I wonder if there’s a stable way to write gradients to disk? For my case I need collecting a lot of snapshots as well as gradients for post processing. For now, I modified tavg plugin to output gradient without time averaging. I wonder if there’s an stable way to do it in case I missed from UserGuide.
Writing gradients is extremely wasteful given that they can be reconstructed from the solution data.
As such the recommended approach is to use the gradient of unstructured data set filter in ParaView (after running through clean to grid). This will typically give more pleasing results than anything PyFR can provide since it does a proper reconstruction across element boundaries. Moreover, the entire calculation is in-memory so avoids needless amounts of expensive disk I/O.