Searcharxiv⌕ Search

arXiv subjects

Minjin Kim

Publications and source records attributed to Minjin Kim.

At least 55 records · Page 3Linked to original sources
arXiv2023

The Seoul National University AGN Monitoring Project III: H$β$ lag measurements of 32 luminous AGNs and the high-luminosity end of the size--luminosity relation

We present the main results from a long-term reverberation mapping campaign carried out for the Seoul National University Active Galactic Nuclei (AGN) Monitoring Project. High-quality data were obtained during 2015-2021 for 32 luminous AGNs (i.e., continuum luminosity in the range of $10^{44-46}$ erg s$^{-1}$) at a regular cadence, of 20-30 days for spectroscopy and 3-5 days for photometry. We obtain time lag measurements between the variability in the H$β$ emission and the continuum for 32 AGNs; twenty-five of those have the best lag measurements based on our quality assessment, examining correlation strength, and the posterior lag distribution. Our study significantly increases the current sample of reverberation-mapped AGNs, particularly at the moderate to high luminosity end. Combining our results with literature measurements, we derive a H$β$ broad line region size--luminosity relation with a shallower slope than reported in the literature. For a given luminosity, most of our measured lags are shorter than the expectation, implying that single-epoch black hole mass estimators based on previous calibrations could suffer large systematic uncertainties.

astro-ph.GA↗
arXiv2023

Dialogue Chain-of-Thought Distillation for Commonsense-aware Conversational Agents

Human-like chatbots necessitate the use of commonsense reasoning in order to effectively comprehend and respond to implicit information present within conversations. Achieving such coherence and informativeness in responses, however, is a non-trivial task. Even for large language models (LLMs), the task of identifying and aggregating key evidence within a single hop presents a substantial challenge. This complexity arises because such evidence is scattered across multiple turns in a conversation, thus necessitating integration over multiple hops. Hence, our focus is to facilitate such multi-hop reasoning over a dialogue context, namely dialogue chain-of-thought (CoT) reasoning. To this end, we propose a knowledge distillation framework that leverages LLMs as unreliable teachers and selectively distills consistent and helpful rationales via alignment filters. We further present DOCTOR, a DialOgue Chain-of-ThOught Reasoner that provides reliable CoT rationales for response generation. We conduct extensive experiments to show that enhancing dialogue agents with high-quality rationales from DOCTOR significantly improves the quality of their responses.

cs.CL↗
arXiv2023

The Structure Function of Mid-infrared Variability in Low-redshift Active Galactic Nuclei

Using the multi-epoch mid-infrared (MIR) photometry from the Wide-field Infrared Survey Explorer spanning a baseline of $\sim10$ yr, we extensively investigate the MIR variability of nearby active galactic nuclei (AGNs) at $0.15 < z < 0.4$. We find that the ensemble structure function in the W1 band ($3.4\ μ$m) can be modeled with a broken power law. Type 1 AGNs tend to exhibit larger variability amplitudes than type 2 AGNs, possibly due to the extinction by the torus. The variability amplitude is inversely correlated with the AGN luminosity, consistent with a similar relation known in the optical. Meanwhile, the slope of the power law increases with AGN luminosity. This trend can be attributed to the fact that the inner radius of the torus is proportional to the AGN luminosity, as expected from the size$-$luminosity relation of the torus. Interestingly, low-luminosity type 2 AGNs, unlike low-luminosity type 1 AGNs, tend to exhibit smaller variability amplitude than do high-luminosity AGNs. We argue that either low-luminosity type 2 AGNs have distinctive central structures due to their low luminosity or their MIR brightness is contaminated by emission from the cold dust in the host galaxy. Our findings suggest that the AGN unification scheme may need to be revised. We find that the variability amplitude of dust-deficient AGNs is systematically larger than that of normal AGNs, supporting the notion that the hot and warm dust in dust-deficient AGNs may be destroyed and reformed according to the strength of the ultraviolet radiation from the accretion disk.

astro-ph.GA↗
arXiv2023

Estimators of Bolometric Luminosity and Black Hole Mass with Mid-infrared Continuum Luminosities for Dust-obscured Quasars: Prevalence of Dust-obscured SDSS Quasars

We present bolometric luminosity ($L_{\rm bol}$) and black hole (BH) mass ($M_{\rm BH}$) estimators based on mid-infrared (MIR) continuum luminosity (hereafter, $L_{\rm MIR}$) that are measured from infrared (IR) photometric data. The $L_{\rm MIR}$-based estimators are relatively immune from dust extinction effects, hence they can be used for dust-obscured quasars. To derive the $L_{\rm bol}$ and $M_{\rm BH}$ estimators, we use unobscured quasars selected from the Sloan Digital Sky Survey (SDSS) quasar catalog, which have wide ranges of $L_{\rm bol}$ ($10^{44.62}$--$10^{46.16}$\,$\rm erg\,s^{-1}$) and $M_{\rm BH}$ ($10^{7.14}$--$10^{9.69}$\,$M_{\odot}$). We find empirical relations between (i) continuum luminosity at 5100\,$\rm{Å}$ (hereafter, L5100) and $L_{\rm MIR}$; (ii) $L_{\rm bol}$ and $L_{\rm MIR}$. Using these relations, we derive the $L_{\rm MIR}$-based $L_{\rm bol}$ and $M_{\rm BH}$ estimators. We find that our estimators allow the determination of $L_{\rm bol}$ and $M_{\rm BH}$ at an accuracy of $\sim$0.2\,dex against the fiducial estimates based on the optical properties of the unobscured quasars. We apply the $L_{\rm MIR}$-based estimators to SDSS quasars at $z \lesssim 0.5$ including obscured ones. The ratios of $L_{\rm bol}$ from the $L_{\rm MIR}$-based estimators to those from the optical luminosity-based estimators become larger with the amount of the dust extinction, and a non-negligible fraction ($\sim$15\,\%) of the SDSS quasars exhibits ratios greater than 1.5. This result suggests that dust extinction can significantly affect physical parameter derivations even for SDSS quasars, and that dust extinction needs to be carefully taken into account when deriving quasar properties.

astro-ph.GA↗
arXiv2023

Photometric Selection of Unobscured QSOs in the Ecliptic Poles: KMTNet in the South Field and Pan-STARRS in the North Field

We search for quasi-stellar objects (QSOs) in a wide area of the south ecliptic pole (SEP) field, which has been and will continue to be intensively explored through various space missions. For this purpose, we obtain deep broadband optical images of the SEP field covering an area of $\sim$$14.5\times14.5$ deg$^2$ with the Korea Microlensing Telescope Network. The 5$σ$ detection limits for point sources in the $BVRI$ bands are estimated to be $\sim$22.59, 22.60, 22.98, and 21.85 mag, respectively. Utilizing data from Wide-field Infrared Survey Explorer, unobscured QSO candidates are selected among the optically point-like sources using the mid-infrared (MIR) and optical-MIR colors. To further refine our selection and eliminate any contamination not adequately removed by the color-based selection, we perform the spectral energy distribution fitting with archival photometric data ranging from optical to MIR. As a result, we identify a total of 2,383 unobscured QSO candidates in the SEP field. We also apply a similar method to the north ecliptic pole field using the Pan-STARRS data and obtain a similar result of identifying 2,427 candidates. The differential number count per area of our QSO candidates is in good agreement with those measured from spectroscopically confirmed ones in other fields. Finally, we compare the results with the literature and discuss how this work will be implicated in future studies, especially with the upcoming space missions.

astro-ph.GA↗
arXiv2023

The Seoul National University AGN Monitoring Project IV: H$α$ reverberation mapping of 6 AGNs and the H$α$ Size-Luminosity Relation

