arXiv subjects

Jingzhe Ma

Publications and source records attributed to Jingzhe Ma.

At least 19 recordsLinked to original sources

arXiv2026

Human Identification at a Distance: Challenges, Methods and Results on the Competition HID 2025

Human identification at a distance (HID) is challenging because traditional biometric modalities such as face and fingerprints are often difficult to acquire in real-world scenarios. Gait recognition provides a practical alternative, as it can be captured reliably at a distance. To promote progress in gait recognition and provide a fair evaluation platform, the International Competition on Human Identification at a Distance (HID) has been organized annually since 2020. Since 2023, the competition has adopted the challenging SUSTech-Competition dataset, which features substantial variations in clothing, carried objects, and view angles. No dedicated training data are provided, requiring participants to train their models using external datasets. Each year, the competition applies a different random seed to generate distinct evaluation splits, which reduces the risk of overfitting and supports a fair assessment of cross-domain generalization. While HID 2023 and HID 2024 already used this dataset, HID 2025 explicitly examined whether algorithmic advances could surpass the accuracy limits observed previously. Despite the heightened difficulty, participants achieved further improvements, and the best-performing method reached 94.2% accuracy, setting a new benchmark on this dataset. We also analyze key technical trends and outline potential directions for future research in gait recognition.

cs.CV↗

arXiv2025

OpenGait: A Comprehensive Benchmark Study for Gait Recognition towards Better Practicality

Gait recognition, a rapidly advancing vision technology for person identification from a distance, has made significant strides in indoor settings. However, evidence suggests that existing methods often yield unsatisfactory results when applied to newly released real-world gait datasets. Furthermore, conclusions drawn from indoor gait datasets may not easily generalize to outdoor ones. Therefore, the primary goal of this paper is to present a comprehensive benchmark study aimed at improving practicality rather than solely focusing on enhancing performance. To this end, we developed OpenGait, a flexible and efficient gait recognition platform. Using OpenGait, we conducted in-depth ablation experiments to revisit recent developments in gait recognition. Surprisingly, we detected some imperfect parts of some prior methods and thereby uncovered several critical yet previously neglected insights. These findings led us to develop three structurally simple yet empirically powerful and practically robust baseline models: DeepGaitV2, SkeletonGait, and SkeletonGait++, which represent the appearance-based, model-based, and multi-modal methodologies for gait pattern description, respectively. In addition to achieving state-of-the-art performance, our careful exploration provides new perspectives on the modeling experience of deep gait models and the representational capacity of typical gait modalities. In the end, we discuss the key trends and challenges in current gait recognition, aiming to inspire further advancements towards better practicality. The code is available at https://github.com/ShiqiYu/OpenGait.

cs.CV↗

arXiv2025

On Denoising Walking Videos for Gait Recognition

To capture individual gait patterns, excluding identity-irrelevant cues in walking videos, such as clothing texture and color, remains a persistent challenge for vision-based gait recognition. Traditional silhouette- and pose-based methods, though theoretically effective at removing such distractions, often fall short of high accuracy due to their sparse and less informative inputs. Emerging end-to-end methods address this by directly denoising RGB videos using human priors. Building on this trend, we propose DenoisingGait, a novel gait denoising method. Inspired by the philosophy that "what I cannot create, I do not understand", we turn to generative diffusion models, uncovering how they partially filter out irrelevant factors for gait understanding. Additionally, we introduce a geometry-driven Feature Matching module, which, combined with background removal via human silhouettes, condenses the multi-channel diffusion features at each foreground pixel into a two-channel direction vector. Specifically, the proposed within- and cross-frame matching respectively capture the local vectorized structures of gait appearance and motion, producing a novel flow-like gait representation termed Gait Feature Field, which further reduces residual noise in diffusion features. Experiments on the CCPG, CASIA-B*, and SUSTech1K datasets demonstrate that DenoisingGait achieves a new SoTA performance in most cases for both within- and cross-domain evaluations. Code is available at https://github.com/ShiqiYu/OpenGait.

