arXiv subjects

Jin Yu

Publications and source records attributed to Jin Yu.

At least 37 records · Page 2Linked to original sources

arXiv2021

Electronic and Optical properties of transition metal dichalcogenides under symmetric and asymmetric field-effect doping

Doping via electrostatic gating is a powerful and widely used technique to tune the electron densities in layered materials. The microscopic details of how these setups affect the layered material are, however, subtle and call for careful theoretical treatments. Using semiconducting monolayers of transition metal dichalcogenides (TMDs) as prototypical systems affected by electrostatic gating, we show that the electronic and optical properties change indeed dramatically when the gating geometry is properly taken into account. This effect is implemented by a self-consistent calculation of the Coulomb interaction between the charges in different sub-layers within the tight-binding approximation. Thereby we consider both, single- and double-sided gating. Our results show that, at low doping levels of $10^{13}$ cm$^{-2}$, the electronic bands of monolayer TMDs shift rigidly for both types of gating, and subsequently undergo a Lifshitz transition. When approaching the doping level of $10^{14}$ cm$^{-2}$, the band structure changes dramatically, especially in the case of single-sided gating where we find that monolayer \ce{MoS2} and \ce{WS2} become indirect gap semiconductors. The optical conductivities calculated within linear response theory also show clear signatures of these doping-induced band structure renormalizations. Our numerical results based on light-weighted tight-binding models indicate the importance of electronic screening in doped layered structures, and pave the way for further understanding gated super-lattice structures formed by mutlilayers with extended Moiré pattern.

cond-mat.mes-hall↗

arXiv2020

DeVLBert: Learning Deconfounded Visio-Linguistic Representations

In this paper, we propose to investigate the problem of out-of-domain visio-linguistic pretraining, where the pretraining data distribution differs from that of downstream data on which the pretrained model will be fine-tuned. Existing methods for this problem are purely likelihood-based, leading to the spurious correlations and hurt the generalization ability when transferred to out-of-domain downstream tasks. By spurious correlation, we mean that the conditional probability of one token (object or word) given another one can be high (due to the dataset biases) without robust (causal) relationships between them. To mitigate such dataset biases, we propose a Deconfounded Visio-Linguistic Bert framework, abbreviated as DeVLBert, to perform intervention-based learning. We borrow the idea of the backdoor adjustment from the research field of causality and propose several neural-network based architectures for Bert-style out-of-domain pretraining. The quantitative results on three downstream tasks, Image Retrieval (IR), Zero-shot IR, and Visual Question Answering, show the effectiveness of DeVLBert by boosting generalization ability.

cs.CV↗

arXiv2020

Strain-induced semiconductor to metal transition in MA2Z4 bilayers

Very recently, a new type of two-dimensional layered material MoSi2N4 has been fabricated, which is semiconducting with weak interlayer interaction, high strength, and excellent stability. We systematically investigate theoretically the effect of vertical strain on the electronic structure of MA2Z4 (M=Ti/Cr/Mo, A=Si, Z=N/P) bilayers. Taking bilayer MoSi2N4 as an example, our first principle calculations show that its indirect band gap decreases monotonically as the vertical compressive strain increases. Under a critical strain around 22%, it undergoes a transition from semiconductor to metal. We attribute this to the opposite energy shift of states in different layers, which originates from the built-in electric field induced by the asymmetric charge transfer between two inner sublayers near the interface. Similar semiconductor to metal transitions are observed in other strained MA2Z4 bilayers, and the estimated critical pressures to realize such transitions are within the same order as semiconducting transition metal dichalcogenides. The semiconductor to metal transitions observed in the family of MA2Z4 bilayers present interesting possibilities for strain-induced engineering of their electronic properties.

