arXiv subjects

Lei Dai

Publications and source records attributed to Lei Dai.

At least 19 recordsLinked to original sources

arXiv2025

An explicit splitting SAV scheme for the kinetic Langevin dynamics

The kinetic Langevin dynamics finds diverse applications in various disciplines such as molecular dynamics and Hamiltonian Monte Carlo sampling. In this paper, a novel splitting scalar auxiliary variable (SSAV) scheme is proposed for the dynamics, where the gradient of the potential $U$ is possibly non-globally Lipschitz continuous with superlinear growth. As an explicit scheme, the SSAV method is efficient, robust and is able to reproduce the energy structure of the original dynamics. By an energy argument, the SSAV scheme is proved to possess an exponential integrability property, which is crucial to establishing the order-one strong convergence without the global monotonicity condition. Moreover, moments of the numerical approximations are shown to have polynomial growth with respect to the time length. This helps us to obtain weak error estimates of order one, with error constants polynomially (not exponentially) depending on the time length. Despite the obtained polynomial growth, the explicit scheme is shown to be computationally effective for the approximation of the invariant distribution of the dynamics with exponential ergodicity. Numerical experiments are presented to confirm the theoretical findings and to show the superiority of the algorithm in sampling.

math.NA↗

arXiv2025

XISM: an eXploratory and Interactive Graph Tool to Visualize and Evaluate Semantic Map Models

Semantic map models visualize systematic relations among semantic functions through graph structures and are widely used in linguistic typology. However, existing construction methods either depend on labor-intensive expert reasoning or on fully automated systems lacking expert involvement, creating a tension between scalability and interpretability. We introduce \textbf{XISM}, an interactive system that combines data-driven inference with expert knowledge. XISM generates candidate maps via a top-down procedure and allows users to iteratively refine edges in a visual interface, with real-time metric feedback. Experiments in three semantic domains and expert interviews show that XISM improves linguistic decision transparency and controllability in semantic-map construction while maintaining computational efficiency. XISM provides a collaborative approach for scalable and interpretable semantic-map building. The system\footnote{https://app.xism2025.xin/} , source code\footnote{https://github.com/hank317/XISM} , and demonstration video\footnote{https://youtu.be/m5laLhGn6Ys} are publicly available.

cs.CL↗

arXiv2024

DRIM-ANN: An Approximate Nearest Neighbor Search Engine based on Commercial DRAM-PIMs

Approximate nearest neighbor search (ANNS) is essential for applications like recommendation systems and retrieval-augmented generation (RAG) but is highly I/O-intensive and memory-demanding. CPUs face I/O bottlenecks, while GPUs are constrained by limited memory. DRAM-based Processing-in-Memory (DRAM-PIM) offers a promising alternative by providing high bandwidth, large memory capacity, and near-data computation. This work introduces DRIM-ANN, the first optimized ANNS engine leveraging UPMEM's DRAM-PIM. While UPMEM scales memory bandwidth and capacity, it suffers from low computing power because of the limited processor embedded in each DRAM bank. To address this, we systematically optimize ANNS approximation configurations and replace expensive squaring operations with lookup tables to align the computing requirements with UPMEM's architecture. Additionally, we propose load-balancing and I/O optimization strategies to maximize parallel processing efficiency. Experimental results show that DRIM-ANN achieves a 2.46x speedup over a 32-thread CPU and up to 2.67x over a GPU when deployed on computationally enhanced PIM platforms.

cs.PF↗

arXiv2024

Higher order numerical methods for SDEs without globally monotone coefficients

In the present work, we delve into further study of numerical approximations of SDEs with non-globally monotone coefficients. We design and analyze a new family of stopped increment-tamed time discretization schemes of Euler, Milstein and order 1.5 type for such SDEs. By formulating a novel unified framework, the proposed methods are shown to possess the exponential integrability properties, which are crucial to recovering convergence rates in the non-global monotone setting. Armed with such exponential integrability properties and by the arguments of perturbation estimates, we successfully identify the optimal strong convergence rates of the aforementioned methods in the non-global monotone setting. Numerical experiments are finally presented to corroborate the theoretical results.

math.NA↗

arXiv2024

An unconditional boundary and dynamics preserving scheme for the stochastic epidemic model

In the present article, we construct a logarithm transformation based Milstein-type method for the stochastic susceptible-infected-susceptible (SIS) epidemic model evolving in the domain (0,N). The new scheme is explicit and unconditionally boundary and dynamics preserving, when used to solve the stochastic SIS epidemic model. Also, it is proved that the scheme has a strong convergence rate of order one. Different from existing time discretization schemes, the newly proposed scheme for any time step size h > 0, not only produces numerical approximations living in the entire domain (0,N), but also unconditionally reproduces the extinction and persistence behavior of the original model, with no additional requirements imposed on the model parameters. Numerical experiments are presented to verify our theoretical findings.

