比例方程组与MPI同步通信静态死锁检测

介绍了一种MPI程序死锁检测的静态方法以及该方法所处理的程序模型.为实现该方法,提出了比例方程组(一种特殊线性方程组)的概念并设计了求解方程组最简解的线性时空复杂度的高效算法.算法由一个四遍扫描过程与一个主控程序构成.主控程序用来处理并行计算节点计算机构成的划分.四遍扫描过程采用深度优先搜索方法确定方程组中各变元之间的比例关系.通过该算法所获得的最简解,任意多个变元之间的比例关系能在常数时间内获得.证明了该算法的正确性,并采用Java语言实现了该算法的标准程序库.该程序库目前已运行于MPI同步通信静态死锁检测的软件框架中.

MPI程序同步通信基本模型死锁检测

本文提出了MPI程序的同步通信模型及三个基本简化模型,给出了判定这些基本模型是否死锁的方法和定理并予以了严格证明.简化模型的死锁检测理论和方法是真实MPI程序死锁检测的必要基础.这些方法基于程序静态分析,必要时进行运行时检测,它们对两种简化模型可以在程序编译前确定是否死锁,对另外一种模型,在编译前可静态确定部分死锁,运行中可确定其他死锁.我们的理论可以证明MPI程序死锁检测主流算法的正确性,其方法可以减少它们对客户源代码或MPI profiling接口的修改量,从而大大降低死锁检测开销,并可在运行前判定死锁.

MPI同步通信顺序模型死锁静态检测算法

静态检测MPI程序同步通信死锁比较困难,通常需要建立程序模型.顺序模型是其他所有复杂模型的基础.通过一种映射方法将顺序模型转化为字符串集合,将死锁检测问题转化为等价的多队列字符申匹配问题,从而设计并实现了一种MPI同步通信顺序模型的静态死锁检测算法.该算法的性能优于通常的环检测方法,并能适应动态消息流.

全节点空间MPI同步通信死锁检测

针对全节点空间上MPI同步通信程序死锁检测的现有穷举方法的弱点,提出了该场景中MPI程序的两种参数化模型:一次参数化模型和二次参数化模型。通过分析一次参数化模型的一类不变式的性质,给出了该模型的死锁检测算法。提出了同构节点空间的一系列概念和性质并由此设计出该算法的几种优化方法,以加快某些特定程序模型死锁检测的过程。仿真实验表明该算法及其优化方法比现有方法的性能提高若干数量级。

Mass resolved MPI spectra of methyl iodide in the 430-490 nm region

The mass resolved multiphoton ionization (MPI) spectra of methyl iodide were obtained in the 430-490 nm region using a time-of-flight (TOF) mass spectrometer. They have the same vibrational structure, which testifies that the fragment species, in the wavelength region under study, are from the photodissociation of multiphoton ionized molecular parent ions. Some features in the spectra are identified as three-photon excitations to 6p and 7s Rydberg states of methyl iodide. Two new vibrational structures of some Rydberg states are observed. The mechanism of ionization and dissociation is also discussed. (C) 2001 Elsevier Science B.V. All rights reserved.

Fast Computing For Lurr Of Earthquake Prediction

The LURR theory is a new approach for earthquake prediction, which achieves good results in earthquake prediction within the China mainland and regions in America, Japan and Australia. However, the expansion of the prediction region leads to the refinement of its longitude and latitude, and the increase of the time period. This requires increasingly more computations, and the volume of data reaches the order of GB, which will be very difficult for a single CPU. In this paper, a new method was introduced to solve this problem. Adopting the technology of domain decomposition and parallelizing using MPI, we developed a new parallel tempo-spatial scanning program.

