The development of efficient models for predicting specific properties through machine learning is of great importance for the innovation of chemistry and material science. However, predicting global electronic structure properties like Frontier molecular orbital highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energy levels and their HOMO–LUMO gaps from the small-sized molecule data to larger molecules remains a challenge. Here, we develop a multilevel attention neural network, named DeepMoleNet, to enable chemical interpretable insights being fused into multitask learning through (1) weighting contributions from various atoms and (2) taking the atom-centered symmetry functions (ACSFs) as the teacher descriptor. The efficient prediction of 12 properties including dipole moment, HOMO, and Gibbs free energy within chemical accuracy is achieved by using multiple benchmarks, both at the equilibrium and nonequilibrium geometries, including up to 110,000 records of data in QM9, 400,000 records in MD17, and 280,000 records in ANI-1ccx for random split evaluation. The good transferability for predicting larger molecules outside the training set is demonstrated in both equilibrium QM9 and Alchemy data sets at the density functional theory (DFT) level. Additional tests on nonequilibrium molecular conformations from DFT-based MD17 data set and ANI-1ccx data set with coupled cluster accuracy as well as the public test sets of singlet fission molecules, biomolecules, long oligomers, and protein with up to 140 atoms show reasonable predictions for thermodynamics and electronic structure properties. The proposed multilevel attention neural network is applicable to high-throughput screening of numerous chemical species in both equilibrium and nonequilibrium molecular spaces to accelerate rational designs of drug-like molecules, material candidates, and chemical reactions.
The mechanical behavior of bone is determined at all hierarchical levels, including lamellae (the basic building block of bone) that are comprised of mineralized collagen fibrils and extrafibrillar matrix. The mechanical behavior of mineralized collagen fibrils has been investigated intensively using both experimental and computational approaches. Yet, the contribution of the extrafibrillar matrix to bone mechanical properties is poorly documented. In this study, we intended to address this issue using a novel cohesive finite element (FE) model, in conjunction with the experimental observations reported in the literature. In the FE model, the extrafibrillar matrix was considered as a nanocomposite of hydroxyapatite (HA) crystals bounded through a thin organic interface modeled as a cohesive interfacial zone. The parameters required by the cohesive FE model were defined based on the experimental data reported in the literature. This hybrid nanocomposite model was tested in two loading modes (i.e. tension and compression) and under two hydration conditions (i.e. wet and dry). The simulation results indicated that (1) the failure modes of the extrafibrillar matrix predicted using the cohesive FE model were closely coincided with those experimentally observed in tension and compression tests; (2) the pre-yield deformation (i.e. internal strain) of HA crystals with respect to the applied strain was consistent with that obtained from the synchrotron X-ray scattering measurements irrespective of the loading modes and hydration status; and (3) the mechanical behavior of the extrafibrillar matrix was dictated by the properties of the organic interface between the HA crystals. Taken together, we postulate that the extrafibrillar matrix plays a major role in the pre-yield deformation and the failure mode of bone, thus, giving rise to important insights in the ultrastructural origins of bone fragility.
A cohesive zone model is employed to simulate the fracture evolution and crack propagation in polycrystalline solids. Numerical simulations of fracture growth with various cohesive zone properties are presented and the simulation results capture the fracture transition from intergranular to transgranular mode. Three different random Voronoi grain cell tessellations are presented to study the grain size effects. The simulation results show that the intergranular to transgranular fracture transition in the polycrystalline solid is sensitive to key cohesive law parameters such as fracture energy and cohesive strength along grain boundaries and in grain cells. This study also provides evidence that tensile strength of polycrystalline solid increases as grain cell size decreases.
Both the frame rate and the latency are crucial to the performance of realtime rendering applications such as video games. Spatial supersampling methods, such as the Deep Learning SuperSampling (DLSS), have been proven successful at decreasing the rendering time of each frame by rendering at a lower resolution. But temporal supersampling methods that directly aim at producing more frames on the fly are still not practically available. This is mainly due to both its own computational cost and the latency introduced by interpolating frames from the future. In this paper, we present ExtraNet, an efficient neural network that predicts accurate shading results on an extrapolated frame, to minimize both the performance overhead and the latency. With the help of the rendered auxiliary geometry buffers of the extrapolated frame, and the temporally reliable motion vectors, we train our ExtraNet to perform two tasks simultaneously: irradiance in-painting for regions that cannot find historical correspondences, and accurate ghosting-free shading prediction for regions where temporal information is available. We present a robust hole-marking strategy to automate the classification of these tasks, as well as the data generation from a series of high-quality production-ready scenes. Finally, we use lightweight gated convolutions to enable fast inference. As a result, our ExtraNet is able to produce plausibly extrapolated frames without easily noticeable artifacts, delivering a 1.5× to near 2× increase in frame rates with minimized latency in practice.
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.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.