cond-mat.mtrl-sci↗

arXiv2020

Poet: Product-oriented Video Captioner for E-commerce

In e-commerce, a growing number of user-generated videos are used for product promotion. How to generate video descriptions that narrate the user-preferred product characteristics depicted in the video is vital for successful promoting. Traditional video captioning methods, which focus on routinely describing what exists and happens in a video, are not amenable for product-oriented video captioning. To address this problem, we propose a product-oriented video captioner framework, abbreviated as Poet. Poet firstly represents the videos as product-oriented spatial-temporal graphs. Then, based on the aspects of the video-associated product, we perform knowledge-enhanced spatial-temporal inference on those graphs for capturing the dynamic change of fine-grained product-part characteristics. The knowledge leveraging module in Poet differs from the traditional design by performing knowledge filtering and dynamic memory modeling. We show that Poet achieves consistent performance improvement over previous methods concerning generation quality, product aspects capturing, and lexical diversity. Experiments are performed on two product-oriented video captioning datasets, buyer-generated fashion video dataset (BFVD) and fan-generated fashion video dataset (FFVD), collected from Mobile Taobao. We will release the desensitized datasets to promote further investigations on both video captioning and general video analysis problems.

cs.CV↗

arXiv2020

A Multi-Semantic Metapath Model for Large Scale Heterogeneous Network Representation Learning

Network Embedding has been widely studied to model and manage data in a variety of real-world applications. However, most existing works focus on networks with single-typed nodes or edges, with limited consideration of unbalanced distributions of nodes and edges. In real-world applications, networks usually consist of billions of various types of nodes and edges with abundant attributes. To tackle these challenges, in this paper we propose a multi-semantic metapath (MSM) model for large scale heterogeneous representation learning. Specifically, we generate multi-semantic metapath-based random walks to construct the heterogeneous neighborhood to handle the unbalanced distributions and propose a unified framework for the embedding learning. We conduct systematical evaluations for the proposed framework on two challenging datasets: Amazon and Alibaba. The results empirically demonstrate that MSM can achieve relatively significant gains over previous state-of-arts on link prediction.

cs.SI↗

arXiv2020

Comprehensive Information Integration Modeling Framework for Video Titling

In e-commerce, consumer-generated videos, which in general deliver consumers' individual preferences for the different aspects of certain products, are massive in volume. To recommend these videos to potential consumers more effectively, diverse and catchy video titles are critical. However, consumer-generated videos seldom accompany appropriate titles. To bridge this gap, we integrate comprehensive sources of information, including the content of consumer-generated videos, the narrative comment sentences supplied by consumers, and the product attributes, in an end-to-end modeling framework. Although automatic video titling is very useful and demanding, it is much less addressed than video captioning. The latter focuses on generating sentences that describe videos as a whole while our task requires the product-aware multi-grained video analysis. To tackle this issue, the proposed method consists of two processes, i.e., granular-level interaction modeling and abstraction-level story-line summarization. Specifically, the granular-level interaction modeling first utilizes temporal-spatial landmark cues, descriptive words, and abstractive attributes to builds three individual graphs and recognizes the intra-actions in each graph through Graph Neural Networks (GNN). Then the global-local aggregation module is proposed to model inter-actions across graphs and aggregate heterogeneous graphs into a holistic graph representation. The abstraction-level story-line summarization further considers both frame-level video features and the holistic graph to utilize the interactions between products and backgrounds, and generate the story-line topic of the video. We collect a large-scale dataset accordingly from real-world data in Taobao, a world-leading e-commerce platform, and will make the desensitized version publicly available to nourish further development of the research community...

cs.CV↗

arXiv2020

Grounded and Controllable Image Completion by Incorporating Lexical Semantics

In this paper, we present an approach, namely Lexical Semantic Image Completion (LSIC), that may have potential applications in art, design, and heritage conservation, among several others. Existing image completion procedure is highly subjective by considering only visual context, which may trigger unpredictable results which are plausible but not faithful to a grounded knowledge. To permit both grounded and controllable completion process, we advocate generating results faithful to both visual and lexical semantic context, i.e., the description of leaving holes or blank regions in the image (e.g., hole description). One major challenge for LSIC comes from modeling and aligning the structure of visual-semantic context and translating across different modalities. We term this process as structure completion, which is realized by multi-grained reasoning blocks in our model. Another challenge relates to the unimodal biases, which occurs when the model generates plausible results without using the textual description. This can be true since the annotated captions for an image are often semantically equivalent in existing datasets, and thus there is only one paired text for a masked image in training. We devise an unsupervised unpaired-creation learning path besides the over-explored paired-reconstruction path, as well as a multi-stage training strategy to mitigate the insufficiency of labeled data. We conduct extensive quantitative and qualitative experiments as well as ablation studies, which reveal the efficacy of our proposed LSIC.

