ART nouveau

Le code ART nouveau est distribué via https://gitlab.com. Pour accéder au code, veuillez me contacter directement et envoyez-moi votre nom d'usager sur gitlab. Je vous donnerai access à l'ensemble des codes.

Détails

The activation-relaxation technique (ART) is an open-ended search method for finding local transition states (G. T. Barkema and N. Mousseau, Event-based relaxation of continuous disordered systems} Phys. Rev. Lett.77, 4358 (1996).). While the original version used an inverted force for converging to saddle point, ART nouveau directly computes the direction of lowest curvature using the Lanczos algorithm, a very efficient iterative method that offers the fastest and most stable way to follow a local eigendirection (R. Malek and N. Mousseau, Dynamics of Lennard-Jones clusters: A characterization of the activation-relaxation technique, Phys. Rev. E 62, 7723-7728 (2000).).

It is interfaced with LAMMPS to give access to a large number of forcefields.

Over the years, we have continued to bring improvements to the method (see, for example Refs. Eduardo Machado-Charry, Laurent Karim Béland, Damien Caliste, Luigi Genovese, Normand Mousseau and Pascal Pochet, Optimized energy landscape exploration using the ab initio based ART-nouveau} J. Chem Phys. 135, 034102 (2011) and M.-C. Marinica, F. Willaime and Normand Mousseau, Energy landscape of small clusters of self-interstitial dumbbells in iron, Phys. Rev. B 83, 094119 (2011).). Besides myself, the latest version incorporates contributions from a number of people, including :

• Gerard Barkema
• Laurent Karim Béland
• Éric Cancès
• Fadwa El-Mellouhi
• Antoine Jay
• Sami Mahmoud
• Mihai-Cosmin Marinica
• Eduardo Machado-Charry
• Mickaël Trochet

License and distribution

The ART nouveau software is freely distributed under GNU General License.

It can be modified and used for any purpose as long as reference is given to the appropriate references :

• G. T. Barkema and N. Mousseau, Event-based relaxation of continuous disordered systems, Phys. Rev. Lett. 77, 4358 (1996).
• R. Malek and N. Mousseau, {Dynamics of Lennard-Jones clusters: A characterization of the activation-relaxation technique}, Phys. Rev. E {{62}}, 7723-7728 (2000).

• Eduardo Machado-Charry, Laurent Karim Béland, Damien Caliste, Luigi Genovese, Normand Mousseau and Pascal Pochet, Optimized energy landscape exploration using the ab initio based ART-nouveau, J. Chem Phys. 135, 034102 (2011).

  ART nouveau linked to the LAMMPS library

It is now possible to use the force library from LAMMPS, a powerful MD code developed at Sandia National Labs. This version is based on version 3.0 but adds access to a large number of forcefields, extending considerably the systems that can be studied out of the box with the code.

Compiling is a bit more delicate, but you get much more flexibility.

You will find all necessary information with the package to compile ART nouveau with LAMMPS. However, to set up the LAMMPS parameters, you will have to look at the documentation prepared by the LAMMPS team.

• Implements a number of modifications to facilitate compiling, to accelerate converge and to use local forces, very useful for large systems (April 22, 2020).
• Corrects a few problems between LAMMPS and ARTn, allows for triclinic boxes and introduces a new compilation procedure inspired by LAMMPS (February 15 2017)
• Brings back the possibility to use more than 1 species with SW potential (including a repulsive potential to simulate a GaAs network) and corrects a few bugs. (December 10th, 2015)
• Version correcting a few problems for compiling and adds an example with Lennard-Jones from LAMMPS (November 29th, 2015)
• ART nouveau coupled to the LAMMPS Library (October 2015):

ART nouveau for empirical potential

This version of ART nouveau incorporates the most recent developments. The difference between this version and that for ab initio codes is mostly in the data management : when running with empirical potentials, one does not have to store configurations at every step and handle a restart at any point.

This version is therefore slightly lighter and more readable than the one in the next section. To use, mostly, for learning the method.

Selected work