A Quickstep forward: development of the CP2K/Quickstep code and application to ice transport processes

密度泛函理论是一种原子尺度的工具，可用于了解物质的结构和行为，特别是当原子或分子相互反应时。例如，它被用来告诉我们当水变成酸或碱时，水的结构如何变化。可以说，它已经彻底改变了我们对科学的许多方面的理解，其中一位创始人（Walter Kohn）因发展了DFT计算机模拟软件的核心基础理论而获得诺贝尔奖（1998年）。自从DFT的第一次实施以来，已经开发了几种DFT，这些方法通常提高了这种方法的准确性，使科学家能够以惊人的准确度计算化学反应的能量。当计算机处理器可以并行使用以将计算划分为小的子计算时，这种方法的有用性增加。目前，英特尔正在销售他们的双核处理器的台式机和笔记本电脑，其中计算机能够分担计算负担的两个处理器。同样的原理也用于国家级超级计算机，在那里，超过1000个处理器可以用来进行非常苛刻的计算（在一个处理器上需要1000年），变成一个更容易管理的任务，在1000个处理器上需要一年，假设程序是完全有效的。在现实中，很难获得这种高效的并行性/需要使用特殊的技巧来使用计算机处理器的性能。这项申请寻求资金来开发一种流行的新软件，它可以在新的国家超级计算机上更有效地运行。一旦开发完成，我们将详细研究冰的结构和纳米级缺陷。了解冰结构中微观缺陷的结构和行为将使我们更好地了解它是如何传导的，但更普遍的是，这些缺陷如何影响冰的稳定性。后者正变得越来越热门和重要，因为我们试图了解冰如何融化，以更好地估计温度对冰川冰盖的影响。

Density Functional Theory is an atomic scale tool which can be used to learn about the structure and behaviour of substances, especially when atoms or molecules react with one another. For instance, it has been used to tell us about how the structure of water changes when it becomes acid or alkali. It is true to say that it has revolutionised our understanding in many aspects of science, and one of the originators (Walter Kohn) was awarded the Nobel Prize (in 1998) for developing the underlying theory which is at the heart of DFT computer simulation software. Since the first implementation of DFT, several flavours of DFT have been developed that have generally increase accuracy of this method, allowing scientists to calculate energies for chemical reactions with amazing accuracy. The usefulness of this method is increased when computer processors can be utilised in parallel to divide up the calculation into small sub-calculations. Currently Intel are marketing their Duo core processors for desktop and notebook computers where the computer is able to split the computational burden over two processors. The same principle is used on national supercomputers, where over 1000 processors can be used to make very demanding calculations (that would take 1000 years on one processor) into a far more manageable task, taking one year on 1000 processors, assuming the program was perfectly efficient. In reality, it is very difficult to obtain such efficient parallelism / special tricks need to be used to use the computer processor performance. This application seeks funding to develop a popular new piece of software that it can run far more efficiently on the new national supercomputer.Once the develpoment has taken place, we will look in detail at the structure and nanoscopic defects in ice. Understanding the structure and behaviour of microscopic imperfections in the ice structure will lead us to better understand how it conducts but more generally, how these defects influence the stability of ice. The latter is becoming ever more topical and important as we seek to understand how ice melts in order to better estimate the influence of temperature on glacial ice sheet.