cs.CV↗

arXiv2019

Tunable magneto-optical properties of single-layer tin diselenide: From GW approximation to large-scale tight-binding calculations

A parameterized tight-binding (TB) model based on the first-principles GW calculations is developed for single layer tin diselenide (SnSe$_2$) and used to study its electronic and optical properties under external magnetic field. The truncated model is derived from six maximally localized wannier orbitals on Se site, which accurately describes the quasi-particle electronic states of single layer SnSe$_2$ in a wide energy range. The quasi-particle electronic states are dominated by the hoppings between nearest wannier orbitals ($t_1$-$t_6$). Our numerical calculation shows that, due to the electron-hole asymmetry, two sets of Landau Level spectrum are obtained when a perpendicular magnetic field is applied. The Landau Level spectrum follows linear dependence on the level index and magnetic field, exhibiting properties of two-dimensional electron gas in traditional semiconductors. The optical conductivity calculation shows that the optical gap is very close to the GW value, and can be tuned by external magnetic field. Our proposed TB model can be used for further exploring the electronic, optical, and transport properties of SnSe$_2$, especially in the presence of external magnetic fields.

physics.comp-ph↗

arXiv2019

Machine Learning Allows Calibration Models to Predict Trace Element Concentration in Soil with Generalized LIBS Spectra

Calibration models have been developed for determination of trace elements, silver for instance, in soil using laser-induced breakdown spectroscopy (LIBS). The major concern is the matrix effect. Although it affects the accuracy of LIBS measurements in a general way, the effect appears accentuated for soil because of large variation of chemical and physical properties among different soils. The purpose is to reduce its influence in such way an accurate and soil-independent calibration model can be constructed. At the same time, the developed model should efficiently reduce experimental fluctuations affecting measurement precision. A univariate model first reveals obvious influence of matrix effect and important experimental fluctuation. A multivariate model has been then developed. A key point is the introduction of generalized spectrum where variables representing the soil type are explicitly included. Machine learning has been used to develop the model. After a necessary pretreatment where a feature selection process reduces the dimension of raw spectrum accordingly to the number of available spectra, the data have been fed in to a back-propagation neuronal networks (BPNN) to train and validate the model. The resulted soilindependent calibration model allows average relative error of calibration (REC) and average relative error of prediction (REP) within the range of 5-6%.

physics.chem-ph↗

arXiv2019

Strain-tunable magnetic and electronic properties of monolayer CrI3

Two-dimensional CrI3 has attracted much attention as it is reported to be a ferromagnetic semiconductor with the Curie temperature around 45K. By performing first-principles calculations, we find that the magnetic ground state of CrI3 is variable under biaxial strain. Our theoretical investigations show that the ground state of monolayer CrI3 is ferromagnetic under compression, but becomes antiferromagnetic under tension. Particularly, the transition occurs under a feasible in-plane strain around 1.8%. Accompanied by the transition of the magnetic ground state, it undergoes a transition from magnetic-metal to half-metal to half-semiconductor to spin-relevant semiconductor when strain varies from -15% to 10%. We attribute these transitions to the variation of the d-orbitals of Cr atoms and the p-orbitals of I atoms. Generally, we report a series of magnetic and electronic phase transition in strained CrI3, which will help both theoretical and experimental researchers for further understanding of the tunable electronic and magnetic properties of CrI3 and their analogous.

physics.comp-ph↗

arXiv2018

Electronic structure of monolayer antimonene nanoribbons under out-of-plane and transverse bias

