Workshop on Teaching Computational Science and Bridging the HPC Talent Gap with Computational Science Research Methods (WTCS) Session 3

Time and Date: 16:20 - 18:00 on 13th June 2017

Room: HG D 1.2

Chair: Angela B. Shiflet

260 Learning Outcomes based Evaluation of HPC Professional Training [abstract]
Abstract: Very often, when evaluating professional training courses and events the analysis is reduced to a set of statistics related to who attended, how many they were, would they come again. The more pertinent questions about the value and relevance of acquired skills and how they could be applied to future work by the attendees tend to be side-tracked. BSC is committed to provide an international high-level education and training program. As part of that commitment, we are conducting a study of the learning outcomes from our courses, and looking into the affect they have on the transfer of knowledge into improved work methodologies/ routines. Supercomputing Centres worldwide, such as BSC, face the challenge of supporting a growing base of HPC users with little or no HPC experience with mainly scientific background. In addition, HPC training centres feel the need to respond to the clearly observed convergence of Data Science and Computational Science research methods into Data and Compute Intensive Science methods by enriching their programs with the necessary course for the diverse audience of domain science researchers needing HPC skills to tackle societal, economic and scientific challenges. The goals of the training evaluation are to capture the impact of the training program and allow an inside in not only the personal progress of the attendees but as well provide understanding how to support the attendees in implementing the learned methodologies/tools in their work. The trainees should be able to bridge the gap between the training classroom and the work/studies after that and to develop the ability to recognise the context for direct implementation or re-design of a methodology/ tools solution. These skills are directly linked to the impact of training and should be perceived as a learning outcome. Usually the skills are developed by using appropriate teaching methods over time, which is possible in the context of a longer-term learning environment, e.g. university degree. The Kirkpatrick model and the related practices the company (http://www.kirkpatrickpartners.com/) suggest as a way to support the on-the-job implementation of training is to build an on-line continuation of a training event which creates a context similar to that of a longer courses and thus facilitate the motivation to implement new skills by changing an established routine. That is time and resource consuming and thus the challenge is to facilitate the needed support of on-the-job implementation with minimal cost impact for the HPC centres.
Nia Alexandrov and Maria-Ribera Sancho
77 Teaching High Performance Computing at a US Regional university: curriculum, resources, student projects. [abstract]
Abstract: In this talk, we focus on the challenges and opportunities in the development of a comprehensive upper undergraduate/first year graduate course in high performance scientific computing at a regional university in the United States. The details of this presentation are based on a new course given at Idaho State University during 2016/2017 academic year. In the first part of the presentation, we focus on the curriculum, discussing the four major parts of the course: serial code optimization; OpenMP, MPI and CUDA programming. In this part of the presentation, we focus on the time requirements and accessibility of the material to the motivated math, science, and engineering students. In the next part of the talk, we focus on resources, and discuss the existing supercomputer educational projects available at the national level if a regional University does not have adequate resources to support a class on HPC. We describe our experience accessing the supercomputer facilities in the United States through the support of national laboratories and federally supported projects at top US Universities. (Mostly, we focus on our collaboration with the Idaho National Laboratory and the National Center for Supercomputing at the University of Illinois at Urbana-Champaign.) In the final part of the presentation, we review the results of the students projects related to the parallel implementations of the well known Krylov-FFT solvers for large sparse algebraic systems appeared in different relevant applications.
Yury Gryazin
583 Data processing as a basic competence of engineering education [abstract]
Abstract: The need in data science specialists is growing very fast, nevertheless the gap in knowledge is mostly bridged with special study programs. Quite often data science skills are needed additional to or embedded in the main study subject. This is especially the case in engineering study programs. In this position paper we discuss several possibilities to fill this gap.
Andreas Pester and Thomas Klinger
542 Towards Data Science Literacy [abstract]
Abstract: Promoting data science represents an increasingly important facet of general education. This paper describes the design and implementation of a course targeted at a non-technical audience and centered on data science literacy, with a focus on collecting, processing, analyzing and using data. The objective is through general education to prime students at an early stage of their college education for the changes in the data-driven society and to provide them with skills to harness the power of data. Our experience and evaluation results indicate that it is realistic for a diverse population of undergraduate students to acquired data science literacy and practical skills through a general education course.
Christo Dichev and Darina Dicheva
431 A Way How to Impart Data Science Skills to Computer Science Students Exemplified by OBDA-Systems Development [abstract]
Abstract: Nowadays to explore and examine data from a variety of angles to tackle Big Data problems, to devise data-driven solutions to the most pressing challenges, it is necessary to build multidisciplinary students’ skills set for innovative methods not only for Masters in Data Science Programs but in traditional Computer Science Programs too. In the paper, we describe how teaching methods and tools, which are used to train students to develop Ontology-Based Data Access systems with natural language interface to relational databases, help Master’s Degree students in Computer Science to collaborate with Data Scientists in real-world interdisciplinary projects and prepare them for a data science career. We use ontology engineering in a combination with Natural Language Processing methods based on lexico-syntactic patterns, in particular, to extract needful data from structured, semi-structured and unstructured datasets in a uniform way to analyze real-world Russian social networks related to new building area.
Svetlana Chuprina, Igor Postanogov and Taisiya Kostareva