Large-scale Computational Resources Research Articles

Accelerating Computational Materials Discovery with Machine Learning and Cloud High-Performance Computing: from Large-Scale Screening to Experimental Validation.

High-throughput computational materials discovery has promised significant acceleration of the design and discovery of new materials for many years. Despite a surge in interest and activity, the constraints imposed by large-scale computational resources present a significant bottleneck. Furthermore, examples of very large-scale computational discovery carried out through experimental validation remain scarce, especially for materials with product applicability. Here, we demonstrate how this vision became reality by combining state-of-the-art machine learning (ML) models and traditional physics-based models on cloud high-performance computing (HPC) resources to quickly navigate through more than 32 million candidates and predict around half a million potentially stable materials. By focusing on solid-state electrolytes for battery applications, our discovery pipeline further identified 18 promising candidates with new compositions and rediscovered a decade's worth of collective knowledge in the field as a byproduct. We then synthesized and experimentally characterized the structures and conductivities of our top candidates, the NaxLi3-xYCl6 (0≤ x≤ 3) series, demonstrating the potential of these compounds to serve as solid electrolytes. Additional candidate materials that are currently under experimental investigation could offer more examples of the computational discovery of new phases of Li- and Na-conducting solid electrolytes. The showcased screening of millions of materials candidates highlights the transformative potential of advanced ML and HPC methodologies, propelling materials discovery into a new era of efficiency and innovation.

Developing Cheap but Useful Machine Learning-Based Models for Investigating High-Entropy Alloy Catalysts.

This work aims to address the challenge of developing interpretable ML-based models when access to large-scale computational resources is limited. Using CoMoFeNiCu high-entropy alloy catalysts as an example, we present a cost-effective workflow that synergistically combines descriptor-based approaches, machine learning-based force fields, and low-cost density functional theory (DFT) calculations to predict high-quality adsorption energies for H, N, and NHx (x = 1, 2, and 3) adsorbates. This is achieved using three specific modifications to typical DFT workflows including: (1) using a sequential optimization protocol, (2) developing a new geometry-based descriptor, and (3) repurposing the already-available low-cost DFT optimization trajectories to develop a ML-FF. Taken together, this study illustrates how cost-effective DFT calculations and appropriately designed descriptors can be used to develop cheap but useful models for predicting high-quality adsorption energies at significantly lower computational costs. We anticipate that this resource-efficient philosophy may be broadly relevant to the larger surface catalysis community.

Building a collaborative cloud platform to accelerate heart, lung, blood, and sleep research.

Research increasingly relies on interrogating large-scale data resources. The NIH National Heart, Lung, and Blood Institute developed the NHLBI BioData CatalystⓇ (BDC), a community-driven ecosystem where researchers, including bench and clinical scientists, statisticians, and algorithm developers, find, access, share, store, and compute on large-scale datasets. This ecosystem provides secure, cloud-based workspaces, user authentication and authorization, search, tools and workflows, applications, and new innovative features to address community needs, including exploratory data analysis, genomic and imaging tools, tools for reproducibility, and improved interoperability with other NIH data science platforms. BDC offers straightforward access to large-scale datasets and computational resources that support precision medicine for heart, lung, blood, and sleep conditions, leveraging separately developed and managed platforms to maximize flexibility based on researcher needs, expertise, and backgrounds. Through the NHLBI BioData Catalyst Fellows Program, BDC facilitates scientific discoveries and technological advances. BDC also facilitated accelerated research on the coronavirus disease-2019 (COVID-19) pandemic.

Calculation of energies and hyperfine-structure constants of U+233 and U233

We carried out calculations of the energies and magnetic dipole hyperfine-structure constants of the low-lying states of $^{233}\mathrm{U}^{+}$ and $^{233}\mathrm{U}$ using two different approaches. With six valence electrons and a very heavy core, uranium represents a major challenge for precision atomic theory even using large-scale computational resources. The first approach combines configuration interaction (CI) with a method allowing us to include core-valence correlations to all orders of the perturbation theory over residual Coulomb interaction. The second approach is a pure CI method which allows the use of different initial approximations. We present a detailed analysis of all calculated properties and discuss the advantages and disadvantages of each of these methods. We report a preliminary value of the U nuclear magnetic moment and outline the need for further experiments.

