REGISTRATION IS CURRENTLY FULL.

The ability to control chemical reactivity continues to dominate the most pressing intellectual and societal challenges of our age. Computational approaches to predicting reactivity are poised for breakthroughs on multiple fronts due to the maturation of physics-based approaches and integration of newer machine learning (ML) strategies. The development of modern semi-empirical (e.g. GFN2,¹ B97-3c²) and ML methods (e.g. ANI-1³) have dramatically advanced the prediction capabilities for reaction discovery. Such methods are beginning to have a transformative impact on predictions of reactivity in extreme conditions, with the potential to couple accurate molecular predictions to multiscale phenomena.⁴ Excited-state reaction prediction remains a challenging frontier, but recent developments in scalable excited state methods and algorithms for exploring excited state surfaces are paving the way for the discovery of excited state reactivity.⁵  Coupling these powerful quantum chemical methodologies for single step reaction prediction with chemical network theory presents the exciting opportunity to construct chemical reaction networks with ML leading to the identification of new reaction pathways and mechanisms in unknown regions of chemical space.^6,7,8

Achieving further breakthroughs that generalize across the field of computational reaction prediction will require concerted methodological advances on several fronts, including ground and excited-state quantum chemistry, reactive force-fields, reaction network optimization, microkinetic modeling, and generative machine-learning. Some specific topics of interest to this workshop involve, but are not limited to:

• What are the practical vs fundamental obstacles to elucidating deep reaction networks involving an open-set of chemical reactions?
• How big is reaction space relative to molecular space? Are there a relatively small number of conserved reaction mechanisms?
• Can requisite function be achieved using well-defined subsets of chemical space and chemical reactions? Can function space be spanned by existing reactivity?
• Are ground-state quantum chemistry methodologies mature enough for polar and pericyclic reaction chemistry discovery?
• Are excited state quantum chemistry methodologies mature enough for excited state reaction discovery and/or tackling the challenge of photostability?
• How do we efficiently model the coupling of chemical reactivity to multiscale phenomena, e.g. reaction-induced phase separation, energetic materials, or molecular doping?
• Are existing computational tools sufficient for addressing reactivity in strongly heterogeneous systems (interfaces, mixtures)?

As this research topic spans multiple disciplines, with requisite expertise distributed globally, this workshop will bring together an international melting pot of physical and chemical scientists, computer scientists, and engineers to attempt to answer the most pressing questions at the frontier of computational reaction prediction. By integrating experts from the fields of reaction network discovery, quantum chemistry, excited state dynamics, transition state and rare events theory, ML, and extreme condition modeling, this workshop will cross-pollinate state-of-the-art advances from each field, leading to new collaborations addressing the most pressing fundamental and applied problems in computational reaction prediction to date