its unique widefast- deep multi-color imaging survey, the Legacy Survey of Space and Time (LSST).1 The LSST will produce approximately 20TB of raw data per night, which will be automatically processed by the LSST Science Pipelines to generate science-ready data products – processed images, catalogs and alerts. To ensure that these data products enable transformative science with LSST, stringent requirements have been placed on their quality and scientific fidelity, for example on image quality and depth, astrometric and photometric performance, and object recovery completeness. In this paper we introduce faro, a framework for automatically and efficiently computing scientific performance metrics on the LSST data products for units of data of varying granularity, ranging from single-detector to full-survey summary statistics. By measuring and monitoring metrics, we are able to evaluate trends in algorithmic performance and conduct regression testing during development, compare the performance of one algorithm against another, and verify that the LSST data products will meet performance requirements by comparing to specifications. We present initial results using faro to characterize the performance of the data products produced on simulated and precursor data sets, and discuss plans to use faro to verify the performance of the LSST commissioning data products.

We predict the sensitivity of the Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST) to faint, resolved Milky Way satellite galaxies and outer-halo star clusters. We characterize the expected sensitivity using simulated LSST data from the LSST Dark Energy Science Collaboration (DESC) Data Challenge 2 (DC2) accessed and analyzed with the Rubin Science Platform as part of the Rubin Early Science Program. We simulate resolved stellar populations of Milky Way satellite galaxies and outer-halo star clusters over a wide range of sizes, luminosities, and heliocentric distances, which are broadly consistent with expectations for the Milky Way satellite system. We inject simulated stars into the DC2 catalog with realistic photometric uncertainties and star/galaxy separation derived from the DC2 data itself. We assess the probability that each simulated system would be detected by LSST using a conventional isochrone matched-filter technique. We find that assuming perfect star/galaxy separation enables the detection of resolved stellar systems with MV = 0 mag and r1/2 = 10 pc with >50% efficiency out to a heliocentric distance of ~250 kpc. Similar detection efficiency is possible with a simple star/galaxy separation criterion based on measured quantities, although the false positive rate is higher due to leakage of background galaxies into the stellar sample. When assuming perfect star/galaxy classification and a model for the galaxy-halo connection fit to current data, we predict that 89 +/- 20 Milky Way satellite galaxies will be detectable with a simple matched-filter algorithm applied to the LSST wide-fast-deep data set. Different assumptions about the performance of star/galaxy classification efficiency can decrease this estimate by ~7-25%, which emphasizes the importance of high-quality star/galaxy separation for studies of the Milky Way satellite population with LSST.

Context. The Legacy Survey of Space and Time (LSST), to be conducted with the Vera C. Rubin Observatory, is poised to revolutionize our understanding of the Solar System by providing an unprecedented wealth of data on various objects, including the elusive interstellar objects (ISOs). Detecting and classifying ISOs is crucial for studying the composition and diversity of materials from other planetary systems. However, the rarity and brief observation windows of ISOs, coupled with the vast quantities of data to be generated by LSST, create significant challenges for their identification and classification. Aims. This study aims to address these challenges by exploring the application of machine learning algorithms to the automated classification of ISO tracklets in simulated LSST data. Methods. We employed various machine learning algorithms, including random forests (RFs), stochastic gradient descent (SGD), gradient boosting machines (GBMs), and neural networks (NNs), to classify ISO tracklets in simulated LSST data. Results. Our results demonstrate that GBM and RF algorithms outperform SGD and NN algorithms in accurately distinguishing ISOs from other Solar System objects. RF analysis shows that many derived Digest2 values are more important than direct observables (right ascension, declination, and magnitude) in classifying ISOs from the LSST tracklets. The GBM model achieves the highest precision, recall, and F1 score, with values of 0.9987, 0.9986, and 0.9987, respectively. Conclusions. These findings lay the foundation for the development of an efficient and robust automated system for ISO discovery using LSST data, paving the way for a deeper understanding of the materials and processes that shape planetary systems beyond our own. The integration of our proposed machine learning approach into the LSST data processing pipeline will optimize the survey’s potential for identifying these rare and valuable objects, enabling timely follow-up observations and further characterization.

