arXiv subjects

Hong Jiang

Publications and source records attributed to Hong Jiang.

At least 37 records · Page 2Linked to original sources

arXiv2025

Rapid Single-Cell Measurement of Transient Transmembrane Water Flow under Osmotic Gradient

While aquaporin (AQP) gating dynamically regulates transmembrane water permeability for cellular homeostasis, its mechanisms remain poorly understood compared to ion channels. A central challenge is the lack of methods to measure water flow through AQPs with the spatiotemporal resolution and sensitivity equivalent to patch-clamp recordings of ion fluxes, a limitation stemming from the electrically silent nature of water transport. We introduce a technique to rapidly detect cytoplasmic flows induced by osmotic-gradient-driven transmembrane water transport in single adherent human cancer cells. This approach enables direct measurement of AQP-mediated water transport and provides a powerful tool to investigate AQP function and regulation and cytoplasmic flow dynamics at the single-cell level.

physics.bio-ph↗

arXiv2025

Insert Anything: Image Insertion via In-Context Editing in DiT

This work presents Insert Anything, a unified framework for reference-based image insertion that seamlessly integrates objects from reference images into target scenes under flexible, user-specified control guidance. Instead of training separate models for individual tasks, our approach is trained once on our new AnyInsertion dataset--comprising 120K prompt-image pairs covering diverse tasks such as person, object, and garment insertion--and effortlessly generalizes to a wide range of insertion scenarios. Such a challenging setting requires capturing both identity features and fine-grained details, while allowing versatile local adaptations in style, color, and texture. To this end, we propose to leverage the multimodal attention of the Diffusion Transformer (DiT) to support both mask- and text-guided editing. Furthermore, we introduce an in-context editing mechanism that treats the reference image as contextual information, employing two prompting strategies to harmonize the inserted elements with the target scene while faithfully preserving their distinctive features. Extensive experiments on AnyInsertion, DreamBooth, and VTON-HD benchmarks demonstrate that our method consistently outperforms existing alternatives, underscoring its great potential in real-world applications such as creative content generation, virtual try-on, and scene composition.

cs.CV↗

arXiv2025

ElaLoRA: Elastic & Learnable Low-Rank Adaptation for Efficient Model Fine-Tuning

Low-Rank Adaptation (LoRA) has become a widely adopted technique for fine-tuning large-scale pre-trained models with minimal parameter updates. However, existing methods rely on fixed ranks or focus solely on either rank pruning or expansion, failing to adapt ranks dynamically to match the importance of different layers during training. In this work, we propose ElaLoRA, an adaptive low-rank adaptation framework that dynamically prunes and expands ranks based on gradient-derived importance scores. To the best of our knowledge, ElaLoRA is the first method that enables both rank pruning and expansion during fine-tuning. Experiments across multiple benchmarks demonstrate that ElaLoRA consistently outperforms existing PEFT methods across different parameter budgets. Furthermore, our studies validate that layers receiving higher rank allocations contribute more significantly to model performance, providing theoretical justification for our adaptive strategy. By introducing a principled and adaptive rank allocation mechanism, ElaLoRA offers a scalable and efficient fine-tuning solution, particularly suited for resource-constrained environments.

cs.LG↗

arXiv2025

Density-Matrix Embedding Based Multi-Configurational Perturbation Theory Approach to Single-Ion Magnets

Multi-configurational wave-function theory (MC-WFT) that combines complete active space self-consistent field (CASSCF) approach with subsequent state interaction (SI) treatment of spin-orbit coupling (SOC), abbreviated as CASSCF-SO, plays important roles in microscopic understanding of single-ion magnets (SIMs) with different central transition metal or lanthanide ions and various coordination environments, but its application to SIMs with complex structure is severely limited due to its highly demanding computational cost. Density-matrix embedding theory (DMET) provides a systematic and mathematically rigorous framework to combine low-level mean field approaches like Hartree-Fock and high-level MC-WFT methods like CASSCF-SO, which is particularly promising to SIMs. As a continuation of our previous work on DMET+CASSCF for $3d$ SIMs (Ai, Sun, and Jiang, J. Phys. Chem. Lett. 2022, 13, 10627), we extend the methodology by considering dynamic correlation on top of CASSCF using the second-order $n$-electron valence perturbation theory (NEVPT2) in the DMET framework, abbreviated as DMET+NEVPT2, and benchmark the accuracy of this approach to molecular magnetic anisotropy in a set of typical transition metal complexes. We found that DMET+NEVPT2 can give the results very close to all-electron treatment, and can be systematically improved for higher accuracy by expanding the region treated as the central cluster, while the computation cost is dramatically reduced due to the reduction of the number of orbitals by DMET construction. Our findings suggest that DMET is capable of accounting for most of the dynamic correlation that is important for magnetic anisotropy in typical SIMs, and can be useful for further high-accuracy spin-phonon study and high-throughput computations.