Classical variational simulation of the Quantum Approximate Optimization Algorithm

A key open question in quantum computing is whether quantum algorithms can potentially offer a significant advantage over classical algorithms for tasks of practical interest. Understanding the limits of classical computing in simulating quantum systems is an important component of addressing this question. We introduce a method to simulate layered quantum circuits consisting of parametrized gates, an architecture behind many variational quantum algorithms suitable for near-term quantum computers. A neural-network parametrization of the many-qubit wavefunction is used, focusing on states relevant for the Quantum Approximate Optimization Algorithm (QAOA). For the largest circuits simulated, we reach 54 qubits at 4 QAOA layers, approximately implementing 324 RZZ gates and 216 RX gates without requiring large-scale computational resources. For larger systems, our approach can be used to provide accurate QAOA simulations at previously unexplored parameter values and to benchmark the next generation of experiments in the Noisy Intermediate-Scale Quantum (NISQ) era.

Accelerated Large-Scale Seismic Damage Simulation With a Bimodal Sampling Approach

Regional damage simulation is a promising method to prepare organizations for the unforeseeable impact of a probable seismic natural hazard. Nonlinear time history analysis (NLTHA) of the finite element models (FEM) of the buildings in a region can provide resembling results to the actual buildings’ damages and responses. This approach requires large-scale computational resources, and to improve efficiency, parallel processing and representing building FEM models with lumped mass models are proposed. However, the computing complexity is still far-reaching when high-performance computing is not available. The building inventory of a region consists of numerous similar buildings with a limited number of distinct structures. In this paper, we propose a data-driven method that runs the NLTHA for the distinct structures exclusively and infers the damage and responses of other buildings using a surrogate model. Considering the skewed distribution of the buildings in a region, a novel informative sample selection method is proposed that is designed for bimodal sampling of the input domain. We use the Gaussian process regression as the surrogate model and compare the performance of different sample selection methods. The proposed method is able to approximate the results of the regional damage simulation regarding total economic loss estimation with 98.99% accuracy while reducing the computational demand to about 1/7th of the simulation processing time.

Impacts of global warming on West African monsoon rainfall: Downscaling by pseudo global warming method

A set of numerical experiments was conducted in order to investigate the impacts of global warming on West African monsoon rainfall for selected five years. The experiments varied different cumulus, microphysics and planetary boundary layer parameterization schemes. Rainfall characteristics over three climatic zones, Guinea Coast, Savannah and Sahel, was analyzed. The potential change associated with global warming is assessed by the pseudo global warming (PGW) downscaling method. Multiple PGW runs were conducted using climate perturbation from the 40-member ensemble of the Community Earth System Model version 1 (CESM1) coupled with Community Atmospheric Model version 5.2 (CAM5.2) component large ensemble project. The model output was compared with TRMM and GPCP rainfall and atmospheric parameters from ECMWF reanalysis datasets. Results show that the rainfall amount in the 2070s estimated from the PGW runs substantially increases, especially in the eastern Sahel due to enhanced moisture convergence, compared to the current climate. The percentage change in simulated total rainfall amount can increase or decrease by 50% in the PGW runs and the theoretical rainfall computed based on Clausius-Clapeyron relation. Also, found is an increase (decrease) in heavy (both light and moderate) rainfall amount. These results, however, depend on the GCM used as the boundary conditions of the RCM. This suggests that the 4 o C change in average surface temperature derived from the 40-member ensemble model strongly influenced the increased rainfall simulated by the PGW experiments. Thus, highlighting the advantage of using the PGW technique to estimate the likely difference between present and future climate with reduced large-scale model biases and computational resources.

Bioinformatics services for analyzing massive genomic datasets.

The explosive growth of next-generation sequencing data has resulted in ultra-large-scale datasets and ensuing computational problems. In Korea, the amount of genomic data has been increasing rapidly in the recent years. Leveraging these big data requires researchers to use large-scale computational resources and analysis pipelines. A promising solution for addressing this computational challenge is cloud computing, where CPUs, memory, storage, and programs are accessible in the form of virtual machines. Here, we present a cloud computing-based system, Bio-Express, that provides user-friendly, cost-effective analysis of massive genomic datasets. Bio-Express is loaded with predefined multi-omics data analysis pipelines, which are divided into genome, transcriptome, epigenome, and metagenome pipelines. Users can employ predefined pipelines or create a new pipeline for analyzing their own omics data. We also developed several web-based services for facilitating downstream analysis of genome data. Bio-Express web service is freely available at https://www.bioexpress.re.kr/.

