scholarly journals Hybrid PDE solver for data-driven problems and modern branching

2017 ◽  
Vol 28 (6) ◽  
pp. 949-972 ◽  
Author(s):  
FRANCISCO BERNAL ◽  
GONÇALO DOS REIS ◽  
GREIG SMITH
Keyword(s):  
High Performance ◽  
Large Scale ◽  
Data Science ◽  
Data Driven ◽  
Stochastic Representation ◽  
Non Linear ◽  
Research Questions ◽  
Stochastic Representations ◽  
Branching Diffusions ◽  
Pde Solvers

The numerical solution of large-scale PDEs, such as those occurring in data-driven applications, unavoidably require powerful parallel computers and tailored parallel algorithms to make the best possible use of them. In fact, considerations about the parallelization and scalability of realistic problems are often critical enough to warrant acknowledgement in the modelling phase. The purpose of this paper is to spread awareness of the Probabilistic Domain Decomposition (PDD) method, a fresh approach to the parallelization of PDEs with excellent scalability properties. The idea exploits the stochastic representation of the PDE and its approximation via Monte Carlo in combination with deterministic high-performance PDE solvers. We describe the ingredients of PDD and its applicability in the scope of data science. In particular, we highlight recent advances in stochastic representations for non-linear PDEs using branching diffusions, which have significantly broadened the scope of PDD. We envision this work as a dictionary giving large-scale PDE practitioners references on the very latest algorithms and techniques of a non-standard, yet highly parallelizable, methodology at the interface of deterministic and probabilistic numerical methods. We close this work with an invitation to the fully non-linear case and open research questions.

Thallo – Scheduling for High-Performance Large-Scale Non-Linear Least-Squares Solvers

2021 ◽  
Vol 40 (5) ◽  
pp. 1-14
Author(s):  
Michael Mara ◽  
Felix Heide ◽  
Michael Zollhöfer ◽  
Matthias Nießner ◽  
Pat Hanrahan
Keyword(s):  
Least Squares ◽  
High Performance ◽  
Large Scale ◽  
Optimization Problems ◽  
Bundle Adjustment ◽  
Large Set ◽  
Linear Least Squares ◽  
Non Linear ◽  
Recent Developments ◽  
Scale Optimization

Large-scale optimization problems at the core of many graphics, vision, and imaging applications are often implemented by hand in tedious and error-prone processes in order to achieve high performance (in particular on GPUs), despite recent developments in libraries and DSLs. At the same time, these hand-crafted solver implementations reveal that the key for high performance is a problem-specific schedule that enables efficient usage of the underlying hardware. In this work, we incorporate this insight into Thallo, a domain-specific language for large-scale non-linear least squares optimization problems. We observe various code reorganizations performed by implementers of high-performance solvers in the literature, and then define a set of basic operations that span these scheduling choices, thereby defining a large scheduling space. Users can either specify code transformations in a scheduling language or use an autoscheduler. Thallo takes as input a compact, shader-like representation of an energy function and a (potentially auto-generated) schedule, translating the combination into high-performance GPU solvers. Since Thallo can generate solvers from a large scheduling space, it can handle a large set of large-scale non-linear and non-smooth problems with various degrees of non-locality and compute-to-memory ratios, including diverse applications such as bundle adjustment, face blendshape fitting, and spatially-varying Poisson deconvolution, as seen in Figure 1. Abstracting schedules from the optimization, we outperform state-of-the-art GPU-based optimization DSLs by an average of 16× across all applications introduced in this work, and even some published hand-written GPU solvers by 30%+.

scholarly journals High-performance computing service for bioinformatics and data science

2018 ◽  
Vol 106 (4) ◽  
Author(s):  
Jean-Paul Courneya ◽  
Alexa Mayo
Keyword(s):  
Open Source ◽  
High Throughput ◽  
High Performance ◽  
Large Scale ◽  
Data Science ◽  
Wet Work ◽  
High Throughput Data ◽  
Guided Learning ◽  
Computational Resources ◽  
Performance Computing

