Markov chain Monte Carlo simulations, combining the sampling of the position of the particles and their chemical nature, are very useful when calculating, for example, average site occupancies at crystalline defects in alloys. Unfortunately, when the relaxations around the solutes are large, the exchange moves can be systematically rejected because of atoms overlapping. As a consequence, the simulations are often trapped in nonphysical configurations. In this paper, the “Smart Darting” method from Andricioaei et al. is adapted and extended to propose a solution to this limitation. The method is tested in a particularly demanding case: the sampling of the arrangements of delocalized vacancies and divacancies in grain boundaries, both in the fcc and the bcc structure. Beyond the methodological aspects, intergranular vacancy clusters are interesting in several contexts, such as ductile fracture, irradiation, or thin-film dewetting, and therefore several properties have been measured that can be useful for mesoscale modeling: segregation energies, effective diffusion barriers in and out of the grain boundaries, vacancy-vacancy binding energies, and elastic dipole tensors.

中文翻译：

---

使用 Smart Darting 对具有较大松弛度的空位配置进行采样

马尔可夫链蒙特卡罗模拟结合了颗粒位置采样及其化学性质，在计算合金中晶体缺陷的平均位点占用率等时非常有用。不幸的是，当溶质周围的弛豫很大时，由于原子重叠，交换运动可能会被系统地拒绝。因此，模拟常常陷入非物理配置中。在本文中，Andricioaei等人的“智能飞镖”方法。被修改和扩展以提出针对此限制的解决方案。该方法在特别苛刻的情况下进行了测试：对 fcc 和 bcc 结构中晶界中的离域空位和双空位的排列进行采样。除了方法论方面之外，晶间空位簇在多种情况下也很有趣，例如延性断裂、辐照或薄膜去湿，因此测量了一些可用于介观建模的属性：偏析能、有效扩散势垒和晶界之外、空位-空位结合能和弹性偶极子张量。