Self-optimization of chemical reactions enables faster optimization of reaction conditions or discovery of molecules with required target properties. The technology of self-optimization has been expanded to discovery of new process recipes for manufacture of complex functional products. A new machine-learning algorithm, specifically designed for multiobjective target optimization with an explicit aim to minimize the number of “expensive” experiments, guides the discovery process. This “black-box” approach assumes no a priori knowledge of chemical system and hence particularly suited to rapid development of processes to manufacture specialist low-volume, high-value products. The approach was demonstrated in discovery of process recipes for a semibatch emulsion copolymerization, targeting a specific particle size and full conversion.

A novel approach to multi-target optimization of expensive-to-evaluate functions is explored that is based on a combined application of Gaussian processes, mutual information and a genetic algorithm. The aim of the approach is to find an approximation to the optimal solution (or the Pareto optimal solutions) within a small budget. The approach is shown to compare favourably with a surrogate based online evolutionary algorithm on two synthetic problems.

No abstract

In this work new methods of processing bio-feedstocks in the formulated consumer products industry are discussed. Our current approach to formulated products design is based on heuristic knowledge of formulators that allows selecting individual compounds from a library of available materials with known properties. We speculate that most of the compounds (or functions) that make up the product to be designed can potentially be obtained from a few bio-sources. In this case, it may be possible to design a sequence of transformations required to convert feedstocks into products with desired properties, analogous to a metabolic pathway of a complex organism. We conceptualize some novel approaches to processing bio-feedstocks with the aim of bypassing the step of a fixed library of ingredients. Two approaches are brought forward: one making use of knowledge-based expert systems and the other making use of applications of metabolic engineering and dynamic combinatorial chemistry.