math.NA↗

arXiv2024

SPEAK: Speech-Driven Pose and Emotion-Adjustable Talking Head Generation

Most earlier researches on talking face generation have focused on the synchronization of lip motion and speech content. However, head pose and facial emotions are equally important characteristics of natural faces. While audio-driven talking face generation has seen notable advancements, existing methods either overlook facial emotions or are limited to specific individuals and cannot be applied to arbitrary subjects. In this paper, we propose a novel one-shot Talking Head Generation framework (SPEAK) that distinguishes itself from the general Talking Face Generation by enabling emotional and postural control. Specifically, we introduce Inter-Reconstructed Feature Disentanglement (IRFD) module to decouple facial features into three latent spaces. Then we design a face editing module that modifies speech content and facial latent codes into a single latent space. Subsequently, we present a novel generator that employs modified latent codes derived from the editing module to regulate emotional expression, head poses, and speech content in synthesizing facial animations. Extensive trials demonstrate that our method ensures lip synchronization with the audio while enabling decoupled control of facial features, it can generate realistic talking head with coordinated lip motions, authentic facial emotions, and smooth head movements. The demo video is available: https://anonymous.4open.science/r/SPEAK-8A22

cs.CV↗

arXiv2023

Perturbation estimates for order-one strong approximations of SDEs without globally monotone coefficients

To obtain strong convergence rates of numerical schemes, an overwhelming majority of existing works impose a global monotonicity condition on coefficients of SDEs. Nevertheless, there are still many SDEs from applications that do not have globally monotone coefficients. As a recent breakthrough, the authors of [Hutzenthaler, Jentzen, Ann. Probab., 2020] originally presented a perturbation theory for stochastic differential equations (SDEs), which is crucial to recovering strong convergence rates of numerical schemes in a non-globally monotone setting. However, only a convergence rate of order 1/2 was obtained there for time-stepping schemes such as a stopped increment-tamed Euler-Maruyama (SITEM) method. An interesting question arises, also raised by the aforementioned work, as to whether a higher convergence rate than 1/2 can be obtained when higher order schemes are used. The present work attempts to give a positive answer to this question. To this end, we develop some new perturbation estimates that are able to reveal the order-one strong convergence of numerical methods. As the first application of the newly developed estimates, we identify the expected order-one pathwise uniformly strong convergence of the SITEM method for additive noise driven SDEs and multiplicative noise driven second order SDEs with non-globally monotone coefficients. As the other application, we propose and analyze a positivity preserving explicit Milstein-type method for Lotka-Volterra competition model driven by multi-dimensional noise, with a pathwise uniformly strong convergence rate of order one recovered under mild assumptions. These obtained results are completely new and significantly improve the existing theory. Numerical experiments are also provided to confirm the theoretical findings.

math.NA↗

arXiv2021

A Granular Sieving Algorithm for Deterministic Global Optimization

A gradient-free deterministic method is developed to solve global optimization problems for Lipschitz continuous functions defined in arbitrary path-wise connected compact sets in Euclidean spaces. The method can be regarded as granular sieving with synchronous analysis in both the domain and range of the objective function. With straightforward mathematical formulation applicable to both univariate and multivariate objective functions, the global minimum value and all the global minimizers are located through two decreasing sequences of compact sets in, respectively, the domain and range spaces. The algorithm is easy to implement with moderate computational cost. The method is tested against extensive benchmark functions in the literature. The experimental results show remarkable effectiveness and applicability of the algorithm.

math.OC↗

arXiv2019

PGU-net+: Progressive Growing of U-net+ for Automated Cervical Nuclei Segmentation

Automated cervical nucleus segmentation based on deep learning can effectively improve the quantitative analysis of cervical cancer. However, accurate nuclei segmentation is still challenging. The classic U-net has not achieved satisfactory results on this task, because it mixes the information of different scales that affect each other, which limits the segmentation accuracy of the model. To solve this problem, we propose a progressive growing U-net (PGU-net+) model, which uses two paradigms to extract image features at different scales in a more independent way. First, we add residual modules between different scales of U-net, which enforces the model to learn the approximate shape of the annotation in the coarser scale, and to learn the residual between the annotation and the approximate shape in the finer scale. Second, we start to train the model with the coarsest part and then progressively add finer part to the training until the full model is included. When we train a finer part, we will reduce the learning rate of the previous coarser part, which further ensures that the model independently extracts information from different scales. We conduct several comparative experiments on the Herlev dataset. The experimental results show that the PGU-net+ has superior accuracy than the previous state-of-the-art methods on cervical nuclei segmentation.