physics.chem-ph↗

arXiv2025

Density Matrix Embedding Theory-Based Multi-Configurational Quantum Chemistry Approach to Lanthanide Single-Ion Magnets

Accurate and efficient theoretical descriptions of lanthanide systems based on ab initio electronic structure theory remain highly challenging due to the complex interplay of strong electronic correlation and significant relativistic effects in 4f electrons. The composite multi-configurational quantum chemistry method, which combines the complete active space self-consistent field (CASSCF) approach with subsequent state interaction (SI) treatment of spin-orbit coupling (SOC), abbreviated as CASSI-SO, has emerged as the preferred method for ab initio studies of lanthanide systems. However, its widespread application is hindered by its substantial computational cost. Building on the success of integrating density-matrix embedding theory (DMET) with CASSI-SO in our previous theoretical study of 3d single-ion magnets (SIMs) (Ai, Sun, and Jiang, J. Phys. Chem. Lett. 2022, 13, 10627), we now extend the DMET+CASSI-SO approach to lanthanide SIM systems. We provide a detailed formulation of the regularized direct inversion of iterative subspace (R-DIIS) algorithm, which ensures obtaining physically correct restricted open-shell Hartree-Fock (ROHF) wavefunctions, a critical factor for the effectiveness of DMET. Additionally, we introduce the subspace R-DIIS (sR-DIIS) algorithm, which proves to be more efficient and robust for lanthanide systems. Using several representative lanthanide single-ion magnets (4f-SIMs) as test cases, we demonstrate the performance of these new algorithms and highlight the exceptional accuracy of the DMET+CASSI-SO approach. We anticipate that this enhanced DMET+CASSI-SO methodology will significantly advance large-scale theoretical investigations of complex lanthanide systems.

physics.chem-ph↗

arXiv2024

MAP-UOT: A Memory-Efficient Approach to Unbalanced Optimal Transport Implementation

Unbalanced optimal transport (UOT) has been widely used as a fundamental tool in many application domains, where it often dominates the application running time. While many researchers have proposed various optimizations for UOT, few have attempted to optimize it from a computer architecture's perspective. In this paper, we first study the performance bottlenecks of UOT through a series of experiments, which reveals that UOT is heavily memory-bound. Guided by these findings, we propose MAP-UOT, a Memory-efficient APproach to the implementation and optimization of UOT on CPU and GPU platforms. Our experimental evaluations show that the proposed strategy consistently and significantly outperforms the state-of-the-art (SOTA) implementations. Specifically, it provides single-threaded performance improvement over POT/COFFEE by up to 2.9X/2.4X, with an average of 1.9X/1.6X. At the same time, it provides parallelized performance improvement over POT/COFFEE by up to 2.4X/1.9X, with an average of 2.2X/1.8X, on Intel Core i9-12900K; and over POT by up to 3.5X, with an average of 1.6X, on Nvidia GeForce RTX 3090 Ti. MAP-UOT also shows great performance improvement on the Tianhe-1 supercomputer.

cs.DC↗

arXiv2024

Optimal coloring and strain-enhanced superconductivity in Li$_n$B$_{n+1}$C$_{n-1}$

Boron-rich lithium borocarbides are promising candidates for phonon-mediated high-temperature superconductors due to their metallic $σ$-bonding electrons. Here, we use the cluster expansion method to identify energetically stable configurations (colorings) of Li$_2$B$_3$C and Li$_3$B$_4$C$_2$, which are characterized by a distinctive pattern of alternating B-B and B-C zigzag chains. Surprisingly, the optimal configuration of Li$_2$B$_3$C exhibits an extremely low superconducting transition temperature of $T_c < 0.03$ K, which is attributed to the suppression of deformation potentials near the Fermi level caused by the specific electron filling of B-B zigzag chains. However, the $σ$-bonding electrons at the Fermi level are highly sensitive to external strain or pressure. Specifically, applying a -5\% compressive uniaxial strain can significantly enhance the electron-phonon coupling and the Eliashberg spectral function, boosting up $T_c$ to 37 K. This work not only presents a novel strategy for achieving phonon-mediated high-temperature superconductivity in Li$_n$B$_{n+1}$C$_{n-1}$ compounds but also provides valuable insights into the complex interplay between electronic structure and superconducting interaction.

