arXiv subjects

Weizhe Liu

Publications and source records attributed to Weizhe Liu.

90 records · Page 5Linked to original sources

arXiv2023

A SPectroscopic survey of biased halos In the Reionization Era (ASPIRE): JWST Reveals a Filamentary Structure around a z=6.61 Quasar

We present the first results from the JWST ASPIRE program (A SPectroscopic survey of biased halos In the Reionization Era). This program represents an imaging and spectroscopic survey of 25 reionization-era quasars and their environments by utilizing the unprecedented capabilities of NIRCam Wide Field Slitless Spectroscopy (WFSS) mode. ASPIRE will deliver the largest ($\sim280~{\rm arcmin}^2$) galaxy redshift survey at 3-4 $μ$m among JWST Cycle-1 programs and provide extensive legacy values for studying the formation of the earliest supermassive black holes (SMBHs), the assembly of galaxies, early metal enrichment, and cosmic reionization. In this first ASPIRE paper, we report the discovery of a filamentary structure traced by the luminous quasar J0305-3150 and ten [OIII] emitters at $z=6.6$. This structure has a 3D galaxy overdensity of $δ_{\rm gal}=12.6$ over 637 cMpc$^3$, one of the most overdense structures known in the early universe, and could eventually evolve into a massive galaxy cluster. Together with existing VLT/MUSE and ALMA observations of this field, our JWST observations reveal that J0305-3150 traces a complex environment where both UV-bright and dusty galaxies are present, and indicate that the early evolution of galaxies around the quasar is not simultaneous. In addition, we discovered 31 [OIII] emitters in this field at other redshifts, $5.3<z<6.7$, with half of them situated at $z\sim5.4$ and $z\sim6.2$. This indicates that star-forming galaxies, such as [OIII] emitters, are generally clustered at high redshifts. These discoveries demonstrate the unparalleled redshift survey capabilities of NIRCam WFSS and the potential of the full ASPIRE survey dataset.

astro-ph.GA↗

arXiv2023

A Comparison of Outflow Properties in AGN Dwarfs vs. Star Forming Dwarfs

Feedback likely plays a crucial role in resolving discrepancies between observed and theoretical predictions of dwarf galaxy properties. Stellar feedback was once believed to be sufficient to explain these discrepancies, but it has thus far failed to fully reconcile theory and observations. The recent discovery of energetic galaxy-wide outflows in dwarf galaxies hosting Active Galactic Nuclei (AGN) suggests that AGN feedback may have a larger role in the evolution of dwarf galaxies than previously suspected. In order to assess the relative importance of stellar versus AGN feedback in these galaxies, we perform a detailed Keck/KCWI optical integral field spectroscopic study of a sample of low-redshift star-forming (SF) dwarf galaxies that show outflows in ionized gas in their SDSS spectra. We characterize the outflows and compare them to observations of AGN-driven outflows in dwarfs. We find that SF dwarfs have outflow components that have comparable widths (W$_{80}$) to those of outflows in AGN dwarfs, but are much less blue-shifted, indicating that SF dwarfs have significantly slower outflows than their AGN counterparts. The outflows in SF dwarfs are spatially resolved and significantly more extended than those in AGN dwarfs. The mass loss rates, momentum and energy rates of SF-driven outflows are much lower than those of AGN-driven outflows. Our results indicate that AGN feedback in the form of gas outflows may play an important role in dwarf galaxies and should be considered along with SF feedback in models of dwarf galaxy evolution.

astro-ph.GA↗

arXiv2022

Multi-view Tracking Using Weakly Supervised Human Motion Prediction

Multi-view approaches to people-tracking have the potential to better handle occlusions than single-view ones in crowded scenes. They often rely on the tracking-by-detection paradigm, which involves detecting people first and then connecting the detections. In this paper, we argue that an even more effective approach is to predict people motion over time and infer people's presence in individual frames from these. This enables to enforce consistency both over time and across views of a single temporal frame. We validate our approach on the PETS2009 and WILDTRACK datasets and demonstrate that it outperforms state-of-the-art methods.

