arXiv subjects

Kartheik Iyer

Publications and source records attributed to Kartheik Iyer.

At least 37 records · Page 2Linked to original sources

arXiv2024

The Lyman Continuum Escape Fraction of Star-forming Galaxies at $2.4\lesssim z\lesssim3.0$ from UVCANDELS

The UltraViolet Imaging of the Cosmic Assembly Near-infrared Deep Extragalactic Legacy Survey Fields (UVCANDELS) survey is a Hubble Space Telescope (HST) Cycle-26 Treasury Program, allocated in total 164 orbits of primary Wide-Field Camera 3 Ultraviolet and Visible light F275W imaging with coordinated parallel Advanced Camera for Surveys F435W imaging, on four of the five premier extragalactic survey fields: GOODS-N, GOODS-S, EGS, and COSMOS. We introduce this survey by presenting a comprehensive analysis of the absolute escape fraction ($f_{\rm esc}^{\rm abs}$) of Lyman continuum (LyC) radiation through stacking the UV images of a population of star-forming galaxies with secure redshifts at $2.4\leq z\leq3.0$. Our stacking benefits from the catalogs of high-quality spectroscopic redshifts compiled from archival ground-based data and HST slitless spectroscopy, carefully vetted by dedicated visual inspection efforts. We develop a robust stacking method to apply to 10 samples of in total 56 galaxies, and perform detailed Monte Carlo (MC) simulations of the intergalactic medium (IGM) attenuation, to take into account the sample variance of the mean IGM transmission when measuring $f_{\rm esc}^{\rm abs}$. The full stack at $z\approx2.44$ from 28 galaxies places a stringent 1-$σ$ upper limit of $f_{\rm esc}^{\rm abs}\lesssim5\%$, whereas the full stack at $z\approx2.72$ of equal number of galaxies gives an upper limit of $f_{\rm esc}^{\rm abs}\lesssim26\%$ at 1-$σ$ confidence level. These new F275W and F435W imaging mosaics from UVCANDELS have been made publicly available on the Barbara A. Mikulski Archive for Space Telescopes (MAST).

astro-ph.GA↗

arXiv2024

Red, hot, and very metal poor: extreme properties of a massive accreting black hole in the first 500 Myr

The James Webb Space Telescope (JWST) has recently discovered a new population of objects at high redshift referred to as `Little Red Dots' (LRDs). Their nature currently remains elusive, despite their surprisingly high inferred number densities. This emerging population of red point-like sources is reshaping our view of the early Universe and may shed light on the formation of high-redshift supermassive black holes. Here we present a spectroscopically confirmed LRD CANUCS-LRD-z8.6 at $z_{\rm spec}=8.6319\pm 0.0005$ hosting an Active Galactic Nucleus (AGN), using JWST data. This source shows the typical spectral shape of an LRD (blue UV and red optical continuum, unresolved in JWST imaging), along with broad H$β$ line emission, detection of high-ionization emission lines (CIV, NIV]) and very high electron temperature indicative of the presence of AGN. This is also combined with a very low metallicity ($Z<0.1 Z_\odot$). The presence of all these diverse features in one source makes CANUCS-LRD-z8.6 unique. We show that the inferred black hole mass of CANUCS-LRD-z8.6 ($M_{\rm BH}=1.0^{+0.6}_{-0.4}\times 10^{8}\rm ~M_\odot$) strongly challenges current standard theoretical models and simulations of black hole formation, and forces us to adopt `ad hoc' prescriptions. Indeed if massive seeds, or light seeds with super-Eddington accretion, are considered, the observed BH mass of CANUCS-LRD-z8.6 at $z=8.6$ can be reproduced. Moreover, the black hole is over-massive compared to its host, relative to the local $M_{\rm BH}-M_*$ relations, pointing towards an earlier and faster evolution of the black hole compared to its host galaxy.

astro-ph.GA↗

arXiv2024

The FENIKS Survey: Stellar-Halo Mass Relationship of Central and Satellite Galaxies in UDS and COSMOS at 0.2 < z < 4.5

