Data Hierarchy

The figure above describes the relationship between StudyGroups, Studys, task groups and Tasks. The concept of a Task is described here. A group of tasks form a task group. Usually, these should be closely related tasks with only minor changes between them. When tasks are restarted using autojob.next, each restarted task is added to the task group of its precedent. A collection of task groups forms a Study, and a collection of studies form a StudyGroup.

The typical use case for this data hierarchy is that every task group will correspond to a single desired output (e.g., relaxed structure, density of states) with its constituent tasks being any repeated attempts to obtain that output. The task groups which comprise a study will typically all be slight variations of the same task (e.g., changed calculation parameters) often linked by a common question. The studies which comprise a study group will typically represent variations of the same question: how does convergence change for given calculation parameters and structures?