Search

This digital garden contains essays, thoughts, random ideas, and references that relate to the practice of scientific research in the digital era, characterized by computers (personal, high-performance, cloud, . . . ), software, the Internet, global collaborations, social networks, and more. They represent exclusively my personal views and in particular not those of my employer. There are many empty pages in this collection, and you may wonder why.

Guix-HPC is a collaborative effort to bring reproducible software deployment to scientific workflows and high-performance computing (HPC). Guix-HPC builds upon the GNU Guix software deployment tool and aims to make it a better tool for HPC practitioners and scientists concerned with reproducible research.This report highlights key achievements of Guix-HPC between our previous report a year ago and today, February 2023, including: developments in Guix and related tools such as the Guix Workflow Language, use cases of Guix in support of reproducible research and open science, and the celebration of ten years of Guix with a three-day conference.; Guix-HPC est un travail collaboratif pour donner les moyens du déploiement logiciel reproductible aux travaux scientifiques et au calcul intensif (HPC). Guix-HPC repose sur l'outil de déploiement logiciel GNU Guix et a pour ambition d’en faire un meilleur outil pour le calcul intensif et pour les scientifiques souhaitant faire des travaux de recherche reproductibles.Ce rapport souligne les réalisations clef de Guix-HPC depuis le précédent rapport, un an auparavant, et aujourd’hui, février 2023, avec notamment : les développements dans Guix et les outils qui lui sont liés, comme le Guix Workflow Language (GWL), les cas d’usage de Guix comme support à la recherche reproductible et à la science ouverte, et enfin la célébration de dix ans de Guix avec un colloque de trois jours.

Talk presented at the SIAM CSE23 conference MS321 An Update on Software Engineering for Science Abstract: Two important tendencies in scientific research are the increasing importance of computational modelling and the nascent Open Science movement. Together, they force us to revise the way scientists and non-scientists use computational tools and models, as all stakeholders must be able to develop trust in these tools and models at their respective levels of domain knowledge and engagement. Today's software engineering techniques, inherited from an industrial context, focus on small development teams creating computational tools for a large number of users that interact with the tools at a much more abstract level. Software developers thus become gatekeepers of the computational models that are embedded in these tools, which mere users can neither fully understand nor safely modify. Open Science requires reducing this epistemic opacity, allowing all stakeholders to interact with computational models at their level of scientific rather than technical expertise. A strategy that I have been exploring is moving the models out of the tools and into the human-computer interface in the form of specifications. The central user interface of such a future scientific computing environment is an authoring tool for a computational medium that can represent models. Computational tools operate on this medium, reading models and parameters from it and adding back their results. I will conclude my presentation with a demonstration of a prototype implementation.

Guix is described as a package manager and a GNU/Linux distribution. While technically correct, this summary hides the fact that Guix is a valuable support tool for reproducible computational research, as an add-on to any GNU/Linux distribution, much like many of us currently use Docker containers. I will attempt to convince you that Guix is a better choice than Docker, and show you how to get started with integrating Guix into your computational work environment.

Guix-HPC is a collaborative effort to bring reproducible software deployment to scientific workflows and high-performance computing (HPC). Guix-HPC builds upon the GNU Guix software deployment tool and aims to make it a better tool for HPC practitioners and scientists concerned with reproducible research. This report highlights key achievements of Guix-HPC between our previous report a year ago and today, February 2022. This report highlights developments on GNU Guix proper, but also downstream on Guix-Jupyter, the Guix Workflow Language, upstream with Software Heritage integration, as well as experience reports on end-to-end reproducible research article authoring pipelines.

Guix-HPC is a collaborative effort to bring reproducible software deployment to scientific workflows and high-performance computing (HPC). Guix-HPC builds upon the GNU Guix software deployment tool and aims to make it a better tool for HPC practitioners and scientists concerned with reproducible research. This report highlights key achievements of Guix-HPC between our previous report a year ago and today, February 2022. This report highlights developments on GNU Guix proper, but also downstream on Guix-Jupyter, the Guix Workflow Language, upstream with Software Heritage integration, as well as experience reports on end-to-end reproducible research article authoring pipelines.