cs.CV↗

arXiv2024

Learned HDR Image Compression for Perceptually Optimal Storage and Display

High dynamic range (HDR) capture and display have seen significant growth in popularity driven by the advancements in technology and increasing consumer demand for superior image quality. As a result, HDR image compression is crucial to fully realize the benefits of HDR imaging without suffering from large file sizes and inefficient data handling. Conventionally, this is achieved by introducing a residual/gain map as additional metadata to bridge the gap between HDR and low dynamic range (LDR) images, making the former compatible with LDR image codecs but offering suboptimal rate-distortion performance. In this work, we initiate efforts towards end-to-end optimized HDR image compression for perceptually optimal storage and display. Specifically, we learn to compress an HDR image into two bitstreams: one for generating an LDR image to ensure compatibility with legacy LDR displays, and another as side information to aid HDR image reconstruction from the output LDR image. To measure the perceptual quality of output HDR and LDR images, we use two recently proposed image distortion metrics, both validated against human perceptual data of image quality and with reference to the uncompressed HDR image. Through end-to-end optimization for rate-distortion performance, our method dramatically improves HDR and LDR image quality at all bit rates.

eess.IV↗

arXiv2024

Grad-CAMO: Learning Interpretable Single-Cell Morphological Profiles from 3D Cell Painting Images

Despite their black-box nature, deep learning models are extensively used in image-based drug discovery to extract feature vectors from single cells in microscopy images. To better understand how these networks perform representation learning, we employ visual explainability techniques (e.g., Grad-CAM). Our analyses reveal several mechanisms by which supervised models cheat, exploiting biologically irrelevant pixels when extracting morphological features from images, such as noise in the background. This raises doubts regarding the fidelity of learned single-cell representations and their relevance when investigating downstream biological questions. To address this misalignment between researcher expectations and machine behavior, we introduce Grad-CAMO, a novel single-cell interpretability score for supervised feature extractors. Grad-CAMO measures the proportion of a model's attention that is concentrated on the cell of interest versus the background. This metric can be assessed per-cell or averaged across a validation set, offering a tool to audit individual features vectors or guide the improved design of deep learning architectures. Importantly, Grad-CAMO seamlessly integrates into existing workflows, requiring no dataset or model modifications, and is compatible with both 2D and 3D Cell Painting data. Additional results are available at https://github.com/eigenvivek/Grad-CAMO.

eess.IV↗

arXiv2024

BigGait: Learning Gait Representation You Want by Large Vision Models

Gait recognition stands as one of the most pivotal remote identification technologies and progressively expands across research and industry communities. However, existing gait recognition methods heavily rely on task-specific upstream driven by supervised learning to provide explicit gait representations like silhouette sequences, which inevitably introduce expensive annotation costs and potential error accumulation. Escaping from this trend, this work explores effective gait representations based on the all-purpose knowledge produced by task-agnostic Large Vision Models (LVMs) and proposes a simple yet efficient gait framework, termed BigGait. Specifically, the Gait Representation Extractor (GRE) within BigGait draws upon design principles from established gait representations, effectively transforming all-purpose knowledge into implicit gait representations without requiring third-party supervision signals. Experiments on CCPG, CAISA-B* and SUSTech1K indicate that BigGait significantly outperforms the previous methods in both within-domain and cross-domain tasks in most cases, and provides a more practical paradigm for learning the next-generation gait representation. Finally, we delve into prospective challenges and promising directions in LVMs-based gait recognition, aiming to inspire future work in this emerging topic. The source code is available at https://github.com/ShiqiYu/OpenGait.

cs.CV↗

arXiv2024

Pedestrian Attribute Editing for Gait Recognition and Anonymization

As a kind of biometrics, the gait information of pedestrians has attracted widespread attention from both industry and academia since it can be acquired from long distances without the cooperation of targets. In recent literature, this line of research has brought exciting chances along with alarming challenges: On the positive side, gait recognition used for security applications such as suspect retrieval and safety checks is becoming more and more promising. On the negative side, the misuse of gait information may lead to privacy concerns, as lawbreakers can track subjects of interest using gait characteristics even under face-masked and clothes-changed scenarios. To handle this double-edged sword, we propose a gait attribute editing framework termed GaitEditor. It can perform various degrees of attribute edits on real gait sequences while maintaining the visual authenticity, respectively used for gait data augmentation and de-identification, thereby adaptively enhancing or degrading gait recognition performance according to users' intentions. Experimentally, we conduct a comprehensive evaluation under both gait recognition and anonymization protocols on three widely used gait benchmarks. Numerous results illustrate that the adaptable utilization of GaitEditor efficiently improves gait recognition performance and generates vivid visualizations with de-identification to protect human privacy. To the best of our knowledge, GaitEditor is the first framework capable of editing multiple gait attributes while simultaneously benefiting gait recognition and gait anonymization. The source code of GaitEditor will be available at https://github.com/ShiqiYu/OpenGait.

