arXiv subjects

Kexin Li

Publications and source records attributed to Kexin Li.

45 records · Page 3Linked to original sources

arXiv2023

Data-aided Underwater Acoustic Ray Propagation Modeling

Acoustic propagation models are widely used in numerous oceanic and other underwater applications. Most conventional models are approximate solutions of the acoustic wave equation, and require accurate environmental knowledge to be available beforehand. Environmental parameters may not always be easily or accurately measurable. While data-driven techniques might allow us to model acoustic propagation without the need for extensive prior environmental knowledge, such techniques tend to be data-hungry and often infeasible in oceanic applications where data collection is difficult and expensive. We propose a data-aided ray physics based high frequency acoustic propagation modeling approach that enables us to train models with only a small amount of data. The proposed framework is not only data-efficient, but also offers flexibility to incorporate varying degrees of environmental knowledge, and generalizes well to permit extrapolation beyond the area where data was collected. We demonstrate the feasibility and applicability of our method through four numerical case studies, and one controlled experiment. We also benchmark our method's performance against classical data-driven techniques.

cs.SD↗

arXiv2023

O2CTA: Introducing Annotations from OCT to CCTA in Coronary Plaque Analysis

Targeted diagnosis and treatment plans for patients with coronary artery disease vary according to atherosclerotic plaque component. Coronary CT angiography (CCTA) is widely used for artery imaging and determining the stenosis degree. However, the limited spatial resolution and susceptibility to artifacts fail CCTA in obtaining lumen morphological characteristics and plaque composition. It can be settled by invasive optical coherence tomography (OCT) without much trouble for physicians, but bringing higher costs and potential risks to patients. Therefore, it is clinically critical to introduce annotations of plaque tissue and lumen characteristics from OCT to paired CCTA scans, denoted as \textbf{the O2CTA problem} in this paper. We propose a method to handle the O2CTA problem. CCTA scans are first reconstructed into multi-planar reformatted (MPR) images, which agree with OCT images in term of semantic contents. The artery segment in OCT, which is manually labelled, is then spatially aligned with the entire artery in MPR images via the proposed alignment strategy. Finally, a classification model involving a 3D CNN and a Transformer, is learned to extract local features and capture dependence along arteries. Experiments on 55 paired OCT and CCTA we curate demonstrate that it is feasible to classify the CCTA based on the OCT labels, with an accuracy of 86.2%, while the manual readings of OCT and CCTA vary significantly, with a Kappa coefficient of 0.113. We will make our source codes, models, data, and results publicly available to benefit the research community.

eess.IV↗

arXiv2022

A comparison of a commercial hydrodynamics TCAD solver and Fermi kinetics transport convergence for GaN HEMTs

Various simulations of a GaN HEMT are used to study the behaviors of two different energy-transport models: the Fermi kinetics transport model and a hydrodynamics transport model as it is implemented in the device simulator Sentaurus from Synopsys. The electron transport and heat flow equations of the respective solvers are described in detail. The differences in the description of electron flux and the discretization methods are highlighted. Next, the transport models are applied to the same simulated device structure using identical meshes, boundary conditions, and material parameters. Static simulations show the numerical convergence of Fermi kinetics to be consistently quadratic or faster, whereas the hydrodynamic model is often sub-quadratic. Further comparisons of large signal transient simulations reveal the hydrodynamic model produces certain anomalous electron ensemble behaviors within the transistor structure. The fundamentally different electron dynamics produced by the two models suggest an underlying cause for their different numerical convergence characteristics.

cond-mat.mes-hall↗

arXiv2022

Selecting Regularization Parameters for nuclear norm type minimization problems

The reconstruction of low-rank matrix from its noisy observation finds its usage in many applications. It can be reformulated into a constrained nuclear norm minimization problem, where the bound $η$ of the constraint is explicitly given or can be estimated by the probability distribution of the noise. When the Lagrangian method is applied to find the minimizer, the solution can be obtained by the singular value thresholding operator where the thresholding parameter $λ$ is related to the Lagrangian multiplier. In this paper, we first show that the Frobenius norm of the discrepancy between the minimizer and the observed matrix is a strictly increasing function of $λ$. From that we derive a closed-form solution for $λ$ in terms of $η$. The result can be used to solve the constrained nuclear-norm-type minimization problem when $η$ is given. For the unconstrained nuclear-norm-type regularized problems, our result allows us to automatically choose a suitable regularization parameter by using the discrepancy principle. The regularization parameters obtained are comparable to (and sometimes better than) those obtained by Stein's unbiased risk estimator (SURE) approach while the cost of solving the minimization problem can be reduced by 11--18 times. Numerical experiments with both synthetic data and real MRI data are performed to validate the proposed approach.

math.OC↗

arXiv2022

Second-order accurate numerical scheme with graded meshes for the nonlinear partial integrodifferential equation arising from viscoelasticity

