What would the difference be if someone chooses to use UltraNest as their Bayesian sampler, but decides not to use MLFriends? How different would the results qualitatively be compared to using the default setting? I ask because threeML apparently allows for an option to turn on/off MLFriends while using UltraNest. Or should it be turned off for specific types of data and not all?
This is more a general question
What would the difference be if someone chooses to use UltraNest as their Bayesian sampler, but decides not to use MLFriends? How different would the results qualitatively be compared to using the default setting? I ask because threeML apparently allows for an option to turn on/off MLFriends while using UltraNest. Or should it be turned off for specific types of data and not all?