-
At First Sight: Zero-Shot Classification of Astronomical Images with Large Multimodal Models
Authors:
Dimitrios Tanoglidis,
Bhuvnesh Jain
Abstract:
Vision-Language multimodal Models (VLMs) offer the possibility for zero-shot classification in astronomy: i.e. classification via natural language prompts, with no training. We investigate two models, GPT-4o and LLaVA-NeXT, for zero-shot classification of low-surface brightness galaxies and artifacts, as well as morphological classification of galaxies. We show that with natural language prompts t…
▽ More
Vision-Language multimodal Models (VLMs) offer the possibility for zero-shot classification in astronomy: i.e. classification via natural language prompts, with no training. We investigate two models, GPT-4o and LLaVA-NeXT, for zero-shot classification of low-surface brightness galaxies and artifacts, as well as morphological classification of galaxies. We show that with natural language prompts these models achieved significant accuracy (above 80 percent typically) without additional training/fine tuning. We discuss areas that require improvement, especially for LLaVA-NeXT, which is an open source model. Our findings aim to motivate the astronomical community to consider VLMs as a powerful tool for both research and pedagogy, with the prospect that future custom-built or fine-tuned models could perform better.
△ Less
Submitted 24 June, 2024;
originally announced June 2024.
-
Environmental Quenching of Low Surface Brightness Galaxies near Milky Way mass Hosts
Authors:
J. Bhattacharyya,
A. H. G. Peter,
P. Martini,
B. Mutlu-Pakdil,
A. Drlica-Wagner,
A. B. Pace,
L. E. Strigari,
Y. -T. Cheng,
D. Roberts,
D. Tanoglidis,
M. Aguena,
O. Alves,
F. Andrade-Oliveira,
D. Bacon,
D. Brooks,
A. Carnero Rosell,
J. Carretero,
L. N. da Costa,
M. E. S. Pereira,
T. M. Davis,
S. Desai,
P. Doel,
I. Ferrero,
J. Frieman,
J. García-Bellido
, et al. (26 additional authors not shown)
Abstract:
Low Surface Brightness Galaxies (LSBGs) are excellent probes of quenching and other environmental processes near massive galaxies. We study an extensive sample of LSBGs near massive hosts in the local universe that are distributed across a diverse range of environments. The LSBGs with surface-brightness $μ_{\rm eff,g}> $24.2 mag arcsec$^{-2}$ are drawn from the Dark Energy Survey Year 3 catalog wh…
▽ More
Low Surface Brightness Galaxies (LSBGs) are excellent probes of quenching and other environmental processes near massive galaxies. We study an extensive sample of LSBGs near massive hosts in the local universe that are distributed across a diverse range of environments. The LSBGs with surface-brightness $μ_{\rm eff,g}> $24.2 mag arcsec$^{-2}$ are drawn from the Dark Energy Survey Year 3 catalog while the hosts with masses $9.0< log(M_{\star}/M_{\odot})< 11.0$ comparable to the Milky Way and the Large Magellanic Cloud are selected from the z0MGS sample. We study the projected radial density profiles of LSBGs as a function of their color and surface brightness around hosts in both the rich Fornax-Eridanus cluster environment and the low-density field. We detect an overdensity with respect to the background density, out to 2.5 times the virial radius for both hosts in the cluster environment and the isolated field galaxies. When the LSBG sample is split by $g-i$ color or surface brightness $μ_{\rm eff,g}$, we find the LSBGs closer to their hosts are significantly redder and brighter, like their high surface-brightness counterparts. The LSBGs form a clear 'red sequence' in both the cluster and isolated environments that is visible beyond the virial radius of the hosts. This suggests a pre-processing of infalling LSBGs and a quenched backsplash population around both host samples. However, the relative prominence of the 'blue cloud' feature implies that pre-processing is ongoing near the isolated hosts compared to the cluster hosts.
△ Less
Submitted 1 December, 2023;
originally announced December 2023.
-
Transformers for scientific data: a pedagogical review for astronomers
Authors:
Dimitrios Tanoglidis,
Bhuvnesh Jain,
Helen Qu
Abstract:
The deep learning architecture associated with ChatGPT and related generative AI products is known as transformers. Initially applied to Natural Language Processing, transformers and the self-attention mechanism they exploit have gained widespread interest across the natural sciences. The goal of this pedagogical and informal review is to introduce transformers to scientists. The review includes t…
▽ More
The deep learning architecture associated with ChatGPT and related generative AI products is known as transformers. Initially applied to Natural Language Processing, transformers and the self-attention mechanism they exploit have gained widespread interest across the natural sciences. The goal of this pedagogical and informal review is to introduce transformers to scientists. The review includes the mathematics underlying the attention mechanism, a description of the original transformer architecture, and a section on applications to time series and imaging data in astronomy. We include a Frequently Asked Questions section for readers who are curious about generative AI or interested in getting started with transformers for their research problem.