cs.CV↗

arXiv2024

An Identity-Preserved Framework for Human Motion Transfer

Human motion transfer (HMT) aims to generate a video clip for the target subject by imitating the source subject's motion. Although previous methods have achieved good results in synthesizing good-quality videos, they lose sight of individualized motion information from the source and target motions, which is significant for the realism of the motion in the generated video. To address this problem, we propose a novel identity-preserved HMT network, termed \textit{IDPres}. This network is a skeleton-based approach that uniquely incorporates the target's individualized motion and skeleton information to augment identity representations. This integration significantly enhances the realism of movements in the generated videos. Our method focuses on the fine-grained disentanglement and synthesis of motion. To improve the representation learning capability in latent space and facilitate the training of \textit{IDPres}, we introduce three training schemes. These schemes enable \textit{IDPres} to concurrently disentangle different representations and accurately control them, ensuring the synthesis of ideal motions. To evaluate the proportion of individualized motion information in the generated video, we are the first to introduce a new quantitative metric called Identity Score (\textit{ID-Score}), motivated by the success of gait recognition methods in capturing identity information. Moreover, we collect an identity-motion paired dataset, $Dancer101$, consisting of solo-dance videos of 101 subjects from the public domain, providing a benchmark to prompt the development of HMT methods. Extensive experiments demonstrate that the proposed \textit{IDPres} method surpasses existing state-of-the-art techniques in terms of reconstruction accuracy, realistic motion, and identity preservation.

cs.CV↗

arXiv2023

SkeletonGait: Gait Recognition Using Skeleton Maps

The choice of the representations is essential for deep gait recognition methods. The binary silhouettes and skeletal coordinates are two dominant representations in recent literature, achieving remarkable advances in many scenarios. However, inherent challenges remain, in which silhouettes are not always guaranteed in unconstrained scenes, and structural cues have not been fully utilized from skeletons. In this paper, we introduce a novel skeletal gait representation named skeleton map, together with SkeletonGait, a skeleton-based method to exploit structural information from human skeleton maps. Specifically, the skeleton map represents the coordinates of human joints as a heatmap with Gaussian approximation, exhibiting a silhouette-like image devoid of exact body structure. Beyond achieving state-of-the-art performances over five popular gait datasets, more importantly, SkeletonGait uncovers novel insights about how important structural features are in describing gait and when they play a role. Furthermore, we propose a multi-branch architecture, named SkeletonGait++, to make use of complementary features from both skeletons and silhouettes. Experiments indicate that SkeletonGait++ outperforms existing state-of-the-art methods by a significant margin in various scenarios. For instance, it achieves an impressive rank-1 accuracy of over 85% on the challenging GREW dataset. All the source code is available at https://github.com/ShiqiYu/OpenGait.

cs.CV↗

arXiv2023

Extended hard X-ray emission in highly obscured AGN

Kilo-parsec scale hard ($>$ 3 keV) X-ray continuum and fluorescent Fe K$α$ line emission has been recently discovered in nearby Compton-thick (CT) active galactic nuclei (AGN), which opens new opportunities to improve AGN torus modeling and investigate how the central supermassive black hole interacts with and impacts the host galaxy. Following a pilot Chandra survey of nearby CT AGN, we present in this paper the Chandra spatial analysis results of five uniformly selected non-CT but still heavily obscured AGN to investigate the extended hard X-ray emission by measuring the excess emission counts, excess fractions, and physical scales. Three of them show extended emission in the 3.0-7.0 keV band detected at $>$ 3$σ$ above the Chandra PSF with total excess fractions ranging from $\sim$8% - 20%. The extent of the hard emission ranges from at least $\sim$250 pc to 1.1 kpc in radius. We compare these new sources with CT AGN and find that CT AGN appear to be more extended in the hard band than the non-CT AGN. Similar to CT AGN, the amounts of extended hard X-ray emission relative to the total emission of these obscured AGN are not negligible. Together with other extended hard X-ray detected AGN in the literature, we further explore potential correlations between the extended hard X-ray component and AGN parameters. We also discuss the implications for torus modeling and AGN feedback. Considering potential contributions from X-ray binaries (XRBs) to the extended emission, we do not see strong XRB contamination in the overall sample.

