@article{22248,
  abstract     = {Many living and artificial systems improve their fitness or performance by adapting to changing environments or diverse training data. However, it remains unclear how environmental variation shapes adaptation, what is learned, and when memory of past conditions is retained. Here we show how cyclic environmental change can produce robust memory. Using a model athermal disordered solid trained by inverse design to attain target elastic properties over a prescribed range, we find that the system evolves toward a marginally absorbing manifold (MAM), meaning that training is reversible within the training range but not beyond it, which encodes a memory of that range. We further propose a general mechanism for MAM formation and memory encoding based on discontinuities in the gradient of the trained quantity. These results provide a simple, broadly applicable physical framework for how adaptive systems learn under changing environments and retain memory of past conditions.},
  author       = {Zu, Mengjie and Goodrich, Carl Peter},
  issn         = {2835-8279},
  journal      = {PRX Life},
  number       = {2},
  publisher    = {American Physical Society},
  title        = {{Learning by Training: Emergent physical memory from cyclically tuning disordered sphere packings}},
  doi          = {10.1103/48k2-cw3b},
  volume       = {4},
  year         = {2026},
}

@article{22288,
  abstract     = {Soft solids and their surface deformations control the response of many natural and artificial systems. Yet, their underlying properties are vigorously debated, particularly for polymer networks. While molecular-scale theories predict no interfacial changes with macroscopic deformation, multiple experiments suggest otherwise. To settle this issue, we measure displacement fields near the interface of a silicone gel, in the limit of small deformations. We discover an unexpected multiscale response. The shear modulus decreases smoothly by half with 20  μ⁢m of the interface. At the same time we observe a surface excess elasticity, that depends on history and outer medium composition. These results reveal the fundamentally multiscale nature of polymeric surfaces, and call for further experimental and theoretical investigations into the basic understanding of soft solid interfaces.},
  author       = {Bain, Nicolas and Wilen, Lawrence A. and Gerber, Dominic and Zu, Mengjie and Goodrich, Carl Peter and Duraivel, Senthilkumar and Varma, Kaarthik and Koganti, Harsha and Style, Robert W. and Dufresne, Eric R.},
  issn         = {2160-3308},
  journal      = {Physical Review X},
  number       = {2},
  publisher    = {American Physical Society},
  title        = {{Multiscale interfacial mechanics of soft solids}},
  doi          = {10.1103/8msx-l8s7},
  volume       = {16},
  year         = {2026},
}

@article{21237,
  abstract     = {Intelligent soft matter lies at the intersection of materials science, physics, and cognitive science, promising to change how we design and interact with materials. This transformative field aims to create materials with life-like capabilities, such as perception, learning, memory, and adaptive behavior. Unlike traditional materials, which typically perform static or predefined functions, intelligent soft matter can dynamically interact with its environment, integrating multiple sensory inputs, retaining past experiences, and making decisions to optimize its responses. Inspired by biological systems, these materials leverage the inherent properties of soft matter such as flexibility, adaptability, and responsiveness to perform functions that mimic cognitive processes. By synthesizing current research trends and projecting their evolution, we present a forward-looking perspective on how intelligent soft matter could be constructed, with the aim of inspiring innovations in areas such as biomedical devices, adaptive robotics, and beyond. We highlight new pathways for integrating sensing, memory and actuation with low-power internal operations, and we discuss key challenges in realizing materials that exhibit truly “intelligent behavior”. These approaches outline a path toward more robust, versatile, and scalable materials that can potentially act, compute, and “think” through their inherent intrinsic material properties—moving beyond traditional smart technologies that rely on external control.},
  author       = {Baulin, Vladimir A. and Giacometti, Achille and Fedosov, Dmitry A. and Ebbens, Stephen and Varela-Rosales, Nydia R. and Feliu, Neus and Chowdhury, Mithun and Hu, Minghan and Füchslin, Rudolf and Dijkstra, Marjolein and Mussel, Matan and van Roij, René and Xie, Dong and Tzanov, Vassil and Zu, Mengjie and Hidalgo-Caballero, Samuel and Yuan, Ye and Cocconi, Luca and Ghim, Cheol-Min and Cottin-Bizonne, Cécile and Miguel, M. Carmen and Esplandiu, Maria Jose and Simmchen, Juliane and Parak, Wolfgang J. and Werner, Marco and Gompper, Gerhard and Hanczyc, Martin M.},
  issn         = {1744-6848},
  journal      = {Soft Matter},
  number       = {21},
  pages        = {4129--4145},
  publisher    = {Royal Society of Chemistry},
  title        = {{Intelligent soft matter: Towards embodied intelligence}},
  doi          = {10.1039/d5sm00174a},
  year         = {2025},
}

@article{19856,
  abstract     = {Unlike in crystals, it is difficult to trace emergent material properties of amorphous solids to their underlying structure. Nevertheless, one can tune features of a disordered spring network, ranging from bulk elastic constants to specific allosteric responses, through highly precise alterations of the structure. This has been understood through the notion of independent bond-level response—the observation that, in many cases, different springs have different effects on different properties. While this idea has motivated inverse design in numerous contexts, it has not been formalized and quantified in a general context that not just informs but enables and predicts inverse design. Here, we show how to quantify independent response by linearizing the simultaneous change in multiple emergent features, and introduce the much stronger notion of fully independent response. Remarkably, we find that the mechanical properties of disordered solids are always fully independent across a wide array of scenarios, regardless of the target features, tunable parameters, system size, dimensionality, and class of interactions. Furthermore, our formulation quantifies the susceptibility of features to parameter changes, which is correlated with the maximum linear tunability. We also demonstrate the implications for multifeature inverse design beyond the linear regime. These results formalize our understanding of a key fundamental difference between ordered and disordered solids while also creating a practical tool to both understand and perform inverse design.},
  author       = {Zu, Mengjie and Desai, Aayush A and Goodrich, Carl Peter},
  issn         = {1079-7114},
  journal      = {Physical Review Letters},
  number       = {23},
  publisher    = {American Physical Society},
  title        = {{Fully independent response in disordered solids}},
  doi          = {10.1103/PhysRevLett.134.238201},
  volume       = {134},
  year         = {2025},
}

@article{17407,
  abstract     = {The ability to control forces between sub-micron-scale building blocks offers significant potential for designing new materials through self-assembly. Traditionally, this involves identifying a crystal structure with a desired property and then designing building-block interactions so that it assembles spontaneously. However, this paradigm fails for structurally disordered solids, which lack a well-defined structure. Here, we show that disordered solids can still be treated from an inverse self-assembly perspective by bypassing structure and directly targeting material properties. Using the Poisson’s ratio as a primary example, we demonstrate how differentiable programming links interaction parameters with emergent behavior, enabling iterative training to achieve the desired Poisson’s ratio. We also tune other properties, including pressure and local 8-fold structural order, and can even control multiple properties simultaneously. This robust, transferable, and scalable approach can handle a wide variety of systems and properties, demonstrating the utility of disordered solids as a practical avenue for self-assembly platforms.},
  author       = {Zu, Mengjie and Goodrich, Carl Peter},
  issn         = {2662-4443},
  journal      = {Communications Materials},
  publisher    = {Springer Nature},
  title        = {{Designing athermal disordered solids with automatic differentiation}},
  doi          = {10.1038/s43246-024-00583-4},
  volume       = {5},
  year         = {2024},
}