What one Environment::step() produces: the next observation, this step's reward, and whether the episode ended.
More...
#include <environment.hpp>
What one Environment::step() produces: the next observation, this step's reward, and whether the episode ended.
- Note
- Plain data, no behavior – same precedent as ReparamGrad (reparameterize.hpp): a small named return-tuple beats an out-parameter or a std::tuple whose members are positional and therefore unreadable at every call site.
-
Deliberately no
truncated flag (gymnasium's five-tuple splits time-limit truncation from genuine termination). This campaign's Phase 1 folds both into done, matching the classic Gym four-tuple; splitting them is a bootstrapping concern for value-function targets that no algorithm in this codebase exists to care about yet, and adding it later is additive.
◆ done
| bool pulsatrix::StepResult::done |
◆ observation
| Tensor pulsatrix::StepResult::observation |
◆ reward
| float pulsatrix::StepResult::reward |
The documentation for this struct was generated from the following file: