Welcome to ExCALIBUR-NEPTUNE’s documentation!

This will be the main location of the documentation for the whole ExCALIBUR-NEPTUNE project.

Acronyms

Acronym	Description
ACM
ADC
ADM	Alternating directions method
AIC	Akaike information criterion
ALM	Augmented Lagrange multiplier
AMR
AMReX
AND
ANL	Argonne National Laboratory
ANN	Artificial neural network
ANN	Artificial neural network
ANOVA	Analysis of Variance
API	Application Programming Interface
ARMA	Autoregressive moving averaoe
ARMAX	Autoregressive moving average with exogenous input
ASQ	Adaptive sparse quadrature
ATS	Advanced Terrestrial Simulator, previously Arctic Terrestrial Simulator
AUXVAL
BD
BG/L
BIC	Bayesian information criterion
BIM	Empirical interpolation method
BOX
BP
BPOD	Balanced proper orthogonal decomposition
CabanaMD
CCA	Canonical correlation analysis
CD
CEA
CESM
CFD	Computational fluid dynamics
CI
CNN	Convolutional neural network
COGENT
COMPAT
CoSaMP	Compressive sampling matching pursuit
COSMO
COSSAN
CPP
CPU	Central Processing Unit
CRC
CS	Compressed sensing
CSE
CSMP
CTO
CUDA	Compute Unified Device Architecture
CWIPI
CWT	Continuous wavelet transform
DAG	Direct Acyclic Graph
DAKOTA
DCT	Discrete cosine transform
DDA
DEIM	Discrete empirical interpolation method
DFT	Discrete Fourier fransform
DFT	Discrete Fourier Transform
DiMDc	Dynamic mode decomposition with control
DL	Deep learning
DMD	Dynamic mode decomposition
DMDc	Dynamic mode decomposition with control
DNS	Direct numerical simulation
DOE	Department of Energy
DOI	Digital Object Identifier
DSL	Domain-Specific Language
DWT	Discrete wavelet transform
ECOG	Electrocorticography
ECP
ECP-copa
eDMD	Extended DMD
EIRENE
EM	Expectation maximization
EOF	Empirical orthogonal functions
ERA	Eigensystem realization algorithm
ESC	Extremum-seeking control
ESI
ESMF
ETS
EU	European Union
FACETS
FCI	Flux-Coordinate Independent (methods)
FEM	Finite Element Method
FEniCS
FFT	Fast Fourier transform
FFT	Fast Fourier Transform
FFTW	Fastest Fourier Transform in the West (library)
FLASH
GA	General Atomics
GBD
GBS
GDB
GMM	Gaussian mixture model
GMRES	Generalized Minimal Residual method
GNU	GNU’s Not Unix!
gPC	Generalised polynomial chaos
GPU	Graphics Processing Unit
GSA	Global sensitivity analysis
GUI	Graphical User Interface
HAVOK	Hankel alternative view of Koopman
HDF5	Hierarchical Data Format (version 5)
HLA
HPC	High Performance Computing
HVAR
ICA	Independent component analysis
ICON
IEEE
IMAS
IMEX	Implicit-Explicit Methods
IO
ITG	Ion Temperature Gradient
ITM	Ion Tearing Mode
ITPA
JET	Joint European Torus
JL	JohnsonLindensfrauss
JOREK
KL	KullbackLeib1er
KLT	Karhunen-Loeve transform
LAD	Least absolute deviations
LAMMPS	Large-scale Atomic/Molecular Massively Parallel Simulator
LANL	Los Alamos National Laboratory
LASH
LASSO	Least absolute shrinkage and selection operator
LASSO	Least Absolute Shrinkage and Selection Operator
LDA	Linear discriminant analysis
LGPL	GNU Lesser General Public License
LHSamp
LLNL	Lawrence Livermore National Laboratory
LQE	Linear quadratic estimator
LQG	Linear quadratic Gaussian controller
LQR	Linear quadratic regulator
LTI	Linear time invariant system
MAP	Maximium A Posteriori
MC	Monte Carlo (methods)
MCMC	Markov Chain Monte Carlo
MCT
MESAGE
MFMA
MFMC	Multi-fidelity Monte Carlo
MHD	Magnetohydrodynamics
MIMC
MIMO	Multiple input, multiple output
MIT	Massachusetts Institute of Technology
ML	Machine Learning
MLC	Machine learning control
MLMC	Multi-level Monte Carlo
MLMF
MMF
MOOSE
MOR	Model Order Reduction
Most	Common Acronyms
MPE	Missing point estimation
MPI	Message Passing Interface
mrDMD	Multi-resolution dynamic mode decomposition
MSSC
MUMPS
MUSCLE 3	Multiscale Coupling Library and Environment 3
NARMAX	Nonlinear autoregressive model with exogenous inputs
NEMO
NEPTUNE
NLS	Nonlinear Schrdinger equation
NUCODE
OASIS
OASIS4
ODE	Ordinary differential equation
OKID	Observer Kalman filter identification
OLYMPUS
OMFIT
OU	Oxford University
OUU	Optimisation under uncertainty
OUUWA
PASTIX
PBH	PopovBelevitchHautus test
PC	Polynomial chaos
PCA	Principal components analysis
PCE	Polynomial chaos expansion
PCP	Principal component pursuit
PDE	Partial differential equation
PDE-FIND	Partial differential equation functional identification of nonlinear dynamics
PDF	Probability distribution function
PFC
PGD
PICPIF
PID	Proportional-integral-derivative control
PIV	Particle image velocimetry
POD	Proper orthogonal decomposition
POOMA
PP20
PRESET
QA
QCG
QMC	Quasi-Monte-Carlo
QoI	Quantity of interest
RIP	Restricted isometry property
RKF23
RKHS	Reproducing kernel Hilbert space
RNG
RNN	Recurrent neural network
ROM	Reduced order model
ROM	Reduced-order model
RPCA	Robust principal components analysis
rSVD	Randomized SVD
SAMRAI
SGD	Stochastic gradient descent
SIAM	Society for Industrial and Applied Mathematics
SINDy	Sparse identification of nonlinear dynamics
SISO	Single input, single output
SLSQT	Sequential Least-Squares’ Thresholding
SMARDDA
SMART
SMITER
SMwiki
SNOWPAC	Stochastic Nonlinear Optimisation with Path-Augmented Constraints (software package)
SOL	Scrape-Off Layer
SOLEDGE
SOLPS
SRC	Sparse representation for classification
SRS
SSA	Singular spectrum analysis
STARWALL
STFT	Short time Fourier transform
STIXGeneral
STLS	Sequential thresholded least-squares
STRUMPACK
SVD	Singular value decomposition
SVD	Singular value decomposition
SVM	Support vector machine
SVM
TAE
TICA	Time-lagged independent component analysis
TM
TOKAM
TOKAM3X
TOMS
TRIMEG
TUM
UK	United Kingdom
UKAEA	United Kingdom Atomic Energy Authority
UKRI	United Kingdom Research and Innovation
UQ	Uncertainty quantification
US
USA
UTF-8
VAC	Variational approach of conformation dynamics
VDE
VECMAtk
VORPAL
XGC1
XML
XMSF

