Organizers: Cecilia Clementi and Shantenu Jha
Dates: 8-9 October 2017
Location: Rice University, Houston, Texas
A workshop to understand the broader cyberinfrastructure requirements of computational molecular sciences was organized at Rice University, Oct 8 and 9. The opening session of the workshop had several talks illustrating the importance of a robust, functional and scalable ecosystem of cyberinfrastructure capabilities that goes beyond simple “single code” performance. These requirements ranged from the ability to run scalable ensemble based simulations to machine learning analysis to improve the parameterization of force-fields. At the conclusion of the first session there was a “talk from the trenches” that conveyed a group perspective of the computational challenges and software development life cycle challenges inherent in molecular sciences. The challenges were perceived to be representative of groups across many size scales. The second session of the workshop was devoted to discussion of the wide range and requirements of workflows; this session did not discuss specific tools or technologies but tried to capture the richness of the requirements and importance of workflow capabilities.
The second day of the workshop had two sessions comprised of invited talks. The first was focussed on a wide range of resource and software capabilities providers; the second day was focused on the data analysis state-of-the-art and software development requirements. Resource providers included Blue Waters, the Open Science Grid and DOE’s OLCF; each of these discussed the resources and how they could be useful for Molecular Sciences community. Resource providers discussed requirements of the community they were well suited to support. Resource providers were in agreement on the importance of community capabilities and activities as a way to improve their ability to support computational molecular science.
The workshop concluded with a robust discussion on the following topics: (i) What are the primary Software Development, deployment challenges that different projects face? And, (ii) From the participants perspective, how and where should/could MolSSI make a difference? An important point that emerged in discussions with the criticality of changing the culture and best practices around software development as much focus on specific software development.
A report on the workshop’s outcomes is available here.
- Oliver Beckstein (Arizona State U.)
- Brett Bode (UIUC)
- John Chodera (Memorial Sloan Kettering Lab)
- Cecilia Clementi (Rice U., MolSSI)
- T. Daniel Crawford (Virginia Tech, MolSSI)
- Erik Engquist (Rice U.)
- Teresa Head-Gordon (U.C. Berkeley, MolSSI)
- Eric Irrgang (U. Michigan)
- Shantenu Jha (Rutgers U., MolSSI)
- Peter Kasson (U. Virginia)
- Lydia Kavraki (Rice U.)
- Matthew Knepley (Rice U.)
- Dmitry Liakh (ORNL)
- Honggao Liu (Texas A&M U.)
- Andre Merzky (Rutgers U.)
- Lauren Michael (U. Wisconsin)
- Julien Michel (University of Edinburgh)
- Jan E Odegard (Rice U.)
- Sanjay Padhi (Amazon, Inc.)
- Sudhakar Pamidighantam (Iowa U.)
- Arvind Ramanathan (ORNL)
- Vivek Singh (U. Rutgers)
- Theresa Windus (Iowa State U., MolSSI)
- David Wright (University College London)