When to switch from Chilli to Pepper

[Tunneller]

Hi, as the rotational correlation slows, I can imagine that a "theoretical" solution of slow-motion should ultimately match the powdered solution. I can also imagine that in practice running chilli with crazy slow rotation could result in numerical errors. I'm looking at vanadyl in a very thick gooey oil.

Chilli with logtcorr = -8 looks close enough to my spectrum that I could imagine doing a numerical fit for g,A, etc. Same is true for pepper (?!) Both simulations are overlaid in the attached g = [1.9860 1.9650], A = [157 480].

But if I run chilli with yet slower rotation, then the results start diverging wildly. At logtcorr = -1 I get a warning message from chilli and an erroneous looking spectrum. At logtcorr = -4 then I dont get a warning but the computation looks equally bad.

Is there a rule of thumb for a "lowest recomended" value of logtcorr after which one should switch to pepper??

Thanks,
[Edit: I guess I mean a "highest recomended" value.]

[Stefan Stoll]

With chili, you should in principle be able to approach the rigid limit by slowing down the rotational correlation time. However, to get converged spectra, it is important to increase the orientational basis size via increasing the values in Opt.LLMK.