△ Less
Submitted 18 October, 2023; v1 submitted 18 October, 2023;
originally announced October 2023.
-
Strong Lensing Parameter Estimation on Ground-Based Imaging Data Using Simulation-Based Inference
Authors:
Jason Poh,
Ashwin Samudre,
Aleksandra Ćiprijanović,
Brian Nord,
Gourav Khullar,
Dimitrios Tanoglidis,
Joshua A. Frieman
Abstract:
Current ground-based cosmological surveys, such as the Dark Energy Survey (DES), are predicted to discover thousands of galaxy-scale strong lenses, while future surveys, such as the Vera Rubin Observatory Legacy Survey of Space and Time (LSST) will increase that number by 1-2 orders of magnitude. The large number of strong lenses discoverable in future surveys will make strong lensing a highly com…
▽ More
Current ground-based cosmological surveys, such as the Dark Energy Survey (DES), are predicted to discover thousands of galaxy-scale strong lenses, while future surveys, such as the Vera Rubin Observatory Legacy Survey of Space and Time (LSST) will increase that number by 1-2 orders of magnitude. The large number of strong lenses discoverable in future surveys will make strong lensing a highly competitive and complementary cosmic probe.
To leverage the increased statistical power of the lenses that will be discovered through upcoming surveys, automated lens analysis techniques are necessary. We present two Simulation-Based Inference (SBI) approaches for lens parameter estimation of galaxy-galaxy lenses. We demonstrate the successful application of Neural Posterior Estimation (NPE) to automate the inference of a 12-parameter lens mass model for DES-like ground-based imaging data. We compare our NPE constraints to a Bayesian Neural Network (BNN) and find that it outperforms the BNN, producing posterior distributions that are for the most part both more accurate and more precise; in particular, several source-light model parameters are systematically biased in the BNN implementation.
△ Less
Submitted 10 November, 2022;
originally announced November 2022.
-
Inferring Structural Parameters of Low-Surface-Brightness-Galaxies with Uncertainty Quantification using Bayesian Neural Networks
Authors:
Dimitrios Tanoglidis,
Aleksandra Ćiprijanović,
Alex Drlica-Wagner
Abstract:
Measuring the structural parameters (size, total brightness, light concentration, etc.) of galaxies is a significant first step towards a quantitative description of different galaxy populations. In this work, we demonstrate that a Bayesian Neural Network (BNN) can be used for the inference, with uncertainty quantification, of such morphological parameters from simulated low-surface-brightness gal…
▽ More
Measuring the structural parameters (size, total brightness, light concentration, etc.) of galaxies is a significant first step towards a quantitative description of different galaxy populations. In this work, we demonstrate that a Bayesian Neural Network (BNN) can be used for the inference, with uncertainty quantification, of such morphological parameters from simulated low-surface-brightness galaxy images. Compared to traditional profile-fitting methods, we show that the uncertainties obtained using BNNs are comparable in magnitude, well-calibrated, and the point estimates of the parameters are closer to the true values. Our method is also significantly faster, which is very important with the advent of the era of large galaxy surveys and big data in astrophysics.
△ Less
Submitted 7 July, 2022;
originally announced July 2022.
-
DeepGhostBusters: Using Mask R-CNN to Detect and Mask Ghosting and Scattered-Light Artifacts from Optical Survey Images
Authors:
Dimitrios Tanoglidis,
Aleksandra Ćiprijanović,
Alex Drlica-Wagner,
Brian Nord,
Michael H. L. S. Wang,
Ariel Jacob Amsellem,
Kathryn Downey,
Sydney Jenkins,
Diana Kafkes,
Zhuoqi Zhang
Abstract:
Wide-field astronomical surveys are often affected by the presence of undesirable reflections (often known as "ghosting artifacts" or "ghosts") and scattered-light artifacts. The identification and mitigation of these artifacts is important for rigorous astronomical analyses of faint and low-surface-brightness systems. However, the identification of ghosts and scattered-light artifacts is challeng…
▽ More
Wide-field astronomical surveys are often affected by the presence of undesirable reflections (often known as "ghosting artifacts" or "ghosts") and scattered-light artifacts. The identification and mitigation of these artifacts is important for rigorous astronomical analyses of faint and low-surface-brightness systems. However, the identification of ghosts and scattered-light artifacts is challenging due to a) the complex morphology of these features and b) the large data volume of current and near-future surveys. In this work, we use images from the Dark Energy Survey (DES) to train, validate, and test a deep neural network (Mask R-CNN) to detect and localize ghosts and scattered-light artifacts. We find that the ability of the Mask R-CNN model to identify affected regions is superior to that of conventional algorithms and traditional convolutional neural networks methods. We propose that a multi-step pipeline combining Mask R-CNN segmentation with a classical CNN classifier provides a powerful technique for the automated detection of ghosting and scattered-light artifacts in current and near-future surveys.
