Computational Science and Engineering (CSE) News Round-Up
30 Sep 2020
No
- Dawn Geatches

 

 

Scientists in the CSE Division of the Scientific Computing Department work on a diverse range of projects:  

No

​​​​​​​​​​​​​

 

​​from publications to funded proposals, virtual meetings and outreach, they have been able to continue their work as seamlessly as ever. (All named staff work within CSE unless designated otherwise).​

Publications

Spin waves in metallic iron and nickel measured by soft x-ray resonant inelastic scattering Dr. Kun Cao and co-authors combined experimental and state-of-the-art computational techniques to explore spin waves in metallic materials. Challenging to measure experimentally, spin waves are nevertheless important in superconductivity, spintronics - potentially more efficient analogues of electronics - and the young and promising field of magnonics and its applications. Combining the complementary computational method with the experimental results shows the potential of applying technically difficult experimental techniques and extracting useful results. With the addition of further experiments, the computational methods could be developed to become predictive tools.

Coupling Molecular Dynamics and Direct Simulation Monte Carlo using a general and high-performance code coupling library Dr. Stephen Longshaw and co-authors developed a code-coupling method that enables the modelling and simulation of mul​​ti-scale processes such as evaporation. The team combined two different time- and length-scale modelling techniques to create a general simulation evironment in which both modelling techniques run simultaneously. Their next step is to add another time- and length-scale and build a fully-coupled tri-scale simulation methodology.

Ontologies for the Virtual Materials Marketplace (VIMMP) lead by ​first author Dr. Martin Horsch, wth co-authors Dr. Silvia Chiacchiera​, Dr. Michael Seaton, Prof. Ilian Todorov and non-STFC co-authors​, describes their work on developing a classification and relational system ​within the VIMMP (Horizon 2020) project. Designed to develop a digital marketplace, i.e., a platform to facilitate exchanges between providers and users in the area of materials modelling, key to which is a series of classification and relational systems called 'ontologies'. Typical goods will include software, workflows, validation data, model parameters, simulation platforms, and services, such as training and translation (from an industrial problem to a modelling solution).

Detecting composite orders in layered models via machine learning​ Dr. Silvia Chiacchiera was involved in a collaboration that used artificial intelligence techniques to determine the phase diagram (i.e., to answer questions such as: does the material behave in the same way or differently at these two temperatures? How many different behaviours does it show?) of layered magnetic systems. Example of magnetic materials are ultra-thin magnetic films (e.g. used in magnetic sensors, recording materials) and superconductors (e.g. used in MagLev, particle accelerators). The authors proposed a method based on comparing microscopic snapshots of the simulation to determine their similarity, and its main advantage is that it does not require any pre-existing knowledge of the macroscopic behaviour of the material.

Potential future developments include the method being used to learn directly from experimental data, and applied to a wider and more complex class of materials​​

Quantifying the impact of disorder on Li-ion and Na-ion transport in perovskite titanate solid electrolytes for solid-state batteries​ Dr. John Purton and co-authors applied large-scale atomistic simulations to analyze the impact of experimental conditions on the disorder and ion transport behavior for perovskite materials used in solid-state batteries. ​They found that the structural disorder induced during their manufacture has a significant impact on their performance​, and suggest guidance to minimize this negative effect of the experimental process.​

Reactive Molecular Dynamics at Constant Pressure via Nonreactive Force Fields: Extending the Empirical Valence Bond Method to the Isothermal-Isobaric Ensemble. As first author and principal researcher, Dr. Ivan Scivetti describes the article (in collaboration with co-authors: Dr. Kakali Sen, Dr. Alin Elena and Prof. Ilian Todorov) as the culmination of the work carried out over the lifetime of his three-year project exploring reactive force-fields and classical molecular mechanics. This work proposes a new methodology for incorporating chemical reactivity within classical molecular dynamics simulations from nonreactive force-fields, thereby taking an important step towards bringing real experimental conditions of chemical reactions within reach of atomistic simulations. The new method 'MD-EVB' is available as additional functionality within DL_POLY_4.


ssDSC_cover_Sept2020_small.jpg

Double Linker Triphenylamine Dyes for Dye-Sensitized Solar Cells  A combined, complementary experimental and computational study probing the atomistic structure of solar cell dyes (funded by EPSRC). Dr. Kakali Sen, Dr. Ya-Wen Hsiao and Dr. Dawn Geatches revealed a level of detail inaccessible to their experimental collaborators (lead by Prof. Peter Holliman of Swansea University). Combining experiments and simulations is leading to a deep understanding of the structure of solar cells that could stimulate significant changes in the design process of solar cell devices. This team's work is highlighted on the front cover of the September issue of​ the journal: Energies.


ARCHER2 embedded Computational Science and Engineering (eCSE)​ funded projects

Multi-Resolution Coupling for Exascale Engineering lead by Dr. A Revell and Dr. A De Rosis (both of the University of Manchester); co-investigators Dr. Jianping Meng, and Dr. Charles Moulinec. The primary objective of this project is to provide a software framework for Exascale computing that enables the computation of complex physics phenomena in industrial engineering simulations. The new framework is highly relevant to the High End Computing consortium: UKCOMES.