We present a comprehensive analysis of the observed Stellar-to-Halo mass relationship (SHMR) spanning redshifts from 0.2 to 4.5. This was enabled through galaxy clustering and abundance measurements from two large (effective area ~ 1.61 deg^2) and homogeneously prepared photometric catalogs - UltraVISTA ultra-deep stripes DR3 (COSMOS) and FENIKS v1 (UDS). To translate these measurements into the SHMR, we introduce a novel halo occupation distribution (HOD) fitting approach (``smooth-$z$'') whereby HOD parameters between neighboring z-bins are connected via physically motivated continuity (smoothing) priors. As a result, the high constraining power at z <~ 2, due to a much wider dynamical range in stellar mass (~ 3 dex), helps constrain the SHMR at z >~ 2, where that range shrinks down to <~ 1 dex. We find that the halo mass is tightly coupled to star formation: the halo mass with peak integrated star-forming efficiency (SFE), M_h^peak remains constant within ~ 10^12.2 - 10^12.4 Msolar throughout the redshifts probed. Furthermore, we show that if we had relied on COSMOS alone (as opposed to COSMOS+UDS), as has been done by many preceding studies, M_h^peak would be systematically lower by up to ~0.15 dex at z < 1.5, highlighting the importance of mitigating cosmic variance. Finally, for the first time, we show how the SFE evolves with redshift as halos grow in mass along their progenitor merger trees, instead of at fixed halo masses.

astro-ph.GA↗

arXiv2024

Disentangling Dense Embeddings with Sparse Autoencoders

Sparse autoencoders (SAEs) have shown promise in extracting interpretable features from complex neural networks. We present one of the first applications of SAEs to dense text embeddings from large language models, demonstrating their effectiveness in disentangling semantic concepts. By training SAEs on embeddings of over 420,000 scientific paper abstracts from computer science and astronomy, we show that the resulting sparse representations maintain semantic fidelity while offering interpretability. We analyse these learned features, exploring their behaviour across different model capacities and introducing a novel method for identifying ``feature families'' that represent related concepts at varying levels of abstraction. To demonstrate the practical utility of our approach, we show how these interpretable features can be used to precisely steer semantic search, allowing for fine-grained control over query semantics. This work bridges the gap between the semantic richness of dense embeddings and the interpretability of sparse representations. We open source our embeddings, trained sparse autoencoders, and interpreted features, as well as a web app for exploring them.

cs.LG↗

arXiv2024

Designing an Evaluation Framework for Large Language Models in Astronomy Research

Large Language Models (LLMs) are shifting how scientific research is done. It is imperative to understand how researchers interact with these models and how scientific sub-communities like astronomy might benefit from them. However, there is currently no standard for evaluating the use of LLMs in astronomy. Therefore, we present the experimental design for an evaluation study on how astronomy researchers interact with LLMs. We deploy a Slack chatbot that can answer queries from users via Retrieval-Augmented Generation (RAG); these responses are grounded in astronomy papers from arXiv. We record and anonymize user questions and chatbot answers, user upvotes and downvotes to LLM responses, user feedback to the LLM, and retrieved documents and similarity scores with the query. Our data collection method will enable future dynamic evaluations of LLM tools for astronomy.

astro-ph.IM↗

arXiv2024

The FENIKS Survey: Multi-wavelength Photometric Catalog in the UDS Field, and Catalogs of Photometric Redshifts and Stellar Population Properties

We present the construction of a deep multi-wavelength PSF-matched photometric catalog in the UDS field following the final UKIDSS UDS DR11 release. The catalog includes photometry in 24 filters, from the MegaCam-uS (0.38 microns) band to the Spitzer-IRAC (8 microns) band, over ~ 0.9 sq. deg. and with a 5-sigma depth of 25.3 AB in the K-band detection image. The catalog, containing approximately 188,564 (136,235) galaxies at 0.2 < z < 8.0 with stellar mass > 10$^{8}$ solar masses and K-band total magnitude K < 25.2 (24.3) AB, enables a range of extragalactic studies. We also provide photometric redshifts, corresponding redshift probability distributions, and rest-frame absolute magnitudes and colors derived using the template-fitting code eazy-py. Photometric redshift errors are less than 3 to 4 percent at z < 4 across the full brightness range in K-band and stellar mass range 10$^{8}$-10$^{12}$ solar masses. Stellar population properties (e.g., stellar mass, star-formation rate, dust extinction) are derived from the modeling of the spectral energy distributions (SEDs) using the codes FAST and Dense Basis.