A systematic study of the electronic properties of single layer Sb (antimonene) nanoribbons is presented. By using a 6-orbital tight-binding Hamiltonian, we study the electronic band structure of finite ribbons with zigzag or armchair termination. We show that there is good agreement between ab initio calculations and the tight-binding model. We study how the size of the gap can be controlled by applying an external bias potential. An electric field applied perpendicular to the antimonene layer is found to increase the band gap, while a transverse bias potential leads to a position dependent reduction of the band gap. Both kinds of bias potential break inversion symmetry of the crystal. This, together with the strong intrinsic spin-orbit coupling of antimonene, leads to spin-splitting of the valence band states.

cond-mat.mes-hall↗

arXiv2018

Tunable electronic and magneto-optical properties of monolayer arsenene from GW approximation to large-scale tight-binding simulations

Monolayers of group VA elements have attracted great attention with the rising of black phosphorus. Here, we derive a simple tight-binding model for monolayer grey arsenic, referred as arsenene (ML-As), based on the first-principles calculations within the partially self-consistent GW0 approach. The resulting band structure derived from the six p-like orbitals coincides with the quasi-particle energy from GW0 calculations with a high accuracy. In the presence of a perpendicular magnetic field, ML-As exhibits two sets of Landau levels linear with respect to the magnetic field and level index. Our numerical calculation of the optical conductivity reveals that the obtained optical gap is very close to the GW0 value and can be effectively tuned by external magnetic field. Thus, our proposed TB model can be used for further large-scale simulations of the electronic, optical and transport properties of ML-As.

cond-mat.mtrl-sci↗

arXiv2018

Effective lattice Hamiltonian for monolayer tin disulphide: tailoring electronic structure with electric and magnetic fields

The electronic properties of monolayer tin dulsulphide (ML-SnS2), a recently synthesized metal dichalcogenide, are studied by a combination of first-principles calculations and tight-binding (TB) approximation. An effective lattice Hamiltonian based on six hybrid sp-like orbitals with trigonal rotation symmetry are proposed to calculate the band structure and density of states for ML-SnS2, which demonstrates good quantitative agreement with relativistic density functional theory calculations in a wide energy range. We show that the proposed TB model can be easily applied to the case of an external electric field, yielding results consistent with those obtained from full Hamiltonian results. In the presence of a perpendicular magnetic field, highly degenerate equidistant Landau levels are obtained, showing typical two-dimensional electron gas behavior. Thus, the proposed TB model provides a simple new way in describing novel properties in ML-SnS2.

physics.comp-ph↗

arXiv2017

Surface magnetism of gallium arsenide nanofilms

Gallium arsenide (GaAs) is the widest used second generation semiconductor with a direct band gap and increasingly used as nanofilms. However, the magnetic properties of GaAs nanofilms have never been studied. Here we find by comprehensive density functional theory calculations that GaAs nanofilms cleaved along the <111> and <100> directions become intrinsically metallic films with strong surface magnetism and magnetoelectric (ME) effect. The surface magnetism and electrical conductivity are realized via a combined effect of transferring charge induced by spontaneous electric-polarization through the film thickness and spin-polarized surface states. The surface magnetism of <111> nanofilms can be significantly and linearly tuned by vertically applied electric field, endowing the nanofilms unexpectedly high ME coefficients, which are tens of times higher than those of ferromagnetic metals and transition metal oxides.

physics.comp-ph↗

arXiv2017

Hyperhoneycomb boron nitride with anisotropic mechanical, electronic and optical properties

Boron nitride structures have excellent thermal and chemical stabilities. Based on state-of-art theoretical calculations, we propose a wide gap semiconducting BN crystal with a three-dimensional hyperhoneycomb structure (Hp-BN), which is both mechanically and thermodynamically stable. Our calculated results show that Hp-BN has a higher bulk modulus and a smaller energy gap as compared to c-BN. Moreover, due to the unique bonding structure, Hp-BN exhibits anisotropic electronic and optical properties. It has great adsorption in the ultraviolet region, but it is highly transparent in the visible and infrared region, suggesting that the Hp-BN crystal could have potential applications in electronic and optical devices.

cond-mat.mtrl-sci↗

arXiv2017

Deciphering intrinsic inter-subunit couplings that lead to sequential hydrolysis of F1-ATPase ring

