اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدActive-learning-based efficient prediction of ab-initio atomic energy: a case study on a Fe random grain boundary model with millions of atoms
79
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
We have developed a method that can analyze large random grain boundary (GB) models with the accuracy of density functional theory (DFT) calculations using active learning. It is assumed that the atomic energy is represented by the linear regression of the atomic structural descriptor. The atomic energy is obtained through DFT calculations using a small cell extracted from a huge GB model, called replica DFT atomic energy. The uncertainty reduction (UR) approach in active learning is used to efficiently collect the training data for the atomic energy. In this approach, atomic energy is not required to search for candidate points; therefore, sequential DFT calculations are not required. This approach is suitable for massively parallel computers that can execute a large number of jobs simultaneously. In this study, we demonstrate the prediction of the atomic energy of a Fe random GB model containing one million atoms using the UR approach and show that the prediction error decreases more rapidly compared with random sampling. We conclude that the UR approach with replica DFT atomic energy is useful for modeling huge GBs and will be essential for modeling other structural defects.
قيم البحث
اقرأ أيضاً
In this work, we propose an efficient computational scheme for first-principle quantum transport simulations to evaluate the open-boundary conditions. Its partitioning differentiates from conventional methods in that the contact self-energy matrices
are constructed on smaller building blocks, principal layers (PL), while conventionally it was restricted to have the same lateral dimensions of the adjoining atoms in a channel region. Here, we obtain the properties of bulk electrodes through non-equilibrium Greens function (NEGF) approach with significant improvements in the computational efficiency without sacrificing the accuracy of results. To exemplify the merits of the proposed method we investigate the carrier density dependency of contact resistances in silicon nanowire devices connected to bulk metallic contacts.
Grain boundaries (GBs), an important constituent of polycrystalline materials, have a wide range of manifestion and significantly affect the properties of materials. Fully understanding the effects of GBs is stalemated due to lack of complete knowled
ge of their structures and energetics. Here, for the first time, by taking graphene as an example, we propose an analytical energy functional of GBs in angle space. We find that an arbitrary GB can be characterized by a geometric combination of symmetric GBs that follow the principle of uniform distribution of their dislocation cores in straight lines. Furthermore, we determine the elusive kinetic effects on GBs from the difference between experimental statistics and energy-dependent thermodynamic effects. This study not only presents an analytical energy functional of GBs which could also be extended to other two-dimensional materials, but also sheds light on understanding the kinetic effects of GBs in material synthesizing processes.
Beyond the conventional trial-and-error method, machine learning offers a great opportunity to accelerate the discovery of functional materials, but still often suffers from difficulties such as limited materials data and unbalanced distribution of t
arget property. Here, we propose the ab initio Bayesian active learning method that combines active learning and high-throughput ab initio calculations to accelerate prediction of desired functional materials with the ultrahigh efficiency and accuracy. We apply it as an instance to a large family (3,119) of two-dimensional hexagonal binary compounds with unbalanced materials property, and accurately screen out the materials with maximal electric polarization and proper photovoltaic band gaps, respectively, whereas the computational costs are significantly reduced by only calculating a few tenths of possible candidates in comparison to the random search. This approach shows enormous advantages for the cases with unbalanced distributions of target property. It can be readily applied to seek for a broad range of advanced materials.
A theoretical study of the surface energy-loss function of freestanding Pb(111) thin films is presented, starting from the single monolayer case. The calculations are carried applying the linear response theory, with inclusion of the electron band st
ructure by means of a first-principles pseudopotential approach using a supercell scheme. Quantum-size effects on the plasmon modes of the thinnest films are found in qualitative agreement with previous work based on the jellium model. For thicker films, results show a dispersionless mode at all thicknesses, in agreeement with electron energy-loss measurements. For sizeable values of the momentum, the raising of the surface plasmon with increasing thickness is retrieved.
We have investigated the initial growth of Fe on GaAs(110) by means of density functional theory. In contrast to the conventionally used (001)-surface the (110)-surface does not reconstruct. Therefore, a flat interface and small diffusion can be expe
cted, which makes Fe/GaAs(110) a possible candidate for spintronic applications. Since experimentally, the actual quality of the interface seems to depend on the growth conditions, e.g., on the flux rate, we simulate the effect of different flux rates by different Fe coverages of the semiconductor surface. Systems with low coverages are highly diffusive. With increasing amount of Fe, i.e., higher flux rates, a flat interface becomes more stable. The magnetic structure strongly depends on the Fe coverage but no quenching of the magnetic moments is observed in our calculations.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