astro-ph.HE↗

arXiv2022

Gas Phase Metallicities of Local Ultra-Luminous Infrared Galaxies Follow Normal Star-Forming Galaxies

Despite advances in observational data, theoretical models, and computational techniques to simulate key physical processes in the formation and evolution of galaxies, the stellar mass assembly of galaxies still remains an unsolved problem today. Optical spectroscopic measurements appear to show that the gas-phase metallicities of local ultra-luminous infrared galaxies (ULIRGs) are significantly lower than those of normal star-forming galaxies. This difference has resulted in the claim that ULIRGs are fueled by metal-poor gas accretion from the outskirts\cite{Mannucci10}. Here we report on a new set of gas-phase metallicity measurements making use of the far-infrared spectral lines of [O{\sc iii}]52 $μ$m, [O{\sc iii}]88 $μ$m, and [N{\sc iii}]57 $μ$m instead of the usual optical lines. Photoionization models have resulted in a metallicity diagnostic based on these three lines that break the electron density degeneracy and reduce the scatter of the correlation significantly. Using new data from SOFIA and archival data from Herschel Space Observatory, we find that local ULIRGs lie on the mass-metallicity relation of star-forming galaxies and have metallicities comparable to other galaxies with similar stellar masses and star formation rates. The lack of a departure suggests that ULIRGs follow the same mass assembly mechanism as luminous star-forming galaxies and $\sim 0.3$ dex under-abundance in metallicities derived from optical lines is a result of heavily obscured metal-rich gas which has a negligible effect when using the FIR line diagnostics.

astro-ph.GA↗

arXiv2021

A Giant Loop of Ionized Gas Emerging from the Tumultuous Central Region of IC 5063

The biconical radiation pattern extending from an active galactic nucleus (AGN) may strongly photoionize the circumnuclear interstellar medium (ISM) and stimulate emission from the narrow line region (NLR). Observations of the NLR may provide clues to the structure of dense material that preferentially obscures the bicone at certain angles, and may reveal the presence of processes in the ISM tied to AGN accretion and feedback. Ground-based integral field units (IFUs) may study these processes via well-understood forbidden diagnostic lines such as [O III] and [S II], but scales of $\sim10$s of pc remain challenging to spatially resolve at these wavelengths for all but the nearest AGN. We present recent narrow filter Hubble Space Telescope (HST) observations of diagnostic forbidden ([O III], [S II]) and Balmer (H$α$, H$β$) lines in the NLR of IC 5063. This AGN's jet inclination into the plane of the galaxy provides an important laboratory for strong AGN-host interactions. We find evidence for a low-ionization loop which emits brightly in [S II] and [N II], and which may arise from plume-like hot outflows that ablate ISM from the galactic plane before escaping laterally. We also present spatially resolved Baldwin-Phillips-Terlevich diagnostic maps of the IC 5063 NLR. These maps suggest a sharp transition to lower-ionization states outside the jet path, and that such emission is dominated by $\sim10-40$ pc clumps and filamentary structure at large (>>25°) angles from the bicone axis. Such emission may arise from precursorless shocks when AGN outflows impact low-density hot plasma in the cross-cone.

astro-ph.GA↗

arXiv2020

Spatially resolved BPT mapping of nearby Seyfert 2 galaxies

We present spatially resolved BPT mapping of the extended narrow line regions (ENLRs) of seven nearby Seyfert 2 galaxies, using HST narrow band filter imaging. We construct the BPT diagrams using $\leq$ 0.1" resolution emission line images of [O III]$λ$5007, H$α$, [S II]$λ$$λ$6717,6731, and H$β$. By mapping these diagnostic lines according to the BPT classification, we dissect the ENLR into Seyfert, LINER, and star-forming regions. The nucleus and ionization cones are dominated by Seyfert-type emission, which can be interpreted as predominantly photoionization by the active galactic nucleus (AGN). The Seyfert nucleus and ionization cones transition to and are surrounded by a LINER cocoon, extending up to $\sim$ 250 pc in thickness. The ubiquity of the LINER cocoon in Seyfert 2 galaxies suggests that the circumnuclear regions are not necessarily Seyfert-type, and LINER activity plays an important role in Seyfert 2 galaxies. We demonstrate that spatially resolved diagnostics are crucial to understanding the excitation mechanisms in different regions and the AGN-host galaxy interactions.

astro-ph.GA↗

arXiv2020

Is extended hard X-ray emission ubiquitous in Compton-thick AGN?