△ Less
Submitted 16 September, 2021;
originally announced September 2021.
-
The Intrinsic Shapes of Low Surface Brightness Galaxies: a Discriminant of LSBG Formation Mechanisms
Authors:
Erin Kado-Fong,
Mihai Petrescu,
Majid Mohammad,
Johnny Greco,
Jenny E. Greene,
Elizabeth A. K. Adams,
Song Huang,
Lukas Leisman,
Ferah Munshi,
Dimitrios Tanoglidis,
Jordan Van Nest
Abstract:
We use the low surface brightness galaxy (LSBG) samples created from the Hyper Suprime-Cam Subaru Strategic Program (HSC-SSP, 781 galaxies), the Dark Energy Survey (DES, 20977 galaxies), and the Legacy Survey (selected via HI detection in the Arecibo Legacy Fast ALFA Survey, 188 galaxies) to infer the intrinsic shape distribution of the low surface brightness galaxy population. To take into accoun…
▽ More
We use the low surface brightness galaxy (LSBG) samples created from the Hyper Suprime-Cam Subaru Strategic Program (HSC-SSP, 781 galaxies), the Dark Energy Survey (DES, 20977 galaxies), and the Legacy Survey (selected via HI detection in the Arecibo Legacy Fast ALFA Survey, 188 galaxies) to infer the intrinsic shape distribution of the low surface brightness galaxy population. To take into account the effect of the surface brightness cuts employed when constructing LSBG samples, we simultaneously model both the projected ellipticity and the apparent surface brightness in our shape inference. We find that the LSBG samples are well-characterized by oblate spheroids, with no significant difference between red and blue LSBGs. This inferred shape distribution is in good agreement with similar inferences made for ultra-diffuse cluster galaxy samples, indicating that environment does not play a key role in determining the intrinsic shape of low surface brightness galaxies. We also find some evidence that LSBGs are more thickened than similarly massive high surface brightness dwarfs. We compare our results to intrinsic shape measures from contemporary cosmological simulations, and find that the observed LSBG intrinsic shapes place considerable constraints on the formation path of such galaxies. In particular, LSBG production via the migration of star formation to large radii produces intrinsic shapes in good agreement with our observational findings.
△ Less
Submitted 28 July, 2021; v1 submitted 9 June, 2021;
originally announced June 2021.
-
DeepShadows: Separating Low Surface Brightness Galaxies from Artifacts using Deep Learning
Authors:
Dimitrios Tanoglidis,
Aleksandra Ćiprijanović,
Alex Drlica-Wagner
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…
▽ More
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.
△ Less
Submitted 24 November, 2020;
originally announced November 2020.
-
Shadows in the Dark: Low-Surface-Brightness Galaxies Discovered in the Dark Energy Survey
Authors:
D. Tanoglidis,
A. Drlica-Wagner,
K. Wei,
T. S. Li,
F. J. Sánchez,
Y. Zhang,
A. H. G. Peter,
A. Feldmeier-Krause,
J. Prat,
K. Casey,
A. Palmese,
C. Sánchez,
J. DeRose,
C. Conselice,
L. Gagnon,
T. M. C. Abbott,
M. Aguena,
S. Allam,
S. Avila,
K. Bechtol,
E. Bertin,
S. Bhargava,
D. Brooks,
D. L. Burke,
A. Carnero Rosell
, et al. (45 additional authors not shown)
Abstract:
We present a catalog of 23,790 extended low-surface-brightness galaxies (LSBGs) identified in $\sim 5000 °^2$ from the first three years of imaging data from the Dark Energy Survey (DES). Based on a single-component Sérsic model fit, we define extended LSBGs as galaxies with $g$-band effective radii $R_{eff}(g) > 2.5''$ and mean surface brightness $\barμ_{eff}(g) > 24.2 \,mag ȧrcsec^{-2}$. We find…
▽ More
We present a catalog of 23,790 extended low-surface-brightness galaxies (LSBGs) identified in $\sim 5000 °^2$ from the first three years of imaging data from the Dark Energy Survey (DES). Based on a single-component Sérsic model fit, we define extended LSBGs as galaxies with $g$-band effective radii $R_{eff}(g) > 2.5''$ and mean surface brightness $\barμ_{eff}(g) > 24.2 \,mag ȧrcsec^{-2}$. 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 $μ_0(g) > 24.0\, mag \,arcsec^{-2}$. The wide-area sample of LSBGs in DES can be used to test the role of environment on models of LSBG formation and evolution.