cond-mat.supr-con↗

arXiv2024

Temperature-dependent Electronic Spectral Functions from Band-Structure Unfolding

The electronic band structure, describing the periodic dependence of electronic quantum states on lattice momentum in reciprocal space, is a fundamental concept in solid-state physics. However, it's only well-defined for static nuclei. To account for thermodynamic effects, this concept must be generalized by introducing the temperature-dependent spectral function, which characterizes the finite-width distributions of electronic quantum states at each reciprocal vector. Many-body perturbation theory can compute spectral functions and associated observables, but it approximates the dynamics of nuclei and its coupling to the electrons using the harmonic approximation and linear-order electron-phonon coupling elements, respectively. These approximations may fail at elevated temperatures or for mobile atoms. To avoid inaccuracies, the electronic spectral function can be obtained non-perturbatively, capturing higher-order couplings between electrons and vibrational degrees of freedom. This process involves recovering the representation of supercell bands in the first Brillouin zone of the primitive cell, a process known as unfolding. In this contribution, we describe the implementation of the band-structure unfolding technique in the electronic-structure theory package FHI-aims and the updates made since its original development.

cond-mat.mtrl-sci↗

arXiv2024

Towards a General Market for Cloud-Edge-IoT Continuum

Recent years have witnessed the proposals aiming at enabling Vertical, two-sided markets with a Single Marketplace (or exchange) (VSMs) for computing and data resources/services (products) offerings in a multi-cloud and crowdsourced IoT-edge sensing environment. A VSM is designed vertically from bottom up with a broker being a built-in component of the marketplace. While preventing seller lock-in and improving efficiency and availability, a VSM suffers from a key weakness from a buyer's perspective, i.e., the broker and the corresponding marketplace lock-in, which may lead to suboptimal shopping experience for buyers, due to marketplace monopoly by the broker and limited choice of products in the marketplace. In this position paper, we argue that a Horizontal two-sided market with Multiple Marketplaces (HMM), resembling the global stock market, should be developed. In an HMM, different marketplaces may be operated by different parties and sell similar and/or different types of products, e.g., computing and/or sensory data products. A broker is no longer a built-in component of any given marketplace. Instead, it may cover multiple marketplaces at the same time and there can be more than one broker in the HMM. Both the number and types of marketplaces and brokers may grow independently or scale horizontally to meet the growing demand. A buyer shops for a broker through whom the buyer gains access to the needed products sold in the marketplace(s) the broker covers and from whom the buyer receives various possible services, e.g., discount, value-added, or full services. An HMM not only overcomes the key weakness of a VSM but also allows the market to grow incrementally and organically. Finally, two example use cases are given to illustrate the benefits of HMM.

eess.SY↗

arXiv2024

Improving density matrix electronic structure method by deep learning

The combination of deep learning and ab initio materials calculations is emerging as a trending frontier of materials science research, with deep-learning density functional theory (DFT) electronic structure being particularly promising. In this work, we introduce a neural-network method for modeling the DFT density matrix, a fundamental yet previously unexplored quantity in deep-learning electronic structure. Utilizing an advanced neural network framework that leverages the nearsightedness and equivariance properties of the density matrix, the method demonstrates high accuracy and excellent generalizability in multiple example studies, as well as capability to precisely predict charge density and reproduce other electronic structure properties. Given the pivotal role of the density matrix in DFT as well as other computational methods, the current research introduces a novel approach to the deep-learning study of electronic structure properties, opening up new opportunities for deep-learning enhanced computational materials study.

physics.comp-ph↗

arXiv2024

Influence Maximization in Hypergraphs by Stratified Sampling for Efficient Generation of Reverse Reachable Sets

