arXiv subjects

Wenjun Cai

Publications and source records attributed to Wenjun Cai.

At least 19 recordsLinked to original sources

arXiv2025

Guidelines for Building Indexes on Partially Cache-Coherent CXL Shared Memory

The \emph{Partial Cache-Coherence (PCC)} model maintains hardware cache coherence only within subsets of cores, enabling large-scale memory sharing with emerging memory interconnect technologies like Compute Express Link (CXL). However, PCC's relaxation of global cache coherence compromises the correctness of existing single-machine software. This paper focuses on building consistent and efficient indexes on PCC platforms. We present that existing indexes designed for cache-coherent platforms can be made consistent on PCC platforms following SP guidelines, i.e., we identify \emph{sync-data} and \emph{protected-data} according to the index's concurrency control mechanisms, and synchronize them accordingly. However, conversion with SP guidelines introduces performance overhead. To mitigate the overhead, we identify several unique performance bottlenecks on PCC platforms, and propose P$^3$ guidelines (i.e., using Out-of-\underline{P}lace update, Re\underline{P}licated shared variable, S\underline{P}eculative Reading) to improve the efficiency of converted indexes on PCC platforms. With SP and P$^3$ guidelines, we convert and optimize two indexes (CLevelHash and BwTree) for PCC platforms. Evaluation shows that converted indexes' throughput improves up to 16$\times$ following P$^3$ guidelines, and the optimized indexes outperform their message-passing-based and disaggregated-memory-based counterparts by up to 16$\times$ and 19$\times$.

cs.OS↗

arXiv2025

Explicit and CPU/GPU parallel energy-preserving schemes for the Klein-Gordon-Schr\"odinger equations

A highly efficient energy-preserving scheme for univariate conservative or dissipative systems was recently proposed in [Comput. Methods Appl. Mech. Engrg. 425 (2024) 116938]. This scheme is based on a grid-point partitioned averaged vector field (AVF) method, allowing for pointwise decoupling and easy implementation of CPU parallel computing. In this article, we further extend this idea to multivariable coupled systems and propose a dual-partition AVF method that employs a dual partitioning strategy based on both variables and grid points. The resulting scheme is decoupled, energy-preserving, and exhibits greater flexibility. For the Klein-Gordon-Schr\"odinger equations, we apply the dual-partition AVF method and construct fully explicit energy-preserving schemes with pointwise decoupling, where the computational complexity per time step is $\mathcal{O}(N^d)$, with $d$ representing the problem dimension and $N$ representing the number of grid points in each direction. These schemes not only enable CPU parallelism but also support parallel computing on GPUs by adopting an update strategy based on a checkerboard grid pattern, significantly improving the efficiency of solving high-dimensional problems. Numerical experiments confirm the conservation properties and high efficiency of the proposed schemes.

math.NA↗

arXiv2023

A novel high-order linearly implicit and energy-stable additive Runge-Kutta methods for gradient flow models

This paper introduces a novel paradigm for constructing linearly implicit and high-order unconditionally energy-stable schemes for general gradient flows, utilizing the scalar auxiliary variable (SAV) approach and the additive Runge-Kutta (ARK) methods. We provide a rigorous proof of energy stability, unique solvability, and convergence. The proposed schemes generalizes some recently developed high-order, energy-stable schemes and address their shortcomings. On the one other hand, the proposed schemes can incorporate existing SAV-RK type methods after judiciously selecting the Butcher tables of ARK methods \cite{sav_li,sav_nlsw}. The order of a SAV-RKPC method can thus be confirmed theoretically by the order conditions of the corresponding ARK method. Several new schemes are constructed based on our framework, which perform to be more stable than existing SAV-RK type methods. On the other hand, the proposed schemes do not limit to a specific form of the nonlinear part of the free energy and can achieve high order with fewer intermediate stages compared to the convex splitting ARK methods \cite{csrk}. Numerical experiments demonstrate stability and efficiency of proposed schemes.

math.NA↗

arXiv2023

Energy stable and maximum bound principle preserving schemes for the Allen-Cahn equation based on the Saul'yev methods

The energy dissipation law and maximum bound principle are significant characteristics of the Allen-Chan equation. To preserve discrete counterpart of these properties, the linear part of the target system is usually discretized implicitly, resulting in a large linear or nonlinear system of equations. The Fast Fourier Transform (FFT) algorithm is commonly used to solve the resulting linear or nonlinear systems with computational costs of $\mathcal{O}(M^d log M)$ at each time step, where $M$ is the number of spatial grid points in each direction, and $d$ is the dimension of the problem. Combining the Saul'yev methods and the stabilized technique, we propose and analyze novel first- and second-order numerical schemes for the Allen-Cahn equation in this paper. In contrast to the traditional methods, the proposed methods can be solved by components, requiring only $\mathcal{O}(M^d)$ computational costs per time step. Additionally, they preserve the maximum bound principle and original energy dissipation law at the discrete level. We also propose rigorous analysis of their consistency and convergence. Numerical experiments are conducted to confirm the theoretical analysis and demonstrate the efficiency of the proposed methods.

