Software Engineering and Problem Solving Environments for Scientific Computing: An Introduction

Abstract

As computational infrastructure becomes more powerful and complex, there is a greater need to provide tools to support the scientific computing community to make better use of such infrastructure. The abscence of such tools is likely to lead to users not taking full advantage of newer functionality made available in computational infrastructure. It is also unlikely that a user (a scientist or a systems administrator) will know the full range of options available – or how a combination of configuration options from elements within the infrastructure (such as compute or data servers) can be usefully specified. This issue has two aspects: (1) allowing such functionality to be made available in tools without requiring the user to know about the functionality – an example includes the ability to support service discovery. In this case, the user does not really need to know where a particular service instance resides, only that a particular type of service is available for use within their application at a given point in time. (2) increasing the reliance of the user on tools – an example includes the ability to automatically compose a set of services based on the requirement outlined by a user. In this case, a user must place greater trust in the infrastructure to deliver a particular result. This aspect has not been fully realised within existing infrastructure, and remains an active research area. It involves both the need to better integrate functionality made available in computational infrastructure, but also requires a “culture change”, in the way that users perceive and make use of computational infrastructure.