Distributed merge forest

Scientific workflows are one of the well-established pillars of modern large-scale computational science. More recently, scientists have started to leverage machine learning (ML) capabilities in their workflows, leading to a new category of scientific workflows, denoted as scientific ML workflows. ML is not only about training and inference, modern ML workflows also involve complex data processing steps before the training can start, which are not often accounted for in most performance studies. In this work, we consider scientific ML workflows, from data pre-processing to training, inference, and model evaluation. We aim to explore (i) how scientific ML workflows differ from more traditional scientific workflows and; (ii) how we can characterize ML workflows both in terms of execution time and data movements when executing on a target cloud platform. We select three representative workflows, ranging from image classification to natural language processing and image segmentation, which have been executed using the academic cloud platform, Chameleon. We build four realistic deployment scenarios for each workflow, which stress data movements during workflow executions. Then, we compare the performance observed when utilizing these different configurations and study how different settings impact overall workflows performance and efficiency when running on cloud infrastructures. Finally, we summarize our findings and discuss performance impacts when augmenting scientific workflows with ML techniques and how traditional workflow management systems can improve their support for such workflows.

In Deep sub-micron design level, interconnect plays an important role in VLSI circuit designs. Traditional interconnects are facing several challenges such as delay and power dissipation. Graphene nanoribbons (GNRs) have been found to be potential alternative for both transistors and interconnects due to its outstanding electrical and thermal properties. The construction of minimum cost global routing trees is a key challenge in deep sub-micron design. The construction of such trees for traditional interconnects based on Required Arrival Time (RAT) has been reported in literature. The construction of global routing trees for GNR routing are different than that of traditional interconnects. In this paper, we propose an algorithm for the construction of cost effective global routing trees for GNR routing based on required arrival time. We formulate the GNR routing problem considering the minimization of the area of triangular grid such that delay from source to each of the sink terminals is within the given time budget. The algorithm is tested on a random set of data and the results obtained are quite encouraging.

Distributed data structures are key to implementing scalable applications for scientific simulations and data analysis. In this paper we look at two implementation styles for distributed data structures: remote direct memory access (RDMA) and remote procedure call (RPC). We focus on operations that require individual accesses to remote portions of a distributed data structure, e.g., accessing a hash table bucket or distributed queue, rather than global operations in which all processors collectively exchange information. We look at the trade-offs between the two styles through microbenchmarks and a performance model that approximates the cost of each. The RDMA operations have direct hardware support in the network and therefore lower latency and overhead, while the RPC operations are more expressive but higher cost and can suffer from lack of attentiveness from the remote side. We also run experiments to compare the real-world performance of RDMA- and RPC-based data structure operations with the predicted performance to evaluate the accuracy of our model, and show that while the model does not always precisely predict running time, it allows us to choose the best implementation in the examples shown. We believe this analysis will assist developers in designing data structures that will perform well on current network architectures, as well as network architects in providing better support for this class of distributed data structures.

In this report, we describe the research accomplished by the OSU team under the Pmodels2 project. The team has worked on various angles: designing high performance MPI implementations on modern networking technologies (Mellanox InfiniBand (including the new ConnectX2 architecture and Quad Data Rate), QLogic InfiniPath, the emerging 10GigE/iWARP and RDMA over Converged Enhanced Ethernet (RoCE) and Obsidian IB-WAN), studying MPI scalability issues for multi-thousand node clusters using XRC transport, scalable job start-up, dynamic process management support, efficient one-sided communication, protocol offloading and designing scalable collective communication libraries for emerging multi-core architectures. New designs conforming to the Argonne’s Nemesis interface have also been carried out. All of these above solutions have been integrated into the open-source MVAPICH/MVAPICH2 software. This software is currently being used by more than 2,100 organizations worldwide (in 71 countries). As of January ’14, more than 200,000 downloads have taken place from the OSU Web site. In addition, many InfiniBand vendors, server vendors, system integrators and Linux distributors have been incorporating MVAPICH/MVAPICH2 into their software stacks and distributing it. Several InfiniBand systems using MVAPICH/MVAPICH2 have obtained positions in the TOP500 ranking of supercomputers in the world. The latest November ’13 ranking include the following systems: 7th ranked Stampede system at TACC with 462,462 cores; 11th ranked Tsubame 2.5 system at Tokyo Institute of Technology with 74,358 cores; 16th ranked Pleiades system at NASA with 81,920 cores; Work on PGAS models has proceeded on multiple directions. The Scioto framework, which supports task-parallelism in one-sided and global-view parallel programming, has been extended to allow multi-processor tasks that are executed by processor groups. A quantum Monte Carlo application is being ported onto the extended Scioto framework. A public release of Global Trees (GT) has been made, along with the Global Chunks (GC) framework on which GT is built. The Global Chunks (GC) layer is also being used as the basis for the development of a higher level Global Graphs (GG) layer. The Global Graphs (GG) system will provide a global address space view of distributed graph data structures on distributed memory systems.