math.NA↗

arXiv2022

Unconditional convergence of conservative spectral Galerkin methods for the coupled fractional nonlinear Klein-Gordon-Schr\"odinger equations

In this work, two novel classes of structure-preserving spectral Galerkin methods are proposed which based on the Crank-Nicolson scheme and the exponential scalar auxiliary variable method respectively, for solving the coupled fractional nonlinear Klein-Gordon-Schr\"odinger equation. The paper focuses on the theoretical analyses and computational efficiency of the proposed schemes, the Crank-Nicoloson scheme is proved to be unconditionally convergent and has the maximum-norm boundness of numerical solutions. The exponential scalar auxiliary variable scheme is linearly implicit and decoupled, but lack of the maximum-norm boundness, also, the energy structure has been modified. Subsequently, the efficient implementations of the proposed schemes are introduced in detail. Both the theoretical analyses and the numerical comparisons show that the proposed spectral Galerkin methods have high efficiency in long-time computations.

math.NA↗

arXiv2022

Linearly implicit energy-preserving integrating factor methods for the 2D nonlinear Schr\"odinger equation with wave operator and convergence analysis

In this paper, we develop a novel class of linear energy-preserving integrating factor methods for the 2D nonlinear Schr\"odinger equation with wave operator (NLSW), combining the scalar auxiliary variable approach and the integrating factor methods. A second-order scheme is first proposed, which is rigorously proved to be energy-preserving. By using the energy methods, we analyze its optimal convergence in the $H^1$ norm without any restrictions on the grid ratio, where a novel technique and an improved induction argument are proposed to overcome the difficulty posed by the unavailability of a priori $L^\infty$ estimates of numerical solutions. Based on the integrating factor Runge-Kutta methods, we extend the proposed scheme to arbitrarily high order, which is also linear and conservative. Numerical experiments are presented to confirm the theoretical analysis and demonstrate the advantages of the proposed methods.

math.NA↗

arXiv2022

On Mixing Distributions Via Random Orthogonal Matrices and the Spectrum of the Singular Values of Multi-Z Shaped Graph Matrices

In this paper, we introduce and analyze a new operation $\circ_{R}$ which mixes two distributions $\Omega$ and $\Omega'$ via a random orthogonal matrix. In particular, we take $\Omega \circ_R \Omega'$ to be the limit as $n \to \infty$ of the distribution of singular values of $DRD'$ where $D$ and $D'$ are $n \times n$ diagonal matrices whose diagonal entries have distributions $\Omega$ and $\Omega'$ respectively and $R$ is a random $n \times n$ orthogonal matrix. We show that $\circ_R$ has several nice properties. We first observe that $\circ_R$ is commutative and associative and compute the moments of $\Omega \circ_R \Omega'$ in terms of the moments of $\Omega$ and $\Omega'$. We then show that $\circ_R$ interacts very nicely with the spectrum of the singular values of Z-shaped and multi-Z-shaped graph matrices. This allows us to answer the question posed by our previous paper of how to describe the spectrum of the singular values of Z-shaped and multi-Z-shaped graph matrices when the input distribution is not $\{-1,1\}$. In our analysis, we show that the moments of our distributions are closely connected to non-crossing partitions and prove a number of new results on non-crossing partitions which may be of independent interest.

math.CO↗

arXiv2021

Partitioned Active Learning for Heterogeneous Systems

Active learning is a subfield of machine learning that focuses on improving the data collection efficiency of expensive-to-evaluate systems. Especially, active learning integrated surrogate modeling has shown remarkable performance in computationally demanding engineering systems. However, the existence of heterogeneity in underlying systems may adversely affect the performance of active learning. In order to improve the learning efficiency under this regime, we propose the partitioned active learning that seeks the most informative design points for partitioned Gaussian process modeling of heterogeneous systems. The proposed active learning consists of two systematic subsequent steps: the global searching scheme accelerates the exploration of active learning by investigating the most uncertain design space, and the local searching exploits the circumscribed information induced by the local GP. We also propose Cholesky update driven numerical remedies for our active learning to address the computational complexity challenge. The proposed method is applied to numerical simulations and two real-world case studies about (i) the cost-efficient automatic fuselage shape control in aerospace manufacturing; and (ii) the optimal design of tribocorrosion-resistant alloys in materials science. The results show that our approach outperforms benchmark methods with respect to prediction accuracy and computational efficiency.

