PrincetonUniversity / PsyNeuLink / 27641887821

84%

master: 85%

Build Type

push

github

Committed by web-flow

Commit Message

feat(pec): allow passing an optuna Study to PECOptimizationFunction (#3559)

## Summary

Adds support for passing an instantiated `optuna.study.Study` object as
the
`method` of `PECOptimizationFunction`, in addition to the existing
options
(`"differential_evolution"`, an optuna sampler instance, or a sampler
class).

Previously the study object was always created under the hood, giving
the user
no opportunity to interact with it. Passing your own study enables
optuna
features that operate on the study directly — e.g.
`study.enqueue_trial(...)`
to "warm start" a sampler with parameter values from a previous fit.

## Usage

```python
study = optuna.create_study(
    sampler=optuna.samplers.RandomSampler(seed=0),
    direction="maximize",  # required for data fitting (MLE)
)
study.enqueue_trial({"param1": value1, "param2": value2})  # optional warm start

optimization_function = PECOptimizationFunction(method=study, max_iterations=10)

Coverage Stats

11017 of 14384 branches covered (76.59%)

Branch coverage included in aggregate %.

9 of 10 new or added lines in 1 file covered. (90.0%)

2 existing lines in 1 file now uncovered.

37653 of 43557 relevant lines covered (86.45%)

0.86 hits per line