cs.CV↗

arXiv2022

First results from the JWST Early Release Science Program Q3D: Turbulent times in the life of a $z \sim 3$ extremely red quasar revealed by NIRSpec IFU

Extremely red quasars, with bolometric luminosities exceeding $10^{47}$ erg s$^{-1}$, are a fascinating high-redshift population that is absent in the local universe. They are the best candidates for supermassive black holes accreting at rates at or above the Eddington limit, and they are associated with the most rapid and powerful outflows of ionized gas known to date. They are also hosted by massive galaxies. Here we present the first integral field unit (IFU) observations of a high-redshift quasar obtained by the Near Infrared Spectrograph (NIRSpec) on board the James Webb Space Telescope (JWST), which targeted SDSSJ165202.64+172852.3, an extremely red quasar at $z=2.94$. JWST observations reveal extended ionized gas - as traced by [OIII]$λ$5007Å- in the host galaxy of the quasar, its outflow, and the circumgalactic medium. The complex morphology and kinematics imply that the quasar resides in a very dense environment with several interacting companion galaxies within projected distances of 10-15 kpc. The high density of the environment and the large velocities of the companion galaxies suggest that this system may represent the core of a forming cluster of galaxies. The system is a good candidate for a merger of two or more dark matter halos, each with a mass of a few $10^{13}$ M$_\odot$ and traces potentially one of the densest knots at $z\sim3$.

astro-ph.GA↗

arXiv2022

Galactic Winds across the Gas-Rich Merger Sequence II. Lyman Alpha Emission and Highly Ionized O VI and N V Outflows in Ultraluminous Infrared Galaxies

This paper is the second in a series aimed at examining the gaseous environments of z$\le$0.3 quasars and ultraluminous infrared galaxies (ULIRGs) as a function of AGN/host galaxy properties across the merger sequence. This second paper focuses on the Ly$α$ emission and O VI and N V absorption features, tracers of highly ionized gas outflows, in ULIRGs observed with HST/COS. Ly$α$ emission is detected in 15 out of 19 ULIRGs, and 12 of the 14 clear Ly$α$ detections show emission with blueshifted velocity centroids and/or wings. The equivalent widths of the Ly$α$ emission increase with increasing AGN luminosities and AGN bolometric fractions. The blueshifts of the Ly$α$ emission correlate positively with those of the [O III] emission, where the latter traces the ionized gas outflows. The Ly$α$ escape fractions tend to be slightly larger in objects with stronger AGN and larger outflow velocities, but they do not correlate with nebular line reddening. Among the 12 ULIRGs with good continuum signal-to-noise ratios, O VI and/or N V absorption features are robustly detected in 6 of them, all of which are blueshifted, indicative of outflows. In the combined ULIRG $+$ quasar sample, the outflows are more frequently detected in the X-ray weak or absorbed sources. The absorption equivalent widths, velocities and velocity dispersions of the outflows are also higher in the X-ray weak sources. No other strong correlations are visible between the properties of the outflows and those of the AGN or host galaxies.

astro-ph.GA↗

arXiv2021

Galactic Winds across the Gas-Rich Merger Sequence: I. Highly Ionized N V and O VI Outflows in the QUEST Quasars

This program is part of QUEST (Quasar/ULIRG Evolutionary Study) and seeks to examine the gaseous environments of z < 0.3 quasars and ULIRGs as a function of host galaxy properties and age across the merger sequence from ULIRGs to quasars. This first paper in the series focuses on 33 quasars from the QUEST sample and on the kinematics of the highly ionized gas phase traced by the N V 1238, 1243 A and O VI 1032, 1038 A absorption lines in high-quality Hubble Space Telescope (HST) Cosmic Origins Spectrograph (COS) data. N V and O VI outflows are present in about 60% of the QUEST quasars and span a broad range of properties, both in terms of equivalent widths (from 20 mA to 25 A) and kinematics (outflow velocities from a few x 100 km/s up to ~10,000 km/s). The rate of incidence and equivalent widths of the highly ionized outflows are higher among X-ray weak or absorbed sources. The weighted outflow velocity dispersions are highest among the X-ray weakest sources. No significant trends are found between the weighted outflow velocities and the properties of the quasars and host galaxies although this may be due to the limited dynamic range of properties of the current sample. These results will be re-examined in an upcoming paper where the sample is expanded to include the QUEST ULIRGs. Finally, a lower limit of ~0.1% on the ratio of time-averaged kinetic power to bolometric luminosity is estimated in the 2-4 objects with blueshifted P V 1117, 1128 absorption features.