Context. The mounting data stream of large time-domain surveys renders the visual inspections of a huge set of transient candidates impractical. Techniques based on deep learning-based are popular solutions for minimizing human intervention in the time domain community. The classification of real and bogus transients is a fundamental component in real-time data processing systems and is critical to enabling rapid follow-up observations. Most existing methods (supervised learning) require sufficiently large training samples with corresponding labels, which involve costly human labeling and are challenging in the early stages of a time-domain survey. One method that can make use of training samples with access to only a limited amount of labels is highly desirable for future large time-domain surveys. These include the forthcoming 2.5-meter Wide-Field Survey Telescope (WFST) six-year survey and the Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST). Aims. Deep-learning-based methods have been favored in astrophysics owing to their adaptability and remarkable performance. They have been applied to the task of the classification of real and bogus transients. Unlike most existing approaches, which necessitate massive and expensive annotated data, we aim to leverage training samples with only 1000 labels and discover real sources that vary in brightness over time in the early stages of the WFST six-year survey. Methods. We present a novel deep learning method that combines active learning and semi-supervised learning to construct a competitive real-bogus classifier. Our method incorporates an active learning stage, where we actively select the most informative or uncertain samples for annotation. This stage aims to achieve higher model performance by leveraging fewer labeled samples, thus reducing annotation costs and improving the overall learning process efficiency. Furthermore, our approach involves a semi-supervised learning stage that exploits the unlabeled data to enhance the model’s performance and achieve superior results, compared to using only the limited labeled data. Results. Our proposed methodology capitalizes on the potential of active learning and semi-supervised learning. To demonstrate the efficacy of our approach, we constructed three newly compiled datasets from the Zwicky Transient Facility, achieving average accuracies of 98.8, 98.8, and 98.6% across these three datasets. It is important to note that our newly compiled datasets only work in terms of testing our deep learning methodology and there may be a potential bias between our datasets and the complete data stream. Therefore, the observed performance on these datasets cannot be assumed to directly translate to the general alert stream for general transient detection in actual scenarios. The algorithm will be integrated into the WFST pipeline, enabling an efficient and effective classification of transients in the early period of a time-domain survey.

Radio surveys typically sample extragalactic sources in higher redshift regimes than is typical for optical surveys, resulting in many radio sources not having a detected optical counterpart. Over the next decade the Legacy Survey of Space and Time (LSST) will be performing the deepest (i < 26.4 mag) wide-area optical survey to date increasing the fraction of radio sources for which we have optical data. In this Research Note we use the Hyper Suprime-Cam survey to analyze how the fraction of radio sources in the Very Large Array Sky Survey (VLASS) with optical detections varies as a function of i-band magnitude and extrapolate to predict the number of optical counterparts we expect LSST to detect. Assuming a final VLASS point source depth of S 3 GHz ≲ 350 μJy, we expect LSST to identify optical counterparts to ∼106 radio sources in VLASS.

Accurate photometric redshift (photo-z) estimation requires support from multiband observational data. However, in the actual process of astronomical observations and data processing, some sources may have missing observational data in certain bands for various reasons. This could greatly affect the accuracy and reliability of photo-z estimation for these sources, and even render some estimation methods unusable. The same situation may exist for the upcoming Chinese Space Station Telescope (CSST). In this study, we employ a deep learning method called generative adversarial imputation networks (GAIN) to impute the missing photometric data in CSST, aiming to reduce the impact of data missing on photo-z estimation and improve estimation accuracy. Our results demonstrate that using the GAIN technique can effectively fill in the missing photometric data in CSST. Particularly, when the data missing rate is below 30 per cent, the imputation of photometric data exhibits high accuracy, with higher accuracy in the g, r, i, z, and y bands compared to the NUV and u bands. After filling in the missing values, the quality of photo-z estimation obtained by the widely used easy and accurate Zphot from Yale (eazy) software is notably enhanced. Evaluation metrics for assessing the quality of photo-z estimation, including the catastrophic outlier fraction (fout), the normalized median absolute deviation ($\rm {\sigma _{NMAD}}$), and the bias of photometric redshift (bias), all show some degree of improvement. Our research will help maximize the utilization of observational data and provide a new method for handling sample missing values for applications that require complete photometry data to produce results.

