Accelerating Science with Open Source

An Introduction to Open-Source Science (OSSci)

May 16, 2024

PyData Prague

Nice planet you got there…

Premise

• Our world is facing many pressing challenges
• Climate, transition to green energy, food security, etc.
• We need more, faster, and better science!
• We know:
• Software and AI accelerate science
• Open source accelerates software and AI

​

However

• The open source ecosystem for science is underdeveloped and lacks maturity
• Without a robust foundation for open source in science, scientific progress will be
• more difficult,
• more costly, and
• take much longer than we can afford.

​

Common Challenges for Open Source in Science

• Abandonware
• Duplicate efforts (a.k.a. reinventing the wheel)
• Hard problems! (materials, climate, energy, etc.)
• Specialized fields, different levels of abstraction, vast difference in software use
• Lack of efficient (centralized) knowledge flows
• Sustainability issues (e.g., lack of funding)
• Misaligned incentives
• Lack of consistent training and onboarding for students and young scientists
• Lack of defined career paths
• Skill/knowledge gaps between domain experts and software engineers
• Fuzzy ROI of open-source scientific software

Open-Source Science (OSSci)

• NumFOCUS program
• Launched in July 2022
• In partnership with IBM
• Goal: to accelerate scientific research and discovery with open-source software and data
• Connects scientists, OSS developers, and other stakeholders
• Share best practices
• Identify common pain points and challenges
• Explore and create solutions together
• Team
• Tim Bonnemann, Community Lead (IBM Research)
• Jonathan Starr, Program Manager (NumFOCUS)

OSSci Interest Groups

• Open for anyone
• Monthly meetings
• Collaboration via Slack

New in 2024:

• Economics IG (June 2024)
• Others (TBD)

Chemistry / Material Science

Life Sciences / Healthcare

Climate and Sustainability

Reproducible Science

Map of Science

What Do the Interest Groups Do? A First Example

Chemistry and Material Science IG

• Reaction prediction resources
• LLMs for chemical applications

Map of Open-Source Science (MOSS)

A new interface to surface the interrelations between existing OSS projects, published research, and the people involved. Goals:

• Improve discoverability
• Encourage collaboration

https://map.opensource.science

​

Other OSSci Activities

Events

• In-person meetups (often co-hosted with PyData)
• Community visits
• Launched in October 2023
• Virtual meetups series
• Launching in June 2024
• Themes and topics exploring open source in science
• Workshops
• Clinics

Content

• Spotlight series
• Learn about science-related OSS projects and their teams
• Event coverage
• Guest posts

Join the OSSci Community

• Stay in the loop
• Email newsletter
• Medium
• Social media (LinkedIn, Twitter, Mastodon, Bsky)
• Join the conversation
• Community forum
• Collaborate
• Join an interest group
• Slack�

https://bit.ly/join-ossci-community

Thanks!

Questions welcome…

https://www.linkedin.com/company/open-source-science-initiative/

https://www.youtube.com/watch?v=jZyLSRCba_M

https://thealliance.ai

Tim Bonnemann

• San José, CA
• Community architect & builder
• IBM Research
• Wannabe trailrunner
• Find me online
• tim@opensource.science
• linkedin.com/in/tbonnemann/
• sfba.social/@planspark
• @planspark.bsky.social

Join the OSSci Community

• Stay in the loop
• Email newsletter
• Medium
• Social media (LinkedIn, Twitter, Mastodon, Bsky)
• Join the conversation
• Community forum
• Collaborate
• Join an interest group
• Slack�

https://bit.ly/join-ossci-community