Jmcclena comp

[jmcclena]

This pull request adds a few pointnames to d3d machine_mapping and gives first implementation of incorporating compensation corrections to signals.

@orso, I'd be grateful any recommendations for improving the compensation strategy.

[jmcclena]

Its not terribly slow. Approximately 8s to generate the constraints plus 0.25s per kfile generated (which I assume a lot of that is in duplicating the file). If there is any obvious way it could be speed up, let me know.

I haven't found a good place for the loop voltage any ideas? Maybe its considered a passive loop or flux loop?

[orso82]

If you parse the kEQDKS once, then you should be able to deepcopy it (not duplicating) and that should be faster (since you'll skip the parsing step).

[orso82]

Something like

template_kEQDSK.load() # force a load
for k in range(10):
    kEQDSK=copy.deepcopy(template_kEQDSK)

[smithsp]

@orso82 There was also a question about loop voltage. Do you recall where you put that for NSTX?

[orso82]

I see a bunch of leftover print statements. @jmcclena please let me know when this is ready for final review and merge.

[jmcclena]

@orso82 This branch is ready for a final review when you have time. Originally, I was hoping to get vloop into omas equilibrium constraints before merging. However, I suspect the correct location for the loop voltage is pf_active since I found written in pf_passive: "any connected loops are expressed as active coil circuits with no power supply attached". I think this will require some special treatment, as vloop information isn't saved in the mhdin.dat file, and would be better as a seperate pull request.

[jmcclena]

Actually, now that I'm thinking about this again. I need to correct the DIII-D error calculation for diagnostics with integrator errors before merging.

[orso82]

@jmcclena do you think you can get to this by the end of the week? As discussed at the OMFIT-users group meeting I'd like to release a new OMAS version by the end of the week, and may be good to have these updates in it.

[jmcclena]

will do

[jmcclena]

@orso82, apparently my concerns with integration errors were unwarranted in this case. I did fix the compensation calculation and it is ready for a final review.