Fix CMOD EFIT nodenames

[yumouwei]

Implemented changes

• Corrected nodenames in efit_cols and cmod/physics.py:

  • li -> ali
  • q0 -> qqmagx
  • qstar -> qsta
  • q95 -> qpsib
  • aminor -> aout/100 (aout: [cm])
  • rmagx -> rmagx/100 (rmagx: [cm])
  • chisq -> tsaisq
  • area -> areao/1e4 (areao: [cm^2])
  • zmagx -> zmagx/100 (zmagx: [cm])

• Removed now-unnecessary unit conversions in cmod/physics.py

• Removed efit_cols_pre_2000 and its references in cmod/efit.py, as all the nodenames in efit_cols have been corrected & match what were in efit_cols_pre_2000.

Remaining issues to be addressed

1. beta_n

• The existing nodename \efit_aeqdsk:betan is an expression Build_Signal(\EFIT_AEQDSK:BETAT / \EFIT_AEQDSK:IN, *, \EFIT_AEQDSK:TIME). IN is the normalized current defined as Ip[MA]/(a*Bt). This expression of beta_n isn't exactly correct as it should be calculated using the total beta instead of beta_t. Also this node is only available for post-2000 shots.

• In addition, there's another block of codes under get_efit_parameters that calculates beta_n for pre-2000 shots. However that code is also wrong as it calculates the total beta and forget to normalize it by IN.

https://github.com/MIT-PSFC/disruption-py/blob/8888f3f601b03cf922d8053beef93e10801d2fe8/disruption_py/machine/cmod/efit.py#L133-L139

2. v_surf

• disruption-py currently calculates v_surf = gradient(SSIBRY)*2pi where SSIBRY is the poloidal flux at the plasma boundary. We need to make sure this is the correct expression.

https://github.com/MIT-PSFC/disruption-py/blob/8888f3f601b03cf922d8053beef93e10801d2fe8/disruption_py/machine/cmod/efit.py#L109-L117

• There is also a node \efit_aeqdsk:vsurfa which is available for all shots but equals 0 for every shot.
• Then there's efit_aeqdsk:vsurf which is an alias for vsurfa for post-2000 shots.

3. v_loop

• For pre-2000 shots, disruption-py calculates v_loop = gradient(SSIMAG)*2pi where SSIMAG is the poloidal flux at the magnetic axis. I'm not sure why this is needed as vloopt is available for both pre-and-post-2000 shots. The results also don't match vloopt.

https://github.com/MIT-PSFC/disruption-py/blob/8888f3f601b03cf922d8053beef93e10801d2fe8/disruption_py/machine/cmod/efit.py#L122-L131

References

• https://github.com/MIT-PSFC/disruption-py/pull/47
• https://github.com/MIT-PSFC/disruption-py/pull/185
• CMOD wiki: EFIT variables
• NSTX: EFIT Signals

[crea-psfc]

Thank you @yumouwei for the thorough review. Generally, all pre-2000 code blocks need to be thoroughly reviewed and validated. Regarding the specific points raised:

1. Let's consult with Bob Granetz to determine the correctness of beta_n.
2. I believe the one you linked is the correct expression for vsurf as C-Mod had one big poloidal Rogowski loop. Bob Granetz is the one that can solve this point as well.
3. same as above, let's raise these points to Bob's attention to determine best course of action.

@gtrevisan there are certainly edge cases for DIII-D as well, but we never went into similar detailed analysis of historical data. I am aware of a strong bifurcation in the availability of the n=1 bradial signal. I talked to @yumouwei at the time, he should be aware. Regarding areao, sure, let's add it to the list of retrieved signals from EFIT. Right now it can be reconstructed via differences between kappa and kappa_area.

[yumouwei]

Following the discussion on 11/4/24, we decided to investigate & address the beta_n, v_surf, & v_loop computations in a separate PR. For now this PR is ready for merging.