Introduction

Professor Jack Dongarra is a recipient of many prestigious awards, including the ACM A.M. Turing Award in 2021. His pioneering contributions to numerical linear algebra, scientific computing, and high-performance software have profound, broad, and lasting impacts on scientific studies in both conventional and emerging fields. This presentation describes Jack’s impact from different perspectives. In addition to valuable mathematical software packages and platforms he brought into existence as the building blocks of modern computing and data analysis, Jack has an extraordinarily large number of coauthors, let alone a much larger number of followers. Via his direct and broad research collaborations, Jack has kept identifying emerging challenges and making paradigm shifts. Furthermore, he has fostered a large, global high-performance computing community across various fields, generations, and cultures.

Technical Description

The impact analysis is based on a document collection, with about 59,000 published articles that directly cite Jack’s work. The collected articles are retrieved from the Semantic Scholar database as updated by April 2023. The collection is incomplete, compared to the estimated total citation number 110,000 by Google as of May 2023, due to limited data sources and access. Three networks are generated from the document collection: (1) a collaborator network, (2) a thematic content network, and (3) a geographical affiliation network. The collaboration network has about 1,660 coauthors with Jack. It is depicted in a constellation structure of 10 clusters by the similarity of publication patterns. The thematic content network is embedded in a 2D spatial space. The articles are clustered by title-abstract content into 23 broad themes. The annotation terms over each theme cluster appear most frequently in the titles and abstracts. For community detection on a network, we use the algorithm BlueRed [1, 2, 3, 4, and 5]. For the spatial embedding of a network with cluster configuration, we use a cluster-enhanced version of the algorithm SG-t-SNE. The maps at the bottom give the geographical locations of about 437 coauthor affiliations. They manifest the global impact of Jack’s contributions and collaboration. The spatial locations are determined by the combined use of Crossref and Google Maps. The data and processing codes are available upon request.

Maps & self-navigated surf-search

We present a few networks and maps generated from the retrieved data. On the thematic content map, every article is represented by a unique dot point. Clicking on the map will activate self-navigated surf-search in the following three modes.

The surf mode (the default mode). One surfs on the map simply by moving the cursor around. When the cursor is on top of an article dot, the article title pops up. Move the cursor around to browse related articles nearby. A click on the dot leads one to the article source (by direct link).
The DOI search mode. Enter a DOI (digital object identification) code in the search window, the article will be located and highlighted on the map. One can then surf around.
The keyword search mode. Enter a keyword in the search window, articles with the key word in the titles or abstracts are highlighted.

Enjoy the surf-and-search!