Technical Specifications

Introduction

Purpose

This document lays out the general technical requirements for the Dissolve project.

Scope

This document lists the high-level technical specifications for the Dissolve project. It outlines general and context-specific requirements for the code, and is recommended reading for users and developers of the code.

Background

See the project description for an overview of the purpose and intention of Dissolve.

General Requirements

Codebase

• The source code should be written in a single language which has wide support and good performance (C++).
• The source code should be platform-independent and compilable with freely-available tools.
• The code should be as self-contained as is practically possible, requiring minimal external dependencies (except where significant functionality is gained by utilising a third-party library).

Technical

• The code should be able to utilise multi-core hardware, spanning single-cpu enclosures up to massively-parallel compute clusters through the use of appropriate parallelisation methodology.
• The code should be memory-efficient such as to allow the simulation of systems containing millions of particles on a mid-range desktop (4 core, 8 Gb memory).
• Output files written by the code should be text-based so as to ensure portability between machine architectures.

End-User

• The code must provide a GUI suitable for the setting up, monitoring, and running of simulations. This includes graphing of calculated data.
• The code should be flexible enough to be able to deal with a wide variety of different systems and use-cases, offering the capability to easily add / remove steps into the general execution pipeline.
• The code should provide a degree of extensibility by users in order to permit the addition of new algorithms, calculation types etc.
• The code should make use of human-readable, non-proprietary file formats for input wherever practically possible, so as to make setup and debugging as straightforward as possible.

Consistency

• The code should implement core algorithms of the existing Empirical Potential Structure Refinement (EPSR) code so as to provide smooth transition in the user community.