This lightning talk presents early work surrounding undergraduate students' experiences in data science and computing classrooms that shape their identities. Identity in learning spaces has been studied extensively in STEM disciplines, but little research has focused on undergraduate students majoring in data science. Since data science sits at the intersection of several disciplines, such as mathematics, statistics, and computer science, it is unclear how data science students navigate those individual discipline identities and/or construct a unique identity that is greater than the sum of those parts. As Michigan State University's Computational Mathematics, Science, and Engineering department offers data science degrees, it is of particular interest to understand the classroom messaging and experiences that shape our students' sense of self in relation to data science and computing. The project will build on various identity frameworks understand how students' data science identities are developed or stifled through classroom experiences. From previous research, we know that prototypical identities of STEM center cisgender, heterosexual, white male students and can exclude students from different demographics and lived experiences. To address this, I will utilize a critical lens to interrogate possible inequity, bias, and inaccessibility that may emerge from students' identity shaping experiences in these classrooms. At the time of this lightning talk, a thematic analysis of the first seven interviews I have conducted will be presented. All are welcome to attend the presentation and provide feedback and insight on initial observations, synthesis, and plans for the future.

Towards a model-driven transformation framework for scientific workflows

The 2nd International Conference on Computing and Data Science (CONF-CDS 2021) was held in Stanford, CA, USA, on January 21-23, 2021. CONF-CDS 2021 is a leading conference on computing technology, machine learning, computer science and data science hosted by Eliwise Academy. CONF-CDS 2021 is open to international participants. It provides an excellent opportunity for scholars interested in computing and data science to share their findings and achievements, thus promoting communication and cooperation.The goal of this conference is to bring together researchers and practitioners from academia and industry to highlight the importance of computing technologies and data science as well as establish new collaborations in these areas. The conference looks for significant contributions to computing, data mining and data science in theoretical and practical aspects.The submission will first be reviewed for its topic and length, then go through an originality check. The peer-review process will begin soon after the paper is found to be qualified. The paper will be sent to have a double-blind peer review by 3 reviewers. They will judge the paper based on the theme, coverage, innovation, integrity, depth, and language. If the paper is accepted, notification of acceptance and review comments will be sent to the author via email. The actual review time may be different under certain situations.Due to the prevention and control of COVID-19 and travel restrictions worldwide, CONF-CDS 2021 adopted the format of virtual conference through network technology to provide a forum for experts and scholars to share real-time information and discuss. The conference included keynote speeches and online discussion, and each presenter has 15-25 minutes (including Q&A).With warmest regards,CONF-CDS 2021 Organizing CommitteeStanford, CA, USA

Scientific research processes are accompanied by burgeoning amounts of research data that serves as findings of the research and provides a platform for the results to be verifiable. Scientific researchers are under pressure to make their data openly available to other researchers for use. Access to scientific data and its curation has become a dominant theme in scientific policy discussions. Besides the funder requirements, researchers are making their data openly available for a number of reasons that include, raising the profiles of their research outputs, to increase their citations, to validate their research outputs and for their authenticity and integrity as researchers. However, not all researchers possess the skills to curate their own data and make it openly available for use in an open environment that is FAIR (Findable, Accessible, Interoperable and Reusable). The primary objective of this paper is to raise awareness of the strategic role of data scientists in the scientific research process. The study seeks to contribute towards a growing and emergent discourse of data science in HEIs. It also seeks to contribute towards an educational model for data scientists in HEIs in South Africa. In-order to realise these objectives the paper asks the following research questions:How are the emergent roles of data scientists conceptualised in HEIs in South Africa?What is the strategic role of a data scientist in the scientific research process?.This study analysed data science programs offered by universities in South Africa to understand how the role of data scientists is conceptualised and to delineate the scope of this emergent field. A qualitative content analysis survey method is used to analyse courses offered by HEIs in South Africa. Preliminary findings from the research revealed that data scientists are drawn from fields like mathematics, statistics, information science, and computer science. The practical jobs of data scientists as conceptualised by the majority of universities employs theories drawn from the information science, computer science and statistics. The data scientists are expected to work closely with the researchers where the research is done. Most universities are offering data science programs at post-graduate level as professional courses for learners coming in from diverse backgrounds. The study recommends that the next generation of data scientists in a South African context should possess a cocktail of skills that enable them to conduct data analysis and enquiry to enable others to work with digital data. Presentation at the UCT E-Research Conference 2019: DataScience4D: Explore the practice, principles and opportunities of data science for development15 – 18 April 2019 held at the University of Cape Town

