. We report on our experiences in building a computational environment for tomographic image analysis for marine seismologists studying the structure and evolution of mid-ocean ridge volcanism. The computational environment is determined by an evolving set of requirements for this problem domain and includes needs for high-performance parallel computing, large data analysis, model visualization, and computation interaction and control. Although these needs are not unique in scientific computing, the integration of techniques for seismic tomography with tools for parallel computing and data analysis into a computational environment was (and continues to be) an interesting, important learning experience for researchers in both disciplines. For the geologists, the use of the environment led to fundamental geologic discoveries on the East Pacific Rise, the improvement of parallel ray tracing algorithms, and a better regard for the use of computational steering in aiding model convergence. ...

In this paper, we present the Ctadel system, a Code-generation Tool for Applications based on Differential Equations using a very high level Language specification. The Ctadel system generates efficient and vectorizable Fortran 77 code automatically from a very high level language description of a model described by partial differential equations (PDEs). The system combines algebraic simplification and powerful global common subexpression elimination to guarantee the generation of efficient code. A prototype implementation has been developed which is currently limited to explicit finite difference methods as solution method. After an informal, but detailed description of the Ctadel system, results of this prototype implementation will be presented for the time-dependent Euler equations to simulate an inviscid, compressible flow and for the calculation of the explicit dynamical tendencies within the hirlam model, which is a production code for limited area numerical weather forecasting....

The simulation of complex physical systems often involves solving a large system of partial differential equations (PDEs). We discuss how solving such a system of PDEs can be done by splitting a domain into several sub-domains and creating of a network of PDE solvers. In such a network several agents are used to: (a) control the execution of individual solvers on each sub-domain, (b) mediate between adjacent sub-domains, and (c) coordinate the execution of the ensemble. This paper presents the advantages of agentbased PDE solving and describes the implementation of a network of Partial Differential Equations, PDE, Solvers using Bond middleware for agent-based computing. 1 Overview Consider the simulation of a complex physical device such as the engine shown in Figure 1. It has several physical phenomena (combustion, waterflow, structure,...) in a complex geometry. One could attack the simulation as follows: (1) separate the device into parts that have the same physical phenome...

The aim of our system is to generate solution code from a high level specification of a financial instrument. Solution code calculates prices and hedge ratios which are partial derivatives of the price with respect to various parameters. The user manipulates a representation of the problem at the domain level, with the complexities of the computer implementation hidden. As many of the problems have no analytical solution, symbolic transformations manipulate the equations specifying the stochastic model and instrument into forms that can be solved numerically. Techniques such as finite differences, spectral methods and Monte Carlo simulation are provided in sequential and parallel versions. The mathematical correctness of the transformation steps is examinable as is the degree of error introduced by approximating transformations. We include examples for the Black-Scholes model and for the Hull White stochastic volatility model. A Linux cluster is used to price an Asian option using the Hull White SV model.

The CTADEL Code-generation Tool for Applications based on Differential Equations using high-level Language specifications is a software environment for generating multi-platform high-performance codes for partial differential equations based problems. The CTADEL system is used as an application driver for the HIRLAM numerical weather forecast system. As such, the CTADEL system can be viewed as a problem-solving environment. With the aid of the CTADEL application driver, efficient High Performance Fortran and Fortran 77 codes with message passing primitives for shared (virtual) and distributed memory parallel computer architectures have been generated from a high-level language specification of the dynamics of the HIRLAM weather forecast model. INTRODUCTION The HIRLAM weather forecast system [8] is a so-called limited area weather forecast system in operational use at several European countries. To automatically generate efficient codes for the HIRLAM system we have developed the CTAD...

The Ctadel system provides an automated means of generating specific high performance scientific codes optimized for serial, vector, or shared virtual memory and distributed memory parallel computer architectures. One of the key elements of this system is the employment of algebraic simplification techniques and powerful methods for global common subexpression elimination to guarantee the generation of efficient high performance codes for various target architectures. In this paper we present the Ctadel Code-generation Tool for Applications based on Differential Equations using high-level Language specifications. A prototype implementation has been developed which is limited to explicit finite difference methods as numerical solution method. Performance results of the codes generated with this prototype implementation will be presented for a limited area numerical weather forecast routine on various hardware platforms. These results show that generation of efficient code is well feasib...

In this paper we briefly describe the symbolic techniques employed by the Ctadel Codegeneration Tool for Applications based on Differential Equations using high-level Language specifications. Performance results of the generated codes for a limited area numerical weather forecast routine will be shown on several types of computer architectures. Keywords: high performance computing; automatic code generation; numerical weather forecasting; software engineering 1 Introduction The arrival of high-performance computer systems based on different architectures, like vector architectures, parallel architectures with shared or distributed memory, makes the adaptation of software to a specific architecture difficult and time-consuming. To achieve an efficient implementation on a given architecture it may be required to modify the software significantly, which results in most cases in a platform dependent version of the code. In particular for production codes running on a variety of architectu...

this paper we will discuss automatic code generation for high performance computer architectures in relation to environmental modeling. We will present the Ctadel code generator for scientific models and show performance results of the generated codes for a meteorological model compared with efficient hand-written production codes.

The simulation and modeling of complex physical systems often involves many components because (1) the physical system itself has components of differing natures, (2) parallel computing strategies require many (somewhat independent) components, and (3) existing simulation software applies only to simpler geometrical shapes and physical situations. We discuss how agent based networks are applied to such multi-component applications. The network agents are used to (a) control the execution of existing solvers on sub-components, (b) mediate between sub-components, and (c) coordinate the execution of the ensemble. This paper focuses on partial differential equation (PDE) models as an instance of the approach and describes the implementation of networks using the PELLPACK problem solving environment for PDEs and the Bond system for agent based computing.

Geographical problem solving environments are becoming the most popular tools which used to solve sophisticated problems. In order to overcome the obstacles caused by mutli-sourced heterogeneous spatial data, a data representation model (DRM) is designed. A data representation model is a description used to provide identification of all data elements within a system, including their attributes and the logical relationships among all data elements. In object-oriented terminology, this is viewed as a class hierarchy, and described through a graphics-based design tool. The DRM provides not only a clear description of the data, but also defines the relationships between the data that are critical to ensuring correct interpretation by users. DRM also provides a common data model to define a data representation structure for traditional data and spatial data. By the experiments, it has shown that it can satisfy the requirements of geographical modelling, analysis for geographical problem solving under today’s open, heterogeneous environment. 1.