Abstract:

This article studies the asymmetric harmonic oscillator with four degrees of freedom. Solve the mode (resonance state) to obtain the relative amplitude of each particle, and the ratio of vibration kinetic energy. At the moment of thermal equilibrium of the oscillator, the total vibrational kinetic energy of each mode is 0.5kT. Kinetic energy of each particle is different, indicating that there is a temperature difference at different positions of the harmonic oscillator. This violates the 2nd and 0th laws of thermodynamics. By placing an AHO in the middle of 1D gas and conducting molecular dynamics analysis, the ratio of the average kinetic energy of the gases on both   sides is 1.41 so violates laws again.  SM assumes that the probability of a system's state depends only on its energy, leading to uniform equilibrium distributions. Dynamic interaction like collision and short-term effect influence probabilities challenging main core.

 

Skyrmions are topologically protected spin textures that have attracted significant attention for their potential use in compact, low-power spintronic devices. Among the most promising platforms for realizing skyrmions are two-dimensional van der Waals magnetic heterostructures, where interfacial interactions can stabilize chiral magnetic states. In this talk, I will focus on a particular system to illustrate how material properties and external conditions influence skyrmion behavior in these layered structures.

 

The proposed research aims to develop ensemble machine learning models to accurately predict the reactivity pulse values in an MTR (Materials Testing Reactor) type nuclear reactor, significantly enhancing operational safety, efficiency, and reliability. When a reactor goes offline, whether for maintenance, refueling, or addressing operational issues, the ability to predict and control reactivity pulses is crucial for ensuring safe shutdowns and minimizing radiation risks. This study utilizes historical data from MTR-type reactors, including reactor configurations, experimental conditions, and other relevant parameters. The proposed ensemble approach integrates multiple machine learning models, such as Random Forests, Gradient Boosting Machines, and Support Vector Machines, to capture complex relationships within the data and improve prediction accuracy and robustness over single-model approaches. This research focuses on careful preprocessing of the dataset, feature selection, and the design of an optimal ensemble framework to enhance the predictive capabilities of the models. By providing accurate reactivity pulse predictions, this research will contribute to better maintenance scheduling, optimized fuel management, minimized radiation risks, regulatory compliance, and reduced economic losses due to unplanned shutdowns. The ensemble model’s performance will be evaluated against traditional methods and single neural network models, using metrics such as prediction accuracy, stability, and interpretability. The outcome will offer a powerful and reliable tool for forecasting reactivity pulse values in MTR-type reactors, supporting safer and more efficient nuclear reactor operations.