crace 1.1.0

crace: Continuous Racing for Automatic Algorithm Configuration

crace: Continuous Racing for Automatic Algorithm Configuration

[ Homepage ] [ Source Code ] [ User Guide (PDF) ] [ Report A Bug ] [ Discussions ]

Maintainers: Leslie Pérez Cáceres , Yunshuang Xiao

Authors: Leslie Pérez Cáceres, Jonas Kuckling , Pablo Contreras , Yunshuang Xiao, Thomas Stützle .

Contributors: Manuel López-Ibáñez .

Contact: [e-mail]

Relevant literature:

• Yunshuang Xiao, Leslie Pérez Cáceres, Manuel López-Ibáñez, and Thomas Stützle. Algorithm Configuration via Continuously Racing: Preliminary Results. In Proceedings of the Companion Conference on Genetic and Evolutionary Computation (GECCO '23 Companion ). Association for Computing Machinery, New York, NY, USA, 1744–1752. doi: 10.1145/3583133.3596408

• Manuel López-Ibáñez, Jérémie Dubois-Lacoste, Leslie Pérez Cáceres, Thomas Stützle, and Mauro Birattari. The irace package: Iterated Racing for Automatic Algorithm Configuration. Operations Research Perspectives, 2016. doi: 10.1016/j.orp.2016.09.002

Introduction

The crace package implements a continuously racing procedure, which is an alternative to irace that performs in each iteration a single race. The continuously racing configurator (crace) evaluates, removes and generates new configurations asynchronously, granting a high level of flexibility regarding the configuration process when compared to the previous iterative scheme. The main use of crace is the automatic configuration of decision and optimization algorithms, that is, finding the most appropriate settings of an algorithm given a set of instances of a problem. However, it may also be useful for configuring other types of algorithms when performance depends on the used parameter settings. It builds upon the race package by Birattari and irace package by López-Ibáñez and it is implemented in Python.

You may also find the cplot package useful for analyzing the output of crace.

Keywords: automatic algorithm configuration, automatic algorithm design, offline tuning, parameter tuning, racing, irace.

Requisites

Python >= 3.6.0 is required for running crace, but you don't need to know the Python language to use it. Python is freely available and you can download it from the Python project website. See section Quick Start for a quick installation guide of Python.

For GNU/Linux and macOS, the command-line executable parallel-crace requires GNU bash. Individual examples may require additional software.

Note: This package is curretly tested with Python versions up to 3.14.3.

User guide

A complete user guide comes with the package. You can access it online or, after installing the crace package, invoking from the terminal console the following command:

crace doc

The following is a quick-start guide. The user guide gives more detailed instructions.

Quick Start

• Install Python (with your favourite package manager, and see more details below).
• Install crace. The crace package can be installed automatically with conda, with pip or from the source code. We advise to use the automatic installation unless particular circumstances do not allow it.
• Once crace is installed, executable command, crace will be added to the active environment. You can call crace to execute a single crace run, crace parallel to execute multiple crace runs and crace doc to check the user guide in Bash shell (Linux and MacOS) as well as in Powershell (Windows).
• You can open the user guide with the following command. This command works on Bash shell (Linux and MacOS) and Powershell (Windows) with the active environment added to PATH (see detailed instructions below).

crace doc

Installing Python

This section gives a quick Python installation guide that will work in most cases. The official instructions are available at https://docs.python.org/3/using/index.html.

GNU/Linux

You should install Python from your package manager. On a Debian/Ubuntu system it will be something like:

sudo apt-get install python3 python3-dev

Once Python is installed, you can use conda, pip or the source code to install the crace package (see section Installing the crace package).

macOS

For macOS 10.9 (Jaguar) up until 12.3 (Catalina) the operating system includes Python 2, which is no longer supported and is not a good choice for development. You should go to do the official downloads page and download the installer.

For newer versions of macOS, Python is no longer included by default and you will have to download and install it. You can refer to the [Python documentation] (https://docs.python.org/3/using/mac.html) for more details on the installation process and getting started.

Alternatively, you can just brew the Python formula from the science tap (unfortunately it does not come already bottled so you need to have Xcode installed to compile it):

# install homebrew if you don’t have it
/bin/bash -c \
  "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

# install python 3
brew tap homebrew/science
brew install python

Once Python is installed, you can use conda, pip or the source code to install the crace package (see section Installing the crace package).