astro-ph.GA↗

arXiv2024

The UV luminosity function at 0.6 < z < 1 from UVCANDELS

UVCANDELS is a HST Cycle-26 Treasury Program awarded 164 orbits of primary ultraviolet (UV) F275W imaging and coordinated parallel optical F435W imaging in four CANDELS fields: GOODS-N, GOODS-S, EGS, and COSMOS, covering a total area of $\sim426$ arcmin$^2$. This is $\sim2.7$ times larger than the area covered by previous deep-field space UV data combined, reaching a depth of about 27 and 28 ABmag ($5σ$ in $0.2"$ apertures) for F275W and F435W, respectively. Along with the new photometric catalogs, we present an analysis of the rest-frame UV luminosity function (LF), relying on our UV-optimized aperture photometry method yielding a factor of $1.5\times$ increase than the H-isophot aperture photometry in the signal-to-noise ratios of galaxies in our F275W imaging. Using well tested photometric redshift measurements we identify 5810 galaxies at redshifts $0.6<z<1$, down to an absolute magnitude of $M_\text{UV} = -14.2$. In order to minimize the effect of uncertainties in estimating the completeness function, especially at the faint-end, we restrict our analysis to sources above $30\%$ completeness, which provides a final sample of 4726 galaxies at $-21.5<M_\text{UV}<-15.5$. We performed a maximum likelihood estimate to derive the best-fit parameters of the UV LF. We report a best-fit faint-end slope of $α= -1.359^{+0.041}_{-0.041}$ at $z \sim 0.8$. Creating sub-samples at $z\sim0.7$ and $z\sim0.9$, we observe a possible evolution of $α$ with redshift. The unobscured UV luminosity density at $M_\text{UV}<-10$ is derived as $ρ_\text{UV}=1.339^{+0.027}_{-0.030}\ (\times10^{26} \text{ergs/s/Hz/Mpc}^3)$ using our best-fit LF parameters. The new F275W and F435 photometric catalogs from UVCANDELS have been made publicly available on the Barbara A. Mikulski Archive for Space Telescopes (MAST).

astro-ph.GA↗

arXiv2024

Stellar Populations With Optical Spectra: Deep Learning vs. Popular Spectrum Fitting Codes

We compare the performance of several popular spectrum fitting codes (Firefly, starlight, pyPipe3D and pPXF), and a deep-learning convolutional neural network (StarNet), in recovering known stellar population properties (mean stellar age, stellar metallicity, stellar mass-to-light ratio M*/L_r and the internal E(B-V)) of simulated galaxy spectra in optical wavelengths. Our mock spectra are constructed from star-formation histories from the IllustrisTNG100-1 simulation. These spectra mimic the Sloan Digital Sky Survey (SDSS) through a novel method of including the noise, sky residuals and emission lines taken directly from SDSS. We find that StarNet vastly outperforms all conventional codes in both speed and recovery of stellar population properties (error scatter < 0.08 dex, average biases < 0.02 dex for all tested quantities), but it requires an appropriate training set. Of the non-machine-learning codes, pPXF was a factor of 3-4 times faster than the other codes, and was the best in recovering stellar population properties (error scatter of < 0.11 dex, average biases < 0.08 dex). However, the errors and biases are strongly dependent on both true and predicted values of stellar age and metallicity, and signal-to-noise ratio. The biases of all codes can approach 0.15 dex in stellar ages, metallicities and log M*/L_r , but remain < 0.05 for E(B-V). Using unrealistic Gaussian noise in the construction of mock spectra will underestimate the errors in the metallicities by a factor of two or more, and mocks without emission lines will underestimate the errors in stellar age and M*/L_r by a factor of two.

astro-ph.GA↗

arXiv2024

The Firefly Sparkle: The Earliest Stages of the Assembly of A Milky Way-type Galaxy in a 600 Myr Old Universe

