Isil Ozfidan

is there a plan to divide the huge one into smaller pieces?

yes!!!

not a bug - this is the correct behaviour. The cutoffcomposite is implemented for precision errors.

the fact that this was an unexpected behaviour for you tells me that I did a bad job describing what the composite does. So yes changing the parameter definition to...

I am against that idea. the main reason for this composite is precision issues.

how about "interactions **_with precision_** below a specified cutoff value"

@JoelPasvolsky I trust your judgement more than my own on this!

so now this sounds like we actually provide a precision for samplers. How about "commensurate with a precision given by cutoff" ? or "bounded by -cutoff < val < cutoff"

I am a bit worried about providing `TooManySamplesWarning`. without knowing the ground states of a problem creating a warning about drawing too many samples is not a great idea. What...

I disagree with that comment. It may simply be a hard problem with a hard landscape that other heuristics will fail too (possibly worse than the QPU).