We present our latest advancements of machine-learned potentials (MLPs) based on the neuroevolution potential (NEP) framework introduced in Fan et al. [Phys. Rev. B 104, 104309 (2021)] and their implementation in the open-source package gpumd. We increase the accuracy of NEP models both by improving the radial functions in the atomic-environment descriptor using a linear combination of Chebyshev basis functions and by extending the angular descriptor with some four-body and five-body contributions as in the atomic cluster expansion approach. We also detail our efficient implementation of the NEP approach in graphics processing units as well as our workflow for the construction of NEP models and demonstrate their application in large-scale atomistic simulations. By comparing to state-of-the-art MLPs, we show that the NEP approach not only achieves above-average accuracy but also is far more computationally efficient. These results demonstrate that the gpumd package is a promising tool for solving challenging problems requiring highly accurate, large-scale atomistic simulations. To enable the construction of MLPs using a minimal training set, we propose an active-learning scheme based on the latent space of a pre-trained NEP model. Finally, we introduce three separate Python packages, viz., gpyumd, calorine, and pynep, that enable the integration of gpumd into Python workflows.
21 June 2010 in Proceedings of the National Academy of Sciences
Solvation of molecules in water is at the heart of a myriad of molecular phenomena and of crucial importance to understanding such diverse issues as chemical reactivity or biomolecular function. Complementing well-established approaches, it has been shown that laser spectroscopy in the THz frequency domain offers new insights into hydration from small solutes to proteins. Upon introducing spatially-resolved analyses of the absorption cross section by simulations, the sensitivity of THz spectroscopy is traced back to characteristic distance-dependent modulations of absorption intensities for bulk water. The prominent peak at ≈200 cm -1 is dominated by first-shell dynamics, whereas a concerted motion involving the second solvation shell contributes most significantly to the absorption at about 80 cm -1 ≈2.4 THz. The latter can be understood in terms of an umbrella-like motion of two hydrogen-bonded tetrahedra along the connecting hydrogen bond axis. Thus, a modification of the hydrogen bond network, e.g., due to the presence of a solute, is expected to affect vibrational motion and THz absorption intensity at least on a length scale that corresponds to two layers of solvating water molecules. This result provides a molecular mechanism explaining the experimentally determined sensitivity of absorption changes in the THz domain in terms of distinct, solute-induced dynamical properties in solvation shells of (bio)molecules—even in the absence of well-defined resonances.
We present a detailed investigation on optical properties of high-pressure phase ZnO in (NaCl) and (CsCl) structures, including dielectric function, refractive index, absorption, and electron energy-loss spectrum. Theoretical calculations are performed using the ab initio pseudopotential density functional method, in which we employ the Perdew-Burke-Eruzerhof form of the generalized gradient approximation available in the CASTEP code together with plane wave basis sets for expanding the periodic electron density. Both structures are optimized under the respective structural phase transition pressures; for the structure it is , which has been verified to be in agreement with theory [Jaffe et al., Phys. Rev. B 62, 1660 (2000)] and experiment [Desgreniers, Phys. Rev. B 58, 14102 (1998)], while for the structure a transition pressure of is predicted in theory [Jaffe et al., Phys. Rev. B 62, 1660 (2000)]. We find that their electronic structures and optical properties under high pressure are quite different from those under ambient pressure.
Machine-learned potentials (MLPs) have exhibited remarkable accuracy, yet the lack of general-purpose MLPs for a broad spectrum of elements and their alloys limits their applicability. Here, we present a promising approach for constructing a unified general-purpose MLP for numerous elements, demonstrated through a model (UNEP-v1) for 16 elemental metals and their alloys. To achieve a complete representation of the chemical space, we show, via principal component analysis and diverse test datasets, that employing one-component and two-component systems suffices. Our unified UNEP-v1 model exhibits superior performance across various physical properties compared to a widely used embedded-atom method potential, while maintaining remarkable efficiency. We demonstrate our approach’s effectiveness through reproducing experimentally observed chemical order and stable phases, and large-scale simulations of plasticity and primary radiation damage in MoTaVW alloys.
We have grown nearly freestanding single-layer on graphitized -SiC(0001) by using molecular beam epitaxy (MBE), and characterized its electronic structure with scanning tunneling microscopy/spectroscopy (STM/STS). The existence of topological edge states at the periphery of single-layer islands was confirmed. Surprisingly, a bulk band gap at the Fermi level and insulating behaviors were also found in single-layer at low temperature, which are likely associated with an incommensurate charge order transition. The realization of two-dimensional topological insulators (2D TIs) in single-layer transition-metal dichalcogenide provides a promising platform for further exploration of the 2D TIs’ physics and related applications.
29 February 2016 in Proceedings of the National Academy of Sciences
Significance: Three-dimensional (3D) Dirac semimetals have attracted a lot of advanced research recently on many exotic properties and their association with crystalline and electronic structures under extreme conditions. As one of the fundamental state parameters, high pressure is an effective, clean way to tune lattice as well as electronic states, especially in quantum states, thus their electronic and magnetic properties. In this paper, by combining multiple experimental probes (synchrotron X-ray diffraction, low-temperature transport under magnetic field) and theoretical investigations, we discover the pressure-induced 3D Dirac semimetal to superconductor transition in ZrTe 5 .