rsr.rsr.get_comp_cond_sys_prob¶
- rsr.rsr.get_comp_cond_sys_prob(refs_mat_upper, refs_mat_lower, probs, comps_st_cond, row_names, s_fun=None, sys_upper_st=1, n_sample=1000000, n_batch=1000000)[source]¶
P(system state | given component states).
‘probs’ is (n_var, n_state) categorical; we condition rows listed in comps_st_cond to one-hot.
We classify samples using refs; for unknowns we call s_fun(comps_dict) to resolve.
Returns probabilities over {‘upper’,’lower’} that sum ~ 1.0.
- Return type:
Dict[str,float]