A5069823171- Seismic Imaging and Inversion Techniques
- Parallel Computing and Optimization Techniques
- Reservoir Engineering and Simulation Methods
- Advanced Data Storage Technologies
- Distributed and Parallel Computing Systems
- Hydraulic Fracturing and Reservoir Analysis
- Medical Imaging Techniques and Applications
- Advanced Numerical Methods in Computational Mathematics
- Computational Physics and Python Applications
- Computer Graphics and Visualization Techniques
- Seismic Waves and Analysis
- Innovations in Concrete and Construction Materials
- Geophysical Methods and Applications
- Computational Geometry and Mesh Generation
- Radiomics and Machine Learning in Medical Imaging
- Drilling and Well Engineering
- Simulation Techniques and Applications
- Embedded Systems Design Techniques
- Advanced MRI Techniques and Applications
- Numerical Methods and Algorithms
- Marine Bivalve and Aquaculture Studies
- Scientific Computing and Data Management
- Oceanographic and Atmospheric Processes
- Physiological and biochemical adaptations
- Marine and coastal ecosystems
European Centre for Medium-Range Weather Forecasts
2018-2021
Imperial College London
2011-2020
The London College
2017-2018
Grantham College
2017
University of British Columbia
2017
Lagrangian analysis is a powerful way to analyse the output of ocean circulation models and other velocity data such as from altimetry. In approach, large sets virtual particles are integrated within three-dimensional, time-evolving fields. Over several decades, variety tools methods for this purpose have emerged. Here, we review state art in field data, starting fundamental kinematic framework with focus on large-scale open applications. Beyond use explicit fields, consider influence...
Firedrake is a new tool for automating the numerical solution of partial differential equations. adopts domain-specific language finite element method FEniCS project, but with pure Python runtime-only implementation centred on composition several existing and abstractions particular aspects scientific computing. The result more complete separation concerns which eases incorporation separate contributions from computer scientists, analysts application specialists. These may add functionality,...
Abstract. As ocean general circulation models (OGCMs) move into the petascale age, where output of single simulations exceeds petabytes storage space, tools to analyse these will need scale up too. Lagrangian analysis, virtual particles are tracked through hydrodynamic fields, is an increasingly popular way OGCM output, by mapping pathways and connectivity biotic abiotic particulates. However, current software stack analysis codes not dynamic enough cope with increasing complexity, for...
Abstract. We introduce Devito, a new domain-specific language for implementing high-performance finite-difference partial differential equation solvers. The motivating application is exploration seismology which methods such as full-waveform inversion and reverse-time migration are used to invert terabytes of seismic data create images the Earth's subsurface. Even using modern supercomputers, it can take weeks process single survey useful subsurface image. computational cost dominated by...
Stencil computations are a key part of many high-performance computing applications, such as image processing, convolutional neural networks, and finite-difference solvers for partial differential equations. Devito is framework capable generating highly optimized code given symbolic equations expressed in Python , specialized in, but not limited to, affine (stencil) codes. The lowering process—from mathematical down to C++ code—is performed by the compiler through series intermediate...
Writing software packages for seismic inversion is a very challenging task because problems such as full-waveform or least-squares imaging are algorithmically and computationally demanding due to the large number of unknown parameters fact that waves propagated over many wavelengths. Therefore, frameworks need combine versatility performance provide geophysicists with means flexibility implement complex algorithms scale exceedingly 3D problems. Following these principles, we have developed...
Abstract. As Ocean General Circulation Models (OGCMs) move into the petascale age, where output from global high-resolution model runs can be of order hundreds terabytes in size, tools to analyse these models will need scale up too. Lagrangian Analysis, virtual particles are tracked through hydrodynamic fields, is an increasingly popular way OGCM output, by mapping pathways and connectivity biotic abiotic particulates. However, current software stack Analysis codes not dynamic enough cope...
To better predict how populations and communities respond to climatic temperature variation, it is necessary understand the shape of response fitness-related rates evolves (the thermal performance curve). Currently, there disagreement about extent which evolution curves constrained. One school thought has argued for prevalence thermodynamic constraints through enzyme kinetics, whereas another argues that adaptation can-at least partly-overcome such constraints. shed further light on this...
