Development of next generation accurate approximate DFT/B methods - Flagship workshop and tutorial
CECAM-DE-MM1P, University of Bremen, Germany
The primary goal of the proposed CECAM workshop is to facilitate communication and collaboration between users and developers of multi-scale quantum methods, with emphasis on approximate DFT and semi-empirical (SE) methods, linear-scaling quantum mechanical and hybrid quantum mechanical/molecular mechanics approaches. The meeting is intended to provide a unique forum to bring together world leaders of different communities in method development to discuss recent advances in current methodologies, to assess their similarities, and to identify critical challenges in the field as well as future directions for research and development of the next generation of computational tools. On the more technical side, this will include the discussion of the state of the art in efficient parametrization techniques, reference databases, and multi-objective optimization. Simultaneously, the workshop will involve people from the DFT and ab initio communities, who are interested in SE/DFTB methods as a supplement to their first principles codes, for example to use the more approximate quantum techniques for screening large numbers of structures in high-dimensional configuration space and for exploring extended trajectories under real environmental conditions (e.g. deMon2k, ATK, ADF, SIESTA).
The proposed workshop should become a forum to brainstorm ideas about solutions to important problems and identify new directions for method development and challenging applications. In this way, we hope to create an exchange mechanism to unite a core of developers in an interactive environment to initiate design of a new generation of integrated software tools for accurate multi-scale quantum modeling. The delivery of this technology to a broad community would facilitate breakthroughs on high-risk, high-impact chemical problems in energy, resources, environment and health. Given the strongly interdisciplinary scope of the workshop, only theoretical modeling groups will be invited in order to ensure a sharp focus on theoretical method developments for next generation high performance quantum-based models for bridging scales in materials/systems sizes and time evolution. About 30 worldwide leading experts in the fields of
● Approximate DFT-methods (Self-consistent charge density functional based tight-binding methods and semi-empirical schemes that reduces the computational overhead in direct DFT methods.)
● Efficient parametrisation techniques (PSO particle swarm optimization and others)
● Lagrangian Born-Oppenheimer Molecular Dynamics, Car-Parrinello MD, and exited state MD
● Order-N electronic structure algorithms extended (Methods that have a computational complexity that expands only linearly with the system size instead of cubically as in traditional methods.)
● Efficient parallel implementation of approximate DFT and semi-empirical schemes for modern petaflop and GPU architectures and planning for the forthcoming move to exaflop systems
● Hybrid coupling approaches QM/MM and QM/EM (Multi-scale approaches where the effect and response of the extended surrounding is modeled with more approximate but computationally faster classical force fields.)
● Exploration of hybrid schemes combining approximate DFT methods and Valence Bond (localized) type Configuration Interaction methods keeping good scaling properties.
● DFTB and novel multi-scale approaches
● Fragment orbital based methods for MD and charge transport
● Time-dependant DFT/B in linear response (frequency domain), in the time domain and TD-transport in open quantum systems
will be invited, and give keynote talks on the related areas to set the stage and define the target for future developments.
We define the following sub-objectives:
1. To share the state of the art at a computational level, regarding performance, scalability, parallelization and hence range of applicability of different approximate quantum based MD techniques.
2. To define common targets for method developments between different communities, focusing on hierarchical and domain multi-scale coupling techniques, interconnection between electronic structure methods, and classical methods.
3. To share the perspectives for the next few years of research in the field, in particular referring to novel emerging issues such as time-dependent and non-equilibrium phenomena in complex systems.
The intended format of the workshop is seven half-days, consisting of 40 minutes talks including ample time for discussion after each talk. To also ensure full and active participation of young emerging researchers attending the meeting, a poster session will be organized.
Fernand Spiegelman (Université Paul Sabatier (UPS) and CNRS) - Organiser
Marcus Elstner (Karlsruhe Institute of Technology) - Organiser
Thomas Frauenheim (University of Bremen) - Organiser