 A Partitioned Fluid-Structure Interaction Framework for Exascale lead by Dr. Alex Skillen; co-investigator Dr. Charles Moulinec; researcher Dr. Wendi Liu. The project aims to build a new framework for Exascale computing that combines established software in the fields of computational fluid dynamics, computational structure mechanics and data exchange. The new framework will facilitate the modelling and simulation of fluid-structure interactions (FSI), with a (post-project) view towards improving understanding of the role of FSI in low carbon energy applications. The new framework is significant for the Collaborative Computational Projects (CCPs): CCP-NTH, and ​CCP-WSI+.

Improving the performance of DL_MONTE for large-scale simulations lead by Prof. Steve Parker (University of Bath), co-investigators Dr. John Purton and Dr. Tina Duran (University of Bath); researcher Dr. Tom Underwood (also University of Bath). The project aims to enc​ode continuous fractional Monte Carlo - a modelling ​method that enables the gradual addition and deletion of molecules in a model system - ​into the software DL_MONTE. Essentially, ​this functionality will enable the insertion of large molecules into small holes, a process that is relevant for metal organic frameworks (MOFs) and covalent organic frameworks (COFs), which are classes of material particularly important for storage of greenhouse gases and catalysis.​

Multi-Layered MPI parallelisation for the R-matrix with time-dependence code lead by Dr Andrew Brown; co-investigator Professor Hugo van der Hart (both of Queens University Belfast); researcher Dr M Plummer. Additional and enhanced parallelization and optimization of the 'RMT' software used to explore laser atom/molecule interactions in attosecond (one quintillionth* of a second) physics, modelling in real-time the behaviour of the atomic electrons irradiated by ultrashort laser pulses. The code makes use of two distinct spatial regions where different numerical methods and algorithms are used. The team are increasing its flexibility and efficiency to enhance capability and ensure that activity in one region does not need to wait for the other region to 'catch-up'. This will also increase the possible size of practical  (computationally viable and efficient) calculations by orders of magnitude (e.g. x10, x100 etc.), to cope with recent code development to describe phenomena (including interactions described using special relativity) in atoms and molecules irradiated by laser pulses whose field polarization may be chosen arbitrarily, ultimately paving the way to more ambitious projects.

​* 1 quintillionth = 0.000000000000000001 ​


Meetings

2020 Edition of the Muon Site Calculation organised by Dr. Leandro Liborio and Dr. Simone Sturniolo was a one day virtual meeting attracting over 50 participants from all over the world.  The purpose of the meeting is to discuss computational techniques used for the interpretation of muon software –a tool that enables the simulation of the behaviour of hydrogen in materials -   and involves a very established community who, although not formally part of the same project, keep in contact and interact on a regular basis. 


Muon_quote.png


2020 Materials Science and Engineering Congress held online saw Dr. Martin Horsch presenting "Ontology-based pragmatic interoperability between open platforms in materials modelling", about the work that he and colleagues Dr. Silvia Chiacchiera, Dr. Michael Seaton and Prof. Ilian Todorov  have accomplished as partners in the VIMMP project.​

CCP5​​ 40th Annual General Meeting was a virtual event attended by 71 registered webinar attendees and a further 200 viewers of the live-streamed video following from the US, Japan, India and Europe. In total, attendance showed nearly a four-fold increase on previous in-person AGMs.

Talks covered computational research on a wide range of topics, from the performance of tyres, through iron deficiency anaemia, to quantum computing for scientific calculations. A highlight of the day was the announcement of the winner of the inaugural biennial CCP5 Prize and Lecture Award: Prof Kostya Trachenko for his '..outstanding contributions to modelling and theory of condensed matter phases including liquid state theory, radiation effects and DL_POLY development'. The award includes £1000, a medal, and expenses to cover attendance at next year's (hopefully in-person) 40th Anniversary celebrations.

 

Outreach-Work Experience Talks

CSE_outreach_quotes.png

Although the pandemic prevented on-site work experience, thanks to the efforts of the Public Engagement Team, it failed to cancel it completely. Instead, students were invited to a series of talks and in mid-August it was the turn of CSE with the help of Evelyn Greeves (Communication and Impact Sandwich Student) who gave an overview of SCD and the importance of communicating impact; Dawn Geatches summarised the Who?, Why?, What? and How? of scientific computational modelling within CSE; Alex Skillen gave an overview of work within the engineering group including a summary of his career path, and Jacob Ward held a virtual tour of the supercomputing machine room at Rutherford Appleton Laboratory.  Gemma Reed organised and hosted the one-hour webinar with technical assistance from Lauren Mowberry.

Fifty-three students attended; 100% said they felt welcome, 81% felt the session was pitched at the right level, and 83% said they'll try to find out more about the subject themselves.

CCP-WSI+​ : A Collaborative Computational Project in Wave/Structure Interaction

CCP-NTH : The Collaborative Computational Project for Nuclear Thermal Hydraulics

UKCOMES : ​UK Consortium on Mesoscale Engineering Sciences

CCP5 : The Collaborative Computational Project for computer simulation of condensed phase materials at length scales spanning from atomistic to mesoscopic levels.


 

 

 

 

 

 

 

 

 



Contact: Geatches, Dawn (STFC,DL,SC)