Description
Computational studies in physics, chemistry, and materials science are frequently characterized by well-parameterized but constantly evolving data schemas. Poor management of these dynamic schemas can significantly impede computational research. Our talk showcases the signac framework, an open- source Python package designed for simple data and workflow management, particularly in high performance computing environments. The framework's flexible data model allows easy adaptation into pre-existing file-based workflows while also providing critical database functionality such as filtering, searching, and grouping data. signac also provides tools to develop complex workflows operating on its data spaces, enabling the simple, efficient, and reproducible execution of computational studies.Presenter(s): Speaker: Vyas Ramasubramani, University of Michigan Author: Carl Adorf, University of Michigan Author: Paul M. Dodd, University of Michigan Author: Bradley Dice, University of Michigan Author: Sharon C. Glotzer, University of Michigan