沿平板可压缩湍流边界层的直接数值模拟

湍流广泛存在于自然界和工程应用中。可压平板湍流是一种典型的壁湍流，对其进行直接数值模拟研究是壁湍流机理研究的基础。本文采用高精度紧致型差分格式求解三维可压N-S方程，利用MPI并行程序，直接数值模拟了来流马赫数为M的＝2．25，雷诺数为Re＝635O00／in．的空间发展的可压缩平板湍流边界层。本文主要研究了如下一些问题。方法研究方面：●针对壁湍流流动的特点，文中提出了与基于非等距网格高精度紧致型差分格式相结合的简化的迭代一隐式时间推进法，建立了可压缩平板湍流边界层直接数值模拟方法。提高了计算效率，并给出了迭代收敛的条件。●研究了不同网格加密方法对数值解分辨率的影响，指出在保证近物面附近物理平面内网格是均匀的条件下，采用坐标变换的方法也可得到近物面高分辨率的数值解，纠正了人们认为只有在基于非等距网格上才能得到高分辨率解的观点。●采用包括前缘在内的二维可压平板层流边界层直接数值模拟结果为三维可压平板湍流边界层计算提供上游边界条件，解决了国外采用的局限性很大的经验公式带来的不相容问题。物理问题的模拟研究：●在国内首次给出了超声速平板边界层从层流经bypass转披直到充分发展湍流的整个过程。所得流场统计特征（如平均速度分布，物面摩阻系数，脉动速度的均方根等）与相关理论和实验符合甚好。这些结果充分证实了本文所建立的数值方法的有效性和所得结果的可靠性。●在此基础上，文中分析了可压平板湍流边界层内的拟序结构的形成过程。研究了初始扰动对平板边界层流动转挨特征的影响，指出上游初始扰动中展向的非对称性是形成bypass转挨的关键因素，且较大的时间扰动频率使得边界层转披位置后移。这些结果为壁湍流控制提供一定依据。●在此基础上研究了可压缩效应对平均流动特征以及湍能的生成和耗散特征的影响，指出在Mc∞=2.25的情况下，Morkovin假设基本成立。但可压缩效应使得边界层内层中过渡层加宽；拟序结构有明显差别。与不可压相比上抛、下扫对不同区域的作用有所不同。在近壁区，湍动能方程中的压力-膨胀项和压力-速度相关项仍是不可忽略的量，这导致声的产生和可压缩效应对湍流的发展所产生的抑制作用。

Mechanisms of femtosecond laser-induced breakdown and damage in MgO

Single-shot laser damage threshold of MgO for 40-986 fs, 800 nm laser pulses is reported. The pump-probe measurements with femtosecond pulses were carried out to investigate the time-resolved electronic excitation processes. A theoretical model including conduction band electrons (CBE) production and laser energy deposition was applied to discuss the roles of multiphoton ionization (MPI) and avalanche ionization in femtosecond laser-induced dielectric breakdown. The results indicate that avalanche ionization plays the dominant role in the femtosecond laser-induced breakdown in MgO near the damage threshold. (c) 2005 Elsevier B.V. All rights reserved.

飞秒激光作用下透明介质材料的反射率演化

利用800nm抽运和400nm探针技术，测量了CaF2和MgO的时间分辨反射率，研究了材料的电子激发和弛豫超快动力学过程。采用耦合动力学模型，探讨了飞秒激光对透明介质材料的激发，以及材料的激发对抽运激光在材料中的传输、分布和反射特性的影响。根据这个理论模型计算了时间分辨反射率的演化，计算结果和实验结果相吻合。研究表明，多光子电离（MPI）和碰撞电离（II）在介质材料的导带电子激发中起着重要的作用。

不同功率下的飞秒激光在水中形成的脉冲压缩和超连续谱

本文研究了飞秒激光脉冲在水中的传输情况.通过改变不同的激光输入功率进行模拟，我们发现从输入功率略高于到远远高于发生自聚焦的临界功率,分别是群速度色散和多光子电离多光子吸收阻止了自聚焦导致的脉冲塌陷，当多光子电离和多光子吸收主导传输时，脉冲能被压缩到几个光学周期.在频域，多光子电离能引起很强的蓝移,而多光子吸收能对这种蓝移起到抑制作用。

Self-compression of femtosecond pulses in argon with a power close to the self-focusing threshold

Self-compression of femtosecond pulses in noble gases with an input power close to the self-focusing threshold has been investigated experimentally and theoretically. It is demonstrated that either multiphoton ionization (MPI) or space time focusing and self-steepening effects can induce pulse shortening, but they predominate at different beam intensities during the propagation. The latter effects play a key role in the final pulse self-compression. By choosing an appropriate focusing parameter, action distance of the space time focusing and self-steepening effects can be lengthened, which can promote a shock pulse structure with a duration as short as two optical cycles. It is also found that, for our calculation cases in which an input pulse power is close to the self-focusing threshold, either group velocity dispersion (GVD) or multiphoton absorption (MPA) has a negligible influence on pulse characteristics in the propagation process.

Nonlinear X-wave formation and conical emission at different powers of a femtosecond laser pulse in water

Nonlinear X-wave formation at different pulse powers in water is simulated using the standard model of nonlinear Schrodinger equation (NLSE). It is shown that in near field X-shape originally emerges from the interplay between radial diffraction and optical Kerr effect. At relatively low power group-velocity dispersion (GVD) arrests the collapse and leads to pulse splitting on axis. With high enough power, multi-photon ionization (NIPI) and multi-photon absorption (MPA) play great importance in arresting the collapse. The tailing part of pulse is first defocused by MPI and then refocuses. Pulse splitting on axis is a manifestation of this process. Double X-wave forms when the split sub-pulses are self-focusing. In the far field, the character of the central X structure of conical emission (CE) is directly related to the single or double X-shape in the near field. (c) 2007 Elsevier B.V. All rights reserved.