Developing DL_POLY Molecular Dynamics Simulation code to tackle challenging problems in science and technology

分子动力学（MD）模拟是当今广泛应用于了解许多系统性质的重要工具，与实验和理论相结合。在一些重要的应用中，有必要模拟接近微米的非常大尺寸的系统。在这些尺度上，许多重要的物理过程在MD模拟中无法研究。重要的是，这些过程包括由于用于封装核废料和未来聚变反应堆的材料中的高能反冲而产生的辐射损伤效应。随着英国政府和其他人意识到能源相关和环境研究的重要性，对这些过程的研究目前非常热门。MD模拟的最新进展和尖端高性能计算的可用性，我们现在能够模拟非常大的系统。然而，对于我们有效运行和分析结果的能力，以及将这些最新进展转化为可操作和有用的模拟工具，存在一些重要的障碍。一个重要的问题是，由于生成的文件大小和写入磁盘的时间不切实际，写入轨迹文件的传统方式不再可行。另一个特定于辐射损伤模拟的问题是，我们目前的通用MD代码DL_POLY没有捕捉到该过程的基本物理特性，即电子能量损失。本提案旨在通过开发我们的高性能MD DL_POLY代码来消除这些障碍。我们将开发代码来动态地分析许多重要的物理属性，从而避免对不切实际大小的文件进行操作。我们还将实现电子能量损失的最新算法作为代码的一个新的通用特征。这将使我们能够在目前被认为是核废料封装基质的材料和未来聚变反应堆中使用的材料中运行高能辐射损伤模拟。因此，我们将获得辐射损伤对材料性能影响的物理见解。结合我们项目合作伙伴的实验，我们的MD模拟结果将提供辐照下材料行为的预测模型，并为这些材料的未来应用建立物理意义。除了对在能源和环境领域工作的学术界和工业界有益之外，我们的研究结果对更广泛的MD模拟社区也很重要。这个社区将受益于能够运行和分析非常大系统结果的MD代码的新特性。

Molecular dynamics (MD) simulation is an important tool widely used nowadays to understand the properties of many systems, alongside with experiments and theory. In several important applications, it is necessary to simulate systems of very large sizes approaching a micrometer. At these scales, many important physical processes operate that could not be studied before in MD simulations. Importantly, these processes include radiation damage effects due high energy recoils in materials used to encapsulate nuclear waste and in future fusion reactors. The research into these processes is currently very topical, as the UK government as well as others realise the importance of energy-related and environmental research.Recent advances in MD simulations and the availability of the cutting-edge high-performance computing, we are now in a position to simulate very large systems. However, there are several important obstacles related to our ability to efficiently run and analyse the results, and to turning these recent advances into an operational and useful simulation tool. One important problem is that the traditional way of writing out the trajectory file is no longer feasible due to the impractical file sizes generated and writing to disk times. Another problem that is specific to radiation damage simulations is that our current general-purpose MD code, DL_POLY, does not capture the essential physics of the process, electronic energy loss.This proposal aims at removing these obstacles by developing our high-performance MD DL_POLY code. We will develop the code to analyse many important physical properties on the fly, thus avoiding the need to operate with files of impractical sizes. We will also implement the recent algorithm of the electronic energy losses as a new general feature of the code. This will enable us to run radiation damage simulations of high energy operable in materials that are currently considered as encapsulation matrices for nuclear waste and in materials to be used in future fusion reactors. As a result, we will obtain physical insights into the impact of radiation damage on material performance. In combination with experiments by our project partners, the results of our MD simulations will provide for the predictive models of materials behaviour under irradiation and establish physical implications for the future applications of these materials.Apart from the benefit to the academic and industrial community working in the area of energy and environment, the results of our research will be important to a wider community of MD simulations. This community will benefit from the new features of the MD code capable of running and analysing the results of very large systems.