RK45 step solver option
mountaindust opened this issue · 1 comments
Currently, all agent movement is Euler. It would be nice to implement a runge-kutta 45 solver to get more accurate results.
This only makes sense when there is no noise, which will not be the case most of the time. It's main use-case is if we are going to implement FTLE as part of this package. More pressing is to get better interpolation methods as an option, such as tricubic or quadcubic. There's a good looking tricubic algorithm at eqtools, and quadcubic is available through ARBInterp, but it's not yet peer reviewed. Closing this in favor of two new issues, one to address interpolation accuracy and another to more specifically address solving the Ito and Stratonovich equations governing simple IBMs with no agent-to-agent interaction.