FUSE meeting notes: Dec 17 2024

[orso82]

Code camp

• agenda
• Ideas
• Great collaborative vibes and feedback
• Few small updates/fixes during code-camp but mostly solid
• We need to merge ability to go from dd to gEQDKS
• People were able to use FUSE on windows (though running Jupyter notebooks via VScode and not Jupyter-Lab)
• There's lots of interest to organize something for onsite people
• Tentative idea is weekly 1h long tutorials with pizza in 15.019, to be recorded an posted. Starting sometimes in January.

Orso

• Working to merge both ActorTroyonBetaNN and ActorFRESCO in dev branch.
• On dev branch also simplified how ActorLimits works
• Played around with idea of using "distributed" coils to identify optimal number of coils
  
• Gearing up for next year's IR&D about develop a plasma pulse simulator to support routine DIII-D operations which will revolve around FUSE

Tim

• Working on restructuring how files of multi-objective optimization are saved to disk. Split into generation sub-folder to avoid issues with NFS.
• Similar for database generator
• Working on PT/NT FPP design with Oak

Joey

• PR for Jardin's sawtooth q-profile model [https://github.com/model to keep q>1 in stationary transport #785]
  • can we do this also for time dependent?
  • We could do it time-averaged and time-resolved
• PR on using gaussian process surrogates in the flux matcher [https://github.com/Gaussian processes surrogates in FluxMatcher #790]

Galina

• Almost finished implementing drifts in PAM.jl
  • Some issue with DifferentialEquations.jl

Min-Gu

• Extended simulation parameters (used for ini and act to accept arbitrary distribution functions) New OptParameterDistribution to use Distributions package SimulationParameters.jl#12
  • Very useful for database generator workflow in FUSE
  • We should add this functionality to the tutorial notebook @TimSlendebroek

Brendan

• FRESCO ready to be merged into FUSE as an Actor
  • We will remove SolovevActor
• Will start looking into setting up a public install of FUSE on OMEGA, with sys-image for fast startup

Tom

• Running MAST-U model on NSTX-U profiles
• For STs it's best to
  • With TGLF: Use SAT1 em AZF-1
  • With QLGYRO: Use SAT3 em AZF-1
• ZGGEV shows an improvement of 2X speed using 6 basis function (in SAT3)
  • Can we look at this in TJLF

Giacomo

• Talked with Chris about first-wall and divertor builders
  • First-wall builder in a notebook, needs to go into FUSE
  • The builders should be heat-flux controlled

Nan

• Working on a workflow to develop high-beta scenarios