MᴀX Webinars

New HPC clusters based on accelerators will soon enter the production phase. These machines are conceptually different from most of those previously employed by Quantum ESPRESSO users. The MaX European Centre of Excellence is working to prepare Quantum ESPRESSO for the new and forthcoming architectures. This webinar will provide an update on the status of the GPU version of Quantum ESPRESSO and on the roadmap for its future evolution. It will also provide a basic set of instructions on how to tune and use the code efficiently on these new HPC systems.

Quantum ESPRESSO is a collection of codes for electronic structure computations widely used in the materials research community. Its HPC users are typically accustomed to machines based on massive MPI parallelism and often need to acquire more familiarity with heterogeneous machines based on GPUs. The webinar will provide a broad view of the different versions of the Quantum ESPRESSO code, where to get them and where to use them. Following a general introduction, the webinar will focus on how to obtain an optimal performance from Quantum ESPRESSO on these new systems. An introductory talk will provide basic instructions on how to manage the resources on GPU based machines, followed by a tutorial explaining how to compile and tune up our codes.

The available computational power has steadily increased over the past decades and continues to do so, with upcoming superclusters going towards a performance on the scale of Exaflops/s. These advances present great new opportunities for computational science but also pose new challenges with respect to how to automate the resources and manage the data that will be produced.

We will give a short introduction to AiiDA, a tool that is designed to help its users leverage high-performance computing resources to automate workflows of HPC codes, such as those developed by MAX, run on those systems and manage the data that they produce. We will detail especially how the recent release AiiDA 1.0 comes with many performance improvements in the workflow engine and database in order to support high-throughput computational loads and the various mechanisms that allow its users to make optimal use of current and future powerful HPC resources. After that, we will exhibit AiiDA lab, a GUI solution on the cloud that makes running these workflows and analyzing the results easy and intuitive, even for non-experts. Finally, we will present the renewed Archive of the Materials Cloud that serves as a dissemination platform that can be used to very easily publish data generated through AiiDA.

In this webinar, we will show how the MAX flagship code AiiDA supports its users to manage their computational workflows and the data that is produced. As opposed to all other flagship codes, AiiDA itself is not a simulation software, but instead is a workflow and data management tool. We will show how it can be used to automate workflows that directly employ the other MAX flagship codes, such as Quantum ESPRESSO, how AiiDA lab provides a user-friendly GUI to the workflows even for non-experts, and how the resulting data can easily be published through the Materials Cloud Archive.

Yambo is an open-source project aimed at studying excited state properties of condensed matter systems from first principles using many-body methods. As input, Yambo requires ground state electronic structure data as computed by density functional theory codes such as Quantum ESPRESSO and Abinit. Yambo’s capabilities include the calculation of linear response quantities, quasi-particle corrections based on the GW formalism, optical absorption, and other spectroscopic quantities. Recent developments range from the electron-phonon interaction and a real-time approach to linear and non-linear optical properties.

New HPC clusters based on accelerators will soon enter the production phase. These machines are conceptually different from most of those previously employed by Yambo users. Indeed, Yambo HPC users are typically accustomed to machines based on massive MPI parallelism and often need to acquire more familiarity with heterogeneous machines based on GPUs.

The MaX European Centre of Excellence is working to prepare Yambo for the forthcoming pre- and exascale machine. The scope of this webinar is, indeed, to provide a broad view of the Yambo code, where to get it from and how to use it. We will also provide an update of the status of the GPU version and outline the roadmap for its future evolution. We will also provide a basic set of instructions on how to tune and use the code efficiently on these new HPC systems.

Following a general introduction on linear response, quasiparticles and excitonic effects the webinar will focus on how to obtain an optimal performance on these new systems. An introductory talk will provide basic instructions on how to manage the resources on GPU-based machines, followed by a tutorial explaining how to compile and tune up the code.

For over two-decades, SIESTA has enabled the treatment of large systems with first-principles electronic-structure methods, bringing new opportunities to many disciplines. At the core of SIESTA's efficiency is the use of a basis of strictly-localized atomic orbitals. The reduced cardinality of the basis and the sparsity of the Hamiltonian means that systems composed of dozens to hundreds of atoms can be treated with modest hardware, and that the programme can employ novel algorithms to extend its applicability to even larger systems. Within the MaX project, which is preparing materials-simulation codes for the upcoming extreme-scale HPC systems, this baseline efficiency has been extended, alongside the domain of applicability of SIESTA with the addition of new features. These extensions will be the focus of this webinar.

A very important aspect of these improvements to the usability and performance of the programme is that they are an example of the power of modularization and code reuse, which have been espoused by MaX and by other international initiatives, such as the Electronic Structure Library and the ELSI project. This webinar will cover in particular the new developments in the area of electronic structure solvers, notably the incorporation of an interface to the ELSI library, which has enabled significant performance enhancements, including GPU acceleration. Another representative area of SIESTA's use domain is ballistic electronic transport. In the webinar we will present an overview of the TranSIESTA built-in module, which implements a formalism based on non-equilibrium Green's functions, and the latest improvements in its functionality (in particular multi-electrode support) and optimisation. The webinar will also highlight the efforts being made in enhancing user support within the SIESTA ecosystem, an area of great relevance in view of the extra complexity that emerges when considering the variety of novel architectures and features.

Within the zoo of different density functional theory methods frequently employed, the full-potential linearized augmented planewave (FLAPW) method is commonly considered as the approach able to provide the most precise results. Among the flagship codes developed within the MaX centre of excellence, the open-source code FLEUR code, implementing this method, can provide reference results and can be utilized to study details of the electronic, magnetic, and atomistic structure of complex materials. The code is able to treat bulk and film systems with all elements of the periodic table. Recently, major advances in the scalability, performance and applicability of the code have been achieved and made available in the MaX releases of the code.

The webinar will focus on the basic features and fundamentals of the FLEUR code. We will include an overview of the different types of simulations possible with the code, including its interfaces to other methods. The use of FLEUR on modern HPC systems including Tier-0 PRACE systems will also be covered in our presentation. Additionally, we aim at providing hints and instructions useful for deploying FLEUR on different systems, to overcome typical challenges and to identify the requirements for the usage of the code. Finally, we will point at possible further sources of information, documentation and support processes and outline our future plans.

The BigDFT project started in 2005 with the aim of testing the advantages of using a Daubechies wavelet basis set for Kohn–Sham DFT with pseudopotentials. This project led to the creation of the BigDFT code, which employs a computational approach with optimal features of flexibility, performance, and precision of the results. In particular, the employed formalism has enabled the implementation of an algorithm able to tackle DFT calculations of large systems, up to many thousands of atoms, with a computational effort that scales linearly with the number of atoms.

In this webinar, we will present some of the features that have been made possible by the peculiar properties of Daubechies wavelets. In particular, we focus our attention on the usage of DFT for large-scale systems. We show how the localized description of the KS problem, emerging from the features of the basis set, is helpful in providing a simplified description of large-scale electronic structure calculations. During the presentation, we will highlight how the MaX consortium enabled the possibility of the implementation of advanced functionalities in the context of pre-exascale computing.