astro-ph.GA↗

arXiv2021

Learning to Align Sequential Actions in the Wild

State-of-the-art methods for self-supervised sequential action alignment rely on deep networks that find correspondences across videos in time. They either learn frame-to-frame mapping across sequences, which does not leverage temporal information, or assume monotonic alignment between each video pair, which ignores variations in the order of actions. As such, these methods are not able to deal with common real-world scenarios that involve background frames or videos that contain non-monotonic sequence of actions. In this paper, we propose an approach to align sequential actions in the wild that involve diverse temporal variations. To this end, we propose an approach to enforce temporal priors on the optimal transport matrix, which leverages temporal consistency, while allowing for variations in the order of actions. Our model accounts for both monotonic and non-monotonic sequences and handles background frames that should not be aligned. We demonstrate that our approach consistently outperforms the state-of-the-art in self-supervised sequential action representation learning on four different benchmark datasets.

cs.CV↗

arXiv2021

Counting People by Estimating People Flows

Modern methods for counting people in crowded scenes rely on deep networks to estimate people densities in individual images. As such, only very few take advantage of temporal consistency in video sequences, and those that do only impose weak smoothness constraints across consecutive frames. In this paper, we advocate estimating people flows across image locations between consecutive images and inferring the people densities from these flows instead of directly regressing them. This enables us to impose much stronger constraints encoding the conservation of the number of people. As a result, it significantly boosts performance without requiring a more complex architecture. Furthermore, it allows us to exploit the correlation between people flow and optical flow to further improve the results. We also show that leveraging people conservation constraints in both a spatial and temporal manner makes it possible to train a deep crowd counting model in an active learning setting with much fewer annotations. This significantly reduces the annotation cost while still leading to similar performance to the full supervision case.

cs.CV↗

arXiv2021

The impact of low luminosity AGN on their host galaxies: A radio and optical investigation of the kpc-scale outflow in MaNGA 1-166919

One way an Active Galactic Nucleus (AGN) influences the evolution of their host galaxy is by generating a large-scale (kpc-scale) outflow. The content, energetics, and impact of such outflows depend on the properties of both the AGN and host galaxy, and understanding the relationship between them requires measuring the properties of all three. In this paper, we do so by analyzing recent radio and optical integral field unit (IFU) spectroscopic observations of MaNGA 1-166919. Our results indicate that the bi-conical outflow in this galaxy is powered by a low-luminosity, low-Eddington ratio AGN ejecting material that drives ~100-200 km/s shocks into the surrounding interstellar medium (ISM) -- producing the hot, ionized gas and relativistic particles associated with the observed outflow. The energetics of the relativistic and ionized gas material produced at this shock are comparable, and both the mass outflow and kinetic power of the ionized gas in this outflow are higher than other AGN with similar bolometric luminosities. Lastly, while the host galaxy's total star formation rate is comparable to that of other star-forming galaxies with a similar stellar mass, there is evidence that the outflow both suppresses and enhances star formation in its immediate surroundings.

astro-ph.GA↗

arXiv2021

Domain Adaptation for Semantic Segmentation via Patch-Wise Contrastive Learning

We introduce a novel approach to unsupervised and semi-supervised domain adaptation for semantic segmentation. Unlike many earlier methods that rely on adversarial learning for feature alignment, we leverage contrastive learning to bridge the domain gap by aligning the features of structurally similar label patches across domains. As a result, the networks are easier to train and deliver better performance. Our approach consistently outperforms state-of-the-art unsupervised and semi-supervised methods on two challenging domain adaptive segmentation tasks, particularly with a small number of target domain annotations. It can also be naturally extended to weakly-supervised domain adaptation, where only a minor drop in accuracy can save up to 75% of annotation cost.