The rotary sequential hydrolysis of metabolic machine F1-ATPase is a prominent feature to reveal high coordination among multiple chemical sites on the stator F1 ring, which also contributes to tight coupling between the chemical reaction and central γ-shaft rotation. High-speed AFM experiments discovered that the sequential hydrolysis was maintained on the F1 ring even in the absence of the γ rotor. To explore how the intrinsic sequential performance arises, we computationally investigated essential inter-subunit couplings on the hexameric ring of mitochondrial and bacterial F1. We first reproduced the sequential hydrolysis schemes as experimentally detected, by simulating tri-site ATP hydrolysis cycles on the F1 ring upon kinetically imposing inter-subunit couplings to substantially promote the hydrolysis products release. We found that it is key for certain ATP binding and hydrolysis events to facilitate the neighbor-site ADP and Pi release to support the sequential hydrolysis. The kinetically feasible couplings were then scrutinized through atomistic molecular dynamics simulations as well as coarse-grained simulations, in which we enforced targeted conformational changes for the ATP binding or hydrolysis. Notably, we detected the asymmetrical neighbor-site opening that would facilitate the ADP release upon the enforced ATP binding, and computationally captured the complete Pi release through charge hopping upon the enforced neighbor-site ATP hydrolysis. The ATP-hydrolysis triggered Pi release revealed in current TMD simulation confirms a recent prediction made from statistical analyses of single molecule experimental data in regard to the role ATP hydrolysis plays. Our studies, therefore, elucidate both the concerted chemical kinetics and underlying structural dynamics of the inter-subunit couplings that lead to the rotary sequential hydrolysis of the F1 ring.

physics.bio-ph↗

arXiv2015

Computational Investigations on Polymerase Actions in Gene Transcription and Replication Combining Physical Modeling and Atomistic Simulations

Polymerases are protein enzymes that move along nucleic acid chains and catalyze template-based polymerization reactions during gene transcription and replication. The polymerases also substantially improve transcription or replication fidelity through the non-equilibrium enzymatic cycles. We briefly review computational efforts that have been made toward understanding mechano-chemical coupling and fidelity control mechanisms of the polymerase elongation. The polymerases are regarded as molecular information motors during the elongation process. It requires a full spectrum of computational approaches from multiple time and length scales to understand the full polymerase functional cycle. We keep away from quantum mechanics based approaches to the polymerase catalysis due to abundant former surveys, while address only statistical physics modeling approach and all-atom molecular dynamics simulation approach. We organize this review around our own modeling and simulation practices on a single-subunit T7 RNA polymerase, and summarize commensurate studies on structurally similar DNA polymerases. For multi-subunit RNA polymerases that have been intensively studied in recent years, we leave detailed discussions on the simulation achievements to other computational chemical surveys, while only introduce very recently published representative studies, including our own preliminary work on structure-based modeling on yeast RNA polymerase II. In the end, we quickly go through kinetic modeling on elongation pauses and backtracking activities. We emphasize the fluctuation and control mechanisms of the polymerase actions, highlight the non-equilibrium physical nature of the system, and try to bring some perspectives toward understanding replication and transcription regulation from single molecular details to a genome-wide scale.

physics.bio-ph↗

arXiv2015

Computational modeling to elucidate molecular mechanisms of epigenetic memory

How do mammalian cells that share the same genome exist in notably distinct phenotypes, exhibiting differences in morphology, gene expression patterns, and epigenetic chromatin statuses? Furthermore how do cells of different phenotypes differentiate reproducibly from a single fertilized egg? These are fundamental problems in developmental biology. Epigenetic histone modifications play an important role in the maintenance of different cell phenotypes. The exact molecular mechanism for inheritance of the modification patterns over cell generations remains elusive. The complexity comes partly from the number of molecular species and the broad time scales involved. In recent years mathematical modeling has made significant contributions on elucidating the molecular mechanisms of DNA methylation and histone covalent modification inheritance. We will pedagogically introduce the typical procedure and some technical details of performing a mathematical modeling study, and discuss future developments.

q-bio.MN↗