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 we 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, GPYUMD, CALORINE, and PYNEP, which enable the integration of GPUMD into Python workflows.

No abstract

No abstract

Molybdenum borides are currently raising great expectations for superhard materials, but their crystal structures and mechanical behaviors are still under discussion. Here, we report an unexpected reduction of stability and hardness from porous hP16-MoB3 and hR18-MoB2 to dense hP20-MoB4 and hR21-Mo2B5, respectively. Furthermore, we demonstrate that this anomalous variation has its electronic origin. These findings not only manifest that the long-recognized hP20-MoB4 (hP3-MoB2) and hR21-Mo2B5 should be hP16-MoB3 and hR18-MoB2, respectively, but also challenge the general design principle for ultrahard materials only pursuing the dense transition-metal borides with high boron content.

Metal−organic frameworks (MOFs) are promising candidates as natural gas adsorbents because of their porous feature and high structural tunability. In the gas adsorption/desorption process, MOFs are often under complicated physical environments, such as varied pressure and temperature; however, limited attention has been paid to the effect of pressure on their thermal properties. In this work, taking ZIF-8 with four different functional groups (−H, −CH 3 , −Cl, and −Br) as an example, we investigate the influence of functional group substitution and pressure on the thermal conductivity of MOFs through equilibrium molecular dynamics simulations. A reduction in thermal conductivity induced by the functional group substitution is observed, which is caused by a damping effect of the acoustic mismatch. Regarding the impact of pressure, the thermal conductivity of ZIF-8 is found to decrease first with increasing hydrostatic pressure. When the pressure exceeds a critical value, a sudden rise is observed in the thermal conductivity of ZIF-8 because a phase transformation from the porous phase to the dense phase is found in this process. The complicated influence of pressure on thermal conductivity is explained by a competition between the aggravation of phonon scattering and the enhancement of volumetric heat capacity in ZIF-8 with increasing pressure. This work is expected to provide molecular insights into the functional group-and pressure-dependent thermal transport of MOFs and thus facilitate their applications in energy storage and gas absorption.