Research Objectives

The vision of the UK High-End Computing Consortium for X-ray Spectroscopy (HPC-CONEXS) is to leverage developments in high-performance computing (HPC) to transform the analysis of X-ray Spectroscopy. By developing, deploying, and distributing codes with excellent scalability over modern computer architectures, we will facilitate collaboration between computational and experimental scientists in all disciplines under the umbrella of X-ray Spectroscopy.

Schematic

The consortium is focused upon delivering exciting new science in three strands:

Strand 1: First-Principles Quantum-Mechanical Simulations:

The development of new methods for first-principles simulations of ’core-level’ spectroscopies is an active area of research. However, urgent advances which provide approaches that are both accurate and scalable remain urgently required. Beside the fundamentals of method development, this project will also seek to exploit efficiently modern computer architectures in order to enhance performance and future-proof our developments by ensuring they are ready for exascale computing facilities. We will seek to fund compute time for projects which:

i) Develop new scalable first-principles quantum-mechanical methodologies.
ii) Adapt existing codes for modern computer architectures to enhance performance.
iii) Implement automation and analysis techniques to streamline workflows.

Strand 2: Experiment-Theory Collaborations:

The development of high-brilliance light sources is transforming the capability of X-ray spectroscopy to deliver highly-detailed information about the local geometric, electronic, and spin structure of matter in a range of different environments. However, in these cases, detailed theoretical studies are essential to provide a firm link between the X-ray spectroscopic observables and their interpretation. In this strand, we will seek to enhance the synergy between experiment and theory by providing compute time for projects which apply state-of-the-art techniques to collaborative theory/experiment investigations.

Strand 3: Data Driven Appraoches:

 Supervised/Unsupervised machine learning models thats are able to extract and learn patterns in big data without hand-coded heuristics have transform many aspects of modern life and can potentially provide a new dimension to the analysis of X-ray spectroscopic data. Work in this strand will focus on the development of machine learning models which can ehance the analysis of X-ray spectroscopic data and address both the forward (property to spectrum) and reverse (spectrum to property) problems. Compute time will be awarded to projects which: 
 
i) Develop new machine learning appraoches to enhance data interpretation.
ii) Analysis of machine learning methodologies to enhance interpretability and quantify uncertainity.
iii) Develop advanced training sets.
iv) Apply machine learning models to interpret complex data.

HPC-CONEXS will also seek to:

Strengthen the UK X-ray Spectroscopy Community

 This will be achieved by:

  • Nurturing a forum for theory and experiment-theory synergy in the area of X-ray spectroscopy.
  • Establishing an annual conference for experimental and theoretical X-ray spectroscopy.
  • Setting up training events for advanced techniques in analysis of X-ray spectroscopy.

Deliver a Virtual Beam line for Diamond Light Source (DLS)

During the CONEXS funding (EP/S022058/1), we developed the web-CONEXS application; this sits alongside existing DLS experimental beamlines as a virtual beamline. The web-CONEXS platform offers users a simple and intuitive web-based graphical user interface (GUI) through which to set up and start calculations.It is aimed at non-experts and designed to perform preliminary calculations before moving to larger-scale facilities.