eess.IV↗

arXiv2019

Particle Energization in Space Plasmas: Towards a Multi-Point, Multi-Scale Plasma Observatory. A White Paper for the Voyage 2050 long-term plan in the ESA's Science Programme

This White Paper outlines the importance of addressing the fundamental science theme < > through a future ESA mission. The White Paper presents five compelling science questions related to particle energization by shocks, reconnection,waves and turbulence, jets and their combinations. Answering these questions requires resolving scale coupling, nonlinearity and nonstationarity, which cannot be done with existing multi-point observations. In situ measurements from a multi-point, multi-scale L-class plasma observatory consisting of at least 7 spacecraft covering fluid, ion and electron scales are needed. The plasma observatory will enable a paradigm shift in our comprehension of particle energization and space plasma physics in general, with very important impact on solar and astrophysical plasmas. It will be the next logical step following Cluster, THEMIS and MMS for the very large and active European space plasmas community. Being one of the cornerstone missions of the future ESA Voyage 2035-2050 science program, it would further strengthen the European scientific and technical leadership in this important field.

physics.space-ph↗

arXiv2019

Hour-scale persistent negative anomaly of atmospheric electrostatic field near the epicenter before earthquake

Although earthquake prediction is a big challenge in the world, some simple observational tools can capture many physical signals and demonstrate that an earthquake (EQ) may be forthcoming in short period. Many researchers have studied the significant variation of atmospheric electrostatic field related to the forthcoming earthquake. However, until now, there is not a compelling physical mechanism which can explain why atmospheric electrostatic abnormal signal could appear just before an earthquake. Here we present a precursor signal and propose a brief physical interpretation. Under fair air conditions, if the near-surface atmospheric electrostatic field Ez (oriented down when it is positive) presents a very stable negative anomaly (from -100 V/m to -5000 V/m), it will forebode that an earthquake (seismicity from 3-8) would take place in the next several to tens of hours within a distance less than 100 km. We name this prediction technique as "DC Ez Determination"(DED). In addition, the mechanism of such abnormal quasi-static electric field before a forthcoming earthquake has been proposed here: (1) Radon gas releases from the rock clefts near the epicenter during seismogenic process. (2) α particles are produced due to the radioactive decay of Radon gas in the air. (3) α particle ionizing radiation creates more positive particles in the air, which is much more effective than that β and γ particles produced during Radon radioactive decay. (4) The new positive particles change formal positive atmospheric electric field (Ez) into stably negative near the earth-surface. (5) The closer the instrument is to the epicenter, the more likely it is to observe the negative Ez signal related to the earthquake. It is recommended to establish an instrument network to capture the reliable precursor and make some warnings before the earthquake disaster.

physics.geo-ph↗

arXiv2019

Cascade Feature Aggregation for Human Pose Estimation

Human pose estimation plays an important role in many computer vision tasks and has been studied for many decades. However, due to complex appearance variations from poses, illuminations, occlusions and low resolutions, it still remains a challenging problem. Taking the advantage of high-level semantic information from deep convolutional neural networks is an effective way to improve the accuracy of human pose estimation. In this paper, we propose a novel Cascade Feature Aggregation (CFA) method, which cascades several hourglass networks for robust human pose estimation. Features from different stages are aggregated to obtain abundant contextual information, leading to robustness to poses, partial occlusions and low resolution. Moreover, results from different stages are fused to further improve the localization accuracy. The extensive experiments on MPII datasets and LIP datasets demonstrate that our proposed CFA outperforms the state-of-the-art and achieves the best performance on the state-of-the-art benchmark MPII.

cs.CV↗

arXiv2019

Evolving generalists via dynamic sculpting of rugged landscapes

Evolving systems, be it an antibody repertoire in the face of mutating pathogens or a microbial population exposed to varied antibiotics, constantly search for adaptive solutions in time-varying fitness landscapes. Generalists correspond to genotypes that remain fit across diverse selective pressures; cross-reactive antibodies are much wanted but rare, while multi-drug resistant microbes are undesired yet prevalent. However, little is known about under what conditions such solutions with a high capacity to adapt would be efficiently discovered by evolution, as environmental changes alter the relative fitness and accessibility of neighboring genotypes. In addition, can epistasis --- the source of landscape ruggedness and path constraints --- play a different role, if the environments are correlated in time? We present a generative model to estimate the propensity of evolving generalists in rugged landscapes that are tunably related and cycling relatively slowly. We find that environment cycling can substantially facilitate the search for fit generalists by dynamically enlarging their effective basins of attraction. Importantly, these high performers are most likely to emerge at an intermediate level of both ruggedness and environmental relatedness, trading diversity for fitness and accessibility. Our work provides a conceptual framework to study evolution in correlated varying complex environments, and offers statistical understanding that suggests general strategies for speeding up the generation of broadly neutralizing antibodies or preventing microbes from evolving multi-drug resistance.

