Keynotes

Artificial Intelligence (AI) holds considerable promise to advance science, and the societal and economic well-being of the citizens. There is much active research in applying the techniques and tools of AI to a wide range of problems, as is evidenced in the programs of a wide range conferences including eScience. The eScience community is rather uniquely positioned in that it investigates the use of AI in eScience infrastructure while also making its tools and services available for others to carry out AI related research. In the NSF funded ICICLE project (Intelligent Cyberinfrastructure for Computational Learning in the Environment) in which I am a co-PI, we call this “AI for Cyberinfrastructure (AI for CI) and Cyberinfrastructure for AI (CI for AI), leading to “AI for CI for AI”.

I discuss technical advancements of the ICICLE AI Institute in AI for CI for AI in the first 3 years of its existence. The project has raised broader questions beyond technical advancement, driven by its tagline “democratizing AI”. While AI research in eScience and eScience infrastructure for AI is broader than AI for CI for AI, I use the latter as a launching point to discuss democratization in the context of how the eScience community carries out its research. We are unique among research communities, I argue, in that significant measures of success include use of our products by others in the research community. This has been the case for decades, but with AI seeping into every layer of our research ecosystem, we proceed with deliberation.

Although Digital Twins have been used since the early 2000’s for Product Lifecycle Management purposes, it is only recently that this concept has gained momentum in the Earth Science domain. It has the potential to revolutionize the way Earth Science research will be conducted in the future, and how results and knowledge from this research will provide information to support decision making and yield impactful societal benefits.

We define an Earth System Digital Twin (or ESDT) as an Information System for Understanding, Forecasting, and Conjecturing the complex interconnections among Earth systems, including anthropomorphic forcings and impacts to humanity.

More specifically, an Earth System Digital Twin is a dynamic and interactive information system that first provides a digital replica of the past and current states of the Earth or Earth system, as accurately and timely as possible (i.e., the “What Now”); second allows for computing forecasts of future states under nominal assumptions and based on the current replica (i.e., the “What Next”); and third offers the capability to investigate many hypothetical scenarios under varying impact assumptions (i.e., the “What If”). The full power of digital twins is that, through an integrated representation and standardized tools and software technologies, the same digital replica can address the needs of multiple users at various resolutions (spatial and temporal) and for various applications (science, economic, policy, etc.) – “from farmer to scientist”.

The challenges of building optimal digital twins are many and complex. In 2020, to study these challenges and to develop the technologies needed to build them, NASA’s Advanced Information Systems Technology (AIST) Program has started a new thrust in Earth System Digital Twins (ESDTs), investigating the benefits of ESDT for NASA Earth Science, studying specific use cases and developing technologies and components as well as a few prototype systems that will help identify building blocks from which future ESDTs will be built. All these various aspects describing NASA ESDT efforts will be presented during the talk.

In addition to revolutionizing our personal lives, Artificial Intelligence (AI) has entered the science arena where it is used to classify celestial galaxies, predict the weather, detect cancer cells, and design new materials among many other applications. Although AI has shown incredible results, it raises technological issues and challenges traditional scientific methods. This talk will explore how the scientific lifecycle that moves from research question generation to result dissemination is changing due to the growing capabilities of AI. Broadly this talk will identify challenges that exist in adopting novel AI-based approaches for the scientific lifecycle at the technological and social levels. It will detail how AI is being integrated into the scientific lifecycle stages, focusing in depth on the computation automation stage. This stage often relies on workflow management systems (WMSs), such as Pegasus, to automate the execution of application workflows on modern cyberinfrastructure. Recent advances in AI provide opportunities to improve workflow performance and resilience. In this context, the talk will explore the following questions: How can computation automation be enhanced with AI technologies? How is the broader scientific lifecycle changing and what are the impacts of AI on the way we will do science in the future?

As the number of qubits in advanced quantum computers is getting larger over 100 qubits, demands for the integration of quantum computers and HPC are gradually growing. The quantum computing technology is a promising component in near future HPC system to accelerate e-Science and computational science. RIKEN R-CCS has been working several projects to build a platform which integrates quantum computers and HPC together. Recently, we have started JHPC quantum project to design and build a quantum-supercomputer hybrid computing platform which integrates different kinds of quantum computers, superconducting quantum computer from IBM and trapped-ion quantum computer from Quantinuum, with supercomputers including Fugaku. In this presentation, the overview and plan of the JHPC quantum project with a perspective of quantum HPC hybrid computing will be presented.