The broad line region (BLR) size-luminosity relation has paramount importance for estimating the mass of black holes in active galactic nuclei (AGNs). Traditionally, the size of the H$β$ BLR is often estimated from the optical continuum luminosity at 5100\angstrom{} , while the size of the H$α$ BLR and its correlation with the luminosity is much less constrained. As a part of the Seoul National University AGN Monitoring Project (SAMP) which provides six-year photometric and spectroscopic monitoring data, we present our measurements of the H$α$ lags of 6 high-luminosity AGNs. Combined with the measurements for 42 AGNs from the literature, we derive the size-luminosity relations of H$α$ BLR against broad H$α$ and 5100\angstrom{} continuum luminosities. We find the slope of the relations to be $0.61\pm0.04$ and $0.59\pm0.04$, respectively, which are consistent with the \hb{} size-luminosity relation. Moreover, we find a linear relation between the 5100\angstrom{} continuum luminosity and the broad H$α$ luminosity across 7 orders of magnitude. Using these results, we propose a new virial mass estimator based on the H$α$ broad emission line, finding that the previous mass estimates based on the scaling relations in the literature are overestimated by up to 0.7 dex at masses lower than $10^7$~M$_{\odot}$.

astro-ph.GA↗
arXiv2023

Deficit of Hot Dust in Low-redshift Active Galactic Nuclei

We assemble a broad-band spectral energy distribution (SED) ranging from optical to mid-infrared of nearby active galactic nuclei at $z < 0.4$. SED fitting analysis is performed using semi-empirical templates derived from Palomar-Green quasars to classify the sample into normal, warm-dust-deficient (WDD), and hot-dust-deficient (HDD) AGNs. Kolmogorov-Smirnov tests reveal that HDD AGNs exhibit, on average higher AGN luminosity than normal and WDD AGNs. HDD fraction, on the other hand, is only weakly correlated with black hole mass and inversely correlated with Eddington ratio. By fixing the other parameters, we conclude that the HDD fraction is primarily connected with the AGN luminosity. It implies that there is a causal connection between the covering factor of the hot dust component and AGN luminosity, possibly due to the sublimation of the innermost dust or the thickening of the intervening gas in the broad-line region. Analysis of the outflow properties traced by the wing of [O III]$\lambda5007$ suggests that outflows may be related to the formation and maintenance of the hot dust component. Finally, we demonstrate through comparison with previous studies that the classification of HDD AGNs requires careful subtraction of the host galaxy light.

astro-ph.GA↗
arXiv2023

Evidence-empowered Transfer Learning for Alzheimer's Disease

Transfer learning has been widely utilized to mitigate the data scarcity problem in the field of Alzheimer's disease (AD). Conventional transfer learning relies on re-using models trained on AD-irrelevant tasks such as natural image classification. However, it often leads to negative transfer due to the discrepancy between the non-medical source and target medical domains. To address this, we present evidence-empowered transfer learning for AD diagnosis. Unlike conventional approaches, we leverage an AD-relevant auxiliary task, namely morphological change prediction, without requiring additional MRI data. In this auxiliary task, the diagnosis model learns the evidential and transferable knowledge from morphological features in MRI scans. Experimental results demonstrate that our framework is not only effective in improving detection performance regardless of model capacity, but also more data-efficient and faithful.

eess.IV↗
arXiv2023

TUTORING: Instruction-Grounded Conversational Agent for Language Learners

In this paper, we propose Tutoring bot, a generative chatbot trained on a large scale of tutor-student conversations for English-language learning. To mimic a human tutor's behavior in language education, the tutor bot leverages diverse educational instructions and grounds to each instruction as additional input context for the tutor response generation. As a single instruction generally involves multiple dialogue turns to give the student sufficient speaking practice, the tutor bot is required to monitor and capture when the current instruction should be kept or switched to the next instruction. For that, the tutor bot is learned to not only generate responses but also infer its teaching action and progress on the current conversation simultaneously by a multi-task learning scheme. Our Tutoring bot is deployed under a non-commercial use license at https://tutoringai.com.

cs.AI↗
arXiv2022

Metallicity-PAH Relation of MIR-selected Star-forming Galaxies in AKARI North Ecliptic Pole-wide Survey