Advanced Data Science and Analytics with Python enables data scientists to continue developing their skills and apply them in business as well as academic settings. The subjects discussed in this book are complementary and a follow-up to the topics discussed in Data Science and Analytics with Python. The aim is to cover important advanced areas in data science using tools developed in Python such as SciKit-learn, Pandas, Numpy, Beautiful Soup, NLTK, NetworkX and others. The model development is supported by the use of frameworks such as Keras, TensorFlow and Core ML, as well as Swift for the development of iOS and MacOS applications. Features: Targets readers with a background in programming, who are interested in the tools used in data analytics and data science Uses Python throughout Presents tools, alongside solved examples, with steps that the reader can easily reproduce and adapt to their needs Focuses on the practical use of the tools rather than on lengthy explanations Provides the reader with the opportunity to use the book whenever needed rather than following a sequential path The book can be read independently from the previous volume and each of the chapters in this volume is sufficiently independent from the others, providing flexibility for the reader. Each of the topics addressed in the book tackles the data science workflow from a practical perspective, concentrating on the process and results obtained. The implementation and deployment of trained models are central to the book. Time series analysis, natural language processing, topic modelling, social network analysis, neural networks and deep learning are comprehensively covered. The book discusses the need to develop data products and addresses the subject of bringing models to their intended audiences – in this case, literally to the users’ fingertips in the form of an iPhone app. About the Author Dr. Jesus Rogel-Salazar is a lead data scientist in the field, working for companies such as Tympa Health Technologies, Barclays, AKQA, IBM Data Science Studio and Dow Jones. He is a visiting researcher at the Department of Physics at Imperial College London, UK and a member of the School of Physics, Astronomy and Mathematics at the University of Hertfordshire, UK.

We are excited to announce the twelve recipients of the ACM SIGHPC Computational and Data Science Fellowships for 2020. The program is intended to increase the diversity of students pursuing graduate degrees in data science and computational science, including women as well as students from racial and ethnic backgrounds that have been historically underrepresented in the computing field. The fellowship provides $15,000 annually for study anywhere in the world.

Scientific workflow technology has become a de facto standard in e-Science. It provides a powerful unifying platform that allows scientists to build arbitrarily complicated applications by combining predefined components. With a wide range of components, with the functions ranging from acquiring data inputs from sensors, querying databases, performing data-mining, performing execution, through to visualising the results, this paradigm allows scientists to automate, manage and execute various steps in scientific research. Although powerful, scientific workflow systems still require some programming knowledge, which might not be optimal for many scientists. Recently, scientific workflows have been integrated into the Web, creating workflow-based science gateways. These gateways allow scientists to launch large-scale computational experiments without concern for the underlying complexity of the infrastructure. This approach delivers scientific workflows to more scientists and is becoming increasingly popular. To date, much workflow-based science gateway development effort has been invested in creating user-friendly and secure interfaces to setup parameters, execute workflows and collect results. However, interaction between users and executing workflows has not received a lot of attention. Scientific workflows have typically been executed in batch mode. This model does not work with human-in-the-loop workflows, which require continuous interaction between users and executing workflows. A fundamental contribution of this thesis is to show that gateways can support complex human- workflow interactions, facilitating monitoring and steering of the workflow execution as it progresses. This is important because scientific workflows increasingly require user decisions and interactions at various steps; human decision-making is an important component of real-world computational science. In order to achieve this goal, the thesis presents the design, implementation and demonstration of WorkWays, an interactive workflow-based science gateway for supporting development and execution of human-in-the-loop scientific workflows. WorkWays consists of two distinct frameworks with well defined APIs and bindings: a framework that enables human- workflow interaction and a framework for workflow development and execution.