Symbols

Symbol	Description
\([a,b]\)	arbitrary finite interval
\(d\)	number of dimensions over which the integral is performed
\(f_0\)	constant in the expansion of \(f\left(x_1,\ldots,x_d\right)\)
\(f\left(x_1,\ldots,x_d\right)\)	joint probability distribution
\(f_i(x_i)\)	coefficient in the expansion of \(f\left(x_1,\ldots,x_d\right)\)
\(f_{ij}(x_i,x_j)\)	coefficient in the expansion of \(f\left(x_1,\ldots,x_d\right)\)
\(p(x)\)	probability distributions
\(r\)	order of higher order term
\(x_i\)	generic parameter or variable
\({\bf x} =\left(x_1,x_2,\dots,x_d\right)\)	is a \(d\)-dimensional vector
\(P(x)\)	Cumulant probability distribution
\(\parallel Q \parallel_E\)	the ‘energy’ norm
\(S_i\)	Sobol sensitivity index, gives a normalised measure of the sensitivity of the distribution of \(f\) to the parameter \(x_i\)
\(S_{ij}\)	Sobol sensitivity index, gives a normalised measure of the sensitivity of the distribution of \(f\) to the parameters \(x_i\) and \(x_j\)
\(\mathrm{Var}(f)\)	variance of the distribution of \(f\) computed by integrating over all variables \(x_i\)
\(V_i\)	variance of the distribution of \(f\) as the parameter \(x_i\) varies
\(V_{ij}\)	variance of the distribution of \(f\) as the parameters \(x_i\) and \(x_j\) vary
\(\mathbb{E}\)	expectation
\(\mathbb{E}_{x_{k\neq i}}\)	expectation computed by integrating over all the \(x_k\) except for \(x_i\)
\(\mathbb{E}_{k\neq i, l\neq j}\)	expectation computed by integrating over all the \(x_k\) except for \(x_i\) and \(x_j\)
\(\xi_i\)	randon number within the unit interval \([0,1]\)

