snap_workflow

snap_workflow is an intuitive, python-based command line tool for setting up, tracking, and organizing modular, systematic computational research on both local and HPC resources. It also includes a few python tools to provide seemless interfacing between the structure it establishes and the scripts and programs unique to each individual research project including non-python based scripts/programs.

Key Features

• Modular Pipelines: Structure your pipeline into decoupled Stages (e.g., mesh prep, solver runs, post-processing) with auto-resolved upstream stage dependencies.

• Automated Parameter Sweeps: Declare variables and generate multivariate parameter grids (SPoints) automatically.

• Resource-Aware Execution: Pack simulation jobs concurrently on local workstations using hardware resource limits (CPUs, GPUs, RAM) or submit worker allocations to Slurm HPC clusters.

• Unified Interfacing: Write stage scripts in Python or any other language, utilizing a simple status context manager for state synchronization.

• Automated Data Collection: Cleanly query, gather, and compile outputs from sweep directories into structured data files.

Getting Started

• Overview
• Quick Start
• Tutorials

Command Line Tools

• workflow

Python Tools

• SPointManager
• DataManager
• SystemParams