Automate reaction-condition optimization with closed-loop laboratories
Year: 2026
Type: preprint
Abstract: This revised article addresses Q0012, 'Automate reaction-condition optimization with closed-loop laboratories?' using a question-specific mathematical construction rather than a generic logarithmic template. The proposed model, termed the Closed-Loop Information-Gain Optimizer, uses yield/selectivity y, uncertainty sigma, experiment cost c, safety penalty h and surrogate-model posterior and is ordered along autonomous experiment iteration. The construction is anchored to external primary research classified as STRONG DIRECT SUPPORT in the validation atlas. The external evidence establishes experimentally measurable mechanisms or target performance; it is not conflated with proof of the new equation itself. The model defines explicit state variables, derives a dimensionless objective or hazard, specifies a prospective calibration procedure, and states falsifiable predictions. The experimental constraint carried forward from the atlas is: Closed-loop self-optimization and scale-up have been demonstrated experimentally. The article incorporates the required correction: Operator is compatible; use iteration-dependent uncertainty.
Author: yıldırım, Begüm
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7234025871