Toward High Fidelity Materials Property Prediction from Multiscale Modeling and Simulation

AbstractThe current approach to materials discovery and design remains dominated by experimental testing, frequently based on little more than trial and error. With the advent of ever more powerful computers, rapid, reliable, and reproducible computer simulations are beginning to represent a feasible alternative. As high performance computing reaches the exascale, exploiting the resources efficiently presents interesting challenges and opportunities. Multiscale modeling and simulation of materials are extremely promising candidates for exploiting these resources based on the assumption of a separation of scales in the architectures of nanomaterials. Examples of hierarchical and concurrent multiscale approaches are presented which benefit from the weak scaling of monolithic applications, thereby efficiently exploiting large scale computational resources. Several multiscale techniques, incorporating the electronic to the continuum scale, which can be applied to the efficient design of a range of nanocomposites, are discussed. Then the work on the development of a software toolkit designed to provide verification, validation, and uncertainty quantification to support actionable prediction from such calculations is discussed.

Learning to Design From Humans: Imitating Human Designers Through Deep Learning

Abstract Humans as designers have quite versatile problem-solving strategies. Computer agents on the other hand can access large-scale computational resources to solve certain design problems. Hence, if agents can learn from human behavior, a synergetic human-agent problem-solving team can be created. This paper presents an approach to extract human design strategies and implicit rules, purely from historical human data, and use that for design generation. A two-step framework that learns to imitate human design strategies from observation is proposed and implemented. This framework makes use of deep learning constructs to learn to generate designs without any explicit information about objective and performance metrics. The framework is designed to interact with the problem through a visual interface as humans did when solving the problem. It is trained to imitate a set of human designers by observing their design state sequences without inducing problem-specific modeling bias or extra information about the problem. Furthermore, an end-to-end agent is developed that uses this deep learning framework as its core in conjunction with image processing to map pixel-to-design moves as a mechanism to generate designs. Finally, the designs generated by a computational team of these agents are then compared with actual human data for teams solving a truss design problem. Results demonstrate that these agents are able to create feasible and efficient truss designs without guidance, showing that this methodology allows agents to learn effective design strategies.

Production experience and performance for ATLAS data processing on a Cray XC-50 at CSCS

Prediction for requirements for the LHC computing for Run 3 and for Run 4 (HL-LHC) over the course of the next 10 year, show a considerable gap between required and available resources, assuming budgets will globally remain flat at best. This will require some radical changes to the computing models for the data processing of the LHC experiments. The use of large scale computational resources at HPC centres worldwide is expected to increase substantially the cost-efficiency of the processing. In order to pave the path towards the HL-LHC data processing, the Swiss Institute of Particle Physics (CHIPP) has taken the strategic decision to migrate the processing of all the Tier-2 workloads for ATLAS and other LHC experiments from a dedicated x86 ̲ 64 cluster that has been in continuous operation and evolution since 2007, to Piz Daint, the current European flagship HPC, which ranks third in the TOP500 at the time of writing. We report on the technical challenges and solutions adopted to migrate to Piz Daint, and on the experience and measured performance for ATLAS in over one year of running in production.

Improving operational flood forecasting in monsoon climates with bias-corrected quantitative forecasting of precipitation

