FUSE meeting notes: Jun 6 2023

[orso82]

Orso

• Performance
• OMAS.jl
• Giacomo may come in
• Thinking about licensing
• QED

David

• Lots of development needs to be done to be used to its full capability
  • TGLF-NN functionality
  • Divertor side
  • Make core functionality robust and work
• Some snaps upgrading to Julia 1.9 and picking up latest update
  • Slurm worker is failing
• Approach: Figure out how to use the optimization
  • Adding some more important constraints (ECH access to the core, LH threshold, divertor challenge)
  • Minimize betan, cost, maximize flattop
  • Need to play around with optimizer settings (exploratory optimization space Vs exploiting solutions)
  • Why confinement factor is not exactly what we requested?
• Simplified (for now)
  • Disabled PF coil optimization
  • 40% power conversion efficiency
  • Power requirements fixed at 20%

Tim

• FluxMatcher on HDB5 database
• Something is funky with the rotation (TGLF-NN? or the solver?)
• Density profiles issues have been solved
  • TGLF-NN was trained better
  • It helps to have a density source
• NOTE: q<1
  • In STEP we add a anomalous current diffusion
• Comparison Vs TAUE-NN
• Regression Vs experimental data looks good
• NBI current drive seems too optimistic

Jackson

• Still working on integration in FUSE

Tyler

• Not much

Kevin

• Got comments back from NeutronTransport.jl and implemented some changes
• Will need to share a example for things to get merged
• Toying around working with threads got much less allocations

Brendan

• TEQUILA working and merged in master
  • Better than Solovev
  • Reduced fidelity version is quite fast
  • PreallocationTools.jl to handle automatic differentiation
  • Can now use automatic differentiation in (parts of) TEQUILA!

Adriana

• Tim and Tom suggested looking at NEO package like TGLF package
  • Important for ITB cases, STs, impurities
  • Making progress on generating input.neo files
• Simplified model for availability
  • open a PR

Amani

• Hirshmar-Sigmar analytical neoclassical model
• GATM working on generating different magnets with different curvatures

Nan

• Still working on convergence
• Are you using some relaxation?