Data science is a discipline that provides principles, methodology, and guidelines for the analysis of data for tools, values, or insights. Driven by a huge workforce demand, many academic institutions have started to offer degrees in data science, with many at the graduate, and a few at the undergraduate level. Curricula may differ at different institutions, because of varying levels of faculty expertise, and different disciplines (such as mathematics, computer science, and business) in developing the curriculum. The University of Massachusetts Dartmouth started offering degree programs in data science from Fall 2015, at both the undergraduate and the graduate level. Quite a few articles have been published that deal with graduate data science courses, much less so dealing with undergraduate ones. Our discussion will focus on undergraduate course structure and function, and specifically, a first course in data science. Our design of this course centers around a concept called the data science life cycle. That is, we view tasks or steps in the practice of data science as forming a process, consisting of states that indicate how it comes into life, how different tasks in data science depend on or interact with others until the birth of a data product or a conclusion. Naturally, different pieces of the data science life cycle then form individual parts of the course. Details of each piece are filled up by concepts, techniques, or skills that are popular in industry. Consequently, the design of our course is both “principled” and practical. A significant feature of our course philosophy is that, in line with activity theory, the course is based on the use of tools to transform real data to answer strongly motivated questions related to the data.

Scientists are confronted with significant datamanagement problems due to the large volume and high complexity of scientific data. In particular, the latter makes data integration a difficult technical challenge. In this paper, we describe our work on semantic mediation and scientific workflows, and discuss how these technologies address integration challenges in scientific data management. We first give an overview of the main data-integration problems that arise from heterogeneity in the syntax, structure, and semantics of data. Starting from a traditional mediator approach, we show how semantic extensions can facilitate data integration in complex, multipleworlds scenarios, where data sources cover different but related scientific domains. Such scenarios are not amenable to conventional schema-integration approaches. The core idea of semantic mediation is to augment database mediators and query evaluation algorithms with appropriate knowledge-representation techniques to exploit information from shared ontologies. Semantic mediation relies on semantic data registration, which associates existing data with semantic information from an ontology. The Kepler scientific workflow system addresses the problem of synthesizing, from existing tools and applications, reusable workflow components and analytical pipelines to automate scientific analyses. After presenting core features and example workflows in Kepler, we present a framework for adding semantic information to scientific workflows. The resulting system is aware of semantically plausible connections between workflow components as well as between data sources and workflow components. This information can be used by the scientist during workflow design, and by the workflow engineer for creating data transformation steps between semantically compatible but structurally incompatible analytical steps. ∗Work supported by NSF/ITR 0225673 (GEON), NSF/ITR 0225676 (SEEK), NIH/NCRR 1R24 RR019701-01 Biomedical Informatics Research Network (BIRN-CC), and DOE SciDAC DE-FC02-01ER25486 (SDM) †San Diego Supercomputer Center, University of California, San Diego, {ludaesch,lin,bowers,efrat,baru}@sdsc.edu ‡Natural Resources of Canada, brodaric@nrcan.gc.ca

