DEVELOPMENT OF TOOLS FOR INTEGRATING AND UTILIZING COMPLEX DATA FROM CHEMICAL SCIENCES, GEOSCIENCES, AND BIOSCIENCES

Fundamental science efforts in the Chemical Sciences, Geosciences, and Biosciences (CSGB) Division (reference 1) produce significant amounts of system-specific data for relevant energy systems. Recent reports (reference 2) pinpoint topical opportunities in which multimodal approaches are needed to increase impact of fundamental research on applied technology. In particular, artificial intelligence/machine learning (AI/ML) tools are needed to facilitate the integration of highly heterogeneous data, promote data sharing for model fine-tuning and generalization, extrapolate to data-poor regions, or extend fundamental theories to complex, many-parameter systems. Grant applications are sought in the following subtopics: a. AI/ML Framework for Multi-Modal Analysis of Complex Data Experimental measurements and computational simulations of physicochemical phenomena produce complex data that are disparate, non-uniform, or unstructured. This presents multiple challenges to data analysis efforts. It can be difficult to extend analytical models based on these data beyond the specific system(s) studied. Moreover, data derived from diverse techniques and research communities can be challenging to integrate. However, when combined, these data may provide an opportunity to generate comprehensive models by combining orthogonal data (e.g., multiple types of spectroscopic and structural characterization data, and reactivity and transport data). Relevant complex data impacted by these challenges are found in many areas of chemical sciences, geosciences and biosciences research including (but not limited to) the following, with references to example databases: Reaction kinetics, compositional, and spectroscopic information on chemical conversions (see reference 3 and 4 for examples) Structural information of active sites and motifs (see reference 5 and 6 for examples) Synthesis protocols (see reference 6 for examples) Geochemical and geophysical data to understand subsurface chemistry Biochemical and biophysical data to understand energy storage and conversion in organisms Since valuable and relevant data are not always present in extant databases, principal investigators are encouraged to consider and identify multiple sources of information such as datasets found in reference 7, experimental and computational publications from diverse research groups, etc. Tools are needed to facilitate the compilation, homogenization, and validation of these data and their distribution among physicochemical modelers working in the chemical sciences, geosciences and biosciences. Advanced tools are also needed to enhance physical models by using AI/ML algorithms, and to facilitate transferability of physical and data models outside their narrow training domains. Applicants should refer to the relevant reports mentioned in reference 2 to identify priority research directions/opportunities in one or more of the areas mentioned above. Applications that propose software solutions alone without addressing one or more of those priority research opportunities or directions would be out of scope for this solicitation. Approaches that do not use the most current AI/ML methods will be considered unresponsive to this topic. Questions Contact: Aaron Holder, Aaron.Holder@science.doe.gov and Raul Miranda, Raul.Miranda@science.doe.gov b. Other In addition to the specific subtopic listed above, the Department invites grant applications in other areas that fall within the scope of the topic description above.