The use of composable abstractions allows the application new and established algorithms to a wide range problems while automatically inheriting benefits well-known performance optimisations. This work highlights composition PETSc DMPlex domain topology abstraction with Firedrake automated finite element system create PDE solving environment that combines expressiveness, flexibility high performance. We describe how utilises provide indirection maps required for assembly, supporting various...
Abstract. In the simulation of complex multi-scale flows arising in weather and climate modelling, one biggest challenges is to satisfy strict service requirements terms time solution budgetary constraints energy solution, without compromising accuracy stability application. These simulations require algorithms that minimise footprint along with required produce a maintain physically level accuracy, are numerically stable, resilient case hardware failure. The European Centre for Medium-Range...
Since its reintroduction by Pratt (1999) , full-waveform inversion (FWI) has gained a lot of attention in geophysical exploration because ability to build high-resolution velocity models more or less automatically areas complex geology. While there is an extensive and growing literature on the topic, publications focus mostly technical aspects, making this topic inaccessible for broader audience due lack simple introductory resources newcomers computational geophysics. We will accomplish...
The distribution of marine species is often modeled using Eulerian approaches, in which changes to population density or abundance are calculated at fixed locations space. Conversely, Lagrangian, individual-based, models simulate the movement individual particles moving continuous space, with broader-scale patterns such as being an emergent property many, potentially adaptive, individuals. These offer advantages examining dynamics across spatiotemporal scales and making comparisons...
Domain specific languages (DSL) have been used in a variety of fields to express complex scientific problems concise manner and provide automated performance optimization for range computational architectures. As such DSLs powerful mechanism speed up Python computation that goes beyond traditional vectorization pre-compilation approaches, while allowing domain scientists build applications within the comforts software ecosystem. In this paper we present Devito, new finite difference DSL...
Domain specific languages have successfully been used in a variety of fields to cleanly express scientific problems as well simplify implementation and performance optimization on different computer architectures. Although large number stencil are available, finite difference domain proved challenging design because most practical use cases require additional features that fall outside the abstraction. Inspired by complexity real-world seismic imaging problems, we introduce Devito, language...
This tutorial is the third part of a full-waveform inversion (FWI) series with step-by-step walkthrough setting up forward and adjoint wave equations building basic FWI framework. For discretizing solving equations, we use Devito ( http://www.opesci.org/devito-public ), Python-based domain-specific language for automated generation finite-difference code Lange et al., 2016 ). The first two parts this Louboutin 2017 , 2018 ) demonstrated how to solve acoustic equation modeling seismic shot...
This is the second part of a three-part tutorial series on full-waveform inversion (FWI) in which we provide step-by-step walk through setting up forward and adjoint wave equation solvers an optimization framework for inversion. In Part 1 ( Louboutin et al., 2017 ), showed how to use Devito http://www.opesci.org/devito-public ) set solve acoustic equations with (impulsive) seismic sources sample wavefields at receiver locations model shot records. Here 2, will discuss and, from that, can...
Abstract. We introduce Devito, a new domain-specific language for implementing high-performance finite difference partial differential equation solvers. The motivating application is exploration seismology where methods such as Full-Waveform Inversion and Reverse-Time Migration are used to invert terabytes of seismic data create images the earth's subsurface. Even using modern supercomputers, it can take weeks process single survey useful subsurface image. computational cost dominated by...
The life-cycle of a partial differential equation (PDE) solver is often characterized by three development phases: the stable numerical discretization, correct (verified) implementation, and optimization implementation for different computer architectures. Often it only after significant time effort has been invested that performance bottlenecks PDE are fully understood, precise details varies between One way to mitigate this issue establish reliable model allows analyst make predictions how...
Inversion and PDE-constrained optimization problems often rely on solving the adjoint problem to calculate gradient of objec- tive function. This requires storing large amounts intermediate data, setting a limit largest that might be solved with given amount memory available. Checkpointing is an approach can reduce required by redoing parts computation instead results. The Revolve checkpointing algorithm o ers optimal schedule trades computational cost for smaller footprints. Integrat- ing...
PETSC2.0 is a software toolkit for portable, parallel (and serial) numerical solution of partial differential equations and minimization problems. It includes the linear nonlinear systems equations. These codes are written in data-structure-neutral manner to enable easy reuse flexibility.