Non-periodic splines used for periodic data in geometry

In trying to use some rather high resolution geqdsk files with GS2 I was encountering reports that we were trying to evaluate R outside the domain. Digging into this it seems the issue was in the boundary spline created in eeq::bound. Here the second derivative of the function being splined (R on the boundary as a function of geometric theta) was coming out as large and highly oscillatory near the +/- pi boundary. This seemed to be resolved by switching to use periodic splines.

We should probably test this further and explore if there are other uses of splines which may be better expressed as periodic splines.

Comments (1)