The most distant galaxies detected by JWST are assembling in a Universe that is less than 5\% of its present age. At these times, the progenitors of galaxies like the Milky Way are expected to be about 10,000 times less massive than they are now, with masses quite comparable to that of massive globular clusters seen in the local Universe. Composed today primarily of old stars and correlating with the properties of their parent dark matter halos, the first globular clusters are thought to have formed during the earliest stages of galaxy assembly. In this article we explore the connection between star clusters and galaxy assembly by showing JWST observations of a strongly lensed galaxy at zspec = 8.304, exhibiting a network of massive star clusters (the 'Firefly Sparkle') cocooned in a diffuse arc. The Firefly Sparkle exhibits the hallmarks expected of a future Milky Way-type galaxy captured during its earliest and most gas-rich stage of formation. The mass distribution of the galaxy seems to be concentrated in ten distinct clusters, with individual cluster masses that straddle the boundary between low-mass galaxies and high-mass globular clusters. The cluster ages suggest that they are gravitationally bound with star formation histories showing a recent starburst possibly triggered by the interaction with a companion galaxy at the same redshift at a projected distance of $\sim$2 kpc away from the Firefly Sparkle. The central star cluster shows nebular-dominated spectra consistent with high temperatures and a top-heavy initial mass function, the product of formation in a very metal poor environment. Combined with abundance matching that suggests that this is likely to be a progenitor of galaxies like our own, the Firefly Sparkle provides an unprecedented case study of a Milky Way-like galaxy in the earliest stages of its assembly in only a 600 million year old Universe.

astro-ph.GA↗

arXiv2024

Bursty star formation and galaxy-galaxy interactions in low-mass galaxies 1 Gyr after the Big Bang

We use CANUCS JWST/NIRCam imaging of galaxies behind the gravitationally-lensing cluster MACS J0417.5-1154 to investigate star formation burstiness in low-mass ($M_\star\sim10^8\ M_\odot$) galaxies at $z\sim4.7-6.5$. Our sample of 123 galaxies is selected using the Lyman break selection and photometric emission-line excess methods. Sixty per cent of the 123 galaxies in this sample have H$α$-to-UV flux ratios that deviate significantly from the range of $η_{1500}$ values consistent with smooth and steady star formation histories. This large fraction indicates that the majority of low-mass galaxies is experiencing bursty star formation histories at high redshift. We also searched for interacting galaxies in our sample and found that they are remarkably common ($\sim40$ per cent of the sample). Compared to non-interacting galaxies, interacting galaxies are more likely to have very low H$α$-to-UV ratios, suggesting that galaxy-galaxy interactions enhance star formation burstiness and enable faster quenching (with timescales of $\lesssim100$ Myr) that follows the rapid rise of star formation activity. Given the high frequency of galaxy-galaxy interactions and the rapid SFR fluctuations they appear to cause, we conclude that galaxy-galaxy interactions could be a leading cause of bursty star formation in low-mass, high-$z$ galaxies. They could thus play a significant role in the evolution of the galaxy population at early cosmological times.

astro-ph.GA↗

arXiv2024

AstroLLaMA-Chat: Scaling AstroLLaMA with Conversational and Diverse Datasets

We explore the potential of enhancing LLM performance in astronomy-focused question-answering through targeted, continual pre-training. By employing a compact 7B-parameter LLaMA-2 model and focusing exclusively on a curated set of astronomy corpora -- comprising abstracts, introductions, and conclusions -- we achieve notable improvements in specialized topic comprehension. While general LLMs like GPT-4 excel in broader question-answering scenarios due to superior reasoning capabilities, our findings suggest that continual pre-training with limited resources can still enhance model performance on specialized topics. Additionally, we present an extension of AstroLLaMA: the fine-tuning of the 7B LLaMA model on a domain-specific conversational dataset, culminating in the release of the chat-enabled AstroLLaMA for community use. Comprehensive quantitative benchmarking is currently in progress and will be detailed in an upcoming full paper. The model, AstroLLaMA-Chat, is now available at https://huggingface.co/universeTBD, providing the first open-source conversational AI tool tailored for the astronomy community.