Despite having an ideal setup in their labs for wet work, researchers often lack the computational infrastructure to analyze the magnitude of data that result from “-omics” experiments. In this innovative project, the library supports analysis of high-throughput data from global molecular profiling experiments by offering a high-performance computer with open source software along with expert bioinformationist support. The audience for this new service is faculty, staff, and students for whom using the university’s large scale, CORE computational resources is not warranted because these resources exceed the needs of smaller projects. In the library’s approach, users are empowered to analyze high-throughput data that they otherwise would not be able to on their own computers. To develop the project, the library’s bioinformationist identified the ideal computing hardware and a group of open source bioinformatics software to provide analysis options for experimental data such as scientific images, sequence reads, and flow cytometry files. To close the loop between learning and practice, the bioinformationist developed self-guided learning materials and workshops or consultations on topics such as the National Center for Biotechnology Information’s BLAST, Bioinformatics on the Cloud, and ImageJ. Researchers apply the data analysis techniques that they learned in the classroom in an ideal computing environment.

scholarly journals Scalability of high-performance PDE solvers

2020 ◽  
Vol 34 (5) ◽  
pp. 562-586 ◽  
Author(s):  
Paul Fischer ◽  
Misun Min ◽  
Thilina Rathnayake ◽  
Som Dutta ◽  
Tzanio Kolev ◽  
...  
Keyword(s):  
Finite Elements ◽  
High Performance ◽  
Degrees Of Freedom ◽  
Large Scale ◽  
Fixed Number ◽  
Peak Performance ◽  
Trade Offs ◽  
Wide Range ◽  
P Type ◽  
Pde Solvers

Performance tests and analyses are critical to effective high-performance computing software development and are central components in the design and implementation of computational algorithms for achieving faster simulations on existing and future computing architectures for large-scale application problems. In this article, we explore performance and space-time trade-offs for important compute-intensive kernels of large-scale numerical solvers for partial differential equations (PDEs) that govern a wide range of physical applications. We consider a sequence of PDE-motivated bake-off problems designed to establish best practices for efficient high-order simulations across a variety of codes and platforms. We measure peak performance (degrees of freedom per second) on a fixed number of nodes and identify effective code optimization strategies for each architecture. In addition to peak performance, we identify the minimum time to solution at 80% parallel efficiency. The performance analysis is based on spectral and p-type finite elements but is equally applicable to a broad spectrum of numerical PDE discretizations, including finite difference, finite volume, and h-type finite elements.

scholarly journals Archival Data Thinking and Practices for Social Media Collections

2021 ◽  
Author(s):  
Lizhou Fan
Keyword(s):  
Social Media ◽  
Large Scale ◽  
Data Science ◽  
Hate Speech ◽  
Research Direction ◽  
Data Driven ◽  
Future Research ◽  
Archival Data ◽  
Data Collections ◽  
Rapid Processing

In the Web 2.0 Era, most social media archives are born digital and large-scale. With an increasing need for processing them at a fast speed, researchers and archivists have started applying data science methods in managing social media data collections. However, many of the current computational or data-driven archival processing methods are missing the critical background understandings like “why we need to use computational methods,” and “how to evaluate and improve data-driven applications.” As a result, many computational archival science (CAS) attempts, with comparatively narrow scopes and low efficiencies, are not sufficiently holistic. In this talk, we first introduce the proposed concept of “Archival Data Thinking” that highlights the desirable comprehensiveness in mapping data science mindsets to archival practices. Next, we examine several examples of implementing “Archival Data Thinking” in processing two social media collections: (i) the COVID-19 Hate Speech Twitter Archive (CHSTA) and (ii) the Counter-anti-Asian Hate Twitter Archive (CAAHTA), both of which are with millions of records and their metadata, and needs for rapid processing. Finally, as a future research direction, we briefly discuss the standards and infrastructures that can better support the implementation of “Archival Data Thinking”.

