Thousands of scientists use PSI's large research facilities every year, where experiments provide key insights into material properties. However, research on complex materials and devices can be significantly accelerated by combining experiments with simulations, particularly using first-principles electronic-structure methods, that have become essential for assisting, complementing, guiding, and interpreting experiments. Nevertheless, even when computational methods are well established and predictive, simulation tools often remain difficult to use, requiring deep expertise to configure parameters and manage workflows, which can grow to be very complex even for a single property. In this talk, I will present the ongoing work of my research group (Materials Software and Data) to simplify access to advanced computational tools. We develop robust, open-source, turn-key workflows using our scalable AiiDA engine (https://www.aiida.net) and combine them with GUIs to facilitate the management of the simulations. Workflows are selected to have clear applications in fields relevant to PSI, such as phonon/IR/Raman spectroscopy, neutron inelastic scattering, muon spectroscopy, and core-level spectroscopy. These tools are then made accessible through our web platform AiiDAlab (https://www.aiidalab.net) to easily launch, manage, and inspect simulations directly from the browser. Our vision is to offer these simulation tools as a digital infrastructure to experimentalists, complementing PSI’s experimental facilities. Additionally, my group is active on the methodological side, working on the verification of DFT implementations to improve their precision, benefiting the broader scientific community that relies on this widespread technique, and establishing PSI as a provider of reference data via Materials Cloud (https://www.materialscloud.org). We are also developing methods to accelerate simulations without compromising their precision, such as automating the computation of Wannier functions and using them to predict material properties. These methods are then applied in my own research, where we explore materials in the quest for candidates with exceptional properties for both fundamental research and new applications (e.g., in nanoelectronics and energy harvesting). Looking ahead, we aim to help establish autonomous laboratories, where automated simulations and robotic experiments, driven by AI, work together to accelerate research. Alongside automation, we are developing tools for open research data management, particularly for integrating experiment and simulation data, ensuring that such future laboratories can generate FAIR (Findable, Accessible, Interoperable, Reusable) data “by design”.
Dr Giovanni Pizzi, Group leader, Materials Software and Data Group