gt-frc / gt3 Goto Github PK

When the LCFS hits a wall, we lose the ability to X-point physics and have to manually set a separatrix. One problem is that we have to decide what the plasma radius becomes. Code is being changed to manually take the left-most R value from the wall_line LineString as the lowest value the inboard midplane can be. Probably need to make this change for other geometries and triangularities.

In CalcErIOL.py, the ionization rate is required in the "iol_term". When there are no neutrals, this term will blow up. Need to suss out what physically is going on here.

The gradient scale lengths calculated for 2D profiles are not done correctly. As a temporary fix, we take the FSA, find the scale length of that, and broadcast back to 2D.

Currently, GT3 requires neutpy to be installed. Since neutpy is fairly extensive, it might be better to have it as an optional dependency. Should be easy to do since RadialTransport has exception handling if it can't find neutrals data.

• read_infile
• core
• iol
• imp_rad
• neutrals
• beamdep
• rad_trans
• thermal_inst

As of matplotlib 2.2.0 _cntr has been deprecated (flatironinstitute/CaImAn#174)

They recommend replacing with contours module. Is this an easy fix?

GT3 current does not have any real unit testing. Should attempt to support.

It would be useful to have a method in GT3 that prints a summary of all the input variables and possibly other relevant information. Maybe toss it in the gt3 class and use a table generator class. I think one is in neutpy.

The output neutrals data file is in JSON format. ReadInFile still tries to read this in and throws errors when it sees JSON data. Probably just need to not even look for this file in the data input routine.

ReadInFile's way of setting instance variables is unnecessary and should be done as it was done in NeutPy.

It would be useful to have a method in GT3 that prints a summary of all the input variables and possibly other relevant information. Maybe toss it in the gt3 class and use a table generator class. I think one is in neutpy.

Currently, GT3 cannot handle 2 X points.

2D values are often tricky because you need to know that they are 2D and you need to remember the weird [:,0] code. I think it would be a good idea to extend 2D profiles to have 1D methods for bringing variables to 1 dimension.