ExCALIBUR NEPTUNE Charter

All members of the ExCALIBUR NEPTUNE team should be aware that to meet the challenges of the NEPTUNE project, and the ExCALIBUR overarching pillars, a distributed team of scientists, software engineers and architecture specialists from different UK institutions will be required to form a community around the NEPTUNE project (and will connect across the overarching ExCALIBUR programme). A high-level objective is to ensure that developed software is of the highest quality, implying a rigid requirement around the production of high-quality documentation and reproducible verification and validation tests for the codebase as it evolves. Since development work may transfer between institutions, it is important that common standards for documentation and testing be available and easy to deploy. The initial NEPTUNE exploratory Proxyapps may be written in a range of languages including for example Python, C++/DPC++, Object Fortran or Julia, however it is envisaged that there will be an emerging steer towards a reduced set of languages and technologies to ensure interoperability across the NEPTUNE software stack, ultimately leading to coupled simulations covering all the physics necessary to deliver an “actionable” simulation for the plasma edge. It is not yet clear for example whether SYCL, Kokkos or OpenMP 5 will offer the most performance portable and sustainable solution for NEPTUNE. The team is therefore expected to be agile and amenable to change once it is clear which are the most promising long–term solutions. For example, a selection of SYCL for the long-term framework/code(s) would force refactoring of any code that is not consistent with a NEPTUNE library and code base instantiated in DPC++, and where feasible, team members should support this process.

Source code for all development should be accessible by the entire NEPTUNE team and all tests should be repeatable by different workers without the need for re-training and/or any possible confusion as to the procedures and metrics needed to declare a test successful.

NEPTUNE will be developed as a sequence of ‘core’ Proxyapps (to be distinguished from other Proxyapps designed to test some novel technique). Core Proxyapps will all need a documentation and testing framework, which must be agreed between all partners for the entire project. This will require developers to work closely with UKAEA and other team members.

A commitment is also expected by all parties to help UKAEA and the Met Office (as SRO for ExCALIBUR) to publicise the project and build a fully connected community across the ExCALIBUR programme, UKRI and Academia, focused upon a team of approximately twenty UK Fusion use case experts. This will be essential for meeting the grand challenge goal of developing a state-of-the-art, Exascale targeted, UK-based plasma physics simulation capability for the tokamak edge plasma (see Science Plan 1).

All Core Proxyapps and related infrastructure/documentation across the NEPTUNE project should meet the demands of the Code Structure and Coordination work package FM-WP4 in so far as the developing project standards:

• adopt a consistent choice of definitions (ontology) of objects or equivalently classes,

• adhere to clearly defined common file formats and interfaces to components for data input and output.

• provide suitably flexible data structures for common use by all developers,

• are established through good scientific software engineering best practice,

• demonstrate performance portability and exploit agreed DSL-like interfaces where possible targeting Exascale-relevant architectures,

• can be integrated into a VVUQ framework and

• are embedded within a coordination and benchmarking framework for correctness testing and performance evaluation.

In order to meet Strategic Priorities Fund terms around eligibility, and to steer the project towards a modular platform where developments across all partners can be integrated into an eventual code or platform available for open use by the European fusion community, a requirement is that all ExCALIBUR NEPTUNE Grant beneficiaries make technology / source code developed through the programme (foreground Intellectual Property) available as open source under a programme–wide permissive license (currently selected as 3-clause BSD for core/foundational infrastructure). Government Digital Service guidance (to which the project subscribes), discussing the benefits of open versus closed technology/software/data can be found in: https://www.gov.uk/government/publications/open-source-guidance/when-code-should-be-open-or-closed and https://www.gov.uk/guidance/be-open-and-use-open-source.

GLOSSARY:

Glossary of terms

Term or Acronym	Definition
C++	Programming language, see https://isocpp.org/
DPC++	Data Parallel C++, Intel compiler for C++ with SYCL extension
DSL	Domain Specific Language, programming language developed for a specific area of resrearch and development
Object Fortran	Programming language, more precisely Fortran 95 or later exploiting object-oriented features of the language, see https://wg5-fortran.org/f2008.html
Python	Programming language, see https://python.org/
SYCL	C++ library for portable high performance computing, see https://www.khronos.org/sycl/
Kokkos	C++ library for portable high performance computing https://cfwebprod.sandia.gov/ cfdocs/CompResearch/docs/Kokkos-Multi-CoE.pdf
OpenMP	Software for parallel programming
SRO	Senior Responsible Owner role in UK government project delivery
Julia	Programming language, see https://julialang.org/
UKRI	United Kingdom Research and Innovation, a non-departmental public body encompassing the research councils and Innovate UK
VVUQ	Verification, Validation and Uncertainty Quantification

1

W. Arter, L. Anton, D. Samaddar, and R. Akers. ExCALIBUR Fusion Modelling System Science Plan. Technical Report CD/EXCALIBUR-FMS/0001, UKAEA, 2019.

Writing Documentation

This documentation is written using reStructuredText (RST) and built using Sphinx on ReadTheDocs. For a good introduction on writing RST, see the reStructuredText primer in the Sphinx docs.

Indices and tables

• Index

• Module Index

• Search Page