Bioinspired nonequilibrium search for novel materials

Abstract

Abstract

Inverse methods for design of soft materials

“…For example, an objective function that periodically switched between two states of reconfigurable circuits 141 and of allosteric networks 142 found solutions prioritizing transitions between states. Murugan and Jaeger recently suggested applying this same "switching" strategy to selfassembling materials, 23 and we agree that this is a potentially fruitful area for further study.…”

“…Inverse methods have proven powerful for designing granular materials, [21][22][23] block copolymer assemblies, [22][23][24] and bio-inspired materials. 23 Recent reviews 25,26 have highlighted inverse techniques that leverage machine learning (ML) to effectively process the high-dimensional data obtained from computer In an example forward approach, materials with new properties may be discovered by repeatedly carrying out the following sequence of steps. Chemical synthesis (A) is used to create material building blocks with effective, coarse-grained interactions (B) that drive their assembly into structures (C) which impart emergent properties on the macroscopic scale (D).…”

“…Inverse methods have proven powerful for designing granular materials, [21][22][23] block copolymer assemblies, [22][23][24] and bio-inspired materials. 23 Recent reviews 25,26 have highlighted inverse techniques that leverage machine learning (ML) to effectively process the high-dimensional data obtained from computer simulations of materials to analyze and design their novel structural and dynamic properties. Inverse methods have also been widely applied in related fields including the design of molecules and chemical reactions.…”

Biologically inspired far-from-equilibrium materials

“…When one searches for materials with improved properties or entirely new functionalities or properties, the process often involves an iteration that is designed to bridge the gap between the initial starting point and the desired target. In their article, Murugan and Jaeger 18 believe that this can be viewed as an optimization problem with a vast search space. However, they take a different tack and point to more recent strategies exploiting knowledge about the material configuration statistics as well as highlighting the advantages of considering time-varying environments, as is often the case in biological structures.…”

Machine learning-assisted design of material properties

“…This challenge has been addressed in part by posing materials design as an inverse problem to be solved using methods of numerical optimization to efficiently navigate the design space. [19][20][21][22][23][24][25] Commonly used algorithms iteratively optimize an objective function formulated based on the desired material properties. At each iteration, the property is measured for the current point in the design space, and the optimizer selects new points to investigate until the algorithm achieves convergence to an optimal solution, within specified tolerances.…”