Andreas - Presentation on PENN
AP: There are many more parameters that are important for the pedestal than the ones used in PENN.Can you handle Ne or N puffing?
- AG: No, there’s not enough data in the database to train the network to predict the effects of seeding.
JF: How’s the separatrix reconstructed in Loreozos fit?
JF: In T17-07 we uses PENN and the pulse you’ve been using. I’ll be comparing PENN with Lorenzo’s scaling law.
AP: Do you have data for AUG?
- Yes, but the AUG db has only 200 pulses, while JET has 2000 pulses.
Rui - Presentation on SOLPSz1
RC: There’s an issue with output_flag, see slack.
- TJ: Ok, I’ll update the TRANSPORT_SOLVER.
AP:
- We now have three different versions of SOLPSz.
- If you’re interesting we’d like to continue to collaborate!
RC:
- SOLPSz1 has an “onion” structure, we could easily update the solpsz-core.
- Inputs to SOLPSz1 can come from either IDSs or code parameter, this is handled by the wrapper and is easy to change.
JF: When will we have interpretative (db) impurities?
- TJ: It’s not hard to implement; depends on the urgency.
- JF: Well, first I’ll test with static impurities.
JF: we need pedestal actor
- TJ: Yes, the plan is to add this in the transport combiner.
- DC: Please tanh function to turn off transport in the edge.
AP: Could be good to have an inertia parameter to damp transient changes in the edge. There’s a natural time scale of the order 10 ms.
- DC: This already exists. The time scale is determined by the parameter called “suger”
- TJ: Perhaps we should use more descriptive code parameter variables in SOLPSz1.
Abreviations:
AP: Alexei Polevoi
AG: Andreas Gillgren
JF: Jorge Ferreira
RC: Rui Coelho
TJ: Thomas Jonsson
DC: David Coster