cs.CV↗

arXiv2021

Leveraging Self-Supervision for Cross-Domain Crowd Counting

State-of-the-art methods for counting people in crowded scenes rely on deep networks to estimate crowd density. While effective, these data-driven approaches rely on large amount of data annotation to achieve good performance, which stops these models from being deployed in emergencies during which data annotation is either too costly or cannot be obtained fast enough. One popular solution is to use synthetic data for training. Unfortunately, due to domain shift, the resulting models generalize poorly on real imagery. We remedy this shortcoming by training with both synthetic images, along with their associated labels, and unlabeled real images. To this end, we force our network to learn perspective-aware features by training it to recognize upside-down real images from regular ones and incorporate into it the ability to predict its own uncertainty so that it can generate useful pseudo labels for fine-tuning purposes. This yields an algorithm that consistently outperforms state-of-the-art cross-domain crowd counting ones without any extra computation at inference time.

cs.CV↗

arXiv2021

Near-Infrared Coronal Line Observations of Dwarf Galaxies hosting AGN-driven Outflows

We have obtained Keck NIR spectroscopy of a sample of nine M$_\star$ $<$ 10$^{10}$ M$_\odot$ dwarf galaxies to confirm AGN activity and the presence of galaxy-wide, AGN-driven outflows through coronal line (CL) emission. We find strong CL detections in 5/9 galaxies (55$\%$) with line ratios incompatible with shocks, confirming the presence of AGN in these galaxies. Similar CL detection rates are found in larger samples of more massive galaxies hosting type 1 and 2 AGN. We investigate the connection between the CLs and galaxy-wide outflows by analyzing the kinematics of the CL region, as well as the scaling of gas velocity with ionization potential of different CLs. In addition, using complementary Keck KCWI observations of these objects, we find that the outflow velocities measured in [Si VI] are generally faster than those seen in [O III]. The galaxies with the fastest outflows seen in [O III] also have the highest [Si VI] luminosity. The lack of $J$-band CN absorption features, which are often associated with younger stellar populations, provides further evidence that these outflows are driven by AGN in low mass galaxies.

astro-ph.GA↗

arXiv2020

Integral-Field Spectroscopy of Fast Outflows in Dwarf Galaxies with AGN

Feedback likely plays a vital role in the formation of dwarf galaxies. While stellar processes have long been considered the main source of feedback, recent studies have revealed tantalizing signs of AGN feedback in dwarf galaxies. In this paper, we report the results from an integral-field spectroscopic study of a sample of eight dwarf galaxies with known AGN and suspected outflows. Outflows are detected in seven of them. The outflows are fast, with 50-percentile (median) velocity of up to $\sim$240 km s$^{-1}$ and 80-percentile line width reaching $\sim$1200 km s$^{-1}$, in clear contrast with the more quiescent kinematics of the host gas and stellar components. The outflows are generally spatially extended on a scale of several hundred pc to a few kpc, although our data do not clearly resolve the outflows in three targets. The outflows appear to be primarily photoionized by the AGN rather than shocks or young, massive stars. The kinematics and energetics of these outflows suggest that they are primarily driven by the AGN, although the star formation activity in these objects may also contribute to the energy input. A small but non-negligible portion of the outflowing material likely escapes the main body of the host galaxy and contributes to the enrichment of the circumgalactic medium. Overall, the impact of these outflows on their host galaxies is similar to those taking place in the more luminous AGN in the low-redshift universe.

astro-ph.GA↗

arXiv2020

Estimating People Flows to Better Count Them in Crowded Scenes

Modern methods for counting people in crowded scenes rely on deep networks to estimate people densities in individual images. As such, only very few take advantage of temporal consistency in video sequences, and those that do only impose weak smoothness constraints across consecutive frames. In this paper, we advocate estimating people flows across image locations between consecutive images and inferring the people densities from these flows instead of directly regressing. This enables us to impose much stronger constraints encoding the conservation of the number of people. As a result, it significantly boosts performance without requiring a more complex architecture. Furthermore, it also enables us to exploit the correlation between people flow and optical flow to further improve the results. We will demonstrate that we consistently outperform state-of-the-art methods on five benchmark datasets.

