I had a chance to talk with Linda Sugiyama (MIT), the physicist in charge of the M3D fusion plasmas code, while at the CPES meeting.  I explained what we were trying to do and she had some suggestions for examples of the kinds of things that would help her out in the IOG/MB space:

1. To measure instability in the plasma, they look for deformations in the shape of the toroid.  The way they do this is to take a single particle within each plane (the mzeta equivalent value for m3d) and average the position around the toroid.  They then compare the average value against the variance for each particle to get an idea of the deformation.  This would require us to filter all of the restart writes to grab the right particle, keep track of a value (or values) across all of them, and know when we have seen them all.

2. I don't understand this one fully, but she said that perform gradients on all of the data to see where spikes and other abnormalities are so we could put in a gradient filter with some parameters to identify the problem areas.

3. With code coupling of xgc0 and m3d, they currently can only do in the xgc0->m3d direction.  To do this, they take the n=0 plane from the xgc0 output and construct an m3d input file.  She'd like to see it also be able to go back the other direction by reconstructing an xgc0 startup file from the m3d output.  They don't/can't do that right now.  The other issue, although not as important as the functionality, is that they hit disk multiple times to do the conversion (once to write the original data, once to put the data out in the proper format).  They are also needing to write things in a geqdisk(?) file.

That was all the time we had so I didn't have a chance, nor did I really have the background to fully understand, the physics better so I could clarify some of the darker corners of these examples.

Jay