FUSE meeting notes: Dec 3 2024

[orso82]

Orso

• Improved flux-surface tracing
• Started work on automated COCOS translations in IMASdd.jl
• Code camp preparation are in full swing
  • Fixed some issues
  • Everyone has badge access
• Registered IMAS.jl as a package
• FACIT has a pure Python implementation in Aurora. Good target for transpiling to Julia.

Tim

• Improved multi-objective generator
  • Ability to resume
• Improved database generator
  • Arbitrary workflow
  • Arbitrary sampling
• FUSE installs no problem on Windows
• Something weird going on with multi-threading under Windows
  • @TimSlendebroek can open an issue and try narrow down where it happens
  • Could it be use of threadid() in TEQUILA?

Joey

• Worked with Tim
  • Validation of neg-T DIII-D Kr shots, looks very good
  • Let's add this neg-T case to DIII-D (@TimSlendebroek @tomneiser)
• Difficult to find higher Z-effective solutions in neg-T because of high radiation

Brendan

• Time dependent free-boundary (FRESCO)
  • Can run VDE's (with reasonable timescales)
  • @lstagner has implemented a control algorithm for vertical position and current in Julia
  • No shape control yet --> closed-boundary feedforward first, then run free-boundary
  • @lstagner can also use TokSys shape controller in FUSE
• Work to run FRESCO with P and J
• TEQUILA + VacuumFields still useful
  • for PF coils design
  • for MHD stability calculations
• Still needs to make FRESCO work with entropy and q to have a proper Grad-Hogan solver (future work)
• Native Julia implementation of the contouring routine, using marching square algorithm
  • Much faster! No allocations (with preallocation)
  • Right now only traces one flux surface
• There's a minimum level of psi that depends on the grid resolution
  • We need to decide how to handle that

Nan

• Database of FPP with ITB for Tom

Tom

• Starting to train a TGLF-NN on new FPP w/ ITB
• Will want to generate a new database for UKAEA STEP
• Will work on generating database for Tokamak Energy
• SAT0 is the best general model
  • SAT4 = (SAT2 linear physics x SAT0 saturation rule) for QLGYRO
• Thinking of making TGLF-EP its own repository
  • Must add Eric and Ben to the repo
  • Eric is interested to be in charge or the repo
• Gary's pushed some update to TGLF to make things faster
  • really only faster for large number of basis function (32 basis functions, like in TGLF-EP)
• @bclyons12 may want to revisit issue with eigen function in TJLF.jl

Min Gu

• Faster precompilation json2imas() (from 20s to 5s, by removing type inference at compilation time)
• Now plot(ids, r"...") allows plotting multiple things (uses findall(...) underneath)