ROSE

What is ROSE:

The RADICAL Optimal & Smart-Surrogate Explorer (ROSE) toolkit is a framework for supporting the concurrent and adaptive execution of simulation and surrogate training and selection tasks on High-Performance Computing (HPC) resources. ROSE is a Python package that provides tools to facilitate the development of active learning methods for scientific applications. It allows users to define simulation and surrogate training tasks and automatically manage their execution on HPC resources.

ROSE also provides tools to facilitate the selection of the best surrogate model for a given simulation based on performance metrics.

ROSE uses RADICAL-Cybertools -- middleware building blocks to facilitate the development of sophisticated scientific workflows on HPC resources.

How to install: pip install .

How to use:

1- Import the necessary pacakges

import os

from rose.learner import ActiveLearner
from rose.engine import Task, ResourceEngine

2- Define the resource engine and Active learner

engine = ResourceEngine({'runtime': 30,
                         'resource': 'local.localhost'})
acl = ActiveLearner(engine)

3- Define the simulation task

@acl.simulation_task
def simulation(*args):
    return Task(executable=f'python3 sim.py')

4- Define the surrogate training task

@acl.training_task
def training(*args):
    return Task(executable=f'python3 train.py')

5- Define and register the active learning task

@acl.active_learn_task
def active_learn(*args):
    return Task(executable=f'python3 active.py')

6- Optionally you can define a stop criterion or run for MAX ITERATIONS

# Defining the stop criterion with a metric (MSE in this case)
@acl.as_stop_criterion(metric_name='mean_squared_error_mse', threshold=0.1)
def check_mse(*args):
    return Task(executable=f'python3 check_mse.py')

7- Finally invoke the tasks and register them with the active learner as a workflow

simul = simulation()
train = training()
active = active_learn()
stop_cond = check_mse()

# Start the teaching loop and break if max_iter = 10 or stop condition is met
acl.teach(max_iter=10)
engine.shutdown()