How do you tell EFIT to use MSE data?
There are two SNAP files available to use with the multi-channel MSEs. Both can be obtained from the EFIT: area on the VAXs or the /link/efit area on the HPs
- efit_snap.dat_mse1
- efit_snap.dat_mse2
The MSE1 SNAP file has one less fitting parameters then MSE2. The current density profiles obtained using MSE1 usually are smooth, whereas those using MSE2 tend to have fine structures. This is similarly reflected in the condition number of the fitting matrix. The condition number from MSE1 is typically a factor of 3 to 5 smaller than that from MSE2, although both are acceptable. Unless MSE1 give excessive high chi2 or convergent error, we recommend using the MSE1 SNAP file rather than the MSE2.
When using MSE data in the fitting, a significant amount of CPU time is spent in setting up the response functions for MSE. The response functions are stored in the file
- RSTARKXX.DAT
after making a run. The CPU time can be reduced by renaming this file appropriately for EFIT to read in
- RSnnnn_01.D3D for Vaxs and HP single precision
- RSnnnn_01.DDD for HP double precision
where nnnn representing the EFIT grid size. For example, 3333 for 33x33, 3365 for 33x65, 129129 for 129x129, etc. This is particularly handly when making time-dependent EFIT runs. It can reduce the processing time by a factor as much as 2.
The fitting information is summarized in the plots. These include the location of the MSE viewing points, q values at these points, chi2 due to MSE, a comparion of the fitted and the measured pitch angles and a plot of the individual chi2.
More on fitting using MSE data. The present MSE system contains little information on the edge current density and quite often it tends to give a negative current density near the edge for many of the H and VH mode shots (shown as J1N on the plot on first page, edge current normalized to I/area).
This sometimes can lead to discrepancy between the MHD separatrix and that as expected from the Thomson measurements. One remedy is to impose a constraint of J1N around 0.1-0.2 at PSIWANT=0.99. There is a sample SNAP file
- EFIT_SNAP.DAT_MSE2_J1
in the EFIT: area on the Vaxs and /link/efit/ area on HPs. Please use it with care.
Again, if accurate results are desired, full kinetic fits are strongly recommended.
A number of rs_*.* files are also stored in the EFIT: area on the Vaxs and /link/efit/ area on the HPs. To use them, copy the appropriate version of the files into your local area to reduce the execution time when MSE data are used in the fits.
.