This paper explores challenges in training Physics-Informed Neural Networks (PINNs), emphasizing the role of loss landscape process. We examine difficulties minimizing PINN function, particularly due to ill-conditioning caused by differential operators residual term. compare gradient-based optimizers Adam, L-BFGS, and their combination \al{}, showing superiority introduce a novel second-order optimizer, NysNewton-CG (NNCG), which significantly improves performance. Theoretically, our work...

To study the influence of parameter uncertainty in small pressurized water reactor (SPWR) once-through steam generator (OTSG), nonlinear mathematical model SPWR is firstly established. Including core model, OTSG and pressurizer model. Secondly, a control strategy that both coolant average temperature secondary-side outlet pressure are constant adopted. Then, quantification method established based on Latin hypercube sampling statistical method. On this basis, quantitative platform for...

ABSTRACT Objective Understanding the impact of sociocultural tendencies on personality development adolescents represents a critical theoretical and practical issue in field adolescent development. In context China's collectivist culture, developmental trajectories interaction between sensation seeking collectivism among remain largely unknown. Method This study examined heterogeneity joint growth patterns their interactions across distinct latent trajectory classes. We collected 3‐year...

Neural network-based approaches have recently shown significant promise in solving partial differential equations (PDEs) science and engineering, especially scenarios featuring complex domains or the incorporation of empirical data. One advantage neural network method for PDEs lies its automatic differentiation (AD), which necessitates only sample points themselves, unlike traditional finite difference (FD) approximations that require nearby local to compute derivatives. In this paper, we...

Trans-inductor can be applied in a transient enhancement topology that uses the secondary windings of transformers to connect each phase multi-phase voltage regulator. The connection makes all phases have response when there is load transient. Based on this, resultant transinductor regulator (TLVR) achieve significantly improved current slew rate. ripple maintains reasonable for phase, making design feasible implement. In this paper, analysis TLVR presented and verified. For an eight-phase...

A 1.02 MHz 5.72-μW 60-ppm/°C relaxation oscillator is demonstrated in a 180-nm CMOS process. The proposed simplifies the logic control circuits to save power consumption and chip area. By employing an improved delay compensation technique, has minimized sensitivity temperature variations. measured results show average frequency drift of about 60 ppm/°C below 6 μW. For supply voltage change from 1.1V 2 V, changes within only ± 1.6%.

Neural network-based methods have emerged as powerful tools for solving partial differential equations (PDEs) in scientific and engineering applications, particularly when handling complex domains or incorporating empirical data. These leverage neural networks basis functions to approximate PDE solutions. However, training such can be challenging, often resulting limited accuracy. In this paper, we investigate the dynamics of solvers with a focus on impact initialization techniques. We...

In this paper, we propose a novel approach to generative modeling using loss function based on elastic interaction energy (EIE), which is inspired by the between defects in crystals. The utilization of EIE-based metric presents several advantages, including its long range property that enables consideration global information distribution. Moreover, inclusion self-interaction term helps prevent mode collapse and captures all modes To overcome difficulty relatively scattered distribution...

In this paper, we investigate the training process of generative networks that use a type probability density distance named particle-based as objective function, e.g. MMD GAN, Cram\'er EIEG GAN. However, these GANs often suffer from problem unstable training. analyze stability perspective dynamics. our framework, regard discriminator $D$ in feature transformation mapping maps high dimensional data into space, while generator $G$ random variables to samples resemble real terms space. This...

Understanding how different player rotations may impact team performance allows basketball coaches to select effective line-ups for specific tactical scenarios. The study aimed i) assess or combinations a team's game performance; ii) explore the variations in line-up utilization among national women's teams; and iii) examine offensive efficiency of each evolves during game. Data from 3,387 ball possessions 23 international games were collected across four major competitions over six years....

Facing with severe environmental constraints, will the promulgation of laws and regulations be conducive to promoting green growth industry in developing country? This paper tries answer it by providing an empirical study China. Based on SBM Directional Distance Function, we first calculated index its decomposition, then investigated heterogeneous effects legislation industrial resource allocation efficiency China's industry. The reveals some following findings. First, local is Meanwhile...

In this paper, a 5.05MHz, 2.06μW/MHz, 22ppm/°C relaxation oscillator working as temperature varies from -40°C to 125°C and supply voltage 1.1V 2V is proposed designed in 180nm CMOS process. By using single comparator control method reduced charging/discharging capacitance, the energy-per-cycle figure of merit reaches 2.06μW/MHz. With on-chip current reference optimized delay times, output frequency has coefficient only 22ppm/°C. The changes for 16% when 2V.

In the context of close integration economic globalization and local production networks, network topology producer service formed by linkage between manufacturing industry has attracted widespread concern. Based on research perspective complex theory, this paper theoretically analyzes impact mechanism how evolution promotes upgrading manufacturing’s value chain. addition, taking Guangdong-Hong Kong-Macao Greater Bay Area as object, further characteristics topology, such breadth, intensity...