This paper establishes and analyzes a second-order accurate numerical scheme for the nonlinear partial integrodifferential equation with a weakly singular kernel. In the time direction, we apply the Crank-Nicolson method for the time derivative, and the product-integration (PI) rule is employed to deal with Riemann-Liouville fractional integral. From which, the non-uniform meshes are utilized to compensate for the singular behavior of the exact solution at $t=0$ so that our method can reach second-order convergence for time. In order to formulate a fully discrete implicit difference scheme, we employ a standard centered difference formula for the second-order spatial derivative, and the Galerkin method based on piecewise linear test functions is used to approximate the nonlinear convection term. Then we derive the existence and uniqueness of numerical solutions for the proposed implicit difference scheme. Meanwhile, stability and convergence are proved by means of the discrete energy method. Furthermore, to demonstrate the effectiveness of the proposed method, we utilize a fixed point iterative algorithm to calculate the discrete scheme. Finally, numerical experiments illustrate the feasibility and efficiency of the proposed scheme, in which numerical results are consistent with our theoretical analysis.

math.NA↗

arXiv2021

ScalaBFS: A Scalable BFS Accelerator on HBM-Enhanced FPGAs

High Bandwidth Memory (HBM) provides massive aggregated memory bandwidth by exposing multiple memory channels to the processing units. To achieve high performance, an accelerator built on top of an FPGA configured with HBM (i.e., FPGA-HBM platform) needs to scale its performance according to the available memory channels. In this paper, we propose an accelerator for BFS (Breadth-First Search) algorithm, named as ScalaBFS, that builds multiple processing elements to sufficiently exploit the high bandwidth of HBM to improve efficiency. We implement the prototype system of ScalaBFS and conduct BFS in both real-world and synthetic scale-free graphs on Xilinx Alveo U280 FPGA card real hardware. The experimental results show that ScalaBFS scales its performance almost linearly according to the available memory pseudo channels (PCs) from the HBM2 subsystem of U280. By fully using the 32 PCs and building 64 processing elements (PEs) on U280, ScalaBFS achieves a performance up to 19.7 GTEPS (Giga Traversed Edges Per Second). When conducting BFS in sparse real-world graphs, ScalaBFS achieves equivalent GTEPS to Gunrock running on the state-of-art Nvidia V100 GPU that features 64-PC HBM2 (twice memory bandwidth than U280).

cs.AR↗

arXiv2019

Ecosystem-bedrock interaction changes nutrient compartmentalization during early oxidative weathering

Ecosystem-bedrock interactions power the biogeochemical cycles of Earth's shallow crust, supporting life, stimulating substrate transformation, and spurring evolutionary innovation. While oxidative processes have dominated half of terrestrial history, the relative contribution of the biosphere and its chemical fingerprints on Earth's developing regolith are still poorly constrained. Here, we report results from a two-year incipient weathering experiment. We found that the mass release and compartmentalization of major elements during weathering of granite, rhyolite, schist and basalt was rock-specific and regulated by ecosystem components. A tight interplay between physiological needs of different biota, mineral dissolution rates, and substrate nutrient availability resulted in intricate elemental distribution patterns. Biota accelerated CO2 mineralization over abiotic controls as ecosystem complexity increased, and significantly modified stoichiometry of mobilized elements. Microbial and fungal components inhibited element leaching (23.4% and 7%), while plants increased leaching and biomass retention by 63.4%. All biota left comparable biosignatures in the dissolved weathering products. Nevertheless, the magnitude and allocation of weathered fractions under abiotic and biotic treatments provide quantitative evidence for the role of major biosphere components in the evolution of upper continental crust, presenting critical information for large-scale biogeochemical models and for the search for stable in situ biosignatures beyond Earth.

astro-ph.EP↗

arXiv2018

Global Existence of Strong Solutions to Compressible Navier-Stokes System with Degenerate Heat Conductivity in Unbounded Domains

In one-dimensional unbounded domains, we prove global existence of strong solutions to the compressible Navier-Stokes system for a viscous and heat conducting ideal polytropic gas, when the viscosity is constant and the heat conductivity $κ$ depends on the temperature $θ$ according to $κ=\barκθ^β(β>0)$. Note that the conditions imposed on the initial data are the same as those of the constant heat conductivity case ([Kazhikhov, A. V. Siberian Math. J. 23 (1982), 44-49]) and can be arbitrarily large. Therefore, our result generalizes Kazhikhov's result for the constant heat conductivity case to the degenerate and nonlinear one.

math.AP↗

arXiv2018

Global Well-Posedness and Large-Time Behavior of 1D Compressible Navier-Stokes System with Density-Depending Viscosity and Vacuum in Unbounded Domains

We consider the Cauchy problem for one-dimensional (1D) barotropic compressible Navier-Stokes equations with density-dependent viscosity and large external force. Under a general assumption on the density-dependent viscosity, we prove that the Cauchy problem admits a unique global strong (classical) solution for the large initial data with vacuum. Moreover, the density is proved to be bounded from above time-independently. As a consequence, we obtain the large time behavior of the solution without external forces.

math.AP↗