Identify and categorize use cases and scenarios motivating SCALE-MS design and packaging.
High level use cases for applications built on the Python client library.
Use cases encountered when implementing user interfaces or high-level applications with the Python data flow client library.
level: scripting construct / algorithm building block
Use case description: If N trajectory segments contain M conformation samples, each, create a reference to the collection of X=N*M samples.
Scenarios: Data transformation and placement scenarios depend on how the reference is consumed.
Variant: For some data flow topologies, this equates to an AllGather step before downstream operations.
Variant: In the absence of any other facilities, this is typically accomplished for GROMACS data by performing :command:`trajcat` on the ensemble trajectory files.
Python code written directly against the data flow client library.
Use cases encountered in the client library implementation, interacting with the task execution middleware in support of of client use cases.
Resource management, data placement, task discovery, and task scheduling use cases. Interactions include various architectural aspects, compute/data/communications abstractions, quality of service guarantees, and performance optimizations.
Units of work (nodes) describe the inputs and expected outputs, and identify a module able to perform the operation.
Use cases encountered when implementing operation code to be executed by the runtime.