.
Efficient thermal management is paramount for enhancing the power density and operational longevity of proton exchange membrane fuel cells. This study systematically investigates twelve feasible coupled coolant–reactant flow-direction configurations within the same three-stage segmented cooling-channel geometry featuring straight, wavy, and tapered sections. We propose a robust data-driven framework that integrates high-fidelity physical modeling, Bayesian surrogate modeling, and multi-objective genetic algorithms to decouple complex thermo-fluid mechanisms. Case 12 is identified as the best-performing flow-direction topology, reducing the average temperature uniformity index by 45.7% relative to Case 8 based on ten Latin-hypercube-sampled coolant velocity–temperature difference combinations under identical geometric and operating conditions. The subsequent optimization determines a synergistic operating window with a velocity of 1.989 m/s and a temperature difference of 6.109 K, yielding an additional 14.53% enhancement in thermal uniformity while sustaining a high current density. The thermo-hydraulic synergistic mechanism involves reshaping the thermal boundary layer via induced secondary Dean vortices and spatial acceleration while successfully avoiding diminishing returns. This work provides a systematic design framework for coordinated flow-orientation and operating-parameter optimization in proton exchange membrane fuel cell thermal management systems.
In recent years, the role of fostering problem-solving abilities in science classrooms has taken on increasing importance, especially in the rural setting, with students often having fewer resources. The traditional teacher approach may support students with tasks in retaining information, but often does not develop student involvement in assessing, evaluating, and investigating. There is no doubt that the effect of Inquiry-Based Science Instruction (IBSI) on problem-solving kids' development in a rural primary school in China, where children of farmers are studying, has been one of the topics of research. The research goes on to take a look at the possibility of different measurements for the engagement, reasoning, and application of scientific ideas under the two teaching methods, one with and one without inquiry. A mixed-methods framework was applied to the study, in which qualitative and quantitative data were considered. The study shaped a pre- and post-test activity to investigate the improvement of problem-solving in the four domains: Improvement of Problem-Solving, Growth Compared, Reasoning & Inquiry Engagement, and diversity of strategies. Statistical approximations of meaning were guided by paired t-tests and effect sizes. For the qualitative data collection, semi-structured student interviews and classroom observations were thematically analysed to maximise the depth of student experience and teacher opinion. Results indicated that statistically meaningful improvements occurred in all four areas explored: Problem-Solving Improvement (1.5238, 1.8270); Comparative Growth (1.3982, 1.7006); Reasoning and Inquiry Engagement (1.4770, 1.7796); and Strategy Diversity (1.4732, 1.7762), which were also supported with large effect sizes. According to the qualitative results, an upsurge in student motivation and self-esteem, along with better teamwork, were the main positive impacts. All in all, the results suggest that IBSI is the one generating an active learning floor that lets children's minds and cooperation through problem-solving to flourish in the rural classrooms.
This paper introduces an intelligent method for identifying parameters of the Black-Scholes system driven by mixed fractional Brownian motion (mfBm), using deep learning techniques. Firstly, a new parameter identifier, BTNN, is proposed, combining BiLSTM's capability to capture local features with Transformer's ability to model long-term dependencies. Then, the mfBm is considered as the random influence on the Black-Scholes system, with all parameters being identified using the proposed BTNN identifier. Finally, the effectiveness and applicability of the BTNN are validated through simulations and empirical analyses. Results show that the BTNN outperforms the existing PENN and CBANN identifiers, offering more accurate parameter identification. Furthermore, this paper also discusses the advantages of the BTNN identifier in terms of its robust generalization ability, optimal selection of training sample sizes, appropriate choices for sequence length and sampling frequency, the comparative advantages of weight selection in the weighted loss function, and the effectiveness of network construction.
Quantum nonlocality, as an invaluable quantum resource, plays an indispensable role in processing numerous quantum information. Accurate characterization and effective detection of nonlocality have always been important and challenging topics in theoretical and experimental quantum information research. How to precisely identify and verify the phenomenon of quantum nonlocality in complex many-body quantum systems, and how to design more efficient detection methods for the nonlocality, have become urgent scientific issues that need to be addressed. This paper is dedicated to the detection of multipartite quantum nonlocality, with a focus on exploring how the Svetlichny inequality can be used to detect it. First, the maximum quantum violation of the Svetlichny inequality is discussed. Through construction, a quantum state p0 and a set of observables A0 are obtained, thereby achieving the maximum quantum violation of the Svetlichny inequality. It is also demonstrated how to construct other quantum states and sets of observables to achieve their maximum violation of the Svetlichny inequality, thereby clarifying that the quantum states and sets of observables that achieve the maximum quantum violation of the Svetlichny inequality are not unique. Second, in order to find more quantum states and sets of observables that violate the Svetlichny inequality, a corresponding Hamiltonian is constructed using the Svetlichny operator. This core issue of finding quantum states that violate the Svetlichny inequality is ingeniously transformed into solving the ground state of this Hamiltonian. Leveraging the powerful function approximation capability of neural networks, neural network quantum states are constructed. Two optimization algorithms, i.e. the Nelder-Mead simplex method and quantum variational Monte Carlo (VMC), are respectively adopted to optimize the network parameters in order to find the ground state energy and ground state of the Hamiltonian, thereby achieving a violation of the Svetlichny inequality and ultimately detecting nonlocal states. To ensure the efficiency and accuracy of the detection method, we conduct a comparative study of different optimization methods. By comparing the Nelder-Mead simplex method with the VMC method, we find that the VMC method is more suitable for nonlocality detection based on neural network quantum states in terms of efficiency and accuracy, providing reliable computational support for detecting many-body quantum nonlocality and the violation of the Svetlichny inequality. To verify the validity and universality of the proposed method, we detect the nonlocality of multipartite quantum pure states by using neural network quantum states and the VMC method under different Hamiltonians. The results indicate that this method successfully captures violations of the Svetlichny inequality in many-body quantum systems, thereby achieving effective detection of multipartite quantum nonlocality. This fully confirms the validity and universal potential of the VMC method in nonlocality detection based on neural network quantum states. This study not only verifies the theoretical and technical feasibility of detecting multipartite quantum nonlocality based on neural network quantum states and the VMC method, but also provides valuable new insights for detecting nonlocality. More importantly, it opens up a new research avenue for using neural networks to solve complex quantum many-body problems.
Effective implementation of Generative AI (GenAI) in education requires deeper insights into its implications for equity. This bibliometric and synthesis study investigates studies on GenAI in English as a Foreign Language (EFL) learning through quantitative bibliometric analysis, topic modelling, and mixed method genre analysis. The study maps the publication panorama, identifies research hotspots, and examines how equity issues are addressed in extant studies. Results find a rapidly increasing publication trend since 2022, with key journals, authors, and highly cited works highlighted. Seven dominant research topics are identified through topic modelling and Equity and Access ranks the fifth, suggesting a lack of sufficient attention on the issue. A further analysis of 20 closely equity-related studies revealed that equity-expressions indicating challenges to equity appear more in Literature review and Findings, whereas contributions mostly appear in Introductions. Only one study focuses on equity issue in Methodology section, suggesting that equity is not a central research concern in the literature analyzed. Overall, GenAI in EFL is framed as both opportunities and challenges to equity, but they are only mentioned rather than enquired. The study calls for more explicit attention to equity to ensure a more inclusive integration of GenAI into EFL learning and teaching.