astro-ph.IM↗

arXiv2023

Outshining by Recent Star Formation Prevents the Accurate Measurement of High-z Galaxy Stellar Masses

In this paper, we demonstrate that the inference of galaxy stellar masses via spectral energy distribution (SED) fitting techniques for galaxies formed in the first billion years after the Big Bang carries fundamental uncertainties owing to the loss of star formation history (SFH) information from the very first episodes of star formation in the integrated spectra of galaxies. While this early star formation can contribute substantially to the total stellar mass of high-redshift systems, ongoing star formation at the time of detection outshines the residual light from earlier bursts, hampering the determination of accurate stellar masses. As a result, order of magnitude uncertainties in stellar masses can be expected. We demonstrate this potential problem via direct numerical simulation of galaxy formation in a cosmological context. In detail, we carry out two cosmological simulations with significantly different stellar feedback models which span a significant range in star formation history burstiness. We compute the mock SEDs for these model galaxies at z=7 via 3D dust radiative transfer calculations, and then backwards fit these SEDs with Prospector SED fitting software. The uncertainties in derived stellar masses that we find for z>7 galaxies motivate the development of new techniques and/or star formation history priors to model early Universe star formation.

astro-ph.GA↗

arXiv2023

AstroLLaMA: Towards Specialized Foundation Models in Astronomy

Large language models excel in many human-language tasks but often falter in highly specialized domains like scholarly astronomy. To bridge this gap, we introduce AstroLLaMA, a 7-billion-parameter model fine-tuned from LLaMA-2 using over 300,000 astronomy abstracts from arXiv. Optimized for traditional causal language modeling, AstroLLaMA achieves a 30% lower perplexity than Llama-2, showing marked domain adaptation. Our model generates more insightful and scientifically relevant text completions and embedding extraction than state-of-the-arts foundation models despite having significantly fewer parameters. AstroLLaMA serves as a robust, domain-specific model with broad fine-tuning potential. Its public release aims to spur astronomy-focused research, including automatic paper summarization and conversational agent development.

astro-ph.IM↗

arXiv2023

A Hierarchy of Normalizing Flows for Modelling the Galaxy-Halo Relationship

Using a large sample of galaxies taken from the Cosmology and Astrophysics with MachinE Learning Simulations (CAMELS) project, a suite of hydrodynamic simulations varying both cosmological and astrophysical parameters, we train a normalizing flow (NF) to map the probability of various galaxy and halo properties conditioned on astrophysical and cosmological parameters. By leveraging the learnt conditional relationships we can explore a wide range of interesting questions, whilst enabling simple marginalisation over nuisance parameters. We demonstrate how the model can be used as a generative model for arbitrary values of our conditional parameters; we generate halo masses and matched galaxy properties, and produce realisations of the halo mass function as well as a number of galaxy scaling relations and distribution functions. The model represents a unique and flexible approach to modelling the galaxy-halo relationship.

astro-ph.GA↗

arXiv2023

Harnessing the Power of Adversarial Prompting and Large Language Models for Robust Hypothesis Generation in Astronomy

This study investigates the application of Large Language Models (LLMs), specifically GPT-4, within Astronomy. We employ in-context prompting, supplying the model with up to 1000 papers from the NASA Astrophysics Data System, to explore the extent to which performance can be improved by immersing the model in domain-specific literature. Our findings point towards a substantial boost in hypothesis generation when using in-context prompting, a benefit that is further accentuated by adversarial prompting. We illustrate how adversarial prompting empowers GPT-4 to extract essential details from a vast knowledge base to produce meaningful hypotheses, signaling an innovative step towards employing LLMs for scientific research in Astronomy.

