MolSSI Sustainable Software Symposium at Spring 2019 ACS Meeting
MolSSI is pleased to host the symposium “Sustainable Software for Computational Molecular Science” at the Spring 2019 ACS Meeting in Orlando from March 31-April 4, 2019. This program brings together leading researchers and software developers from academia and industry in the computational molecular sciences for a 5 day, 9 session meeting.
Symposium Description
The field of computational molecular sciences (CMS) broadly encompasses quantum chemistry, materials science, and biomolecular simulation, and computational molecular scientists worldwide have made innumerable vital contributions to scientific discovery. Indeed, modern molecular dynamics and quantum chemical models have achieved such a level of robustness and accuracy that they are often considered “computational experiments,” in many cases with greater reliability than laboratory measurements, with direct impact on the design of more efficient combustion systems, development of new chiral drugs, elucidation of the functionalities of biological macromolecules, the development of more advanced materials for solar-energy storage, technology for CO2 sequestration, and many others. These scientific breakthroughs have been made possible by the evolution of hundreds of community codes, some with lifetimes reaching back to the earliest days of computing. In the last several years, a “software revolution” has begun to grow as more and more molecular scientists adopt the tools, methods, and even philosophy of modern software engineering. Although still in its earliest phases, such a revolution promises to change fundamentally the level of sophistication, sustainability, and interoperability of molecular science codes, ultimately leading to a dramatic increase in the complexity of scientific problems that the field can address. This symposium will include topics such as best practices in software engineering, reproducibility in CMS, new software tools for machine learning, and reaching advanced hardware.
Symposium Sessions
Interoperability and Reproducibility in the Computational Molecular Sciences
In principle, quantum and classical simulations can be exactly reproduced within statistical uncertainty given the input to a simulation software and its source code. However, reproducibility and validation in computational molecular sciences remains a challenge. Software and hardware heterogeneity, missing data, lack of validation with respect to previous published data or lack of reproducibility across simulation software hinder the replication of simulation studies. Topics in this session include relevant software tools and approaches for enhancing reproducibility and validation in molecular simulation.
Date and Time: Sunday, Mar 31 8:30 AM
Invited Speakers
- Harold W. Hatch, National Institutes of Standards and Technology
- Michael Shirts, CU Boulder
- Giulia Galli, University of Chicago
- Julia Rice, IBM
See the session abstracts here.
Workflow technologies simplify the processes of developing reliable computational methods, deploying reproducible software, exploiting scalable computing, and sharing standardized best practices. This session will explore the available workflow software for CMS, and the associated advantages and current limitations.
Date and Time: Sunday, Mar 31 1:30 PM
Invited Speakers
- Jharrod Lafon, OpenEye Scientific
- Oliver Beckstein, Arizona State University
- Alberto Gobbi, Genentech
- Chaya Stern, Memorial Sloan Kettering Cancer Center
- Shantenu Jha, Rutgers University
- Marcus Hanwell, Kitware Inc
See the session abstracts here.
Experiences and Challenges Developing Open and Modular Software
This session focuses on experiences developing software for computational chemistry for several well known codes. Speakers share strategies for developing software, and lessons learned including when to pursue formal software development, maintaining large projects, code refactoring, and making codes flexible and sustainable.
Date and Time: Monday, Apr 1 8:30 AM
Invited Speakers
- Theresa Windus, Iowa State University
- Edward Maginn, University of Notre Dame
- Sharon Glotzer, University of Michigan
- Steve Plimptom, Sandia National Laboratories
- Shyam Dwaraknath, Lawrence Berkeley National Labs
See the session abstracts here.
Best Practices in Software Development from CMS Communities and Beyond
High Performance and Massively Parallel Chemistry
The purpose of this session is to discuss the challenges and opportunities facing the molecular sciences community as the pace and importance of software development increases against a backdrop of frequently-changing hardware and an increasing desire for software performance, reliability, and reproducibility. Solving these issues will require new programming paradigms and practices in order for software to remain flexible and relevant into the future.
Date and Time: Tuesday, Apr 2 8:30 AM
Invited Speakers
- Robert Harrison, Stony Brook University
- Pavel Pokhilko, University of Southern California, Los Angeles
- Jeff Hammond, Intel
See the session abstracts here.
In recent years, machine learning and AI have become widely used in commercial settings for applications such as character and facial recognition, email spam filtering, and even autonomous vehicles. Machine learning is now emerging as a new paradigm in chemistry, as there is significant interest in harnessing the same methods in tasks such as drug discovery, protein structure prediction, or predicting potential energy surfaces. All machine learning relies on accurate and available data for training models. This session will explore software tools for data and machine learning for chemistry.
Date and Time: Wednesday, Apr 3 8:30 AM
Invited Speakers
- Patrick Rinke, Aalto University, Novel Materials Discovery Laboratory (NOMAD)
- Bryce Meredig, Citrine Informatics
- Adrian Roitberg, University of Florida
- Levi Naden, The Molecular Sciences Software Institute
See the session abstracts here.
Software Tools: Molecular Mechanics
We may not know exactly how the computational chemistry will change in the near future, however, we do know that it will be critical to most developments moving forward. This session looks at the classical molecular mechanics and QM/MM fields to showcase developing and near-future tools which will improve the capability of existing experiments, and possibly be used in the design of future ones. Presenters at this session range expansions on existing tools such as Amber, to standalone tools like a polarizable force field for classical and QM/MM, to integrated software for areas such as multi-body dynamics and sumulations of soft matter.
Date and Time: Wednesday, Apr 3 1:30 PM
Invited Speakers
- Peter Cummings, Vanderbilt University
- Marc Riera Riambau, University of California, San Diego
See the session abstracts here.
Software Tools: Quantum Mechanics
This session focuses on software development supporting capabilities in quantum chemistry. Topics include automated property-driven molecule discovery, real-time quantum chemistry for spectroscopy, large-scale parallel linear-response calculations, automated calibration of QM/MM models, standardized components and interfaces for quantum chemistry, cloud-based quantum chemistry, stochastic simulations of electron correlation, and the automation of combustion chemistry modeling.
Date and Time: Thursday, Apr 4 8:30 AM
Invited Speakers
- Geoff Hutchison, University of Pittsburgh
- Xiaosong Li, University of Washington
- Todd Martinez, Stanford
See the session abstracts here.
This session contains a variety of theoretical and computational developments. The theoretical results include more efficient high-order expansions in the fast multipole method, the use of symmetries to reduce the cost of symmetry-preserving nudged-elastic-band calculations, a flexible thermostat for both classical and path-integral molecular dynamics, and rare-event modeling for slow chemical reactions. The computational results include precise simulations of uniform few-electron gases and comparative benchmarks of electronic structure solvers within a unified software framework.
Date and Time: Thursday, Apr 4 1:30 PM
See the session abstracts here.