Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Unable to reproduce results before and after a study is deepcopied

See original GitHub issue

The behavior is different between 0.19.0 and 1.x.x.

Environment

Python version: 3.6.5
OS: Windows 10

Examples

import copy

import optuna


def objective(trial):
    x = trial.suggest_uniform('x', -10, 10)

    return (x - 2) ** 2


study = optuna.create_study(sampler=optuna.samplers.TPESampler(seed=0))
study.optimize(objective, n_trials=1)

study = optuna.create_study(sampler=optuna.samplers.TPESampler(seed=0))
study = copy.deepcopy(study)
study.optimize(objective, n_trials=1)

Logs

v0.19.0

[I 2020-02-12 02:37:41,933] Finished trial#0 resulted in value: 1.0480229520791995. Current best value is 1.0480229520791995 with parameters: {'x': 0.9762700785464951}.
[I 2020-02-12 02:37:41,990] Finished trial#0 resulted in value: 1.0480229520791995. Current best value is 1.0480229520791995 with parameters: {'x': 0.9762700785464951}.

v1.0.0

[I 2020-02-12 02:38:44,010] Finished trial#0 resulted in value: 1.0480229520791995. Current best value is 1.0480229520791995 with parameters: {'x': 0.9762700785464951}.
[I 2020-02-12 02:38:44,067] Finished trial#0 resulted in value: 46.90638356377688. Current best value is 46.90638356377688 with parameters: {'x': 8.848823516763801}.

v1.1.0

[I 2020-02-12 02:39:39,168] Finished trial#0 resulted in value: 1.0480229520791995. Current best value is 1.0480229520791995 with parameters: {'x': 0.9762700785464951}.
[I 2020-02-12 02:39:39,226] Finished trial#0 resulted in value: 2.756825098011473. Current best value is 2.756825098011473 with parameters: {'x': 3.6603689644207016}.

Issue Analytics

State:
Created 4 years ago
Reactions:1
Comments:6 (5 by maintainers)

Top GitHub Comments

1reaction

silecommented, May 8, 2020

@Y-oHr-N Since this problem seems to have been solved by #968, I’ll close this issue. Feel free to reopen it if you still have any problems.

0reactions

github-actions[bot]commented, Apr 18, 2020

This issue has not seen any recent activity.

Top Results From Across the Web

Why can't I reproduce the experiments in the original paper that ...

I wrote some python code to reproduce this paper's purported results. My code very efficiently optimizes simple smooth functions like bowls, ...

What Happens When Someone Fails to Reproduce ... - Enago

If you weren't the Principal Investigator (PI) on the study, did your PI get adequate time to review the project before releasing the...

Why does deepcopy fail when copying a complex object

So the idea is create a full copy of the original object in a new class with the added members. orig_obj = SomeSqlObject.get_root()...

optuna.study. - trials - Read the Docs

Enqueue a trial with given parameter values. get_trials ([deepcopy, states]). Return all trials in the study. optimize (func[, n_trials, timeout, n_jobs, .

ReCrash: Making Software Failures Reproducible by ...

It is very hard to fix a software failure without being able to reproduce ... Reproducing failures (in Java, these often result from...