Aims. We aim to use multiband imaging from the Phase-3 Verification Data of the J-PLUS survey to derive accurate photometric redshifts (photo-z) and look for potential new members in the surroundings of the nearby galaxy clusters A2589 (z = 0.0414) &amp; A2593 (z = 0.0440), using redshift probability distribution functions (PDFs). The ultimate goal is to demonstrate the usefulness of a 12-band filter system in the study of largescale structure in the local Universe. Methods. We present an optimized pipeline for the estimation of photometric redshifts in clusters of galaxies. This pipeline includes a PSF-corrected photometry, specific photometric apertures capable of enhancing the integrated signal in the bluest filters, a careful recalibration of the photometric uncertainties and accurate upper-limit estimations for faint detections. To foresee the expected precision of our photo-z beyond the spectroscopic sample, we designed a set of simulations in which real cluster galaxies are modeled and reinjected inside the images at different signal-to-noise ratio (S/N) levels, recomputing their photometry and photo-z estimates. Results. We tested our photo-z pipeline with a sample of 296 spectroscopically confirmed cluster members with an averaged magnitude of ⟨r⟩ = 16.6 and redshift ⟨z⟩ = 0.041. The combination of seven narrow and five broadband filters with a typical photometric-depth of r ~ 21.5 provides δz/(1 + z) = 0.01 photo-z estimates. A precision of δz/(1 + z) = 0.005 is obtained for the 177 galaxies brighter than magnitude r &lt; 17. Based on simulations, a δz/(1 + z) = 0.02 and δz/(1 + z) = 0.03 is expected at magnitudes ⟨r⟩ = 18 and ⟨r⟩ = 22, respectively. Complementarily, we used SDSS/DR12 data to derive photo-z estimates for the same galaxy sample. This exercise demonstrates that the wavelength-resolution of the J-PLUS data can double the precision achieved by SDSS data for galaxies with a high S/N. Based on the Bayesian membership analysis carried out in this work, we find as much as 170 new candidates across the entire field (~5 deg2). The spatial distribution of these galaxies may suggest an overlap between the systems with no evidence of a clear filamentary structure connecting the clusters. This result is supported by X-ray Rosat All-Sky Survey observations suggesting that a hypothetical filament may have low density contrast on diffuse warm gas. Conclusions. We prove that the addition of the seven narrow-band filters make the J-PLUS data deeper in terms of photo-z-depth than other surveys of a similar photometric-depth but using only five broadbands. These preliminary results show the potential of J-PLUS data to revisit membership of groups and clusters from nearby galaxies, important for the determination of luminosity and mass functions and environmental studies at the intermediate and low-mass regime.

We present ANNz2, a new implementation of the public software for photometric redshift (photo-z) estimation of Collister & Lahav, which now includes generation of full probability distribution functions (PDFs). ANNz2 utilizes multiple machine learning methods, such as artificial neural networks and boosted decision/regression trees. The objective of the algorithm is to optimize the performance of the photo-z estimation, to properly derive the associated uncertainties, and to produce both single-value solutions and PDFs. In addition, estimators are made available, which mitigate possible problems of non-representative or incomplete spectroscopic training samples. ANNz2 has already been used as part of the first weak lensing analysis of the Dark Energy Survey, and is included in the experiment's first public data release. Here we illustrate the functionality of the code using data from the tenth data release of the Sloan Digital Sky Survey and the Baryon Oscillation Spectroscopic Survey. The code is available for download at http://github.com/IftachSadeh/ANNZ.

Photometric redshift (photo-z) estimates are playing an increasingly important role in extragalactic astronomy and cosmology. Crucial to many photo-z applications is the accurate quantification of photometric redshift errors and their distributions, including identification of likely catastrophic failures in photo-z estimates. We consider several methods of estimating photo-z errors and propose new training-set based error estimators based on spectroscopic training set data. Using data from the Sloan Digital Sky Survey and simulations of the Dark Energy Survey as examples, we show that this method provides a robust, relatively unbiased estimate of photo-z errors. We show that culling objects with large, accurately estimated photo-z errors from a sample can reduce the incidence of catastrophic photo-z failures.

We present a deep learning model based on neural posterior estimation (NPE) for comprehensive extraction of astrophysical parameters from galaxy-scale strong gravitational lenses. The unprecedentedly large amount of galaxy-scale strong lenses expected in future cosmological surveys (${\cal O}(10^5)$) promises to enable valuable statistical constraints in various studies ranging from galaxy formation to the nature of dark matter, but it also poses a significant challenge for traditional modelling pipelines. To this end, our automated model includes several new, state-of-the-art features and approaches leveraging the framework of simulation-based inference (SBI). We infer a total of 20 parameters describing the mass and light profiles of both lens and source galaxies, using simulated raw multi-band data modelled under noise and observing conditions expected by the Legacy Survey of Space and Time (LSST), with its summary statistics generated by a residual network. We examine the efficacy of multi-band data in extracting nearly 20 model parameters simultaneous from strong lensing images including lens light. Finally, We perform a comprehensive set of diagnostics for SBI models, evaluating the model's prediction accuracy, stability, and uncertainty quantification.

We provide a code and data repository that can be used to facilitate planning for AGN science with the upcoming Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST). For this purpose, we have produced a common exploratory dataset that can be used to develop tools for parameterization of AGN light curves, AGN selection, and AGN photometric redshifts

Upcoming imaging surveys will allow for high signal-to-noise measurements of galaxy clustering at small scales. In this work, we present the results of the Rubin Observatory Legacy Survey of Space and Time (LSST) bias challenge, the goal of which is to compare the performance of different nonlinear galaxy bias models in the context of LSST Year 10 (Y10) data. Specifically, we compare two perturbative approaches, Lagrangian perturbation theory (LPT) and Eulerian perturbation theory (EPT) to two variants of Hybrid Effective Field Theory (HEFT), with our fiducial implementation of these models including terms up to second order in the bias expansion as well as nonlocal bias and deviations from Poissonian stochasticity. We consider a variety of different simulated galaxy samples and test the performance of the bias models in a tomographic joint analysis of LSST-Y10-like galaxy clustering, galaxy-galaxy-lensing and cosmic shear. We find both HEFT methods as well as LPT and EPT combined with non-perturbative predictions for the matter power spectrum to yield unbiased constraints on cosmological parameters up to at least a maximal scale of k max = 0.4 Mpc-1 for all samples considered, even in the presence of assembly bias. While we find that we can reduce the complexity of the bias model for HEFT without compromising fit accuracy, this is not generally the case for the perturbative models. We find significant detections of non-Poissonian stochasticity in all cases considered, and our analysis shows evidence that small-scale galaxy clustering predominantly improves constraints on galaxy bias rather than cosmological parameters. These results therefore suggest that the systematic uncertainties associated with current nonlinear bias models are likely to be subdominant compared to other sources of error for tomographic analyses of upcoming photometric surveys, which bodes well for future galaxy clustering analyses using these high signal-to-noise data.