Die Mitglieder der AG Notebook interessieren sich für die Art und Weise von FAIR(e), insbesondere für die Möglichkeiten der Wiederverwendung oder Reproduzierbarkeit eines digitalen Prozesses, der mit der Produktion von Wissen verbunden oder seines Produkts ist, aber auch für die Interoperabilität von digitalen Systemen. Diese Prozesse des Zugangs zu Daten oder Wissen wären insofern inkorporiert, als sie im Laufe des Prozesses ihrer Entstehung (verstanden) werden oder sich zu erkennen geben, auch wenn sie Computercode mobilisieren, der mit den Fingerspitzen gebildet wird. Der hier untersuchte Fall ist der Prozess der Erstellung eines R Markdown-Notebook., The members of the Notebook WG are interested in the FAIR(e) way, in particular re-use or reproducibility of digital processes associated with the production of knowledge, or of its product, but also the interoperability of digital systems. The data or knowledge accessing process would be embodied in that they would be under(stood)taken or say, would reveal themselves as knowledge during the very process of their genesis, even though when computer code is used, formed with the fingertips. The case studied here is the production process of a R markdown notebook., Les membres du GT Notebook s’intéressent à la manière de FAIR(e), en particulier les possibilités de ré-usage ou de reproductibilité d’un processus digital associé à la production de connaissances, ou de son produit, mais aussi l’interopérabilité de systèmes digitaux. Ces processus d’accès aux données ou connaissances seraient incorporées en ce qu’elles seraient (com)prises ou se donneraient à connaître au cours du processus même de leur genèse, y compris lorsque celles-ci mobilisent du code informatique, formé du bout des doigts. Le cas étudié ici est le processus de production d’un notebook R markdown .

International audience

International audience; Over the last years, the need for better user interfaces in scientific computing has become apparent by the enormous growth in popularity of computational notebooks. However, the user experience provided by these notebooks is very limited compared to the live programming environments of Smalltalk, and first studies on how they are actually used in research studies point out various weaknesses. We present the first steps we have taken towards building a scientific workbench in Pharo, based on the Glamorous Toolkit, and outline future developments towards that goal.

International audience; In the course of only a few decades computers have revolutionized scientific research and more and more scientists are writing computer programs for doing their work. In spite of the ubiquitous use of computers in science, few researchers in the natural sciences have any schooling in computer science, software engineering, or numerical analysis. They usually acquire their computing knowledge 'on the job' and often feel overwhelmed by the amount of computing knowledge they must absorb.Computation in Science provides a background in computation for scientists who use computational methods. The book explains how computing is used in the natural sciences and provides a high-level overview of relevant aspects of computer science and software engineering with a focus on concepts, results, and applications. The goal of this book is to explain these basic principles, and to show how they relate to the tasks of a scientist's daily work in a language familiar to them. Its unique feature is in connecting the dots between computational science, the theory of computation and information, and software engineering. It will compensate for the general lack of any formal training in computer science and information theory allowing readers to better understand how they use computers in their work, and how computers work. Readers will learn to use computers with more confidence, and to see computing technologies in a different light, evaluating them based on how they contribute to doing science.This new edition has been significantly updated and extended to reflect developments in scientific computing, including new examples and references. It also includes a new chapter on reproducibility which reflects the importance that computational reproducibility plays. Accompanied by a website maintained by the author which hosts companion code and supplementary material, it is intended for both graduate students and experienced scientists. Some hands-on experience with computing is highly desirable, but no competence in any specific computing technology is expected.

International audience

Guix-HPC is a collaborative effort to bring reproducible software deployment to scientific workflows and high-performance computing (HPC). Guix-HPC builds upon the GNU Guix software deployment tool and aims to make it a better tool for HPC practitioners and scientists concerned with reproducible research.This report highlights key achievements of Guix-HPC between our previous report a year ago and today, February 2020. This year was marked by a major milestone: the release in May 2019 of GNU Guix 1.0, seven years and more than 40,000 commits after its inception.

Guix-HPC is a collaborative effort to bring reproducible software deployment to scientific workflows and high-performance computing (HPC). Guix-HPC builds upon the GNU Guix software deployment tool and aims to make it a better tool for HPC practitioners and scientists concerned with reproducible research.This report highlights key achievements of Guix-HPC between our previous report a year ago and today, February 2020. This year was marked by a major milestone: the release in May 2019 of GNU Guix 1.0, seven years and more than 40,000 commits after its inception.

Serverless computing is a new way of managing computations in the cloud. We show how it can be put to work for scientific data analysis. For this, we detail our serverless architecture for an application analyzing data from one of the instruments onboard the ESA Mars Express orbiter, and then, we compare it with a traditional server solution.