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.