Given a hypergraph, influence maximization (IM) is to discover a seed set containing $k$ vertices that have the maximal influence. Although the existing vertex-based IM algorithms perform better than the hyperedge-based algorithms by generating random reverse researchable (RR) sets, they are inefficient because (i) they ignore important structural information associated with hyperedges and thus obtain inferior results, (ii) the frequently-used sampling methods for generating RR sets have low efficiency because of a large number of required samplings along with high sampling variances, and (iii) the vertex-based IM algorithms have large overheads in terms of running time and memory costs. To overcome these shortcomings, this paper proposes a novel approach, called \emph{HyperIM}. The key idea behind \emph{HyperIM} is to differentiate structural information of vertices for developing stratified sampling combined with highly-efficient strategies to generate the RR sets. With theoretical guarantees, \emph{HyperIM} is able to accelerate the influence spread, improve the sampling efficiency, and cut down the expected running time. To further reduce the running time and memory costs, we optimize \emph{HyperIM} by inferring the bound of the required number of RR sets in conjunction with stratified sampling. Experimental results on real-world hypergraphs show that \emph{HyperIM} is able to reduce the number of required RR sets and running time by orders of magnitude while increasing the influence spread by up to $2.73X$ on average, compared to the state-of-the-art IM algorithms.

cs.SI↗

arXiv2024

LearnedFTL: A Learning-Based Page-Level FTL for Reducing Double Reads in Flash-Based SSDs

We present LearnedFTL, a new on-demand page-level flash translation layer (FTL) design, which employs learned indexes to improve the address translation efficiency of flash-based SSDs. The first of its kind, it reduces the number of double reads induced by address translation in random read accesses. LearnedFTL proposes three key techniques: an in-place-update linear model to build learned indexes efficiently, a virtual PPN representation to obtain contiguous PPNs for sorted LPNs, and a group-based allocation and model training via GC/rewrite strategy to reduce the training overhead. By tightly integrating the aforementioned key techniques, LearnedFTL considerably speeds up address translation while reducing the number of flash read accesses caused by the address translation. Our extensive experiments on a FEMU-based prototype show that LearnedFTL can reduce up to 55.5\% address translation-induced double reads. As a result, LearnedFTL reduces the P99 tail latency by 2.9$\times$ $\sim$ 12.2$\times$ with an average of 5.5$\times$ and 8.2$\times$ compared to the state-of-the-art TPFTL and LeaFTL schemes, respectively.

cs.AR↗

arXiv2023

Insomnia impairs muscle function via regulating protein degradation and muscle clock

Background: Insomnia makes people more physically unable of doing daily duties, which results in a lack of strength, leads to lacking in strength. However, the effects of insomnia on muscle function have not yet been thoroughly investigated. So, the objectives of this study were to clarify how insomnia contributes to the decrease of muscular function and to investigate the mechanisms behind this phenomenon. Methods: To understand how insomnia influence muscle function, we analyzed the expression level of factors associated with muscle protein degradation, muscle protein synthesis , protein synthesis and degradation pathways and muscle clock. Results: The results showed that lower BMI and grip strength were observed in insomnia patients. The mice in the sleep deprivation(SD) group saw a 7.01 g loss in body mass. The SD group's tibialis anterior and gastrocnemius muscle mass decreased after 96 h of SD). The grip strength reduced in SD group. Using the RT-PCR approaches, we found a significant increase in muscle degradation factors expression in SD group versus normal control group. Conclusions: Insomnia can impair muscle function. The mechanism may be associated with the increased expression of muscle degradation related factors , as well as the abnormal expression of Clock gene.

q-bio.NC↗

arXiv2023

DSSP: A Distributed, SLO-aware, Sensing-domain-privacy-Preserving Architecture for Sensing-as-a-Service

In this paper, we propose DSSP, a Distributed, SLO-aware, Sensing-domain-privacy-Preserving architecture for Sensing-as-a-Service (SaS). DSSP addresses four major limitations of the current SaS architecture. First, to improve sensing quality and enhance geographic coverage, DSSP allows Independent sensing Administrative Domains (IADs) to participate in sensing services, while preserving the autonomy of control and privacy for individual domains. Second, DSSP enables a marketplace in which a sensing data seller (i.e., an IAD) can sell its sensing data to more than one buyer (i.e., cloud service provider (CSP)), rather than being locked in with just one CSP. Third, DSSP enables per-query tail-latency service-level-objective (SLO) guaranteed SaS. Fourth, DSSP enables distributed, rather than centralized, query scheduling, making SaS highly scalable. At the core of DSSP is the design of a budget decomposition technique that translates: (a) a query tail-latency SLO into exact task response time budgets for sensing tasks of the query dispatched to individual IADs; and (b) the task budget for a task arrived at an IAD into exact subtask queuing deadlines for subtasks of the task dispatched to individual edge nodes in each IAD. This enables IADs to allocate their internal resources independently and accurately to meet the task budgets and hence, query tail-latency SLO, based on a simple subtask-budget-aware earliest-deadline-first queuing (EDFQ) policy for all the subtasks. The performance and scalability of DSSP are evaluated and verified by both on-campus testbed experiment at small scale and simulation at large scale.

