|
LIRRA
1.0.0
|
Public Member Functions | |
| __init__ (self, RecedingProblem problem, ISTLSettings settings=ISTLSettings(time_limit=2.0), float|np.ndarray noise=0.0, int seed=0) | |
| np.ndarray | __call__ (self, ProposalContext ctx) |
Public Attributes | |
| problem | |
| settings | |
| noise | |
| rng | |
| last | |
Solves the synthesis problem itself (rooted at the measured state): an ideal policy. Used to generate behaviour-cloning data and as an upper bound on what a learned proposer can achieve. ``noise`` (scalar or per input) adds Gaussian noise to its output, clipped to the input box (imperfect-learner stand-in).
| rl_stl.evaluation.patrol.ExpertProposer.__init__ | ( | self, | |
| RecedingProblem | problem, | ||
| ISTLSettings | settings = ISTLSettings(time_limit=2.0), |
||
| float | np.ndarray | noise = 0.0, |
||
| int | seed = 0 |
||
| ) |
| np.ndarray rl_stl.evaluation.patrol.ExpertProposer.__call__ | ( | self, | |
| ProposalContext | ctx | ||
| ) |
| rl_stl.evaluation.patrol.ExpertProposer.last |
| rl_stl.evaluation.patrol.ExpertProposer.noise |
| rl_stl.evaluation.patrol.ExpertProposer.problem |
| rl_stl.evaluation.patrol.ExpertProposer.rng |
| rl_stl.evaluation.patrol.ExpertProposer.settings |