Zheng Weimin

Potential-based hierarchical clustering

When performing hierarchical clustering, some metric must be used to determine the similarity between pairs of clusters. Traditional similarity metrics either can only deal with simple shapes or are very sensitive to outliers. We propose two potential-based similarity metrics, APES and AMAPES, inspired by the concept of electric potential in physics. The main features of these metrics are: they have strong anti-jamming capability; and they are capable of finding clusters of complex irregular shapes.

The compiler for supporting multithreading in cyclic register windows

This paper presents the compilation technique to support the P&M (pipelining and multithreading) multithreaded program execution model on the SPARC processor, which has cyclic register windows. The program execution model improves the performance by running multiple user-level threads on each processor and filling the idle cycles that would otherwise be wasted waiting for messages. The integrated compilation paradigm is described systematically. The generated code is efficient, which obtains rapid context switching without sacrificing the benefit of fast procedure call.

A new numerical method on American option pricing

Mathematically, the Black-Scholes model of American option pricing is a free boundary problem of partial differential equation. It is well known that this model is a nonlinear problem, and it has no closed form solution. We can only obtain an approximate solution by numerical method, but the precision and stability are hard to control, because the singularity at the exercise boundary near expiration date has a great effect on precision and stability for numerical method. We propose a new numerical method, FDA method, to solve the American option pricing problem, which combines advantages the Semi-Analytical Method and the Front-Fixed Difference Method. Using the FDA method overcomes the difficulty resulting from the singularity at the terminal of optimal exercise boundary. A large amount of calculation shows that the FDA method is more accurate and stable than other numerical methods.

Research on data replica distribution pattern for web server clusters

Today more and more WWW applications ask for basically non-stop services around the clock. Finding a good highly available scheme for a cluster web server becomes very important. Web server clusters with different data replica distribution pattern have different data availability level. Therefore our research focuses on data replica distribution pattern. After making theoretical analysis on data replica distribution patterns, an optimized one is put forward in this paper.

Design and Implementation of an Efficient Multi-version File System

The benefit of multi-version file system is high reliability. File versioning can be used for file-oriented recovery from deletion and limiting exposure to data losses during file system failure. The problem is that accesses to versions in the distant past may be prohibitively expensive. We adopt the hierarchical architecture in version space to reduce the time spent on search; We present a new structure named red black tree embedded in inode to organize versions of a directory and another new structure named red black tree with weight and link to build index for versions of a file. Using aforementioned technologies, we implement an efficient multi-version file system named thvfs. The experiment results show that the average time of searching old versions in thvfs is 34.4% lower than that in ext3cow, the famous multi-version file system.

Orthogonal replication algorithm on cluster storage system

This paper presents a replication algorithm, which can be embedded in storage APIs provided by cluster storage systems and can improve the availability of storage. Compared with previous methods, this orthogonal algorithm is independent of data types and upper-level applications. It implements three data-consistency criteria for developers to select for their applications. Availability is object based and can be dynamically adjusted. Moreover, low latency of commands is obtained by reducing the communication among replicas as much as possible. We implemented this method partially in TODS, which is a distributed object persistent system running on COCs.

A Java Virtual Machine design based on hybrid concurrent compilation model

The shift from interpretation to just-in-time (JIT) compilation has improved Javas performance greatly. However, further amelioration is still possible. A new execution model of Java-the hybrid concurrent compilation model (HCCM)-is proposed in this paper, which possesses the potential to surpass JIT by overlapping the production of native code with program execution through multithreaded control. We also present the design of our ongoing HCCM-based Java Virtual Machine (JVM), discuss some newly-emerged technical issues in the implementation of such a JVM and make some predictions about its performance.

Parallel computing for lattice Monte Carlo simulation of large-scale thin film growth

This paper proposes two viable computing strategies for distributed parallel systems: domain division with sub-domain overlapping and asynchronous communication. We have implemented a parallel computing procedure for simulation of Ti thin film growing process of a system with 1000 × 1000 atoms by means of the Monte Carlo (MC) method. This approach greatly reduces the computation time for simulation of large-scale thin film growth under realistic deposition rates. The multi-lattice MC model of deposition comprises two basic events: deposition, and surface diffusion. Since diffusion constitutes more than 90% of the total simulation time of the whole deposition process at high temperature, we concentrated on implementing a new parallel diffusion simulation that reduces communication time during simulation. Asynchronous communication and domain overlapping techniques are used to reduce the waiting time and communication time among parallel processors. The parallel algorithms we propose can simulate the thin film growth of a system with many more particles than before under realistic deposition rates, and can provide a more efficient means for computer simulation of thin film growth.

Task scheduling of parallel programs to optimize communications for cluster of SMPs

This paper discusses the compile time task scheduling of parallel program running on cluster of SMP workstations. Firstly, the problem is stated formally and transformed into a graph partition problem and proved to be NP-Complete. A heuristic algorithm MMP-Solver is then proposed to solve the problem. Experiment result shows that the task scheduling can reduce communication overhead of parallel applications greatly and MMP-Solver outperforms the existing algorithms.This paper discusses the compile time task scheduling of parallel program running on cluster of SMP workstations. Firstly, the problem is stated formally and transformed into a graph partition problem and proved to be NP-Complete. A heuristic algorithm MMPSolver is then proposed to solve the problem. Experiment result shows that the task scheduling can reduce communication overhead of parallel applications greatly and MMPSolver outperforms the existing algorithms.

On communication patterns of parallel objects

Object-oriented programming is proving to be quite useful in many areas which have not been served well by more conventional approaches. Object-oriented systems show great potential for exploiting concurrence/parallelism. But many object-oriented languages currently in use are sequential. How to implement object-oriented languages on parallel systems is still a promising research topic. A parallel object-oriented system is a single-user system which is implemented on multi-processors. Objects can be active concurrently in one processor or do things on different processors in parallel. There are two communication patterns: synchronous and asynchronous communication. We put emphasis on communication patterns of parallel objects. We also discuss several ways to achieve performance in object-oriented languages and how to implement communication patterns on transputer-based multicomputer systems.<<ETX>>