Add option for continuous sampling

[randomir]

Keeping N samples in memory is expensive if we only want the k best ones!

Open question is the interface -- num_reads is synonymous with num_samples returned in sample set for all samplers in Ocean. Also, initial_states (even for random states) are expanded to num_reads input samples -- something we would also want to avoid in this case.

To retain compatible behavior with existing samplers, perhaps we could introduce a parameter like num_resample. Also, resample_reduce_method (min/max). And to support the k best samples use case, we'll need a parameter like num_samples.

So, something like:

ss = greedy.sample(bqm, num_reads=1, num_resample=1000, num_samples=3, resample_reduce_method='k-best')

---

Another take on this would be async sampler interface. In which case a caller would yield as many samples are needed, all with minimal memory overhead!

[jackraymond]

DWaveSampler uses the parameters answer_mode and max_answers in addition to num_reads. For compatibility these may be good naming convention in other dimod samplers, addressing related issues. max_samples for the QPU assumes 'k-best', I don't think there are (many) interesting use cases beyond k-best

[randomir]

Is there a way, using these existing parameters, to express: "run N descends, but return only k best ones"? You answered in your edit already. 😄

[JoelPasvolsky]

uses the parameters 'sampling_mode'

Do you mean answer_mode and max_answers?

[jackraymond]

Names here: https://docs.dwavesys.com/docs/latest/c_solver_parameters.html - I corrected the initial comment as well.