05/17/2022:
Summer School on Non-Orthogonal Configuration Interaction and
its Parallel and GPU Accelerated Implementation
Schedule
Monday, May 23, 2022
13:00 UTC: Lecture “The GNOME Algorithm”
14:00 UTC: Lecture “GronOR Building, Installing and Executing”
15:00 UTC: Tutorial Example 1 “Benzene Dimer”
15:15 UTC: Hands-On
17:00 UTC: Adjourn
Tuesday, May 24, 2022
13:00 UTC: Lecture “Keeping hte NICO Calculations Tractable”
14:00 UTC: Lecture “Getting Started with OpenMolcas and GronOR”
15:00 UTC: Hands-On
17:00 UTC: Adjourn
Wednesday, May 25, 2022
13:00 UTC: Lecture “MEBF Generation and Electron Correlation”
14:00 UTC: Lecture “GronOR Code Structure and Flow Chart”
14:30 UTC: Lecture “Parallel and Accelerated Implementation of GronOR”
15:00 UTC: Tutorial Example 2 “Magnetism”
15:15 UTC: Hands-On
17:00 UTC: Adjourn
Thursday, May 26, 2022
13:00 UTC: Lecture “Overlapping Fragments and Ab Initio Frenkel Davydov”
14:00 UTC: Tutorial Example 3 “Overlapping Fragments”
14:15 UTC: Hands-On
17:00 UTC: Adjourn
Friday, May 27, 2022
13:00 UTC: Lecture “Case Studies”
14:00 UTC: Lecture “GronOR Scalability and Accelerated Performance”
15:00 UTC: Hands-On
17:00 UTC: Adjourn
12/01/2021:
Summer School on Non-Orthogonal Configuration Interaction and
its Parallel and GPU Accelerated Implementation
We are happy to announce our on-line Summer School on Non-Orthogonal Configuration Interaction and its massively parallel
and GPU-accelerated implementation, which will be held May 23-27, 2022.
Over the last decade there has been a renewed interest in electronic structure calculations based on non-orthogonal orbitals.
Despite the larger computational complexity of these methods, they have important advantages (such as the full orbital
relaxation and the intuitive interpretation of the results), which makes non-orthogonal CI an interesting alternative to standard
electronic structure calculations in certain cases.
The focus of the Summer School is threefold: In the first part, electronic structure calculations with non-orthogonal orbitals will
be discussed to get acquainted with the advantages (and disadvantages) of lifting the orthogonality restrictions common to
standard molecular orbital theory. In the second part of the course, you will get to know the basic concepts of parallelization
and GPU-acceleration taking our implementation of NOCI in GronOR as a showcase. The third part will include a tutorial and
hands-on sessions in which the concepts will be put to practice using the OpenMolcas and GronOR software applications, and
computer access will be provided through support by SURF (Dutch National Supercomputer).
The Summer School will be entirely on-line. There will be no registration fee, but the number of registrants will be limited. If
you want to be kept informed, please let us through this link: https://lnkd.in/eTyu3DCh
Topics of the Summer School will include:
Part 1, Theory behind NOCI
•
Non-orthogonal CI, Valence Bond theory and MO theory: a historical overview
•
The GNOME algorithm
•
MEBFs: general spin coupling and multi fragment NOCI
•
NOCI-F, step by step using some case studies
•
How to keep the calculations tractable: common MO basis, selection of determinant pairs
•
Ab Initio Davydov Frenkel calculations
Part 2, Parallelism and GPU acceleration of GronOR
•
Intra-node parallelism: OpenMP
•
Inter-node parallelism: MPI
•
GPU acceleration: OpenACC, CUDA
•
Libraries: MKL, Cusolver
•
Scalability and accelerated performance
•
Task-based algorithm and fault resiliency
Part 3, Tutorial and hands-on sessions
•
Installing GronOR
•
Preparing the fragments with OpenMolcas
•
GronOR, flowchart
•
Some small NOCI-F calculations
11/01/2021:
University Rovira i Virgili receives PRACE and INCITE awards
for computational research of organic photovoltaic materials
Professor Coen de Graaf of the Quantum Chemistry Group of the Department of Physical and Inorganic Chemistry has been
awarded computational resources for a PRACE project to be carried out on Juwels at the Jülich Supercomputer Center in
Germany and an INCITE project to be carried out on Summit at the Oak Ridge Leadership Computing Facility in the United
States. These two machines are the fastest and largest supercomputers for open science in Europe and the USA, respectively.
Led by ICREA/URV researcher de Graaf as the Principal Investigator, the two projects will focus on computational
investigations of novel organic photovoltaic materials as alternatives for traditional silicon-based solar cells, with potential
advantages such as lower production costs and better portability, flexibility, and reduced weight. Such novel materials could
be applied in situations where the heavy, non-flexible silicon cells are difficult to use. Detailed computations of the
electronic structure of these materials will guide development of design rules for materials with improved efficiency, thereby
making the use of organic photovoltaics an attractive alternative to traditional solar cells, and contribute to ways to convert
an even larger part of incoming solar radiation into electricity.
In addition to URV, the international team includes researchers from the Theoretical Chemistry group of the University of
Groningen in the Netherlands, the National Center for Computational Sciences at the Oak Ridge National Laboratory in the
USA, and the Department of Physical Chemistry at the University of Barcelona. This team has developed the highly scalable
and accelerated software GronOR required for the projects.
The European project is awarded in the PRACE (PaRtnership for Advanced Computing in Europe) program and is the first
ever project to be awarded to URV in this program. It will provide 243.550 node hours on the Juwels Booster for the
computational study of multiple exciton generation and intermolecular Coulombic decay in molecules of the acene family in
which some carbon atoms are substituted with boron or nitrogen. The American project is awarded by the US Department of
Energy in the Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program and is the first
project ever awarded to a Spanish research group in the almost twenty years of the INCITE program. It will provide
1.080.000 node hours over two years on Summit for the development of a better understanding of singlet fission in
perylenediimide and indole derivatives, plus the electron transfer process in photocatalytic complexes.
More info:
www.gronor.org
www.doeleadershipcomputing.org/about/
www.doeleadershipcomputing.org/wp-content/uploads/2022INCITEFactSheets.pdf
prace-ri.eu/
prace-ri.eu/hpc-access/project-access/project-access-awarded-projects/projects-awarded-under-prace-project-access-call-
23/#ChemicalSciencesAndMaterials