2022-02-02 SciML Software
Work
led by Kuangdai Leng developing new software for small angle scattering
data analysis has recently been released as the ffsas Python package.
FFSAS is a python library for the inversion of parameter distributions
of a polydisperse system in small-angle scattering (SAS) experiments. In
FFSAS, the formulation of the inverse problem is physics-independent,
covering SAS models with an arbitrary number of polydisperse parameters
and both 1D and 2D intensity observations. Employing a versatile
trust-region method as the underlying NLP solver, it simultaneously
optimises all the polydisperse parameters in free form, achieving high
accuracy and efficiency based on a series of theoretical and
computational enhancements. The figure shows a large-scale synthetic
test on polydisperse cylinders, where we accurately recover the four
model parameters on the left using the theoretically-predicted intensity
image on the right.
2021-11-10 SciML Publication
Our latest research looks at combining graph neural networks (GNNs) with Gaussian processes to perform active learning of materials properties. GNNs provide a powerful intuitive route to featurising material structures, while Gaussian processes provide estimates and uncertanties that allow us to identify the optimal next material to investigate in order to improve the model generalisability. Read more in J. Chem. Phys.
2021-10-21 SciML at I2NS
2021-09-01 Welcoming new group members
The SciML team are delighted to welocme our latest group members. Hattie Stewart, Michael Norman and Ben White from the Data Intensive CDT, as well as Andy Sode Anker from Copenhagen University are all joining us for 6 month projects to work on a varity of exciting topics. We are really looking forward to the exciting work that we will do over the next 6 months.
2021-07-28 GraphCore System & Training Workshop
In
anticipation of the arrival of a new Graphcore system at STFC, RAL, the
SciML team has collectively attended a three day virtual training on
how to effectively leverage the GraphCore's TensorFlow and PyTorch
software frameworks to significantly speed up machine learning
workflows. Throughout the course, the team was able to road test two
GraphCore systems and gain experience scaling ML workflows. This
training will be crucial for utilising the forthcoming Graphcore
IPU-M2000 system with 4 x Colossus MK2 IPUs. This system will augment
SCD’s and STFC’s AI computing capabilities, namely, PEARL and SCD-Cloud,
and is expected to provide an additional 1 PetaFlops of computing
power towards AI applications.
2021-06-28 - SciML & Diamond Scientific Software Workshop
Network diagram from the talk "Enhanced Analysis of Diffraction and Microscopy with Machine Learning"
Scientific Machine Learning (SciML) and the Diamond Scientific Software
(SciSoft) groups hosted a joint event to highlight some of the Machine
Learning powered projects that are ongoing on the Rutherford campus.
This event marked the beginning of a series of joint workshops
with the overall aim of sharing information about ongoing Machine
Learning endeavours between the two groups. The first workshop aimed at
providing a broad overview of a number of projects that are of mutual
interest, with follow-up meetings more focussed on project discussions
and knowledge exchange to follow.
2021-06-01 - SciML @ BNL
Keith
Butler from SciML recently gave an invited talk at a workshop held
virtually in Berkeley National Laboratory. The three day workshop on the
theme of autonomous discovery was held in April (
website).
Keith presented our collaborative work with ISIS Neutron and Muon
Source at the session devoted to autonomous discovery using neutron
scattering.
