Ophidian Complete!
The red points are the grid points just inside the plasma, and the blue points are the interpolated position of the last closed flux surface. This would have been a simple problem, except the Python NURBS library doesn't work and hasn't been updated in years. This wasn't an actual wheel re-inventing event, but I did find that all current wheel implementations were but shattered monuments, and was forced to chisel my own by studying the shards I was able to excavate from the hardening loam of the Internet's sedimentary layers.
Once the grid is represented as a non-uniform rational B-spline (which just sounds cooler than "NURBS"), finding the plasma boundary is just an exercise in projecting the surface onto a plane and finding roots of some cubic equations. So, after spending a few hours mixing up the array indexes, I managed to write a beautiful solution to my problem. Most importantly, it got rid of the nasty wedges of points where the psi-solver would generate bad data.
The only issue left on my list of bugs is that the boundary finder will get confused if there are any regions of vacuum flux that exceed zero flux. That tends to happen near vertical field coils, especially when computing equilibria for highly shaped plasmas. Unfortunately, plasma shaping is very important, so I'll probably have to fix this eventually. For this paper, though, I'm going to stick to circular, large aspect ratio equilibria. Nothing too exotic.
Posts
