A5087979024- Distributed and Parallel Computing Systems
- Advanced Data Storage Technologies
- Advanced Chemical Physics Studies
- Parallel Computing and Optimization Techniques
- Spectroscopy and Quantum Chemical Studies
- Scientific Computing and Data Management
- Catalysis and Oxidation Reactions
- Precipitation Measurement and Analysis
- Quantum and electron transport phenomena
- Quantum many-body systems
- Semiconductor materials and devices
- Distributed systems and fault tolerance
- Quantum Computing Algorithms and Architecture
- Meteorological Phenomena and Simulations
- Advanced NMR Techniques and Applications
- Machine Learning in Materials Science
- Semiconductor Lasers and Optical Devices
- Analytical Chemistry and Sensors
- Atmospheric Ozone and Climate
- Cloud Computing and Resource Management
- Iterative Learning Control Systems
- Big Data and Business Intelligence
- Magnetism in coordination complexes
- Catalytic Processes in Materials Science
- Traffic Prediction and Management Techniques
European Centre for Medium-Range Weather Forecasts
2017-2025
Max Planck Institute for Solid State Research
2015-2020
University of Waterloo
2020
University of Iowa
2020
University of Cambridge
2012-2020
Data61
2020
Commonwealth Scientific and Industrial Research Organisation
2020
The Dodd-Walls Centre for Photonic and Quantum Technologies
2020
Massey University
2020
King's College London
2020
A novel stochastic Complete Active Space Self-Consistent Field (CASSCF) method has been developed and implemented in the Molcas software package. two-step procedure is used, which CAS configuration interaction secular equations are solved stochastically with Full Configuration Interaction Quantum Monte Carlo (FCIQMC) approach, while orbital rotations performed using an approximated form of Super-CI method. This new does not suffer from strong combinatorial limitations standard MCSCF...
For many decades, quantum chemical method development has been dominated by algorithms which involve increasingly complex series of tensor contractions over one-electron orbital spaces. Procedures for their derivation and implementation have evolved to require the minimum amount logic rely heavily on computationally efficient library-based matrix algebra optimised paging schemes. In this regard, recent exact stochastic reduce computational scaling memory overhead requires a contrasting...
We expand upon the recent semi-stochastic adaptation to full configuration interaction quantum Monte Carlo (FCIQMC). present an alternate method for generating deterministic space without a priori knowledge of wave function and stochastic efficiencies variety both molecular lattice systems. The algorithmic details efficient implementation are presented, with particular consideration given effect that has on parallel performance in FCIQMC. further demonstrate benefit calculation reduced...
We present a new approach to calculate excited states with the full configuration interaction quantum Monte Carlo (FCIQMC) method. The uses Gram-Schmidt procedure, instantaneously applied stochastically evolving distributions of walkers, orthogonalize higher energy against lower ones. It can thus be used study several lowest-energy system within same symmetry. This additional step is particularly simple and computationally inexpensive, requiring only small change underlying FCIQMC algorithm....
We present NECI, a state-of-the-art implementation of the Full Configuration Interaction Quantum Monte Carlo (FCIQMC) algorithm, method based on stochastic application Hamiltonian matrix sparse sampling wave function. The program utilizes very powerful parallelization and scales efficiently to more than 24 000 central processing unit cores. In this paper, we describe core functionalities NECI its recent developments. This includes capabilities calculate ground excited state energies,...
We provide a spin-adapted formulation of the Full Configuration Interaction Quantum Monte Carlo (FCIQMC) algorithm, based on Graphical Unitary Group Approach (GUGA), which enables exploitation SU(2) symmetry within this stochastic framework. Random excitation generation and matrix element calculation Shavitt graph GUGA can be efficiently implemented via biasing procedure branching diagram. The use spin-pure basis explicitly resolves different spin-sectors ensures that stochastically sampled...
Memory and I/O performance bottlenecks in supercomputing simulations are two key challenges that must be addressed on the road to Exascale. The new byte-addressable persistent non-volatile memory technology from Intel, DCPMM, promises an exciting opportunity break with status quo, unprecedented levels of capacity at near-DRAM speeds. Here, we explore potential DCPMM context high-performance scientific applications terms outright performance, efficiency usability for both its App Direct...