We investigate the variation in the mid-infrared spectral energy distributions of 373 low-redshift ($z<0.4$) star-forming galaxies, which reflects a variety of polycyclic aromatic hydrocarbon (PAH) emission features. The relative strength of PAH emission is parameterized as $q_\mathrm{PAH}$, which is defined as the mass fraction of PAH particles in the total dust mass. With the aid of continuous mid-infrared photometric data points covering 7-24$μ$m and far-infrared flux densities, $q_\mathrm{PAH}$ values are derived through spectral energy distribution fitting. The correlation between $q_\mathrm{PAH}$ and other physical properties of galaxies, i.e., gas-phase metallicity ($12+\mathrm{log(O/H)}$), stellar mass, and specific star-formation rate (sSFR) are explored. As in previous studies, $q_\mathrm{PAH}$ values of galaxies with high metallicity are found to be higher than those with low metallicity. The strength of PAH emission is also positively correlated with the stellar mass and negatively correlated with the sSFR. The correlation between $q_\mathrm{PAH}$ and each parameter still exists even after the other two parameters are fixed. In addition to the PAH strength, the application of metallicity-dependent gas-to-dust mass ratio appears to work well to estimate gas mass that matches the observed relationship between molecular gas and physical parameters. The result obtained will be used to calibrate the observed PAH luminosity-total infrared luminosity relation, based on the variation of MIR-FIR SED, which is used in the estimation of hidden star formation.

astro-ph.GA↗
arXiv2022

KMTNet Nearby Galaxy Survey: Overview and a Survey Description

Recently, there has been an increasing demand for deep imaging surveys to investigate the history of the mass assembly of galaxies in detail by examining the remnants of mergers and accretions, both of which have very low surface brightness (LSB). In addition, the nature of star formation in LSB regions, such as galaxy outer disks, is also an intriguing topic in terms of understanding the physical mechanisms of disk evolution. To address these issues, this study conducts a survey project, called the Korea Microlensing Telescope Network (KMTNet) Nearby Galaxy Survey to construct a deep imaging data set of nearby galaxies in the southern hemisphere using KMTNet. It provides deep and wide-field images with a field-of-view of $\sim$12 deg$^2$ for 13 nearby galaxies drawn from the Carnegie-Irvine Galaxy Survey catalog, in optical broadbands ($BRI$) and an H$α$ narrowband. Through a dedicated data reduction, the surface brightness limit in 10$^{\prime\prime}\times10^{\prime\prime}$ boxes was found to reach as deep as $μ_{1σ}\sim29$-31 mag arcsec$^{-2}$ in the optical broadbands and $f_{1σ}\sim1$-$2\times 10^{-18}$ erg s$^{-1}$ cm$^{-2}$ arcsec$^{-2}$ in the H$α$ narrowband. To conclude the paper, several possible scientific applications for this data set are described.

astro-ph.GA↗
arXiv2022

A Mid-infrared Flare in the Seyfert Galaxy NGC 3786: A Changing-look Event Triggered by an Obscured Tidal Disruption Event?

We report an exceptional mid-infrared flare in the Seyfert 1.8 NGC 3786. In the multi-epoch data from the Wide-field Infrared Survey Explorer, the nuclear mid-infrared brightness of NGC 3786 appears to vary substantially up to $0.5-0.8$ mag around mid-2020. However, there is no evidence of significant variation in the corresponding light curve of the optical band from the Zwicky Transient Facility. This implies that the flare may have been heavily obscured by nuclear dust. Through follow-up spectroscopic observations with Gemini-North after the flare, we find that broad emission lines in ${\rm Pa}α$ and ${\rm Pa}β$ newly appear, while the broad ${\rm H}β$ emission is marginally detected in the post-flare spectrum. In addition, their central wavelengths are systematically redshifted up to 900 km s$^{-1}$ with respect to the narrow emission lines. This reveals that the flare is associated with the changing-look phenomenon from type 1.8 to type 1. Based on these findings, we argue that the flare is likely to originate from an obscured tidal disruption event, although extreme variation in the accretion rate may not be ruled out completely.

astro-ph.GA↗
arXiv2022

The Infrared Medium-deep Survey. IX. Discovery of Two New $z\sim6$ Quasars and Space Density down to $M_{1450}\sim-23.5$ mag

