Handling Times

I cannot see how to do this in asimple way withot subclassing your whole framework:

I want to write a message lets say every 1 hour to a file if my study is running.

start = time.time()
start_data= datetime.now()

def objective(trial):

  print("MULTIPLE OF 1 MIN: ", (time.time() - start) // 60)

  if (time.time() - start) // 60 == 1:
    print("1 MIN OVERb Since")
    print(start_data)
    start = time.time()
    start_data= datetime.now()

    ....
    
   return score


study = optuna.create_study(direction="maximize")
study.optimize(objective, n_trials=300, timeout=None)

But since start is outside the function it is unknown inside the trials…

UnboundLocalError: local variable 'start' referenced before assignment How can I handover a start time to measure how much time is elapsed? I just can start the time each trial…

[nzw0301]

I think the simplest way to define your own objective as in this page. The objective stores the start_time and checks your condition in each call.

So how about this?

import time
from datetime import datetime

import optuna


class Objective(object):
    def __init__(self, min_x, max_x):
        self.min_x = min_x
        self.max_x = max_x
        self.start_time = time.time()

    def __call__(self, trial):
        time.sleep(0.5)  # for this example

        # Calculate an objective value by using the extra arguments.
        x = trial.suggest_uniform("x", self.min_x, self.max_x)

        current_time = time.time()

        if current_time - self.start_time >= 1.0:  # check the evaluation takes more than 1 second since ether the last message or the initialisation
            print("1 MIN OVER Since", self.start_time)
            self.start_time = current_time

        return (x - 2) ** 2


study = optuna.create_study(direction="maximize")
study.optimize(Objective(-1., 1.), n_trials=10, timeout=None)

[nzw0301]

Please feel free to re-open the issue if the question was not solved.