Make the parameters in the NG force law of Marsden+73 user-defined

[federico-spada]

The parameters "alpha," "nk," "nm," "nn," and "r0" in the g(r) function defining the NG force are currently hard-coded.

For work on long-period comets orbits, it would be very helpful if these parameters could be user-defined, perhaps through extras.particle_params. Calculating derivatives wouldn't probably be necessary, as these parameters are fixed in typical use-cases focused on fitting A1, A2, and A3 for specific values of alpha, nk, nm, nn, r0 (e.g., 0.1113, 4.6142, 2.15, 5.093, 2.808; Marsden et al. 1973).

[hannorein]

Does your use case require you to use a different set of values for each particle or is one set for the entire simulation good enough? The former is a bit harder to implement in an efficient way.

[federico-spada]

For my use case, one set of values for the entire simulation is perfectly fine.

[hannorein]

Then one could just add them directly in the assist_extras struct (not in the particle_params struct). Are you ok doing that yourself? If not, I can provide a patch. Note that if you're using python, you also need to update the structure in the python code.

FWIW, there are better ways to do this (see e.g. how REBOUNDx handles particle specific parameters) but I'd hesitate to do anything like this here. The amount of extra work does not seem justified. And anything more complicated could affect the performance.

[federico-spada]

Hi Hanno, thank you very much for your prompt replies and for offering your help! I would really appreciate it if you could make the changes yourself when you have some time. I do not feel comfortable with handling the interactions between C and Python.

[hannorein]

No problem. See these changes here:

https://github.com/matthewholman/assist/pull/113/files

[federico-spada]

Perfect, thank you very much!