We present the result of the Infrared Medium-deep Survey (IMS) $z\sim6$ quasar survey, using the combination of the IMS near-infrared images and the Canada-France-Hawaii Telescope Legacy Survey (CFHTLS) optical images. The traditional color-selection method results in 25 quasar candidates over $86$ deg$^{2}$. We introduce the corrected Akaike Information Criterion (AICc) with the high-redshift quasar and late-type star models to prioritize the candidates efficiently. Among the color-selected candidates, seven plausible candidates finally passed the AICc selection of which three are known quasars at $z\sim6$. The follow-up spectroscopic observations for the remaining four candidates were carried out, and we confirmed that two out of four are $z\sim6$ quasars. With this complete sample, we revisited the quasar space density at $z\sim6$ down to $M_{1450}\sim-23.5$ mag. Our result supports the low quasar space density at the luminosity where the quasar's ultraviolet ionizing emissivity peaks, favoring a minor contribution of quasars to the cosmic reionization.

astro-ph.GA↗
arXiv2022

Radio Spectra of Luminous, Heavily Obscured WISE-NVSS Selected Quasars

We present radio spectra spanning $0.1 - 10$ GHz for the sample of heavily obscured luminous quasars with extremely red mid-infrared-optical colors and compact radio emission. The spectra are constructed from targeted 10 GHz observations and archival radio survey data, which together yield $6-11$ flux density measurements for each object. Our suite of Python tools for modeling the radio spectra is publicly available on Github. Our primary result is that most (61%) of the sample have peaked or curved radio spectra and many (36%) could be classified as Gigahertz Peaked Spectrum (GPS) sources. This indicates compact emission regions likely arising from recently triggered radio jets. Assuming synchrotron self-absorption (SSA) generates the peaks, we infer compact source sizes ($3 - 100$ pc) with strong magnetic fields ($6 - 100$ mG) and young ages ($30 - 10^4$ years). Conversely, free-free absorption (FFA) could also create peaks due to the high column densities associated with the deeply embedded nature of the sample. However, we find no correlations between the existence or frequency of the peaks and any parameters of the MIR emission. The high-frequency spectral indices are steep ($α\approx -1$) and correlate, weakly, with the ratio of MIR photon energy density to magnetic energy density, suggesting that the spectral steepening could arise from inverse Compton scattering off the intense MIR photon field. This study provides a foundation for combining multi-frequency and mixed-resolution radio survey data for understanding the impact of young radio jets on the ISM and star formation rates of their host galaxies.

astro-ph.GA↗
arXiv2022

Relation between Black Hole Mass and Bulge Luminosity in Hard X-ray selected Type 1 AGNs

Using $I$-band images of 35 nearby ($z<0.1$) type 1 active galactic nuclei (AGNs) obtained with Hubble Space Telescope, selected from the 70-month Swift-BAT X-ray source catalog, we investigate the photometric properties of the host galaxies. With a careful treatment of the point-spread function (PSF) model and imaging decomposition, we robustly measure the $I$-band brightness and the effective radius of bulges in our sample. Along with black hole (BH) mass estimates from single-epoch spectroscopic data, we present the relation between BH mass and $I$-band bulge luminosity ($M_{\rm BH}-M_{I,\rm bul}$ relation) of our sample AGNs. We find that our sample lies offset from the $M_{\rm BH}-M_{I,\rm bul}$ relation of inactive galaxies by 0.4 dex, i.e., at a given bulge luminosity, the BH mass of our sample is systematically smaller than that of inactive galaxies. We also demonstrate that the zero point offset in the $M_{\rm BH}-M_{I,\rm bul}$ relation with respect to inactive galaxies is correlated with the Eddington ratio. Based on the Kormendy relation, we find that the mean surface brightness of ellipticals and classical bulges in our sample is comparable to that of normal galaxies, revealing that bulge brightness is not enhanced in our sample. As a result, we conclude that the deviation in the $M_{\rm BH}-M_{I,\rm bul}$ relation from inactive galaxies is possibly because the scaling factor in the virial BH mass estimator depends on the Eddington ratio.

astro-ph.GA↗
arXiv2022

Mid-infrared Variability of Low-redshift Active Galactic Nuclei : Constraints on a Hot Dust Component with a Variable Covering Factor