cs.LG↗

arXiv2021

A Degradation Performance Model With Mixed-type Covariates and Latent Heterogeneity

Successful modeling of degradation performance data is essential for accurate reliability assessment and failure predictions of highly reliable product units. The degradation performance measurements over time are highly heterogeneous. Such heterogeneity can be partially attributed to external factors, such as accelerated/environmental conditions, and can also be attributed to internal factors, such as material microstructure characteristics of product units. The latent heterogeneity due to the unobserved/unknown factors shared within each product unit may also exists and need to be considered as well. Existing degradation models often fail to consider (i) the influence of both external accelerated/environmental conditions and internal material information, (ii) the influence of unobserved/unknown factors within each unit. In this work, we propose a generic degradation performance modeling framework with mixed-type covariates and latent heterogeneity to account for both influences of observed internal and external factors as well as unobserved factors. Effective estimation algorithm is also developed to jointly quantify the influences of mixed-type covariates and individual latent heterogeneity, and also to examine the potential interaction between mixed-type covariates. Functional data analysis and data augmentation techniques are employed to address a series of estimation issues. A real case study is further provided to demonstrate the superior performance of the proposed approach over several alternative modeling approaches. Besides, the proposed degradation performance modeling framework also provides interpretable findings.

stat.AP↗

arXiv2020

NP-ODE: Neural Process Aided Ordinary Differential Equations for Uncertainty Quantification of Finite Element Analysis

Finite element analysis (FEA) has been widely used to generate simulations of complex and nonlinear systems. Despite its strength and accuracy, the limitations of FEA can be summarized into two aspects: a) running high-fidelity FEA often requires significant computational cost and consumes a large amount of time; b) FEA is a deterministic method that is insufficient for uncertainty quantification (UQ) when modeling complex systems with various types of uncertainties. In this paper, a physics-informed data-driven surrogate model, named Neural Process Aided Ordinary Differential Equation (NP-ODE), is proposed to model the FEA simulations and capture both input and output uncertainties. To validate the advantages of the proposed NP-ODE, we conduct experiments on both the simulation data generated from a given ordinary differential equation and the data collected from a real FEA platform for tribocorrosion. The performances of the proposed NP-ODE and several benchmark methods are compared. The results show that the proposed NP-ODE outperforms benchmark methods. The NP-ODE method realizes the smallest predictive error as well as generates the most reasonable confidence interval having the best coverage on testing data points.

cs.LG↗

arXiv2020

Arbitrary high-order linearly implicit energy-preserving algorithms for Hamiltonian PDEs

In this paper, we present a novel strategy to systematically construct linearly implicit energy-preserving schemes with arbitrary order of accuracy for Hamiltonian PDEs. Such novel strategy is based on the newly developed exponential scalar variable (ESAV) approach that can remove the bounded-from-blew restriction of nonlinear terms in the Hamiltonian functional and provides a totally explicit discretization of the auxiliary variable without computing extra inner products, which make it more effective and applicable than the traditional scalar auxiliary variable (SAV) approach. To achieve arbitrary high-order accuracy and energy preservation, we utilize the symplectic Runge-Kutta method for both solution variables and the auxiliary variable, where the values of internal stages in nonlinear terms are explicitly derived via an extrapolation from numerical solutions already obtained in the preceding calculation. A prediction-correction strategy is proposed to further improve the accuracy. Fourier pseudo-spectral method is then employed to obtain fully discrete schemes. Compared with the SAV schemes, the solution variables and the auxiliary variable in these ESAV schemes are now decoupled. Moreover, when the linear terms are of constant coefficients, the solution variables can be explicitly solved by using the fast Fourier transform. Numerical experiments are carried out for three Hamiltonian PDEs to demonstrate the efficiency and conservation of the ESAV schemes.

math.NA↗

arXiv2020

An explicit and practically invariants-preserving method for conservative systems

An explicit numerical strategy that practically preserves invariants is derived for conservative systems by combining an explicit high-order Runge-Kutta (RK) scheme with a simple modification of the standard projection approach, which is named the explicit invariants-preserving (EIP) method. The proposed approach is shown to have the same order as the underlying RK method, while the error of invariants is analyzed in the order of $\mathcal{O}\left(h^{2(p+1)}\right),$ where $h$ is the time step and $p$ represents the order of the method. When $p$ is appropriately large, the EIP method is practically invariants-conserving because the error of invariants can reach the machine accuracy. The method is illustrated for the cases of single and multiple invariants, with regard to both ODEs and high-dimensional PDEs. Extensive numerical experiments are presented to verify our theoretical results and demonstrate the superior behaviors of the proposed method in a long time numerical simulation. Numerical results suggest that the fourth-order EIP method preserves much better the qualitative properties of the flow than the standard fourth-order RK method and it is more efficient in practice than the fully implicit integrators.

