A5005925476- Machine Learning in Materials Science
- Fuel Cells and Related Materials
- Protein Structure and Dynamics
- Advanced Chemical Physics Studies
- Viral Infections and Vectors
- X-ray Diffraction in Crystallography
- Topic Modeling
- Quantum, superfluid, helium dynamics
- Advanced Memory and Neural Computing
- Diamond and Carbon-based Materials Research
- Erythropoietin and Anemia Treatment
- Meningioma and schwannoma management
- Advanced NMR Techniques and Applications
- Intracranial Aneurysms: Treatment and Complications
- Bayesian Modeling and Causal Inference
- Aortic Disease and Treatment Approaches
- Virology and Viral Diseases
- Infective Endocarditis Diagnosis and Management
- Nursing Education, Practice, and Leadership
- Model-Driven Software Engineering Techniques
- Nuclear physics research studies
- Cold Atom Physics and Bose-Einstein Condensates
- Respiratory viral infections research
- Iron Metabolism and Disorders
- Infrastructure Maintenance and Monitoring
Xi'an Medical University
2024
Air Force Medical University
2020-2024
Peking University
1984-2024
Southeast University
2022-2023
Xijing Hospital
2019
Wuhan University of Science and Technology
2015
Institute of Theoretical Physics
1984
DeePMD-kit is a powerful open-source software package that facilitates molecular dynamics simulations using machine learning potentials known as Deep Potential (DP) models. This package, which was released in 2017, has been widely used the fields of physics, chemistry, biology, and material science for studying atomistic systems. The current version offers numerous advanced features, such DeepPot-SE, attention-based hybrid descriptors, ability to fit tensile properties, type embedding, model...
For 35 years, ab initio molecular dynamics (AIMD) has been the method of choice for modeling complex atomistic phenomena from first principles. However, most AIMD applications are limited by computational cost to systems with thousands atoms at most. We report that a machine learning based simulation protocol (Deep Potential Molecular Dynamics), while retaining accuracy, can simulate more than 1 nanosecond-long trajectory over 100 million per day, using highly optimized code (GPU DeePMD-kit)...
Machine-learning-based interatomic potential energy surface (PES) models are revolutionizing the field of molecular modeling. However, although much faster than electronic structure schemes, these suffer from costly computations via deep neural networks to predict and atomic forces, resulting in lower running efficiency as compared typical empirical force fields. Herein, we report a model compression scheme for boosting performance Deep Potential (DP) model, learning-based PES model. This...
High-performance computing, together with a neural network model trained from data generated first-principles methods, has greatly boosted applications of ab initio molecular dynamics in terms spatial and temporal scales on modern supercomputers. Previous state-of-the-art can achieve 1 -- 2 nanoseconds simulation per day for 100-million atoms the entire Summit supercomputer. In this paper, we have significantly reduced memory footprint computational time by comprehensive approach both...
ABACUS (Atomic-orbital Based Ab-initio Computation at USTC) is an open-source software for first-principles electronic structure calculations and molecular dynamics simulations. It mainly features density functional theory (DFT) compatible with both plane-wave basis sets numerical atomic orbital sets. serves as a platform that facilitates the integration of various methods, such Kohn-Sham DFT, stochastic orbital-free real-time time-dependent etc. In addition, aid high-performance computing,...
We perform a systematic study on the structure and dynamics of warm dense aluminum (Al) at temperatures ranging from 0.5 to 5.0 eV with molecular utilizing both density functional theory (DFT) deep potential (DP) method. On one hand, unlike Thomas-Fermi kinetic energy (KEDF), we find that orbital-free DFT method Wang-Teter non-local KEDF yields properties Al agree well Kohn-Sham method, enabling accurate simulations relatively low temperatures. other DP constructs neural network has high...