Windows

You can install Python from the official website. We recommend that you install Python on a filesystem path without spaces, special characters or long names, such as C:\Python.

Note that Python 3.5 - 3.8 cannot be used on Windows XP or earlier and Python 3.9 - 3.13 cannot be used on Windows 7 or earlier.

Once Python is installed, you can use conda, pip or the source code to install the crace package (see section Installing the crace package).

Installing the crace package

Install automatically with conda

If you use Conda, execute the following line at the shell to install the package:

# Best practice, use an environment rather than install in the base env
conda create -n my-env
conda activate my-env
# If you want to install from conda-forge
conda config --env --add channels conda-forge
# The actual install command
conda install crace

Alternatively, within the Conda graphical interface (Anaconda Navigator), you may search and install crace.

Install automatically with pip

If you use pip, execute the following line at the shell to install the package:

pip3 install crace

Also when using pip, it's good practice to use a virtual environment and here is the guide for using the virtual environment.

Create a new virtual environment.

venv (for Python 3) allows you to manage separate package installations for different projects. It creates a “virtual” isolated Python installation. When you switch projects, you can create a new virtual environment which is isolated from other virtual environments. You benefit from the virtual environment since packages can be installed confidently and will not interfere with another project’s environment.

To create a virtual environment, go to your project’s directory and run the following command. This will create a new virtual environment in a local folder named .venv:

# Unix/macOS
python3 -m venv .venv

# Windows
py -m venv .venv

The second argument is the location to create the virtual environment. Generally, you can just create this in your project and call it $.venv$.

venv will create a virtual Python installation in the .venv folder.

Activate a virtual environment.

Before you can start installing or using packages in your virtual environment you’ll need to activate it. Activating a virtual environment will put the virtual environment-specific python and pip executables into your shell’s PATH.

# Unix/macOS
source .venv/bin/activate

# Windows
.venv\Scripts\Activate.PS1

To confirm the virtual environment is activated, check the location of your Python interpreter:

# Unix/macOS
which python

# Windows
where python

While the virtual environment is active, the above command will output a filepath that includes the .venv directory, by ending with the following:

# Unix/macOS
.venv/bin/python

# Windows
.venv\Scripts\python

While a virtual environment is activated, pip will install packages into that specific environment. This enables you to import and use packages in your Python application.

Deactivate a virtual environment.

If you want to switch projects or leave your virtual environment, deactivate the environment:

deactivate

Reactivate a virtual environment.

If you want to reactivate an existing virtual environment, follow the same instructions about activating a virtual environment. There’s no need to create a new virtual environment.

Manual download and installation from source

If the previous installation instructions fail because of insufficient permissions and you do not have sufficient admin rights to install crace system-wide, then you need to force a local installation. From the crace package PyPI website, search and download one of the three versions available depending on your operating system:

• crace_xxx.tar.gz (Unix/Linux/BSD/macOS)
• crace-xxx-py3-none-any.whl (python3, any platform)

Alternatively, you can use wget or curl from the command line to download the file:

# PyPI
wget https://files.pythonhosted.org/packages/source/c/crace/crace-xxx.tar.gz
# Github
wget https://github.com/race-autoconfig/crace/

To install the package on GNU/Linux, macOS or Windows using the formats .tar.gz, .tar or .zip, you should first extract the source files and then navigate to the directory containing the extracted files and run the following command to install the package:

pip3 install .

To install the package on any platform using the Wheel format, you should run the following command to install the package:

pip3 install /path/to/crace-xxx-py3-none-any.whl

Testing the installation and invoking crace

Once crace has been installed, it can be listed by using command line at the shell or load the package and test that the installation was successful by opening a Python console and executing:

# List installed packages at the shell
pip3 list
# Show the information of crace at the shell
pip3 show crace

# Load and test at the Python console
import crace
print(crace.__version__)

Checking the installation path of crace

Once crace is installed, the executable command will be added to the environment. To check the installation path of crace using command line at the shell:

which crace

The output of this line must be $PYTHON_HOME/bin/crace, which means the path before /bin/crace is $PYTHON_HOME. Here, $PYTHON_HOME should be:

CRACE_HOME=$PYTHON_HOME/lib/python3.X/site-packages/crace

Also, you can check the installation path of crace opening a Python console and executing:

import importlib.util
print(importlib.util.find_spec('crace').submodule_search_locations)

This command must print out the filesystem path where crace is installed. In the remainder of this guide, the variable $CRACE_HOME is used to denote this path. When executing any provided command that includes the $CRACE_HOME variable, do not forget to replace it with the true installation path of crace.

