FUSE meeting notes: Nov 5 2024

[orso82]

! VOTE !

Orso

• Setup a dev branch now and pledges to use that and not master
• Overhauled how PF/OH coils are assigned to different layers
  • More flexibility
  • More robust
• Fixed issue with long precompile times
  • Could be made smaller if we limited to IDSs that we actually use
• Faster access to time dependent arrays/structures
• Updating some internals: @checkin and @chechout
• Registration for code camp is now closed
• Quite some interest by ICF community in using FUSE

Tim

• Testing time dependence on SPARC case
  • Works very well
  • Now trying to move calculations at earlier times, ideally right after breakdown

Min-Gu

• No wall beta limit
  • Yuequiang Vs student version
  • 15% error on test case
  • New version for conversion of j to q
  • Adriana will take on the development of the actor in FUSE
• Generating database of TEQUILA
  • Orso: may want to look at the GACODE local equilibrium (expand it to MXH)
• Where can I find TEQUILA documentation?
  • public API is documented
  • everything with _ is internal
  • some details in ArXiv paper
  • work with Brendan for more details
• Improved visualization of tree information in IMAS with min/avg/max
• Added unicode plots (not using Plots recipes)
  • Orso: can we make it use Plots recipes!?

Brendan

• Profiling identified issue with get_time_array
• Still issue with getproperty() allocating data
  • https://discourse.julialang.org/t/fragile-type-inference-in-custom-getproperty/122268
  • Orso: write things in LLVM directly?
• FRESCO
  • can now compute flux surface averages needed to do a q-solver
  • Luke is working on equilibrium controller for rmaxis, zmaxis, and ip
• Still some issue with timescales for vertical stability

Adriana

• New actor tutorial for the code-camp
  • Need a way to be able to expand act without recompiling: for Orso
• Andrew Merris asked to give a talk to MIT to the risk group
  • Probably something for 2025
• Fusion materials changes reviewed and can be merged
• Rabbit neutron rates
  • Only for one specific reaction
• Risk PRs need to be merged

Giacomo

• PECAN code EM forces for EU-DEMO, TE, etc.
  • The real value is the framework
  • Both 2D and 3D
• EM stresses are the driver for VV
• Early career award

Galina

• Validation of 2D source against OMFIT
  • Some discrepancy that goes away if OMFIT grid resolution becomes smaller
• Some small issue in Julia version, when using very small pellets
  • Maybe due to some underlying issue in flux surface averaging?
• Implemented simple ITER scaling for drift
  • OMFIT version does not seem to have the time dependency?