Time Integrators more compatible with DiffEq

[htsnowden]

trafficstars

To create a new framework where we break down our problem into nuclear and electronic propagation with seperate (coupled) integrators. This allows for a few benefits

• Variable timestepping between the different systems
• Utilisation of the full suite of DiffEq integrators where applicable, e.g. out of the box Tsit5 for electrons and VelocityVerlet for nuclei
• GPU compatible solvers for systems where we want
• Automatic compatibility with future DiffEq versions (prevent the fsalfirstlast issue from returning)

A new style of coupled integrator has been created by myself and the next step is to write the codes dynamics and problem setup in a way which is compatible with this new style. This will be begun on a branch that I shall link here