7. Conclusion
Amidst the explosion of data in all walks of science, engineering, biology and biomedical science, it is useful to seek an interpretable basis for the emergence of function. How can geometry, physics, and engineering best inform biology or lead to the discoveries of new functional advanced materials? The complex multiscale interactions that characterize the dynamic behavior of biological systems [170] and advanced materials have limited our ability to understand the fundamental mechanisms behind the emergence of function to relatively idealized systems [171].
ML integrated with MSM is poised to enhance the capabilities of standard MSM approaches profoundly, particularly in the face of increasing problem complexity and data intensiveness. The contemporary research problems warrant an interdisciplinary environment to tackle emerging scientific and technological grand-challenge problems that carry substantial societal impact. Research projects, while posed across varied application domains in the broad STEM field, often have common features: (1) the problem/solution spans diverse length and timescales and benefits from MSM, (2) ML methods integrate into the MSM methods to define the new approaches at the frontiers of MSM development, (3) tools of data science are effectively leveraged to integrate experimental data with the proposed model, and (4) the implementation of the model will utilize HPC methods and/ or platforms. Foundational training for future scholars should ideally provide: (1) working knowledge in fundamental science and modeling methodologies at multiple lengths and timescales spanning the molecular to process scales; (2) the requisite skills to integrate, and couple multiple scales into a multiscale paradigm; (3) learnings to exploit elements of data science, including machine learning methods and tools of data integration from cloud-based, data-rich repositories in order to validate and test computational models and software; (4) learnings to combine the rich tools of ML with MSM methods to define the next-generation of MSM methods; (5) experience to adopt, and implement best practices in software architecture to leverage modern computational infrastructure and develop efficient sustainable codes. With these foundations and skill-sets in the arsenal of the emerging researcher, the potential to blend MSM, HPC, and ML presents opportunities for unbound innovation and represents the future of MSM and explainable ML that will likely define the fields in the 21st century.