For 35 years, {\it ab initio} molecular dynamics (AIMD) has been the method of choice for modeling complex atomistic phenomena from first principles. However, most AIMD applications are limited by computational cost to systems with thousands atoms at most. We report that a machine learning-based simulation protocol (Deep Potential Molecular Dynamics), while retaining accuracy, can simulate more than 1 nanosecond-long trajectory over 100 million per day, using highly optimized code (GPU...
Abstract Molecular dynamics (MD) is an indispensable atomistic-scale computational tool widely-used in various disciplines. In the past decades, nearly all ab initio MD and machine-learning have been based on general-purpose central/graphics processing units (CPU/GPU), which are well-known to suffer from their intrinsic “memory wall” “power bottlenecks. Consequently, nowadays calculations with accuracy extremely time-consuming power-consuming, imposing serious restrictions simulation size...
Iron homeostasis is crucial for optimal cardiac function.Iron deficiency and overload have been linked to the development of cardiomyopathy heart failure (HF) via intricate mechanisms.Although role SLC40A1 in iron metabolism by facilitating efflux cellular has confirmed, its specific molecular functions cardiovascular diseases remain poorly understood.In this study, we generated mice with inducible cardiomyocyte-specific overexpression first time.The cardiomyocytes adult resulted significant...
A detailed understanding of the material response to rapid compression is challenging and demanding. For instance, element gold under dynamic exhibits complex phase transformations where there exist some large discrepancies between experimental theoretical studies. Here, we combined large-scale molecular dynamics simulations with a deep potential elucidate processes from an atomic level. The constructed by accurately reproducing free energy surfaces density-functional-theory calculations for...
DeePMD-kit is a powerful open-source software package that facilitates molecular dynamics simulations using machine learning potentials (MLP) known as Deep Potential (DP) models. This package, which was released in 2017, has been widely used the fields of physics, chemistry, biology, and material science for studying atomistic systems. The current version offers numerous advanced features such DeepPot-SE, attention-based hybrid descriptors, ability to fit tensile properties, type embedding,...
Machine-learning-based interatomic potential energy surface (PES) models are revolutionizing the field of molecular modeling. However, although much faster than electronic structure schemes, these suffer from costly computations via deep neural networks to predict and atomic forces, resulting in lower running efficiency as compared typical empirical force fields. Herein, we report a model compression scheme for boosting performance Deep Potential (DP) model, learning based PES model. This...
The clinical features and therapeutic measures of vestibular schwannoma (VS) radiation-related aneurysm (RRA) have not been well described. We reported the first VS RRA case admitted for acute anterior inferior cerebella artery (AICA) ischemic symptoms. Literature was reviewed to present research fruits about RRAs, some advices were given. A 54-year-old woman who had undergone GKS 10 years previously a right our hospital in 2018 because sudden onset severe vertigo vomiting, accompanied with...
NFTs (Non-Fungible Tokens) are a type of digital asset based on blockchain technology that has become an attractive investment tool for many investors. This paper proposes model knowledge graphs and LSTM (Long Short-Term Memory) networks to provide investors with ranking prediction future NFT returns, assisting traders in making more accurate decisions the market. To verify usability accuracy model, we collected data from 345 different types OpenSea platform conducted experimental...
探讨3D打印技术辅助主动脉瓣狭窄拟行经皮主动脉瓣置换术(TAVR)术前评估的可行性和有效性。首先获取主动脉瓣狭窄患者CT数据，利用Mimics 21.0软件分析重建介入路径，3D打印主动脉狭窄瓣膜模型，行术前评估与手术体外模拟。然后进行DSA造影下经皮介入置入主动脉瓣膜，术后即刻行超声心动图评价瓣膜释放情况，术后3个月CT结果评价瓣膜情况，并利用CT数据重建3D打印模型，进行术后评估。结果发现3D打印模型不仅能够准确辅助选择介入主动脉瓣类型及型号，而且能够评估支架不同位置导致的主动脉瓣形变情况，从而辅助手术方案选择。所以，3D打印技术辅助重度主动脉瓣狭窄治疗方案选择具有可行性与有效性。