Zhang Linbo

Direct numerical simulation of transition and turbulence in compressible mixing layer

The three-dimensional compressible Navier-Stokes equations are approximated by a fifth order upwind compact and a sixth order symmetrical compact difference relations combined with three-stage Ronge-Kutta method. The computed results are presented for convective Mach numberMc = 0.8 andRe = 200 with initial data which have equal and opposite oblique waves. From the computed results we can see the variation of coherent structures with time integration and full process of instability, formation of A -vortices, double horseshoe vortices and mushroom structures. The large structures break into small and smaller vortex structures. Finally, the movement of small structure becomes dominant, and flow field turns into turbulence. It is noted that production of small vortex structures is combined with turning of symmetrical structures to unsymmetrical ones. It is shown in the present computation that the flow field turns into turbulence directly from initial instability and there is not vortex pairing in process of transition. It means that for large convective Mach number the transition mechanism for compressible mixing layer differs from that in incompressible mixing layer.

Model adjointization and its cost

In this article, the least program behavior decomposition method (LPBD) is put forward from a program structure point of view. This method can be extensively used both in algorithms of automatic differentiation (AD) and in tools design, and does not require programs to be evenly separable but the cost in terms of operations count and memory is similar to methods using checkpointing. This article starts by summarizing the rules of adjointization and then presents the implementation of LPBD. Next, the definition of the separable program space, based on the fundamental assumptions (FA) of automatic differentiation, is given and the differentiation cost functions are derived. Also, two constants of fundamental importance in AD, σ and μ, are derived under FA. Under the assumption of even separability, the adjoint cost of simple and deep decomposition is subsequently discussed quantitatively using checkpointing. Finally, the adjoint costs in terms of operations count and memory through the LPBD method are shown to be uniformly dependent on the depth of structure or decomposition.

Parallel algorithm design on some distributed systems

Some testing results on DAWNING-1000, Paragon and workstation cluster are described in this paper. On the home-made parallel system DAWNING-1000 with 32 computational processors, the practical performance of 1.117 Gflops and 1.58 Gflops has been measured in solving a dense linear system and doing matrix multiplication, respectively. The scalability is also investigated. The importance of designing efficient parallel algorithms for evaluating parallel systems is emphasized.

Massively parallel computing in nano-VLSI interconnect modeling and lithography simulation

VLSI is large in scale, and complex in structure. In todays nano-VLSI, serious process variations induced by the complex nanometer integrated circuit process technology may result in severe degradation of the integrated circuit performance. These factors present ever increasing challenges for present day nano-scale VLSI design. Interconnect modeling and lithography simulations rely on numerical approaches for solving large-scale Maxwells equations, of which the computational cost is extremely high. In this paper, several massively parallel computing approaches for interconnect modeling and lithography simulation are surveyed, based on adaptive finite element theory and a parallel hierarchical grid (PHG) platform. Regarding interconnect modeling, we first review the parallel adaptive finite-element method ParAFEMCap for parasitic capacitance extraction, which achieves a parallel efficiency of 75.7\% on 1536 CPU cores. In addition, we review a hybridization of the boundary integral equation method and the random walk on spheres method (BIE-WOS) for surface charge density computations for conductors or dielectric mediums. The proposed method proves to be superior to existing methods for massively parallel computing. On a supercomputer with 5120 CPU cores, BIE-WOS can achieve almost a linear parallel efficiency. Regarding lithography simulation, we propose a parallel adaptive finite-element framework method by adopting the PHG (parallel hierarchical grid) platform and a perfectly matched anisotropy uniaxial layer to handle scattering boundary conditions.

The toolbox PHG and its applications

PHG (parallel hierarchical grid) is an open source toolbox, developed by the State Key Laboratory of Scientific and Engineering Computing at the Chinese Academy of Sciences, for writing parallel adaptive finite element programs. Among the many adaptive finite element toolboxes and software packages, PHG is characterized by the following features: 1)\ The ability to provide the most recent vertex bisection-based parallel local mesh refinement of unstructured conforming tetrahedral meshes. 2)\ It supports

New Consideration on the Evaluation Model of Cluster Area Network

h

COMPARISON AND ANALYSIS OF THE EFFECTIVE OBSERVATION AREA OF THE TENSOR CSAMT BY NUMERICAL MODELING

-

並列適応有限要素ソフトウェアプラットフォームPHGとその応用【JST・京大機械翻訳】

p

ナノ集積回路配線モデリングとリソグラフィーシミュレーションにおける大規模並列計算法【JST・京大機械翻訳】

adaptive finite element computations. 3)\ It incorporates massively parallel with transparent dynamical load balancing. 4)\ It provides flexible preconditioning and solutions of large sparse linear systems of equations. In this paper, the main modules and some core algorithms of PHG are described, and some parallel applications based on PHG, including the computation of iron-loss for large power transformers, simulation of ionic transport in ion channels, parasitic extraction of interconnects in integrated circuits, ice sheet simulation, and spectral element simulation of elastic waves with PML, are introduced.