The recent Chandra discovery of extended $\sim$kpc-scale hard ($>$ 3 keV) X-ray emission in nearby Compton-thick (CT) active galactic nuclei (AGN) opens a new window to improving AGN torus modeling and investigating how the central super massive black hole interacts with and impacts the host galaxy. Since there are only a handful of detections so far, we need to establish a statistical sample to determine the ubiquity of the extended hard X-ray emission in CT AGN, and quantify the amount and extent of this component. In this paper, we present the spatial analysis results of a pilot Chandra imaging survey of 7 nearby ($0.006 < z < 0.013$) CT AGN selected from the Swift-BAT spectroscopic AGN survey. We find that five out of the seven CT AGN show extended emission in the 3-7 keV band detected at $>$ 3$σ$ above the Chandra PSF with $\sim$12% to 22% of the total emission in the extended components. ESO 137-G034 and NGC 3281 display biconical ionization structures with extended hard X-ray emission reaching kpc-scales ($\sim$ 1.9 kpc and 3.5 kpc in diameter). The other three show extended hard X-ray emission above the PSF out to at least $\sim$360 pc in radius. We find a trend that a minimum 3-7 keV count rate of 0.01 cts/s and total excess fraction $>$20% is required to detect a prominent extended hard X-ray component. Given that this extended hard X-ray component appears to be relatively common in this uniformly selected CT AGN sample, we further discuss the implications for torus modeling and AGN feedback.

astro-ph.GA↗

arXiv2020

Emergence of an Ultra-Red Ultra-Massive Galaxy Cluster Core at $z=4$

Recent simulations and observations of massive galaxy cluster evolution predict that the majority of stellar mass build up happens within cluster members by $z=2$, before cluster virialization. Protoclusters rich with dusty, star-forming galaxies (DSFGs) at $z>3$ are the favored candidate progenitors for these massive galaxy clusters at $z\sim0$. We present here the first study analyzing stellar emission along with cold dust and gas continuum emission in a spectroscopically confirmed $z=4.002$ protocluster core rich with DSFGs, the Distant Red Core (DRC). We combine new HST and Spitzer data with existing Gemini, Herschel, and ALMA observations to derive individual galaxy-level properties, and compare them to coeval field and other protocluster galaxies. All of the protocluster members are massive ($>10^{10}$ M$_\odot$), but not significantly more so than their coeval field counterparts. Within uncertainty, all are nearly indistinguishable from galaxies on the star-forming vs. stellar mass main-sequence relationship, and on the star formation efficiency plane. Assuming no future major influx of fresh gas, we estimate that these gaseous DSFGs will deplete their gas reservoirs in $\sim300$ Myr, becoming the massive quiescent ellipticals dominating cluster cores by $z\sim3$. Using various methodologies, we derive a total $z=4$ halo mass of $\sim10^{14}$ M$_\odot$, and estimate that the DRC will evolve to become an ultra-massive cluster core of mass $\gtrsim10^{15}$ M$_\odot$ by $z=0$.

