INSPIRE: Adaptive Multi-Scale Modeling of Plasmas

该INSPIRE项目由数学和物理科学局物理司的等离子体物理和计算物理方案、地球科学局大气和地球空间科学司的磁层物理方案和综合活动办公室共同资助。电离气体，或科学术语等离子体，是宇宙中最常见的物质状态。例如，在太阳系中，发生太阳喷发的日冕、将喷发的等离子体和磁场从太阳带到地球的太阳风、环绕地球并保护我们免受喷发有害影响的磁层，以及通过其传播无线电通信和GPS信号并受到干扰的电离层，都由等离子体组成。了解等离子体对于预测和减轻空间天气的影响至关重要。等离子体在工程中也发挥着重要作用，例如在聚变型反应堆的设计中，聚变型反应堆有望为人类提供取之不尽的清洁能源。由于空间和时间尺度的不同以及系统行为的复杂性，等离子体动力学的计算建模是非常具有挑战性的。该项目旨在将现有等离子体模拟模型的效率提高1000倍甚至更多。如果成功，新模型将为目前甚至无法在最大的超级计算机上建模的系统提供准确和负担得起的模拟。等离子体建模有不同的方法，都有优点和缺点。最精确的动力学方法通过描述六维相空间中的全分布函数来描述等离子体的所有重要效应，但它们的计算量很大。即使在今天的超级计算机上，用动力学方法对一个大型三维系统进行建模也是遥不可及的。替代的流体类型方法用几个矩来描述等离子体的分布函数，例如密度、速度和压力。除了磁场外，求解这些量可以非常有效地完成，事实上，人们可以使用全球流体模型和合理的计算资源来模拟太阳日冕、太阳风和磁层。不幸的是，在大多数系统中，域的某些部分的流体描述不够充分，这可能会对全局解产生影响。该项目旨在以自适应和动态的方式将运动型和流体型方法结合起来。昂贵的动力学模型将被限制在流体描述不够准确的领域的一小部分，而高效的流体方法将被应用于领域的绝大多数。这种混合方法有望以完全动力学模型的极小一部分成本提供精确的解决方案。预计速度将提高1000倍甚至更多。这将使全球等离子体系统的建模具有前所未有的准确性，并极大地提高我们的理解和预测能力。

This INSPIRE project is jointly funded by the Plasma Physics and Computational Physics programs in the Physics Division in the Mathematical and Physical Sciences Directorate, the Magnetospheric Physics program in the Atmospheric and Geospace Sciences Division in the Directorate for Geosciences, and the Office of Integrative Activities. Ionized gas, or in scientific terms plasma, is the most common state of matter in the Universe. In the solar system, for example, the solar corona where solar eruptions occur, the solar wind that carries the erupted plasma and magnetic field from the Sun to the Earth, the magnetosphere surrounding the Earth and protecting us from the harmful effects of the eruption, and the ionosphere through which radio communications and GPS signals propagate and get disturbed, all consist of plasma. Understanding plasma is crucial for predicting and mitigating the effects of space weather. Plasmas also play an important role in engineering, for example in the design of fusion type reactors that promise to provide an inexhaustible source of clean energy for humanity. Computational modeling of plasma dynamics is very challenging due to the different spatial and temporal scales and the complex behavior of the system. The project is aimed at improving the efficiency of present plasma simulation models by a factor of 1000 or even more. If successful, the new model will provide accurate and affordable simulations for systems that currently cannot be modeled even on the largest supercomputers.There are different approaches for plasma modeling that all have advantages and drawbacks. The most accurate kinetic methods describe all the important effects of plasma by describing the full distribution function in a six dimensional phase space, but they have tremendous computational cost. Even on today's supercomputers, modeling a large three-dimensional system with kinetic methods is far out of reach. Alternative fluid-type methods describe the plasma distribution function with a handful of moments, such as density, velocity and pressure. Solving for these quantities in addition to the magnetic field can be done quite efficiently, and in fact one can model the solar corona, the solar wind, and the magnetosphere with global fluid models with reasonable computational resources. Unfortunately, in most systems there are some parts of the domain where the fluid description is not sufficient, and this can have consequences for the global solution. The project aims at combining the kinetic and fluid type methods in an adaptive and dynamic fashion. The expensive kinetic model will be restricted to the small parts of the domain where the fluid description is not accurate enough, while the efficient fluid methods will be employed in the vast majority of the domain. This hybrid approach promises to provide accurate solutions at a tiny fraction of the cost of the fully kinetic models. A speed up of factor of 1000 or even more is expected. This will allow modeling global plasma systems with unprecedented accuracy and vastly improve our understanding and predictive capabilities.