scholarly journals Manifold Learning and Clustering for Automated Phase Identification and Alignment in Data Driven Modeling of Batch Processes

2020 ◽  
Vol 2 ◽  
Author(s):  
Carlos André Muñoz López ◽  
Satyajeet Bhonsale ◽  
Kristin Peeters ◽  
Jan F. M. Van Impe
Keyword(s):  
Manifold Learning ◽  
Large Scale ◽  
Linear Method ◽  
Phase Changes ◽  
Data Driven ◽  
Support Vector ◽  
Phase Identification ◽  
Scale Production ◽  
Non Linear ◽  
Data Driven Modeling

Processing data that originates from uneven, multi-phase batches is a challenge in data-driven modeling. Training predictive and monitoring models requires the data to be in the right shape to be informative. Only then can a model learn meaningful features that describe the deterministic variability of the process. The presence of multiple phases in the data, which display different correlation patterns and have an uneven duration from batch to batch, reduces the performance of the data-driven modeling methods significantly. Therefore, phase identification and alignment is a critical step and can lead to an unsuccessful modeling exercise if not applied correctly. In this paper, a novel approach is proposed to perform unsupervised phase identification and alignment based on the correlation patterns found in the data. Phase identification is performed via manifold learning using t-Distributed Stochastic Neighbor Embedding (t-SNE), which is a state-of-the-art machine learning algorithm for non-linear dimensionality reduction. The application of t-SNE to a reduced cross-correlation matrix of every batch with respect to a reference batch results in data clustering in the embedded space. Models based on support vector machines (SVMs) are trained to, 1) reproduce the manifold learning obtained via t-SNE, and 2) determine the membership of the data points to a process phase. Compared to previously proposed clustering approaches for phase identification, this is an unsupervised, non-linear method. The perplexity parameter of the t-SNE algorithm can be interpreted as the estimated duration of the shortest phase in the process. The advantages of the proposed method are demonstrated through its application on an in-silico benchmark case study, and on real industrial data from two unit-operations in the large scale production of an active pharmaceutical ingredients (API). The efficacy and robustness of the method are evidenced in the successful phase identification and alignment obtained for these three distinct processes, displaying smooth, sudden and repetitive phase changes. Additionally, the low complexity of the method makes feasible its online implementation.

The Structure and Properties of MoSi2 Thin Film in Mos Process

1980 ◽  
Vol 38 ◽  
pp. 326-327
Author(s):  
C.K. Wu ◽  
P. Chang ◽  
N. Godinho
Keyword(s):  
Thin Film ◽  
Integrated Circuits ◽  
High Performance ◽  
Large Scale ◽  
Process Development ◽  
Structure And Properties ◽  
Metal Silicides ◽  
High Oxidation ◽  
Important Approach ◽  
High Oxidation Resistance

Recently, the use of refractory metal silicides as low resistivity, high temperature and high oxidation resistance gate materials in large scale integrated circuits (LSI) has become an important approach in advanced MOS process development (1). This research is a systematic study on the structure and properties of molybdenum silicide thin film and its applicability to high performance LSI fabrication.

About digital transformation of the transport industry in Russia

2020 ◽  
Vol 19 (3) ◽  
pp. 95-105
Author(s):  
Natalia V. Vysotskaya ◽  
T. V. Kyrbatskaya
Keyword(s):  
Artificial Intelligence ◽  
Business Model ◽  
Data Science ◽  
Digital Transformation ◽  
Digital Culture ◽  
Blockchain Technology ◽  
Community Approach ◽  
Non Linear ◽  
Transport Industry ◽  
Artificial Intelligence Systems

The article is devoted to the consideration of the main directions of digital transformation of the transport industry in Russia. It is proposed in the process of digital transformation to integrate the community approach into the company's business model using blockchain technology and methods and results of data science; complement the new digital culture with a digital team and new communities that help management solve business problems; focus the attention of the company's management on its employees and develop those competencies in them that robots and artificial intelligence systems cannot implement: develop algorithmic, computable and non-linear thinking in all employees of the company.