math.NA↗

arXiv2020

The Spectrum of the Singular Values of Z-Shaped Graph Matrices

Graph matrices are a type of matrix which has played a crucial role in analyzing the sum of squares hierarchy on average case problems. However, except for rough norm bounds, little is known about graph matrices. In this paper, we take a step towards better understanding graph matrices by determining the limiting distribution of the spectrum of the singular values of Z-shaped graph matrices. We then give a partial generalization of our results for $m$-layer Z-shaped graph matrices.

math.CO↗

arXiv2020

A linearly implicit structure-preserving scheme for the Camassa-Holm equation based on multiple scalar auxiliary variables approach

In this paper, we present a linearly implicit energy-preserving scheme for the Camassa-Holm equation by using the multiple scalar auxiliary variables approach, which is first developed to construct efficient and robust energy stable schemes for gradient systems. The Camassa-Holm equation is first reformulated into an equivalent system by utilizing the multiple scalar auxiliary variables approach, which inherits a modified energy. Then, the system is discretized in space aided by the standard Fourier pseudo-spectral method and a semi-discrete system is obtained, which is proven to preserve a semi-discrete modified energy. Subsequently, the linearized Crank-Nicolson method is applied for the resulting semi-discrete system to arrive at a fully discrete scheme. The main feature of the new scheme is to form a linear system with a constant coefficient matrix at each time step and produce numerical solutions along which the modified energy is precisely conserved, as is the case with the analytical solution. Several numerical results are addressed to confirm accuracy and efficiency of the proposed scheme.

math.NA↗

arXiv2020

A linearly implicit energy-preserving exponential integrator for the nonlinear Klein-Gordon equation

In this paper, we generalize the exponential energy-preserving integrator proposed in the recent paper [SIAM J. Sci. Comput. 38(2016) A1876-A1895] for conservative systems, which now becomes linearly implicit by further utilizing the idea of the scalar auxiliary variable approach. Comparing with the original exponential energy-preserving integrator which usually leads to a nonlinear algebraic system, our new method only involve a linear system with constant coefficient matrix. Taking the nonlinear Klein-Gordon equation for example, we derive the concrete energy-preserving scheme and demonstrate its high efficiency through numerical experiments.

math.NA↗

arXiv2020

Sequential Selection for Accelerated Life Testing via Approximate Bayesian Inference

Accelerated life testing (ALT) is typically used to assess the reliability of material's lifetime under desired stress levels. Recent advances in material engineering have made a variety of material alternatives readily available. To identify the most reliable material setting with efficient experimental design, a sequential test planning strategy is preferred. To guarantee a tractable statistical mechanism for information collection and update, we develop explicit model parameter update formulas via approximate Bayesian inference. Theories show that our explicit update formulas give consistent parameter estimates. Simulation study and a case study show that the proposed sequential selection approach can significantly improve the probability of identifying the material alternative with best reliability performance over other design approaches.

stat.ME↗

arXiv2019

Two novel classes of arbitrary high-order structure-preserving algorithms for canonical Hamiltonian systems

In this paper, we systematically construct two classes of structure-preserving schemes with arbitrary order of accuracy for canonical Hamiltonian systems. The one class is the symplectic scheme, which contains two new families of parameterized symplectic schemes that are derived by basing on the generating function method and the symmetric composition method, respectively. Each member in these schemes is symplectic for any fixed parameter. A more general form of generating functions is introduced, which generalizes the three classical generating functions that are widely used to construct symplectic algorithms. The other class is a novel family of energy and quadratic invariants preserving schemes, which is devised by adjusting the parameter in parameterized symplectic schemes to guarantee energy conservation at each time step. The existence of the solutions of these schemes is verified. Numerical experiments demonstrate the theoretical analysis and conservation of the proposed schemes.

math.NA↗

arXiv2019

A structure-preserving algorithm for the fractional nonlinear Schrödinger equation based on the SAV approach

The main objective of this paper is to present an efficient structure-preserving scheme, which is based on the idea of the scalar auxiliary variable approach, for solving the space fractional nonlinear Schrödinger equation. First, we reformulate the equation as a Hamiltonian system, and obtain a new equivalent system via introducing a scalar variable. Then, we construct a semi-discrete energy-preserving scheme by using the Fourier pseudo-spectral method to discretize the equivalent system in space direction. After that, applying the Crank-Nicolson method on the temporal direction gives a linear implicit scheme in the fully-discrete version. As expected, the proposed scheme can preserve the energy exactly and more efficient in the sense that only decoupled equations with constant coefficients need to be solved at each time step. Finally, numerical experiments are provided to demonstrate the effectiveness and conservation of the scheme.

math.NA↗