On GNU/Linux or macOS, you can let the operating system know where to find crace by defining the $PYTHON_HOME variable and adding it to the system PATH. Append the following commands to ~/.bash_profile, ~/.bashrc or ~/.profile:

# Replace <CRACE_HOME> with the crace installation path
export CRACE_HOME=<CRACE_HOME>
# Tell operating system where to find crace
export PATH=${PYTHON_HOME}/bin/:$PATH

Then, open a new terminal and launch crace as follows:

crace --help

Alternatively, you may directly invoke crace within the Python console by executing:

import crace
crace.run("--help")

Usage

1. Create a directory for storing the tuning scenario setup (Bash shell):

   mkdir ./tuning
   cd ./tuning
   

2. Initialize your tuning directory with template config files (Bash shell):

   crace --init
   

3. Modify the generated files following the instructions found within each file. In particular,

   • The script file target-runner should be executable. The output of target-runner is minimized by default. If you wish to maximize it, just multiply the value by -1 within the script.

   • If you are using Windows, or if the script file 'target-runner' is in .bat, .py or .Rscript format, please use option targetRunnerLauncher instead of targetRunner and set targetRunnerLauncher="cmd file.bat".

   • In scenario.txt, uncomment and assign only the parameters for which you need a value different from the default one. For example, you may need to set trainInstancesDir="./Instances/".

4. Put the instances in ./tuning/Instances/. In addition, you can create a file that specifies which instances from that directory should be run and which instance-specific parameters to use. See scenario.txt and instances-list.txt for examples. The command crace will not attempt to create the execution directory (execDir), so it must exist before calling crace. The default execDir is the current directory.

5. Calling the command in the Bash shell to perform one run of Continuous Race.:

   cd ./tuning/ && crace
   

See the output of crace --help for additional available crace parameters. Command-line parameters override the scenario setup specified in the scenario.txt file.

Many tuning runs in parallel

On GNU/Linux or macOS, several repetitions of crace in parallel is allowed. Call the program crace parallel from the Bash shell:

cd ./tuning/ && crace parallel N

where N is the number of repetitions. By default, the execution directory of each run of crace will be set to ./exp-xx, where xx is a number padded with zeroes.

Be careful, crace parallel will create these directories from scratch, deleting them first if they already exist.

Check the help of crace parallel by running it without parameters.

Parallelize one tuning

A single run of crace can be done much faster by executing the calls to targetRunner (the runs of the algorithm being tuned) in parallel:

crace --parallel N

where N is the number of slaves to execute the target algorighm in parallel. See the user guide for more details.

License

The crace package is Copyright © 2026 and distributed under the GNU General Public License version 3.0.

This program is free software (software libre): you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

IMPORTANT NOTE: Please be aware that the fact that this program is released as Free Software does not excuse you from scientific propriety, which obligates you to give appropriate credit! If you write a scientific paper describing research that made substantive use of this program, it is your obligation as a scientist to (a) mention the fashion in which this software was used in the Methods section; (b) mention the algorithm in the References section. The appropriate citation is:

• Yunshuang Xiao, Leslie Pérez Cáceres, Manuel López-Ibáñez, and Thomas Stützle. 2023. Algorithm Configuration via Continuously Racing: Preliminary Results. In Proceedings of the Companion Conference on Genetic and Evolutionary Computation (GECCO '23 Companion ). Association for Computing Machinery, New York, NY, USA, 1744–1752. doi: 10.1145/3583133.3596408

The crace package incorporates code under the GPL from the irace package is Copyright (C) 2010-2025 Manuel Lopez-Ibanez, Jeremie Dubois-Lacoste, Leslie Perez Caceres.

Frequently Asked Questions

The user guide contains a list of frequently asked questions.