cs.CV↗

arXiv2020

Using Depth for Pixel-Wise Detection of Adversarial Attacks in Crowd Counting

State-of-the-art methods for counting people in crowded scenes rely on deep networks to estimate crowd density. While effective, deep learning approaches are vulnerable to adversarial attacks, which, in a crowd-counting context, can lead to serious security issues. However, attack and defense mechanisms have been virtually unexplored in regression tasks, let alone for crowd density estimation. In this paper, we investigate the effectiveness of existing attack strategies on crowd-counting networks, and introduce a simple yet effective pixel-wise detection mechanism. It builds on the intuition that, when attacking a multitask network, in our case estimating crowd density and scene depth, both outputs will be perturbed, and thus the second one can be used for detection purposes. We will demonstrate that this significantly outperforms heuristic and uncertainty-based strategies.

cs.CV↗

arXiv2019

Geometric and Physical Constraints for Drone-Based Head Plane Crowd Density Estimation

State-of-the-art methods for counting people in crowded scenes rely on deep networks to estimate crowd density in the image plane. While useful for this purpose, this image-plane density has no immediate physical meaning because it is subject to perspective distortion. This is a concern in sequences acquired by drones because the viewpoint changes often. This distortion is usually handled implicitly by either learning scale-invariant features or estimating density in patches of different sizes, neither of which accounts for the fact that scale changes must be consistent over the whole scene. In this paper, we explicitly model the scale changes and reason in terms of people per square-meter. We show that feeding the perspective model to the network allows us to enforce global scale consistency and that this model can be obtained on the fly from the drone sensors. In addition, it also enables us to enforce physically-inspired temporal consistency constraints that do not have to be learned. This yields an algorithm that outperforms state-of-the-art methods in inferring crowd density from a moving drone camera especially when perspective effects are strong.

cs.CV↗

arXiv2019

Context-Aware Crowd Counting

State-of-the-art methods for counting people in crowded scenes rely on deep networks to estimate crowd density. They typically use the same filters over the whole image or over large image patches. Only then do they estimate local scale to compensate for perspective distortion. This is typically achieved by training an auxiliary classifier to select, for predefined image patches, the best kernel size among a limited set of choices. As such, these methods are not end-to-end trainable and restricted in the scope of context they can leverage. In this paper, we introduce an end-to-end trainable deep architecture that combines features obtained using multiple receptive field sizes and learns the importance of each such feature at each image location. In other words, our approach adaptively encodes the scale of the contextual information required to accurately predict crowd density. This yields an algorithm that outperforms state-of-the-art crowd counting methods, especially when perspective effects are strong.

cs.CV↗

arXiv2019

Elliptical Galaxy in the Making: The Dual Active Galactic Nuclei and Metal-enriched Halo of Mrk 273

A systematic analysis of the X-ray emission from the nearby ultraluminous infrared galaxy Mrk 273 was carried out by combining new 200 ksec Chandra data with archived 44 ksec data. The active galactic nucleus (AGN) associated with the Southwest nucleus is confirmed by the new data, and a secondary hard X-ray (4-8 keV) point source is detected, coincident with the Northeast nucleus at a projected distance of 0.75 kpc from the Southwest nucleus. The hard X-ray spectrum of the Northeast nucleus is consistent with a heavily absorbed AGN, making Mrk 273 another example of a dual AGN in a nearby galaxy merger. Significant 1-3 keV emission is found along the ionization cones and outflowing gas detected in a previous study. The data also map the giant X-ray nebula south of the host galaxy with unprecedented detail. This nebula extends on a scale of $\sim$ 40 kpc $\times$ 40 kpc, and is not closely related to the well-known tidal tail seen in the optical. The X-ray emission of the nebula is best described by a single-temperature gas model, with a temperature of $\sim$ 7 million K and a super-solar $α$/Fe ratio. Further analysis suggests that the southern nebula has most likely been heated and enriched by multiple galactic outflows generated by the AGN and/or circumnuclear starburst in the past, on a time scale of $\lesssim$0.1 Gyr, similar to the merger event itself.

astro-ph.GA↗