Workflow, Automation, and Data Management for Lattice Gauge Theory

30 Jul 2018, 08:00 → 31 Jul 2018, 18:00 US/Central

Theory Conference Room Wilson Hall 3 NW (Fermilab)

Ethan Neil (University of Colorado, Boulder), james simone (Fermilab)

With lattice simulations and the groups undertaking them both growing in scale over time, there are a number of emerging challenges related to sharing and management of the associated data, from primary results like gauge configurations and propagators to the various intermediate and final products associated with each analysis. A more sophisticated approach to managing data alongside the workflow of lattice simulation can enhance collaboration, ease publication of data products alongside research papers, and enable automation of common analysis tasks. This will be an informal workshop focused on discussion of best practices and software development opportunities.

Talks and discussions will be solicited from workshop participants; suggestions for topics are welcome.

Directions to Fermilab http://www.fnal.gov/pub/visiting/directions/index.html

Guest Internet Access Visitor Internet Access

Monday, 30 July

09:00 → 12:00 Data storage and design

Design and use of centralized repositories such as SQL databases in lattice calculation workflows

12:00 → 13:00 Lunch

13:00 → 17:00 Structured data storage

Use of structured outputs (XML, YAML, HDF5, etc.) to organize and store data

18:00 → 19:30 Dinner

Tuesday, 31 July

09:00 → 12:00 Workflow management tools

Workflow management software such as Pegasus, Apache Taverna, Apache Spark, and others

12:00 → 13:00 Lunch

13:00 → 16:30 Automation

Automation of common lattice "tuning" tasks (e.g. HMC parameter tuning, phase diagram exploration)