PFHub Ideas

Daniel Wheeler

NIST

CHiMaD Phase Field Methods XVIII

2025-04-23

PFHub is Broken 

• I don't have the bandwidth to work on PFHub
• Result submission are no longer active
• PFHub is stuck between two states of functionality (both broken) 
  • Javascript to Python CLI transition was not finished
• Links are broken due to security issues
• AI Benchmark is proposed but not added (also other benchmarks)

LONG term vision for Pfhub

• PFHub is not currenlty supported
• Is it worth the effort to continue?
• Requires volunteer and/or formal support/funding
• Long term vision?
  • A central registry of phase field results with coherent submission process
  • A formal/automated method to submit new benchmark problems
  • Establish a standard for benchmark problem statements and results
  • Build metadata standard for data into codes

RO-Crate Metadata Standard

• The main bottleneck for PFHub is a lack of a metadata standard
• MaRDA working group to develop a standard metadata schema for phase field data
• Converged on RO-Crate as a standard
  • Lightweight system for packaging research data using schema.org as starting point 
• MatCore is another emerging standard, but does not address meso-scale material modeling

Way Forward

• Find a volunteer with computer science / ontology / data guru + domain knowledge 
• Generate examples for RO-Crate metadata standard
  • Communicate with RO-Crate metadata working group
  • Also interface with MatCore
• Investigate using a variety of services for research results (WorkflowHub, MDF, Zenodo)
  • Port PFHub entries to alternative database

Way Forward

• Explore generating a problem specification standard
• Formalize new  benchmark submission process
• Build registry of results and problem statements
• Build codes to access data from variety of databases using the RO-Crate standard
• Join MatCore / promote RO-Crate / focus on meso-scale modeling metadata
• What projects can we piggyback off?