RECOMMENDATIONS FOR THE CHOICE OF MILKING INSTALLATIONS IN LOOSE HOUSING SYSTEMS OF COWS

2020 ◽  
Author(s):  
В.В. ГОРДЕЕВ ◽  
В.Е. ХАЗАНОВ
Keyword(s):  
Dairy Cows ◽  
High Performance ◽  
Large Scale ◽  
Dairy Farms ◽  
Economic Indicators ◽  
Technical Level ◽  
Housing Systems ◽  
Working Shift ◽  
Technical And Economic Indicators

При выборе типа доильной установки и ее размера необходимо учитывать максимальное планируемое поголовье дойных коров и размер технологической группы, кратность и время одного доения, продолжительность рабочей смены дояров. Анализ технико-экономических показателей наиболее распространенных на сегодняшний день типов доильных установок одинакового технического уровня свидетельствует, что наилучшие удельные показатели имеет установка типа «Карусель» (1), а установка типа «Елочка» (2) требует более высоких затрат труда и средств. Установка «Параллель» (3) занимает промежуточное положение. Из анализа пропускной способности и количества необходимых операторов: установка 2 рекомендована для ферм с поголовьем дойного стада до 600 голов, 3 — не более 1200 дойных коров, 1 — более 1200 дойных коров. «Карусель» — наиболее рациональный, высокопроизводительный, легко автоматизируемый и, следовательно, перспективный способ доения в залах, особенно для крупных молочных ферм. The choice of the proper type and size of milking installations needs to take into account the maximum planned number of dairy cows, the size of a technological group, the number of milkings per day, and the duration of one milking and the operator's working shift. The analysis of technical and economic indicators of currently most common types of milking machines of the same technical level revealed that the Carousel installation had the best specific indicators while the Herringbone installation featured higher labour inputs and cash costs. The Parallel installation was found somewhere in between. In terms of the throughput and the required number of operators Herringbone is recommended for farms with up to 600 dairy cows, Parallel — below 1200 dairy cows, Carousel — above 1200 dairy cows. Carousel was found the most practical, high-performance, easily automated and, therefore, promising milking system for milking parlours, especially on the large-scale dairy farms.

scholarly journals Statistical and machine learning models for optimizing energy in parallel applications

2019 ◽  
Vol 33 (6) ◽  
pp. 1079-1097 ◽  
Author(s):  
Mark Endrei ◽  
Chao Jin ◽  
Minh Ngoc Dinh ◽  
David Abramson ◽  
Heidi Poxon ◽  
...  
Keyword(s):  
Machine Learning ◽  
Energy Efficiency ◽  
High Performance ◽  
Large Scale ◽  
Energy Use ◽  
Parallel Applications ◽  
Learning Models ◽  
Trade Off ◽  
Time Required ◽  
Machine Learning Models

Rising power costs and constraints are driving a growing focus on the energy efficiency of high performance computing systems. The unique characteristics of a particular system and workload and their effect on performance and energy efficiency are typically difficult for application users to assess and to control. Settings for optimum performance and energy efficiency can also diverge, so we need to identify trade-off options that guide a suitable balance between energy use and performance. We present statistical and machine learning models that only require a small number of runs to make accurate Pareto-optimal trade-off predictions using parameters that users can control. We study model training and validation using several parallel kernels and more complex workloads, including Algebraic Multigrid (AMG), Large-scale Atomic Molecular Massively Parallel Simulator, and Livermore Unstructured Lagrangian Explicit Shock Hydrodynamics. We demonstrate that we can train the models using as few as 12 runs, with prediction error of less than 10%. Our AMG results identify trade-off options that provide up to 45% improvement in energy efficiency for around 10% performance loss. We reduce the sample measurement time required for AMG by 90%, from 13 h to 74 min.

Sign in / Sign up

Export Citation Format

Share Document