astro-ph.GA↗

arXiv2019

Spitzer catalog of Herschel-selected ultrared dusty, star-forming galaxies

The largest Herschel extragalactic surveys, H-ATLAS and HerMES, have selected a sample of "ultrared" dusty, star-forming galaxies (DSFGs) with rising SPIRE flux densities ($S_{500} > S_{350} > S_{250}$; so-called "500 $μ$m-risers") as an efficient way for identifying DSFGs at higher redshift ($z > 4$). In this paper, we present a large Spitzer follow-up program of 300 Herschel ultrared DSFGs. We have obtained high-resolution ALMA, NOEMA, and SMA data for 63 of them, which allow us to securely identify the Spitzer/IRAC counterparts and classify them as gravitationally lensed or unlensed. Within the 63 ultrared sources with high-resolution data, $\sim$65% appear to be unlensed, and $\sim$27% are resolved into multiple components. We focus on analyzing the unlensed sample by directly performing multi-wavelength spectral energy distribution (SED) modeling to derive their physical properties and compare with the more numerous $z \sim 2$ DSFG population. The ultrared sample has a median redshift of 3.3, stellar mass of 3.7 $\times$ 10$^{11}$ $M_{\odot}$, star formation rate (SFR) of 730 $M_{\odot}$yr$^{-1}$, total dust luminosity of 9.0 $\times$ 10$^{12}$ $L_{\odot}$, dust mass of 2.8 $\times$ 10$^9$ $M_{\odot}$, and V-band extinction of 4.0, which are all higher than those of the ALESS DSFGs. Based on the space density, SFR density, and stellar mass density estimates, we conclude that our ultrared sample cannot account for the majority of the star-forming progenitors of the massive, quiescent galaxies found in infrared surveys. Our sample contains the rarer, intrinsically most dusty, luminous and massive galaxies in the early universe that will help us understand the physical drivers of extreme star formation.

astro-ph.GA↗

arXiv2019

Infrared Contributions of X-Ray Selected Active Galactic Nuclei in Dusty Star-Forming Galaxies

We investigate the infrared contribution from supermassive black hole activity versus host galaxy emission in the mid to far-infrared (IR) spectrum for a large sample of X-ray bright active galactic nuclei (AGN) residing in dusty, star-forming host galaxies. We select 703 AGN with L_X = 10^42-46 ergs/s at 0.1 < z < 5 from the Chandra XBootes X-ray Survey with rich multi-band observations in the optical to far-IR. This is the largest sample to date of X-ray AGN with mid and far-IR detections that uses spectral energy distribution (SED) decomposition to determine intrinsic AGN and host galaxy infrared luminosities. We determine weak or nonexistent relationships when averaging star-formation activity as a function of AGN activity, but see stronger positive trends when averaging L_X in bins of star-forming activity for AGN at low redshifts. We estimate an average dust covering factor of 33% based on infrared SEDs and bolometric AGN luminosity, corresponding to a Type 2 AGN population of roughly a third. We also see a population of AGN that challenge the inclination based unification model with individual dust covering factors that contradict the nuclear obscuration expected from observed X-ray hardness ratios. We see no strong connection between AGN fractions in the IR and corresponding total infrared, 24 um, or X-ray luminosities. The average rest-frame AGN contribution as a function of IR wavelength shows significant (~80%) contributions in the mid-IR that trail off at lambda > 30 um. Additionally, we provide a relation between observed L_X and pure AGN IR output for high-z AGN allowing future studies to estimate AGN infrared contribution using only observed X-ray flux density estimates.

astro-ph.GA↗

arXiv2019

Imaging the Molecular Interstellar Medium in a Gravitationally Lensed Star-forming Galaxy at z=5.7

Aims: We present and study spatially resolved imaging obtained with the Atacama Large Millimeter/submillimeter Array (ALMA) of multiple $^{12}$CO($J=$6$-$5, 8$-$7 and 9$-$8) and two H$_2$O(2$_{02}-$1$_{11}$ and 2$_{11}-$2$_{02}$) emission lines and cold dust continuum toward the gravitationally lensed dusty star forming galaxy SPT0346-52 at z=$5.656$. Methods: Using a visibility-domain source-plane reconstruction we probe the structure and dynamics of the different components of the interstellar medium (ISM) in this galaxy down to scales of 1 kpc in the source plane. Results: Measurements of the intrinsic sizes of the different CO emission lines indicate that the higher J transitions trace more compact regions in the galaxy. Similarly, we find smaller dust continuum intrinsic sizes with decreasing wavelength, based on observations at rest-frame 130, 300 and 450$μ$m. The source shows significant velocity structure, and clear asymmetry where an elongated structure is observed in the source plane with significant variations in their reconstructed sizes. This could be attributed to a compact merger or turbulent disk rotation. The differences in velocity structure through the different line tracers, however, hint at the former scenario in agreement with previous [CII] line imaging results. Measurements of the CO line ratios and magnifications yield significant variations as a function of velocity, suggesting that modeling of the ISM using integrated values could be misinterpreted. Modeling of the ISM in SPT0346-52 based on delensed fluxes indicate a highly dense and warm medium, qualitatively similar to that observed in high redshift quasar hosts.

astro-ph.GA↗