We utilize mid-infrared multi-epoch data from Wide-field Infrared Survey Explorer (WISE) over a $\sim10$-year period in W1 (3.4$μ$m) and W2 (4.6$μ$m) bands to investigate the structure of dusty torus in low-redshift ($0.15<z\le 0.4$) active galactic nuclei (AGN). We calculate a Spearman correlation coefficient ($r_{12}$) between W1 magnitude and W1$-$W2 color based on the light curve in individual objects. Interestingly, $r_{12}$ spans a broad range from $-1$ to $1$, and is detected to be correlated with mean W1$-$W2 color and AGN bolometric luminosity, in the sense that objects with blue W1$-$W2 color and low AGN luminosity tend to become redder (bluer) with increasing (decreasing) W1 brightness in the light curve (i.e. $r_{12} < 0$), although the correlation of $r_{12}$ with the bolometric luminosity is relatively weak. The fit for the spectral energy distribution reveals a significant contribution from the host galaxy in the W1 and W2 bands. However, the dependencies of $r_{12}$ on W1$-$W2 color and AGN luminosity still persist even after careful elimination of the host light contribution. We propose that this result can be explained if covering factor of the hot dust component decreases as AGN luminosity increases.

astro-ph.GA↗
arXiv2021

Reversible Photo-Induced Phase Segregation and Origin of Long Carrier Lifetime in Mixed-Halide Perovskite Films

Mixed-halide based hybrid perovskite semiconductors have attracted tremendous attention as a promising candidate for high efficient photovoltaic and light-emitting devices. However, these advanced perovskite materials may undergo phase segregation under light illumination due to halide ion migration, affecting their optoelectronic properties. Here, we report photo-excitation induced phase segregation in triple-cation mixed-halide perovskite films that yields to red-shift in photoluminescence response. We demonstrate that photo-excitation induced halide ion migration leads to the formation of smaller-bandgap iodide-rich and larger-bandgap bromide-rich domains in the perovskite film, where the phase segregation rate is found to follow the excitation power-density as a power law. Results confirm that charge carrier lifetime increases with redshift in photoluminescence due to the trapping of photo-excited carriers in the segregated smaller-bandgap iodide-rich domains. Interestingly, we found that these photo-induced changes are fully reversible and thermally activated when the excitation power is turned off. A significant difference in activation energies for halide ion migration is observed during phase segregation and recovery process under darkness. Additionally, we have investigated the emission linewidth broadening as a function of temperature which is governed by the exciton-optical phonon coupling. The mechanism of photo-induced phase segregation is interpreted based on excitonphonon coupling strength in both mixed and demixed (segregated) states of perovskite film.

physics.app-ph↗
arXiv2021

A Hubble Space Telescope Imaging Survey of Low-Redshift Swift-BAT Active Galaxies

We present initial results from a Hubble Space Telescope snapshot imaging survey of the host galaxies of Swift-BAT active galactic nuclei (AGN) at z<0.1. The hard X-ray selection makes this sample sample relatively unbiased in terms of obscuration compared to optical AGN selection methods. The high-resolution images of 154 target AGN enable us to investigate the detailed photometric structure of the host galaxies, such as the Hubble type and merging features. We find that 48% and 44% of the sample is hosted by early-type and late-type galaxies, respectively. The host galaxies of the remaining 8% of the sample are classified as peculiar galaxies because they are heavily disturbed. Only a minor fraction of host galaxies (18%-25%) exhibit merging features (e.g., tidal tails, shells, or major disturbance). The merging fraction increases strongly as a function of bolometric AGN luminosity, revealing that merging plays an important role in triggering luminous AGN in this sample. However, the merging fraction is weakly correlated with the Eddington ratio, suggesting that merging does not necessarily lead to an enhanced Eddington ratio. Type 1 and type 2 AGN are almost indistinguishable in terms of their Hubble type distribution and merging fraction. However, the merging fraction of type 2 AGN peaks at a lower bolometric luminosity compared with those of type 1 AGN. This result may imply that the triggering mechanism and evolutionary stages of type 1 and type 2 AGN are not identical.

astro-ph.GA↗