Structure prediction of nanoclusters from global optimization techniques: computational strategies and connection to experiments.
IPREM, Technopole Helioparc 2 avenue du Président Angot 64000 Pau, FRANCE
This workshop intends to gather both theoreticians of the field and a few representative experimentalists in the field of structural optimization of nanoclusters and nanoparticles (of course the panel of experimentalists cannot be exhaustive for such a workshop).
The main scientific objectives of this workshop is focused on the confrontation of the global optimization techniques and the recent developments within the main classes of methods mentioned above: simulated annealing (SA) and the developments concerned with ergodicity, such as multi-tempering, parallel-tempering in Molecular Dynamics or Monte Carlo schemes, genetic algorithms (GA) schemes (and the variety of declinations for crossing changes, mixing, etc..), Basin-hopping techniques and derived techniques. The scope is clearly to discuss the ability of various methods to be operative for nanoparticles in the range from a few tenths up to a few hundred atoms or molecules.
Obviously, discussion of the workshops will include the compared quality of those various schemes both in terms of results (success in finding the global minimum and the low energy isomers) and in terms of numerical efficiency especially in view of new developments in terms of massively parallel computing and existing or forthcoming computer architectures (exascale, petascale).
Another focus will be put in the combination of those global optimization schemes, namely examining combination with the various methods used to compute the forces. This should include (i) force fields with increased complexities to represent real systems, such as for instance polarizable force fields (molecular clusters), reactive force fields (covalent clusters), or other kinds of many-body potentials (metal nanoclusters) (ii) approximate and semi-empirical methods generally, still explicitly quantum and derived from ab initio methods (Tight-Binding, CNDO, Density Functional Tight Binding), and naturally (iii) First principle methods such as Density Functional Theory calculations, which is the most reliable method is this size range. Schemes involving multi-method approaches combining some or several of the afore-mentioned approaches should obviously be discussed, either in sequential or in adaptative combinations, from instance in on-the fly parametrization of lower status parametric methods against DFT.
Finally, the third scope is to establish close connections between theory and experiments during this workshop, discussing : (i) the dependence of various properties onto either the detailed or the global shapes of the nanoclusters/nanoparticles (in other words : does structure matter and to which extend for a given observable /experiment ?), (ii) alternatively which experiments can really be conducted to probe the structure and (iii) the possibly interactive schemes based on observable driven optimization and their comparison with standard energy-driven schemes.
From a practical point of view, and despite the fact that the workshop will address specialists of the subject, we will strive to encourage the participants to prepare their oral presentations in a clear and understandable way for the colleagues developing different experimental and theoretical approaches, still developing in details the core features. This will facilitate the discussion and the sharing of ideas between theoreticians and experimentalists but also between theoreticians.
Concerning the objective to overcome the theoretical (and also experimental) limits, we think first that the previous objective can help. Second, most of the nanoclusters are, from an experimental point of view, placed in an environment: adsorbed on surface or stabilized by ligands. Thus, additional strategies emerge to pave the route toward the structure prediction of (sub)nanoclusters as they are observed “in the real world”. In parallel, new experimental setups, or combinations of available but still separate experimental techniques might be designed to better address the knowledge of nanocluster structural features in link with their physical/chemical properties.
Philippe Carbonniere ( University of Pau and Pays de l''Adour ) - Organiser
Michel Rérat ( Université de Pau et des Pays de l'Adour ) - Organiser
Fernand Spiegelman ( Université Paul Sabatier (UPS) and CNRS ) - Organiser