ABSTRACTFor flood-prone countries subject to large-scale and seasonal flooding, precipitation forecasting is the single most important factor for improving the skill of flood forecasting for such large river basins dominated by the monsoon. Several flood forecasting agencies in South and Southeast Asia, where monsoon floods dominate (e.g. Bangladesh, Pakistan, India, Thailand and Vietnam), are currently using quantitative precipitation forecast (QPF) from numerical weather prediction (NWP) models. Although there are numerous studies reported in the literature to evaluate QPF precipitation performance, there appears to be lack of studies about the impact on the flood forecasting skill. In this study, we demonstrate tangible improvements in flood forecasting based on NWP precipitation forecast using an approach that is operationally feasible in resource-limited settings of many flood agencies. Our improvement is based on a bias correction methodology for enhancing the skill of QPF using observed and QPF climatology. The proposed approach can be applied to any type of QPF dataset such as those dynamically downscaled from regional NWP. We demonstrate clear and consistent improvement in the enhancement of flood forecasting skill at longer lead times of up to 7 days in three river basins of Ganges, Brahmaputra and Mekong by about 50% (reduction in RMSE) or 25% improvement in correlation when compared to the forecasts obtained from uncorrected QPF. Furthermore, our proposed bias correction methodology yields significantly higher skill improvement in flood forecast for global (non-downscaled) QPF than those dynamically downscaled QPFs for the macroscale hydrologic model used for forecasting stream flows. The simplicity of the QPF bias correction methodology along with the numerical efficiency can be of tremendous appeal to operational flood forecasting agencies of the developing world faced with large-scale monsoonal flooding and limited computational resources and time for disaster response.

Cloud computing for genomic data analysis and collaboration.

Next-generation sequencing has made major strides in the past decade. Studies based on large sequencing data sets are growing in number, and public archives for raw sequencing data have been doubling in size every 18 months. Leveraging these data requires researchers to use large-scale computational resources. Cloud computing, a model whereby users rent computers and storage from large data centres, is a solution that is gaining traction in genomics research. Here, we describe how cloud computing is used in genomics for research and large-scale collaborations, and argue that its elasticity, reproducibility and privacy features make it ideally suited for the large-scale reanalysis of publicly available archived data, including privacy-protected data.

Baryon interactions from lattice QCD with physical masses — strangeness S = -1 sector —

We present our recent results of baryon interactions with strangeness S = −1 based on Nambu-Bethe-Salpeter (NBS) correlation functions calculated fromlattice QCD with almost physical quark masses corresponding to (mk,mk) ≈ (146, 525) MeV and large volume (La)4 ≈ (96a)4 ≈ (8.1 fm)4. In order to perform a comprehensive study of baryon interactions, a large number of NBS correlation functions from NN to ΞΞ are calculated simultaneously by using large scale computer resources. In this contribution, we focus on the strangeness S = −1 channels of the hyperon interactions by means of HAL QCD method. Four sets of three potentials (the 3S1 − 3 D1 central, 3S1 − 3 D1 tensor, and the 1S0 central potentials) are presented for the ∑N − ∑N (the isospin I = 3/2) diagonal, the ∧N − ∧N diagonal, the ∧N → ∑N transition, and the ∑N − ∑N (I = 1/2) diagonal interactions. Scattering phase shifts for ∑N (I = 3/2) system are presented.

On the Effect of Mate Selection Scheme of Multi-objective Evolutionary Algorithms with Large Population on Constrained Multi-objective Optimization Problems

Multi-objective evolutionary algorithms (MOEAs) are active research topics for multi-objective design problems. MOEAs are population-based global optimization algorithms and it is said that the performance of the MOEAs depends on the population size. Meanwhile, one of the recent trends of computer development is in the large-scale many-core architectures and massive parallel computation is getting feasible in more companies and laboratories. One of the simple yet effective ways of MOEAs for utilizing the large-scale computational resources would be to increase the population size. This study examines the effect of the population size on MOEAs' performance on a real-world-derived benchmarking optimization problem, with large population size. In this paper, three mate selection schemes with different degree of elitist strategy are adapted to M2M-decomposed NSGA-II. The experimental results show that the elitist strategy can efficiently make use of the effect of the large population size, therefore can reduce the turn-around time of an MOEA.