We employ the recently developed full configuration interaction quantum Monte Carlo (FCIQMC) method to compute π → π* vertical excitation energies of ethene and all-trans butadiene. These excitations have been subject extensive theoretical studies, their location with respect corresponding absorption band maximum is source a long lingering debate. Here, we reliably estimate butadiene by performing FCIQMC calculations for spaces as large 10(18) 10(29) Slater determinants, respectively. For...
As urban areas face increasing threats from climate change, citizen science has emerged as an important tool to engage communities in monitoring and responding environmental challenges, thus filling the gap which existing tools are not addressing appropriately. Citizen initiatives essential for engaging citizens action, involving them observations human impacts. These participatory between society have gained popularity across various fields, including sociology, astronomy, protection. By...
In numerical weather prediction and high-performance computing, the primary computational bottleneck has gradually evolved from floating-point arithmetic to throughput of data storage. This phenomenon is commonly referred as I/O performance gap. We present MultIO, a set software libraries that provide two mechanisms mitigate this effect: an asynchronous I/O-server decouple output model computations, user-programmable processing pipelines operate on directly.
Numerical Weather Prediction (NWP) and Climate simulations sit in the intersection between classically understood High Performance Computing (HPC) Big Data / Analytics (HPDA) communities. Driven by ever more ambitious scientific goals, both size number of output data elements generated as part NWP operations has grown several orders magnitude, will continue to grow into future. The total amount is expected exponentially with time, over last 30 years this increase been approximately 40% per...
Numerical Weather Prediction (NWP) and Climate simulations sit at the intersection between classically understood High Performance Computing (HPC) Big Data / Analytics (HPDA). Driven by ever more ambitious scientific goals, both size number of output data-elements generated as part NWP operations have grown several orders magnitude, are expected to continue growing exponentially in future. Over last 30 years this increase has been approximately 40% per year. To cope with projected growth,...
Object storage solutions potentially address long-standing performance issues with POSIX file systems for certain I/O workloads, and new technologies offer promising characteristics data-intensive use cases.In this work, we present a preliminary assessment of Intel's Distributed Asynchronous Store (DAOS), an emerging high-performance object store, in conjunction non-volatile evaluate its potential HPC storage. We demonstrate DAOS can provide the required performance, bandwidth scaling...
High-performance object stores are an emerging technology which offers alternative solution in the field of HPC storage, with potential to address long-standing scalability issues traditional distributed POSIX file systems due excessive consistency assurance and metadata prescriptiveness.In this paper we assess performance storing object-like data within a standard system, where configuration access mechanisms have not been optimised for behaviour, compare investigate benefits using storage...
Operational Numerical Weather Prediction (NWP) workflows are highly data-intensive. Data volumes have increased by many orders of magnitude over the last 40 years, and expected to continue do so, especially given upcoming adoption Machine Learning in forecast processes. Parallel POSIX-compliant file systems been dominant paradigm data storage exchange HPC for years. This paper presents ECMWF's move beyond POSIX paradigm, implementing a backend their library support DAOS -- novel...
Operational Numerical Weather Prediction (NWP) workflows are highly data-intensive. Data volumes have increased by many orders of magnitude over the last 40 years, and expected to continue do so, especially given upcoming adoption Machine Learning in forecast processes. Parallel POSIX-compliant file systems been dominant paradigm data storage exchange HPC for years. This paper presents ECMWF's move beyond POSIX paradigm, implementing a backend their library support DAOS --- novel...
Distributed Asynchronous Object Store (DAOS) is a novel software-defined object store leveraging Non-Volatile Memory (NVM) devices, designed for high performance. It provides number of interfaces applications to undertake I/O, ranging from native storage API DAOS FUSE module seamless compatibility with existing using POSIX file system APIs. In this paper we discuss these and the options they provide, exercise through them various I/O benchmarks, analyse observed We also briefly compare...
Object storage solutions potentially address long-standing performance issues with POSIX file systems for certain I/O workloads, and new technologies offer promising characteristics data-intensive use cases. In this work, we present a preliminary assessment of Intel's Distributed Asynchronous Store (DAOS), an emerging high-performance object store, in conjunction non-volatile evaluate its potential HPC storage. We demonstrate DAOS can provide the required performance, bandwidth scaling...
High-performance object stores are an emerging technology which offers alternative solution in the field of HPC storage, with potential to address long-standing scalability issues traditional distributed POSIX file systems due excessive consistency assurance and metadata prescriptiveness. In this paper we assess performance storing object-like data within a standard system, where configuration access mechanisms have not been optimised for behaviour, compare investigate benefits using storage...