A workshop on creation of an artificially intelligent organic chemist: applied machine learning for multiphase chemistry
Abstract
Accurate prediction of chemistry given reactants and a set of conditions is a desirable capability in many fields, yet is only possible for simple systems. Environmental Chemistry and specifically, multiphase organic chemistry deals with the physicochemical transformations between gaseous, liquid and solid matter on scales ranging from nanoseconds to millennia, molecules to solar systems. Evolution of organic compounds in multiphase systems is inherently more complex than single-phase counterparts because there are simultaneous sequential and parallel chemical corridors in addition to phase transitions. These multi-phase organic transformations are essential for all life on Earth. While the field has some understanding of the multiphase organic phenomena in relatively simple systems, little Is known under realistic complex conditions and there are almost no data on radical-driven chemistry in thin film systems when complex, often highly concentrated and non?ideal mixtures of water, organics and inorganics are present. Furthermore, in such complex systems not only the composition and chemistry can change on the nanoscale, but also the phases involved. Historically predictive approaches for multiphase chemistry describe relatively simple mechanisms that explain oxidation or hydrolysis of a very limited number (or single) species, or a simplified partitioning system at equilibrium due, in part, to computational constraints. Machine learning lends itself to such systems because traditional chemical approaches, such as explicit hand-written mechanisms, are impossible when systems are highly non-linear and the probabilistic nature is non-trivial. The overall purpose is to identify strategies that capitalize on computational approaches to solve complex chemistry. A major theme is to bring together a group of 20-25 researchers representing a cross-section across multiphase chemistry and computation to identify critical open questions in computational chemistry for multiphase systems, as well as formulate protocols for answering those questions. A combination of experiments involving theory, numerical simulations, lab and field experimentation is anticipated. The steering committee will prepare five open and fundamental questions for presentation on the first day. Each question will be presented by an invitees paired as chemist and another researcher (e.g., computer scientist). Each presenter will critique and refine the question from their unique science perspective. As a group and in breakouts attendees will conduct thought experiments.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 14, 2019
- Source ID
- W911NF1810349
Entities
People
- Ann Marie Carlton
Organizations
- Army Contracting Command
- United States Army
- University of California, Irvine