Virtual Research Environment for Materials Modelling and Data Analysis
Scientific workflows enable automation of complex simulation and data analysis, persist provenance information that improves reproducibility of research, and allow finding, analysis, and reuse of produced data. The currently most used workflow management systems are generic and require extensive programming and command-line skills. On the other hand, research in computational materials science is currently performed using general-purpose computer languages, such as C++, Fortran and Python whereby domain-specific components are used via application user interfaces (API). The productivity of scientists in managing the lifecycles of model and data can be significantly improved by using a domain-specific language instead of general-purpose languages.
In this project, we go beyond this state of the art by developing a use-case driven domain-specific language (DSL) covering common objects and frequent tasks in three methodic domains of electronic structure, molecular dynamics and phase-field modeling. The newly developed DSL will provide specific expressions for the use cases and will abstract from knowledge of the underlying workflow systems, and modeling and analysis tools. Supporting tools such as parsers, interpreters, validators, editors etc., that we will implement and provide, will ensure the productive exploitation of the DSL. We expect that existing models and analyses in these three fields will be easily accessible for a broader user community of scientists from the materials research domain.
Based on JupyterLab, we will seamlessly integrate the available computing resources, workflow systems into an interactive end-user environment. In a dedicated repository, we will provide established models and analyses in the form of Python notebooks. At a later stage, we will integrate the DSL supporting tools in to the new environment and enable using DSL notebooks. Furthermore, we will provide translations of the models and analyses in the repository based on the selected use cases into DSL notebooks. All models and analyses in the repository will be continuously verified against existing output data, using continuous integration methods and tools, and provided for productive use within the virtual research environment.
|Ivan Kondov||Steinbuch Centre for Computing (SCC)|
|Marvin Müller||Steinbuch Centre for Computing (SCC)|