We conduct a detailed analysis of the photometric redshift requirements for the proposed Dark Energy Survey (DES) using two sets of mock galaxy simulations and an artificial neural network code - ANNz. In particular, we examine how optical photometry in the DES grizY bands can be complemented with near infra-red photometry from the planned VISTA Hemisphere Survey (VHS) in the JHK_s bands. We find that the rms scatter on the photometric redshift estimate over 1<z<2 is sigma_z=0.2 from DES alone and sigma_z=0.15 from DES+VISTA, i.e. an improvement of more than 30%. We draw attention to the effects of galaxy formation scenarios such as reddening on the photo-z estimate and using our neural network code, calculate the extinction, A_v for these reddened galaxies. We also look at the impact of using different training sets when calculating photometric redshifts. In particular, we find that using the ongoing DEEP2 and VVDS-Deep spectroscopic surveys to calibrate photometric redshifts for DES, will prove effective. However we need to be aware of uncertainties in the photometric redshift bias that arise when using different training sets as these will translate into errors in the dark energy equation of state parameter, w. Furthermore, we show that the neural network error estimate on the photometric redshift may be used to remove outliers from our samples before any kind of cosmological analysis, in particular for large-scale structure experiments. By removing all galaxies with a neural network photo-z error estimate of greater than 0.1 from our DES+VHS sample, we can constrain the galaxy power spectrum out to a redshift of 2 and reduce the fractional error on this power spectrum by ~15-20% compared to using the entire catalogue.

Layer-by-layer description of the core is performed to understand the regularities of the structure of the geological section, predict the development of reservoirs, clarify stratigraphic boundaries and obtain calculation parameters for assessing hydrocarbon reserves. In this case, the name of the breed is one of the key parameters determined in the layer-by-layer description. This paper presents a comparative analysis of two approaches to determining the breed using machine learning methods: based on graphical identifiers and convolutional neural networks. The original sample contained photographs of core samples from the Tyumenskaya suite fields (8 fields, 15 wells, more than 2 km of core) under daylight. For the analysis, 4 main classes of rocks (siltstones, mudstones, sandstones, coals) were selected. For these rocks, windows of 5 × 5 cm were formed and compressed to 299 × 299 pixels. The total sample exceeded 90,000 windows: 70% — training sample (60,359 windows) and 30% — test (31,140 windows). The training and test samples contain photographs of core samples from different fields. The comparison was made between convolutional neural networks (ResNet, ResNeXt, Inception, etc.) and a classifier (such as XGBoost) based on graphic identifiers of two types: color (average color, dominant colors) and texture (entropy, Euler’s number, contrast, dissimilarity, uniformity, energy, correlation). According to the results of the experiments, the model based on convolutional neural networks turned out to be more sensitive to implicit features and made it possible to reduce the error in the weighted average f1-measure with respect to the ensemble of weak classifiers by 12.5% on the test sample even without optimization of hyperparameters. Thus, we can conclude that the model based on convolutional neural networks is more sensitive to implicit features that are difficult to extract using known graphic identifiers. On the other hand, the approach based on graphic identifiers and an ensemble of weak classifiers can be used without specialized computing power (video cards).

The aim of this paper is to investigate ways to optimize the accuracy of photometric redshifts for a Supernova Acceleration Probe (SNAP)-like mission. We focus on how the accuracy of the photometric redshifts depends on the magnitude limit and signal-to-noise ratio (S/N), wavelength coverage, and the number of filters and their shapes and observed galaxy type. We use simulated galaxy catalogs constructed to reproduce observed galaxy luminosity functions from the Great Observatories Origins Deep Survey and derive photometric redshifts using a template fitting method. By using a catalog that resembles real data, we can estimate the expected number density of galaxies for which photometric redshifts can be derived. We find that the accuracy of the photometric redshifts is strongly dependent on the S/N (i.e., S/N &gt; 10 is needed for accurate photometric redshifts). The accuracy of the photometric redshifts is also dependent on galaxy type, with smaller scatter for earlier-type galaxies. Comparing results using different filter sets, we find that including the U-band and near-IR bands is important for decreasing the fraction of outliers, i.e., "catastrophic failures." Using broad overlapping filters with resolution ~4 gives better photometric redshifts compared to narrower filters (resolution ≳ 5) with the same integration time. We find that filters with square response curves result in a slightly higher scatter, mainly due to a higher fraction of outliers at faint magnitudes. We also compare a 9-filter set to a 17-filter set, where we assume that the available exposure time per filter in the latter set is half that of the first set. We find that the 9-filter set gives more accurate redshifts for a larger number of objects and reaches higher redshift, while the 17-filter set gives better results at bright magnitudes.