DIRECT CORONARY COUPLING APPROACH FOR COMPUTING FFRCT

With the advances in computational fluid dynamics (CFD) and image-based modeling techniques, fractional flow reserve (FFR) can be computed from coronary computed tomography angiography (CTA) scans (FFRCT). However, this non-invasive approach requires large-scale computational resources, which limits its application in routine clinical setting. A 3D–0D coupling approach is proposed to improve the coupling efficiency of FFRCT. Aortic–root is modeled by a lumped parameter model and connected with the models of left ventricle and systemic circulation. With this approach, the interested coronary regions can be directly coupled to the lumped parameter model, resulting in a significant reduction (up to 20 times reduction) in the volume of the CFD computing domain. The proposed approach is applied to a patient-specific model and compared with previous non-reduced approach. Results show that the computed coronary flow rates, pressure waveforms and FFRCT contours by the proposed approach are consistent well with that of the non-reduced approach. These results demonstrate that the proposed approach can reduce the CFD computing domain of FFRCT significantly while maintaining the similar accuracy as compared with the non-reduced approach, and it can be further employed to promote FFRCT in routine clinical setting.

Using Wavelet Analysis To Assist in Identification of Significant Events in Molecular Dynamics Simulations.

Long time scale molecular dynamics (MD) simulations of biological systems are becoming increasingly commonplace due to the availability of both large-scale computational resources and significant advances in the underlying simulation methodologies. Therefore, it is useful to investigate and develop data mining and analysis techniques to quickly and efficiently extract the biologically relevant information from the incredible amount of generated data. Wavelet analysis (WA) is a technique that can quickly reveal significant motions during an MD simulation. Here, the application of WA on well-converged long time scale (tens of μs) simulations of a DNA helix is described. We show how WA combined with a simple clustering method can be used to identify both the physical and temporal locations of events with significant motion in MD trajectories. We also show that WA can not only distinguish and quantify the locations and time scales of significant motions, but by changing the maximum time scale of WA a more complete characterization of these motions can be obtained. This allows motions of different time scales to be identified or ignored as desired.

Distributed multi-scale muscle simulation in a hybrid MPI–CUDA computational environment

We present Mexie, an extensible and scalable software solution for distributed multi-scale muscle simulations in a hybrid MPI–CUDA environment. Since muscle contraction relies on the integration of physical and biochemical properties across multiple length and time scales, these models are highly processor and memory intensive. Existing parallelization efforts for accelerating multi-scale muscle simulations imply the usage of expensive large-scale computational resources, which produces overwhelming costs for the everyday practical application of such models. In order to improve the computational speed within a reasonable budget, we introduce the concept of distributed calculations of multi-scale muscle models in a mixed CPU–GPU environment. The concept is applied to a two-scale muscle model, in which a finite element macro model is coupled with the microscopic Huxley kinetics model. Finite element calculations of a continuum macroscopic model take place strictly on the CPU, while numerical solutions of the partial differential equations of Huxley’s cross-bridge kinetics are calculated on both CPUs and GPUs. We present a modular architecture of the solution, along with an internal organization and a specific load balancer that is aware of memory boundaries in such a heterogeneous environment. Solution was verified on both benchmark and real-world examples, showing high utilization of involved processing units, ensuring high scalability. Speed-up results show a boost of two orders of magnitude over any previously reported distributed multi-scale muscle models. This major improvement in computational feasibility of multi-scale muscle models paves the way for new discoveries in the field of muscle modeling and future clinical applications.

Future opportunities and trends for e-infrastructures and life sciences: going beyond the grid to enable life science data analysis

With the increasingly rapid growth of data in life sciences we are witnessing a major transition in the way research is conducted, from hypothesis-driven studies to data-driven simulations of whole systems. Such approaches necessitate the use of large-scale computational resources and e-infrastructures, such as the European Grid Infrastructure (EGI). EGI, one of key the enablers of the digital European Research Area, is a federation of resource providers set up to deliver sustainable, integrated and secure computing services to European researchers and their international partners. Here we aim to provide the state of the art of Grid/Cloud computing in EU research as viewed from within the field of life sciences, focusing on key infrastructures and projects within the life sciences community. Rather than focusing purely on the technical aspects underlying the currently provided solutions, we outline the design aspects and key characteristics that can be identified across major research approaches. Overall, we aim to provide significant insights into the road ahead by establishing ever-strengthening connections between EGI as a whole and the life sciences community.

41 A full exploration of the conformational ensembles of nucleic acid structural motifs

Ensembles of optimized molecular dynamics simulations engines on CPUs and GPUs, coupled with access to large-scale computational resources available through Blue Waters and XSEDE, have enabled the ...