physics.bio-ph↗

arXiv2015

In situ Evidence of Breaking the Ion Frozen-in Condition via the Non-gyrotropic Pressure Effect in Magnetic Reconnection

For magnetic reconnection to proceed, the frozen-in condition for both ion fluid and electron fluid in a localized diffusion region must be violated by inertial effects, thermal pressure effects, or inter-species collisions. It has been unclear which underlying effects unfreeze ion fluid in the diffusion region. By analyzing in-situ THEMIS spacecraft measurements at the dayside magnetopause, we present clear evidence that the off-diagonal components of the ion pressure tensor is mainly responsible for breaking the ion frozen-in condition in reconnection. The off-diagonal pressure tensor, which corresponds to a nongyrotropic pressure effect, is a fluid manifestation of ion demagnetization in the diffusion region. From the perspective of the ion momentum equation, the reported non-gyrotropic ion pressure tensor is a fundamental aspect in specifying the reconnection electric field that controls how quickly reconnection proceeds.

physics.plasm-ph↗

arXiv2013

Dynamics of a producer-parasite ecosystem on the brink of collapse

Ecosystems can undergo sudden shifts to undesirable states, but recent studies with simple single species ecosystems have demonstrated that advance warning can be provided by the slowing down of population dynamics near a tipping point. However, it is not clear how this effect of critical slowing down will manifest in ecosystems with strong interactions between their components. Here we probe the dynamics of an experimental producer parasite ecosystem as it approaches a catastrophic collapse. Surprisingly, the producer population grows in size as the environment deteriorates, highlighting that population size can be a misleading measure of ecosystem stability. By analyzing the oscillatory producer parasite dynamics for over ~100 generations in multiple environmental conditions, we found that the collective ecosystem dynamics slows down as the tipping point is approached. Analysis of the coupled dynamics of interacting populations may therefore be necessary to provide advance warning of collapse in complex communities.

q-bio.PE↗

arXiv2013

THEMIS Observations of the Magnetopause Electron Diffusion Region: Large Amplitude Waves and Heated Electrons

We present the first observations of large amplitude waves in a well-defined electron diffusion region at the sub-solar magnetopause using data from one THEMIS satellite. These waves identified as whistler mode waves, electrostatic solitary waves, lower hybrid waves and electrostatic electron cyclotron waves, are observed in the same 12-sec waveform capture and in association with signatures of active magnetic reconnection. The large amplitude waves in the electron diffusion region are coincident with abrupt increases in electron parallel temperature suggesting strong wave heating. The whistler mode waves which are at the electron scale and enable us to probe electron dynamics in the diffusion region were analyzed in detail. The energetic electrons (~30 keV) within the electron diffusion region have anisotropic distributions with T_{e\perp}/T_{e\parallel}>1 that may provide the free energy for the whistler mode waves. The energetic anisotropic electrons may be produced during the reconnection process. The whistler mode waves propagate away from the center of the 'X-line' along magnetic field lines, suggesting that the electron diffusion region is a possible source region of the whistler mode waves.

physics.space-ph↗

arXiv2010

Mode Splitting for Efficient Plasmoinc Thin-film Solar Cell

We propose an efficient plasmonic structure consisting of metal strips and thin-film silicon for solar energy absorption. We numerically demonstrate the absorption enhancement in symmetrical structure based on the mode coupling between the localized plasmonic mode in Ag strip pair and the excited waveguide mode in silicon slab. Then we explore the method of symmetry-breaking to excite the dark modes that can further enhance the absorption ability. We compare our structure with bare thin-film Si solar cell, and results show that the integrated quantum efficiency is improved by nearly 90% in such thin geometry. It is a promising way for the solar cell.

physics.optics↗

arXiv2010

Electromagnetically induced transparency in hybrid plasmonic-dielectric system

We present theoretical and numerical analysis of a plasmonic-dielectric hybrid system for symmetric and asymmetric coupling between silver cut-wire pairs and silicon grating waveguide with periodic grooves. The results show that both couplings can induce electromagnetically-induced transparency (EIT) analogous to the quantum optical phenomenon. The transmission spectrum shows a single transparency window for the symmetric coupling. The strong normal phase dispersion in the vicinity of this transparent window results in the slow light effect. However, the transmission spectrum appears an additional transparency window for asymmetry coupling due to the double EIT effect, which stems from an asymmetrically coupled resonance (ACR) between the dark and bright modes. More importantly, the excitation of ACR is further associated with remarkable improvement of the group index from less than 40 to more than 2500 corresponding to a high transparent efficiency by comparing with the symmetry coupling. This scheme provides an alternative way to develop the building blocks of systems for plasmonic sensing, all optical switching and slow light applications.

physics.optics↗