here's the old plotting for reference... @arwelHughes could you elaborate on what you think looks a bit horrible about it? I can't really see a difference between it and the old plots

one thing I will note is that the aspect ratio seems a bit funny - we'd usually expect each item in the plot to be square, but i don't know if that's just how it's been scaled in this specific case

one thing I will note is that the aspect ratio seems a bit funny - we'd usually expect each item in the plot to be square, but i don't know if that's just how it's been scaled in this specific case

The aspect ratio is square when it draws initially but it does not preserve the ratio when the figure is scaled just like the former

Ideally, we'd love to fit ellipses (as an option), but I've never worked out how! e.g. [Learned harmonic mean estimation of the Bayesian evidence with normalizing flows]

@arwelHughes we can look at this as a future issue but please let us know what looks horrible? The major issue was that 2 different axes were being drawn on the top left and bottom right of the plot. We could revert the code and fix only this if you feel confident the plot has changed?

@arwelHughes This is a side by side of both plots the new is on the left and the old is on the right