• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Computational methods in science and engineering: advances in computational science >Symposium on 'Development and Applications of Linear Scaling Techniques'
【24h】

Symposium on 'Development and Applications of Linear Scaling Techniques'

机译：“线性缩放技术的开发与应用”专题讨论会

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

The predictive strength of modern ab initio techniques is now widely accepted. Various properties can nowadays be determined with satisfactory accuracy, just to mention equilibrium molecular structures, electric and magnetic properties or electronic and vibrational spectra. The most successful are the methods that account for electron correlation. However, the scaling of electron correlation treatments with the number of so-called basis functions (N) is far from being linear, i.e. it usually scales as O(Nt), where t > 5. The computational cost of the Hartree-Fock and the Kohn-Sham methods in their basic formulation scales at least as O(N3). According to Moore's law, every two years the speed of computers approximately doubles. This, however, does not compensate the growing need for resources as the size of the systems under study grows rapidly every year. This is the rationale for introducing various approximations. The ultimate goal is to delevop methods that scale linearly with the system size without significant loss of accuracy. The symposium entitled "Development and application of linear scaling techniques" was organized in order to gather researchers working on the development of linear- and low-scaling methods. Most of the presentations were devoted to recent progress in the computational algorithms used in the density functional theory and their applications to large molecular systems (Thomas A. Niehaus, GuanHua Chen, Tsuyoshi Miyazaki, Lin-Wang Wang, Branislav Jansik and Takahito Nakajima). Michelle Ceriotti presented an efficient technique of the finite-temperature density matrix evaluation; recent developments in low-scaling ab initio methods were discussed by Marcin Ziolkowski and Christian Ochsenfeld; the fundamental constraints on linear scaling methods were presented by Paul G. Mezey. This volume includes summaries of six out often lectures given during the symposium.
机译：现代从头算技术的预测强度现已被广泛接受。如今，可以以令人满意的精度确定各种性质，仅提及平衡分子结构，电和磁性质或电子和振动光谱。最成功的是解释电子相关性的方法。但是，电子相关性处理与所谓的基函数（N）的缩放比例远非线性关系，即，通常按O（Nt）缩放，其中t>5。Hartree-Fock和Kohn-Sham方法的基本公式化规模至少为O（N3）。根据摩尔定律，计算机的速度每两年大约翻一番。但是，由于所研究系统的规模每年都在快速增长，因此无法满足对资源不断增长的需求。这是引入各种近似值的理由。最终目标是开发出能够随系统大小线性缩放而又不会显着降低精度的方法。为了举办研究线性和低尺度方法的研究人员，组织了名为“线性尺度技术的开发和应用”的研讨会。大多数演讲都致力于密度泛函理论中使用的计算算法的最新进展及其在大分子系统中的应用（Thomas A. Niehaus，ChenHua Chen，Tyoshi Miyazaki，Lin-Wang Wang，Branislav Jansik和Takahito Nakajima）。 Michelle Ceriotti提出了一种有效的有限温度密度矩阵评估技术。 Marcin Ziolkowski和Christian Ochsenfeld讨论了低尺度从头计算方法的最新发展。 Paul G. Mezey提出了线性缩放方法的基本约束。该卷包括研讨会期间进行的六次常讲的总结。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号