△ Less
Submitted 16 December, 2020; v1 submitted 7 June, 2020;
originally announced June 2020.
-
Turnaround density as a probe of the cosmological constant
Authors:
Vasiliki Pavlidou,
Giorgos Korkidis,
Theodore Tomaras,
Dimitrios Tanoglidis
Abstract:
Spherical collapse predicts that a single value of the turnaround density (average matter density within the scale on which a structure detaches from the Hubble flow) characterizes all cosmic structures at the same redshift. It has been recently shown by Korkidis et al. that this feature persists in complex non-spherical galaxy clusters identified in N-body simulations. Here we show that the low-r…
▽ More
Spherical collapse predicts that a single value of the turnaround density (average matter density within the scale on which a structure detaches from the Hubble flow) characterizes all cosmic structures at the same redshift. It has been recently shown by Korkidis et al. that this feature persists in complex non-spherical galaxy clusters identified in N-body simulations. Here we show that the low-redshift evolution of the turnaround density constrains the cosmological parameters, and that it can be used to derive a local constraint on $Ω_Λ$ alone, independent of $Ω_m$. The turnaround density thus provides a promising new way to exploit upcoming large cosmological datasets.
△ Less
Submitted 9 April, 2020;
originally announced April 2020.
-
Optimizing galaxy samples for clustering measurements in photometric surveys
Authors:
Dimitrios Tanoglidis,
Chihway Chang,
Joshua Frieman
Abstract:
When analyzing galaxy clustering in multi-band imaging surveys, there is a trade-off between selecting the largest galaxy samples (to minimize the shot noise) and selecting samples with the best photometric redshift (photo-z) precision, which generally include only a small subset of galaxies. In this paper, we systematically explore this trade-off. Our analysis is targeted towards the third year d…
▽ More
When analyzing galaxy clustering in multi-band imaging surveys, there is a trade-off between selecting the largest galaxy samples (to minimize the shot noise) and selecting samples with the best photometric redshift (photo-z) precision, which generally include only a small subset of galaxies. In this paper, we systematically explore this trade-off. Our analysis is targeted towards the third year data of the Dark Energy Survey (DES), but our methods hold generally for other data sets. Using a simple Gaussian model for the redshift uncertainties, we carry out a Fisher matrix forecast for cosmological constraints from angular clustering in the redshift range $z = 0.2-0.95$. We quantify the cosmological constraints using a Figure of Merit (FoM) that measures the combined constraints on $Ω_m$ and $σ_8$ in the context of $Λ$CDM cosmology. We find that the trade-off between sample size and photo-z precision is sensitive to 1) whether cross-correlations between redshift bins are included or not, and 2) the ratio of the redshift bin width $δz$ and the photo-z precision $σ_z$. When cross-correlations are included and the redshift bin width is allowed to vary, the highest FoM is achieved when $δz \sim σ_z$. We find that for the typical case of $5-10$ redshift bins, optimal results are reached when we use larger, less precise photo-z samples, provided that we include cross-correlations. For samples with higher $σ_{z}$, the overlap between redshift bins is larger, leading to higher cross-correlation amplitudes. This leads to the self-calibration of the photo-z parameters and therefore tighter cosmological constraints. These results can be used to help guide galaxy sample selection for clustering analysis in ongoing and future photometric surveys.
△ Less
Submitted 3 January, 2020; v1 submitted 19 August, 2019;
originally announced August 2019.
