Parameterising the INVERT approach

This paper gives more detail about the ‘INVERT’ approach and outlines both the challenges faced in trying to use data-driven methods for amorphous structural solution and some of the possible solutions to this problem.

Figure: Ordinarily, the problem solving the structure of amorphous materials from diffraction data is both underconstrained (there are too many solutions) and taxing to reach (the fitting landscape is poorly shaped). This is shown on the left and is compared to an idealised ’target’ landscape on the right.