Purpose of research. The purpose of the paper is to describe the possibilities of including tasks for the design, development, use and experimental evaluation of “inverted” educational resources in the professional training of future teachers of mathematics and computer science for the development of their subject-methodical readiness. The formation and development of subject-methodical readiness is the main goal of professional training of future lecturers at a pedagogical university. The process of creating “inverted” educational resources requires from their author quite serious subject and methodological training. Potentially, the inclusion of tasks for the design, development, expertise, practical use and evaluation of “inverted” educational resources in the professional training of future teachers of mathematics and computer science can create conditions conducive to the formation of competencies of future teachers to create a comprehensive educational process in a digital environment. Since, on the one hand, in such didactic means, the educational content should be presented systematized and structured, have logic and hierarchical connections. On the other hand, the problematic issues included in the resources should reflect specific subject educational results in accordance with a given taxonomic model in the curriculum.Materials and methods. Based on the analysis of theoretical and methodological sources, the article substantiates the relevance of the problem of finding new ways and means of implementing additional subject training in mathematics and computer science for both schoolchildren and future teachers in the context of digitalization of education. The analysis of examples of “inverted” educational resources, which are the results of the implementation of project tasks by undergraduate students of the Krasnoyarsk State Pedagogical University named after V.P. Astafiev, is presented. Through theoretical and methodological substantiation, the main possibilities of including tasks for working with such resources at different stages of professional training of future teachers of mathematics and computer science at a pedagogical university are synthesized.Results. In this study, the problems related to the possibility of including “inverted” educational resources as digital didactic tools in the training programs of future teachers of mathematics and computer science were updated. The authors’ position on the issue of the purpose of such teaching tools in the context of the digitalization of education is presented and substantiated. Examples of the results of the implementation of project tasks by senior students during the 2021-2022 academic year were demonstrated. The key didactic and information technology features, possible tools for creating “inverted” educational resources are given. The definition of the subject-methodical readiness of the future teacher of mathematics and computer science as an essential component of professional readiness is concretized. The possibilities of strengthening subject-methodical training by including tasks for work with “inverted” educational resources are described: analytical – in junior courses when mastering disciplines related to the use of digital technologies in education; project – in the middle courses when mastering methodological disciplines; experimental and expert - in the senior courses when passing industrial (pedagogical) practice and mastering the final disciplines of the professional educational program of higher pedagogical education.Conclusion. On a final note, the results and conclusions obtained in the paper are summarized; the key positions of the authors on the inclusion of “inverted” educational resources in the professional training of future teachers of mathematics and computer science are given. The materials of the article may be of interest to lecturers of pedagogical universities who train future teachers, representatives of schools cooperating with organizations of higher pedagogical education, researchers in the field of digitalization and digital transformation of education.

The advancement of science is increasingly intertwined with complex computational processes [1]. Scientific workflows are at the heart of this evolution, acting as essential orchestrators for a vast range of experiments. Specifically, these workflows are central to the field of computational Earth Sciences, where they orchestrate a diverse range of activities, from cloud-based data preprocessing pipelines in environmental modeling to intricate multi-facility instrument-to-edge-to-HPC computational frameworks for seismic data analysis and geophysical simulations [2]. The emergence of continuum and cross-facility workflows marks a significant evolution in computational sciences [3]. Continuum workflows represent continuous computing access required for analysis pipelines, while cross-facility workflows extend across multiple sites, integrating experiments andcomputing facilities. These cross-facility workflows, crucial for real-time applications, offer resiliency and stand as solutions for the demands of continuum workflows. Addressing continuum and cross-facility computing requires a focus on data, ensuring workflow systems are equipped to handle diverse datarepresentations and storage systems. As we navigate the computing continuum, the pressing needs of contemporary scientific applications in Earth Sciences call for a dual approach: the recalibration of existing systems and the innovation of new workflow functionalities. This recalibration involves optimizing data-intensive operations andincorporating advanced algorithms for spatial data analysis, while innovation may entail the integration of machine learning techniques for predictive modeling and real-time data processing in earth sciences. We offer a comprehensive overview of cutting-edge advancements in this dynamic realm, with a focus on computational Earth Sciences, including managing the increasing volume and complexity of geospatial data, ensuring the reproducibility of large-scale simulations, and adapting workflows to leverage emerging computational architectures.   [1] Ferreira da Silva, R., Casanova, H., Chard, K., Altintas, I., Badia, R. M., Balis, B., Coleman, T., Coppens, F., Di Natale, F., Enders, B., Fahringer, T., Filgueira, R., et al. (2021). A Community Roadmap for Scientific Workflows Research and Development. 2021 IEEE Workshop on Workflows in Support of Large-Scale Science (WORKS), 81–90. DOI: 10.1109/WORKS54523.2021.00016 [2] Badia Sala, R. M., Ayguadé Parra, E., & Labarta Mancho, J. J. (2017). Workflows for science: A challenge when facing the convergence of HPC and big data. Supercomputing frontiers and innovations, 4(1), 27-47. DOI: 10.14529/jsfi170102 [3] Antypas, K. B., Bard, D. J., Blaschke, J. P., Canon, R. S., Enders, B., Shankar, M. A., ... & Wilkinson, S. R. (2021, December). Enabling discovery data science through cross-facility workflows. In 2021 IEEE International Conference on Big Data (Big Data) (pp. 3671-3680). IEEE. DOI: 10.1109/BigData52589.2021.967142