-
Resolving Dark Matter Subhalos With Future Sub-GeV Gamma-Ray Telescopes
Authors:
Ti-Lin Chou,
Dimitrios Tanoglidis,
Dan Hooper
Abstract:
Annihilating dark matter particles in nearby subhalos could generate potentially observable fluxes of gamma rays, unaccompanied by emission at other wavelengths. Furthermore, this gamma-ray emission is expected to be spatially extended, providing us with a powerful way to discriminate dark matter subhalos from other astrophysical gamma-ray sources. Fermi has detected two dark matter subhalo candid…
▽ More
Annihilating dark matter particles in nearby subhalos could generate potentially observable fluxes of gamma rays, unaccompanied by emission at other wavelengths. Furthermore, this gamma-ray emission is expected to be spatially extended, providing us with a powerful way to discriminate dark matter subhalos from other astrophysical gamma-ray sources. Fermi has detected two dark matter subhalo candidates which exhibit a statistically significant degree of spatial extension (3FGL J2212.5+0703 and 3FGL J1924.8-1034). It has been argued that the most likely non-dark matter interpretation of these observations is that they are each in fact multiple nearby point sources, too close to one another on the sky to be individually resolved. In this study, we consider the ability of next generation gamma-ray telescopes to spatially resolve the gamma-ray emission from subhalo candidates, focusing on the proposed e-ASTROGAM mission. We find that such an instrument could significantly clarify the nature of Fermi's dark matter subhalo candidates, and provide an unprecedented level of sensitivity to the presence of annihilating dark matter in nearby subhalos.
△ Less
Submitted 25 September, 2017;
originally announced September 2017.
-
Turnaround overdensity as a cosmological observable: the case for a local measurement of $Λ$
Authors:
Dimitrios Tanoglidis,
Vasiliki Pavlidou,
Theodore Tomaras
Abstract:
We demonstrate that, in the context of the $Λ$CDM model, it is in principle possible to measure the value of the cosmological constant by tracing, across cosmic time, the evolution of the turnaround radius of cosmic structures. The novelty of the presented method is that it is local, in the sense that it uses the effect of the cosmological constant on the relatively short scales of cosmic structur…
▽ More
We demonstrate that, in the context of the $Λ$CDM model, it is in principle possible to measure the value of the cosmological constant by tracing, across cosmic time, the evolution of the turnaround radius of cosmic structures. The novelty of the presented method is that it is local, in the sense that it uses the effect of the cosmological constant on the relatively short scales of cosmic structures and not on the dynamics of the Universe at its largest scales. In this way, it can provide an important consistency check for the standard cosmological model and can give signs of new physics, beyond $Λ$CDM.
△ Less
Submitted 14 January, 2016;
originally announced January 2016.
-
Testing Lambda CDM cosmology at turnaround: where to look for violations of the bound?
Authors:
Dimitrios Tanoglidis,
Vasiliki Pavlidou,
Theodore Tomaras
Abstract:
In $Λ$CDM cosmology, structure formation is halted shortly after dark energy dominates the mass/energy budget of the Universe. A manifestation of this effect is that in such a cosmology the turnaround radius has an upper bound. Recently, a new, local, test for the existence of dark energy in the form of a cosmological constant was proposed based on this turnaround bound. Before designing an experi…
▽ More
In $Λ$CDM cosmology, structure formation is halted shortly after dark energy dominates the mass/energy budget of the Universe. A manifestation of this effect is that in such a cosmology the turnaround radius has an upper bound. Recently, a new, local, test for the existence of dark energy in the form of a cosmological constant was proposed based on this turnaround bound. Before designing an experiment that, through high-precision determination of masses and turnaround radii, will challenge $Λ$CDM cosmology, we have to answer two important questions: first, when turnaround-scale structures are predicted to be close enough to their maximum size, so that a possible violation of the bound may be observable. Second, which is the best mass scale to target for possible violations of the bound. Using the Press-Schechter formalism, we find that turnaround structures have in practice already stopped forming, and consequently, the turnaround radius of structures must be very close to the maximum value today. We also find that the mass scale of $\sim 10^{13} M_\odot$ characterizes turnaround structures that start to form in a statistically important number density today. This mass scale also separates turnaround structures with different cosmological evolution: smaller structures are no longer readjusting their mass distribution inside the turnaround scale, they asymptotically approach their ultimate abundance from higher values, and they are common enough to have, at some epoch, experienced major mergers with structures of comparable mass; larger structures exhibit the opposite behavior. We call this mass scale the transitional mass scale and we argue that it is the optimal for the purpose outlined above. As a corollary result, we explain the different accretion behavior of small and larger structures observed in already conducted numerical simulations.
△ Less
Submitted 26 November, 2015; v1 submitted 20 December, 2014;
originally announced December 2014.