astro-ph.IM↗

arXiv2023

JWST catches the assembly of a $z\sim5$ ultra-low-mass galaxy

Using CANUCS imaging we found an apparent major merger of two $z\sim5$ ultra-low-mass galaxies ($M_\star\sim10^{7}M_\odot$ each) that are doubly imaged and magnified $\sim$12-15$\times$ by the lensing cluster MACS 0417. Both galaxies are experiencing young ($\sim$100 Myr), synchronised bursts of star formation with $\log({\rm sSFR/Gyr^{-1}} )\sim$1.3-1.4, yet SFRs of just $\sim 0.2 M_\odot\ {\rm yr}^{-1}$. They have sub-solar ($Z\sim0.2Z_\odot$) gas-phase metallicities and are connected by an even more metal-poor star-forming bridge. The galaxy that forms from the merger will have a mass of at least $M_\star\sim 2\times10^7 M_\odot$, at least half of it formed during the interaction-induced starburst. More than half of the ionizing photons produced by the system (before and during the merger) will have been produced during the burst. This system provides the first detailed look at a merger involving two high-$z$ ultra-low-mass galaxies of the type believed to be responsible for reionizing the Universe. It suggests that such galaxies can grow via a combination of mass obtained through major mergers, merger-triggered starbursts, and long-term in-situ star formation. If such high-$z$ mergers are common, then merger-triggered starbursts could be significant contributors to the ionizing photon budget of the Universe.

astro-ph.GA↗

arXiv2023

Spectroscopy from Photometry: A Population of Extreme Emission Line Galaxies at $1.7 \lesssim z \lesssim 6.7$ Selected with JWST Medium Band Filters

We use JWST/NIRCam medium band photometry in a single pointing of the CAnadian NIRISS Unbiased Cluster Survey (CANUCS) to identify 118 Extreme Emission Line Galaxies (EELGs) over $1.7 \lesssim z \lesssim 6.7$, selected using a set of color cuts that target galaxies with extreme $\text{[OIII] + H}β$ and H$α$ emission. We show that our medium band color selections are able to select galaxies based on emission line equivalent width (EW), which is advantageous to more commonly used selections since it does not require strong continuum emission, and can select galaxies with faint or red continuum fluxes. The median EWs of our sample is $EW(\text{H}α) = 893 $ Å and $ EW(\text{[OIII] + H}β) = 1255 $ Å, and includes some objects with $EW(\text{[OIII] + H}β) \sim 3000$ Å. These systems are mostly compact with low stellar mass (median $\log(M_\star/M_\odot) = 8.03$), low metallicity (median $Z = 0.14 Z_\odot$), little dust (median $A_V = 0.18$ mag) and high SSFR (median $SSFR = 1.18 \times 10^{-8}/yr$). Additionally, galaxies in our sample show increasing EW(\Ha) and EW(\OIIIHb) with redshift, an anti-correlation of EW(\Ha) with stellar mass, and no correlation between EW(\OIIIHb) and stellar mass. Finally, we present NIRSpec spectroscopy of 15 of the EELGs in our sample. These spectra confirm the redshifts and EWs of the EELGs calculated from the medium bands, which demonstrates the accuracy and efficiency of our color selections. Overall, we show that there are significant advantages to using medium band photometry to identify and study EELGs at a wide range of redshifts.

astro-ph.GA↗

arXiv2023

The first large catalogue of spectroscopic redshifts in Webb's First Deep Field, SMACS J0723.3$-$7327

We present a spectroscopic redshift catalogue of the SMACS J0723.3$-$7327 field ("Webb's First Deep Field") obtained from JWST/NIRISS grism spectroscopy and supplemented with JWST/NIRSpec and VLT/MUSE redshifts. The catalogue contains a total of 190 sources with secure spectroscopic redshifts, including 156 NIRISS grism redshifts, 123 of which are for sources whose redshifts were previously unknown. These new grism redshifts are secured with two or more spectroscopic features (64 sources), or with a single spectral feature whose identity is secured from the object's nine-band photometric redshift (59 sources). These are complemented with 17 NIRSpec and 48 MUSE redshifts, including six new NIRSpec redshifts identified in this work. In addition to the $z_{\rm cl}=0.39$ cluster galaxy redshifts (for which we provide $\sim$40 new NIRISS absorption-line redshifts), we also find three prominent galaxy overdensities at higher redshifts - at $z=1.1$, $z=1.4$, and $z=2.0$ - that were until now not seen in the JWST/NIRSpec and VLT/MUSE data. The paper describes the characteristics of our spectroscopic redshift sample and the methodology we have employed to obtain it. Our redshift catalogue is made available to the community at https://niriss.github.io/smacs0723.

astro-ph.GA↗