cs.DC↗

arXiv2023

Basis-set-error-free RPA correlation energies for atoms based on the Sternheimer equation

The finite basis set errors for all-electron random-phase approximation (RPA) correlation energy calculations are analyzed for isolated atomic systems. We show that, within the resolution-of-identity (RI) RPA framework, the major source of the basis set errors is the incompleteness of the single-particle atomic orbitals used to expand the Kohn-Sham eigenstates, instead of the auxiliary basis set (ABS) to represent the density response function $χ^0$ and the bare Coulomb operator $v$. By solving the Sternheimer equation for the first-order wave function on a dense radial grid, we are able to eliminate the major error -- the incompleteness error of the single-particle atomic basis set -- for atomic RPA calculations. The error stemming from a finite ABS can be readily rendered vanishingly small by increasing the size of the ABS, or by iteratively determining the eigenmodes of the $χ^0 v$ operator. The variational property of the RI-RPA correlation energy can be further exploited to optimize the ABS in order to achieve a fast convergence of the RI-RPA correlation energy. These numerical techniques enable us to obtain basis-set-error-free RPA correlation energies for atoms, and in this work such energies for atoms from H to Kr are presented. The implications of the numerical techniques developed in the present work for addressing the basis set issue for molecules and solids are discussed.

physics.chem-ph↗

arXiv2023

Machine learning guided discovery of superconducting calcium borocarbides

Pursuit of superconductivity in light-element systems at ambient pressure is of great experimental and theoretical interest. In this work, we combine a machine learning (ML) method with first-principles calculations to efficiently search for the energetically favorable ternary Ca-B-C compounds. Three new layered borocarbides (stable CaBC5 and metastable Ca2BC11 and CaB3C3) are predicted to be phonon-mediated superconductors at ambient pressure. The hexagonal CaB3C3 possesses the highest Tc of 26.05 K among the three compounds. The σ-bonging bands around the Fermi level account for the large electron-phonon coupling (λ = 0.980) of hexagonal CaB3C3. The ML-guided approach opens up a way for greatly accelerating the discovery of new high-Tc superconductors.

cond-mat.mtrl-sci↗

arXiv2023

Efficient hybrid density functional calculation by deep learning

Hybrid density functional calculation is indispensable to accurate description of electronic structure, whereas the formidable computational cost restricts its broad application. Here we develop a deep equivariant neural network method (named DeepH-hybrid) to learn the hybrid-functional Hamiltonian from self-consistent field calculations of small structures, and apply the trained neural networks for efficient electronic-structure calculation by passing the self-consistent iterations. The method is systematically checked to show high efficiency and accuracy, making the study of large-scale materials with hybrid-functional accuracy feasible. As an important application, the DeepH-hybrid method is applied to study large-supercell Moiré twisted materials, offering the first case study on how the inclusion of exact exchange affects flat bands in the magic-angle twisted bilayer graphene.

cond-mat.mtrl-sci↗

arXiv2023

Diagnostics for plasmon satellites and Hubbard bands in transition metal oxides

Coulomb correlations between the electrons imprint characteristic signatures to the spectral properties of materials. Among others, they are at the origin of a rich phenomenology of satellite features, either stemming from atomic-like multiplets or from interactions with particle-hole excitations or plasmons. While in many cases the latter lie at considerably higher energies than the former, suggesting clear distinction criteria, this picture has recently become blurred by indications that satellites of different types can coexist in the same energy range. It is now generally accepted that the identification of the nature of spectral features is a highly non-trivial task. In this article we propose a general procedure for tracing the origin of satellites of different types within modern ab initio calculations. As an illustration, we analyze the ternary transition metal oxides SrVO$_3$ and SrMoO$_3$, which are drosophila compounds for the coexistence of Hubbard and plasmonic satellites, reconciling previous seemingly contradictory findings in an unexpected manner.

cond-mat.str-el↗