- Award ID(s):
- 2119103
- NSF-PAR ID:
- 10460982
- Date Published:
- Journal Name:
- arXivorg
- ISSN:
- 2331-8422
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
For decades, atomistic modeling has played a crucial role in predicting the behavior of materials in numerous fields ranging from nanotechnology to drug discovery. The most accurate methods in this domain are rooted in first-principles quantum mechanical calculations such as density functional theory (DFT). Because these methods have remained computationally prohibitive, practitioners have traditionally focused on defining physically motivated closed-form expressions known as empirical interatomic potentials (EIPs) that approximately model the interactions between atoms in materials. In recent years, neural network (NN)-based potentials trained on quantum mechanical (DFT-labeled) data have emerged as a more accurate alternative to conventional EIPs. However, the generalizability of these models relies heavily on the amount of labeled training data, which is often still insufficient to generate models suitable for general-purpose applications. In this paper, we propose two generic strategies that take advantage of unlabeled training instances to inject domain knowledge from conventional EIPs to NNs in order to increase their generalizability. The first strategy, based on weakly supervised learning, trains an auxiliary classifier on EIPs and selects the best-performing EIP to generate energies to supplement the ground-truth DFT energies in training the NN. The second strategy, based on transfer learning, first pretrains the NN on a large set of easily obtainable EIP energies, and then fine-tunes it on ground-truth DFT energies. Experimental results on three benchmark datasets demonstrate that the first strategy improves baseline NN performance by 5% to 51% while the second improves baseline performance by up to 55%. Combining them further boosts performance.more » « less
-
Abstract Developing an accurate interatomic potential model is a prerequisite for achieving reliable results from classical molecular dynamics (CMD) simulations; however, most of the potentials are biased as specific simulation purposes or conditions are considered in the parameterization. For developing an unbiased potential, a finite‐temperature dynamics machine learning (FTD‐ML) approach is proposed, and its processes and feasibility are demonstrated using the Buckingham potential model and aluminum (Al) as an example. Compared with conventional machine learning approaches, FTD‐ML exhibits three distinguished features: 1) FTD‐ML intrinsically incorporates more extensive configurational and conditional space for enhancing the transferability of developed potentials; 2) FTD‐ML employs various properties calculated directly from CMD, for ML model training and prediction validation against experimental data instead of first‐principles data; 3) FTD‐ML is much more computationally cost effective than first‐principles simulations, especially when the system size increases over 103atoms as employed in this research for ensuring reliable training data. The Al Buckingham potential developed by the FTD‐ML approach exhibits good performance for general simulation purposes. Thus, the FTD‐ML approach is expected to contribute to a fast development of interatomic potential model suitable for various simulation purposes and conditions, without limitation of model type, while maintaining experimental‐level accuracy.
-
Abstract Machine learning interatomic potentials (IPs) can provide accuracy close to that of first-principles methods, such as density functional theory (DFT), at a fraction of the computational cost. This greatly extends the scope of accurate molecular simulations, providing opportunities for quantitative design of materials and devices on scales hitherto unreachable by DFT methods. However, machine learning IPs have a basic limitation in that they lack a physical model for the phenomena being predicted and therefore have unknown accuracy when extrapolating outside their training set. In this paper, we propose a class of Dropout Uncertainty Neural Network (DUNN) potentials that provide rigorous uncertainty estimates that can be understood from both Bayesian and frequentist statistics perspectives. As an example, we develop a DUNN potential for carbon and show how it can be used to predict uncertainty for static and dynamical properties, including stress and phonon dispersion in graphene. We demonstrate two approaches to propagate uncertainty in the potential energy and atomic forces to predicted properties. In addition, we show that DUNN uncertainty estimates can be used to detect configurations outside the training set, and in some cases, can serve as a predictor for the accuracy of a calculation.
-
null (Ed.)Abstract Transition metal dichalcogenides (TMDs) offer superior properties over conventional materials in many areas such as in electronic devices. In recent years, TMDs have been shown to display a phase switching mechanism under the application of external mechanical strain, making them exciting candidates for phase change transistors. Molybdenum ditelluride (MoTe2) is one such material that has been engineered as a strain-based phase change transistor. In this work, we explore various aspects of the mechanical properties of this material by a suite of computational and experimental approaches. First, we present parameterization of an interatomic potential for modeling monolayer as well as multilayered MoTe2 films. For generating the empirical potential parameter set, we fit results from density functional theory calculations using a random search algorithm known as particle swarm optimization. The potential closely predicts structural properties, elastic constants, and vibrational frequencies of MoTe2 indicating a reliable fit. Our simulated mechanical response matches earlier larger scale experimental nanoindentation results with excellent prediction of fracture points. Simulation of uniaxial tensile deformation by molecular dynamics shows the complete non-linear stress-strain response up to failure. Mechanical behavior, including failure properties, exhibits directional anisotropy due to the variation of bond alignments with crystal orientation. Furthermore, we show the deterioration of mechanical properties with increasing temperature. Finally, we present computational and experimental evidence of an extended c-axis strain transfer length in MoTe2 compared to TMDs with smaller chalcogen atoms.more » « less
-
Abstract The Li-Sn binary system has been the focus of extensive research because it features Li-rich alloys with potential applications as battery anodes. Our present re-examination of the binary system with a combination of machine learning and ab initio methods has allowed us to screen a vast configuration space and uncover a number of overlooked thermodynamically stable alloys. At ambient pressure, our evolutionary searches identified an additional stable Li3Sn phase with a large BCC-based hR48 structure and a possible high-
T LiSn4ground state. By building a simple model for the observed and predicted Li-Sn BCC alloys we constructed an even larger viable hR75 structure at an exotic 19:6 stoichiometry. At 20 GPa, low-symmetry 11:2, 5:1, and 9:2 phases found with our global searches destabilize previously proposed phases with high Li content. The findings showcase the appreciable promise machine-learning interatomic potentials hold for accelerating ab initio prediction of complex materials.