Computer Methods for Physiological Problems

项目摘要 该项目通过以下方式提供了神经元用户社区所需的核心设施 管理开发、维护、传播和教学。神经元是一个开放的 神经元和神经电路的源模拟器，运行在OS X、MS Windows和 Unix/Linux，硬件范围从笔记本电脑到大规模并行超级计算机。更多 1900多家出版物报道了用它完成的工作。许多项目对神经元的 能力，以及更多依赖于其持续发展和支持的能力。 我们的开发计划解决了神经元用户在使用 实验驱动的细胞和网络模型，其不断增加的复杂性构成了 不断增加的计算负担。这些型号非常适合大型计算机集群，但 通过利用新硬件的体系结构，可以实现更高的性能 作为NVIDIA图形处理器和多核Intel PHI。将神经元的求解器引擎分解为 独立的CoreNEURON程序将内存使用量减少了一个数量级，并使其 可以专注于特定于体系结构的内存布局，以实现最佳性能。这一定是 现在被转换为动态可加载的库，以允许标准的解释器控制 交互模型开发和仿真分析所需的。 增加模型的复杂性还需要采取特殊措施来方便施工， 基于经验的计算模型的分析和再利用。我们将扩展神经元的 用于显示用神经元模型描述的亚细胞模型规范的ModelView工具 描述语言(NMODL)。这些规范还将用于生成等价的 符合NeuroML的低熵建模系统(LEMS)标准的模型描述。 这些变化将允许将整个细胞模型描述从神经元转换为 NeuroML，并实现神经元与其他兼容NeuroML的完全互操作 模拟器。 传播和支助活动将包括：必要的软件维护 确保用户继续能够在OS X、MS的所有组合下运行神经元 Windows、Unix/Linux以及任何版本或没有版本的Python和MPI；积极支持用户 通过电子邮件和神经元论坛；创建新的文件和出版 文章和报告；介绍课程和组织神经元使用者的会议。

Project Summary This project provides the core facility needed by the NEURON user community through management of development, maintenance, dissemination, and teaching. NEURON is an open source simulator for neurons and neural circuits that runs under OS X, MS Windows, and UNIX/Linux, on hardware ranging from laptops to massively parallel supercomputers. More than 1900 publications report work done with it. Many projects contribute to NEURON's capabilities, and many more depend on its continued development and support. Our development plan addresses challenges faced by NEURON users in working with experimentally-driven models of cells and networks, whose increasing complexity poses a growing computational burden. These models are well suited to large computer clusters, but much greater performance is achievable by exploiting the architecture of new hardware such as the NVIDIA GPU and the many core Intel PHI. Factoring out NEURON’s solver engine into a standalone CoreNEURON program cut memory use by an order of magnitude and makes it possible to focus on architecture-specific memory layout for optimal performance. This must now be transformed into a dynamically loadable library to allow the standard interpreter control required for interactive model development and simulation analysis. Increased model complexity also requires special measures to facilitate the construction, analysis, and reuse of empirically based computational models. We will extend NEURON’s ModelView tool to display subcellular model specifications described with NEURON’s model description language (NMODL). These specifications will also be used to generate equivalent model descriptions compliant with NeuroML’s Low Entropy Modeling System (LEMS) standard. These changes will allow entire cell model descriptions to be translated from NEURON to NeuroML, and enable full interoperability between NEURON and other NeuroML compatible simulators. Dissemination and support activities will include: software maintenance necessary to ensure that users continue to be able to run NEURON under all combinations of OS X, MS Windows, UNIX/Linux, and with any or no version of Python and MPI; active support of users through email and the NEURON Forum; creation of new documentation and publication of articles and reports; presenting courses and organizing meetings of NEURON users.