Abstract
This review article highlights state-of-the-art data-driven techniques todiscover, encode, surrogate, or emulate constitutive laws that describe thepath-independent and path-dependent response of solids. Our objective is toprovide an organized taxonomy to a large spectrum of methodologies developed inthe past decades and to discuss the benefits and drawbacks of the varioustechniques for interpreting and forecasting mechanics behavior across differentscales. Distinguishing between machine-learning-based and model-free methods,we further categorize approaches based on their interpretability and on theirlearning process/type of required data, while discussing the key problems ofgeneralization and trustworthiness. We attempt to provide a road map of howthese can be reconciled in a data-availability-aware context. We also touchupon relevant aspects such as data sampling techniques, design of experiments,verification, and validation.