The severe shuttle effect of soluble polysulfides hinders the development of lithium–sulfur batteries. Herein, we develop a three-dimensionally ordered macro/mesoporous (3DOM) Nb2O5/Nb4N5 heterostructure, which combines the strong adsorption of Nb2O5 and remarkable catalysis effect of Nb4N5 by the promotion “adsorption-transformation” mechanism in sulfur reaction. Furthermore, the high electrocatalytic activity of Nb4N5 facilitates ion/mass transfer during the charge/discharge process. As a result, cells with the S-Nb2O5/Nb4N5 electrode delivered outstanding cycling stability and higher discharge capacity than its counterparts. Our work demonstrates a new routine for the multifunctional sulfur host design, which offers great potential for commercial high-performance lithium–sulfur batteries.
We propose a new class of random feature methods for linearizing softmax and Gaussian kernels called hybrid random features (HRFs) that automatically adapt the quality of kernel estimation to provide most accurate approximation in the defined regions of interest. Special instantiations of HRFs lead to well-known methods such as trigonometric (Rahimi & Recht, 2007) or (recently introduced in the context of linear-attention Transformers) positive random features (Choromanski et al., 2021b). By generalizing Bochner's Theorem for softmax/Gaussian kernels and leveraging random features for compositional kernels, the HRFmechanism provides strong theoretical guarantees -unbiased approximation and strictly smaller worst-case relative errors than its counterparts. We conduct exhaustive empirical evaluation of HRF ranging from pointwise kernel estimation experiments, through tests on data admitting clustering structure to benchmarking implicit-attention Transformers (also for downstream Robotics applications), demonstrating its quality in a wide spectrum of machine learning problems.
In the subcooled boiling flow under low-pressure conditions, bubble characteristic diameter is of great influence on the surface heat transfer coefficient. However, large errors are still found in calculations using traditional mechanistic models or empirical correlations, especially for wide experimental condition. In this paper, we propose a widely applicable data-driven model using artificial neural networks (ANN) to predict the bubble maximum diameter and investigate the effect of experimental conditions. After a series of analyses on structural parameters and input parameters, the ANN model is established and validated based on six available experimental databases. The result shows that the relative error is around 14%. Uncertainty analysis is carried out for the four experimental conditions and two structural conditions. The results show the measuring accuracy of pressure is one of the most sensitive parameters on the prediction of bubble maximum diameter in the subcooled boiling flow under 1.0 MPa, especially for the bubble sizes larger than 0.5 mm. According to the results of uncertainty analysis, a new correlation is proposed for coefficients C and φ, which are used to express the effect of pressure and fluid dynamic. The new correlation works well for all the experimental databases, and the error for bubble datasets of large size is also modified. Furthermore, another independent validation with a low relative error to 14% is provided to prove the accuracy of the new correlation.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.