Being able to predict when and where a material will fracture is a key issue with important industrial implications in the area of device and component monitoring. Researchers from the Center for Complexity and Biosystems and the “Aldo Pontremoli” Department of Physics at the University of Milan in collaboration with colleagues from the Department of Materials Science at the Friedrich-Alexander University of Erlangen-Nuremberg in Germany, in a paper recently published in Nature Communications, have shown that through artificial intelligence it is possible to predict the fracture of silica glasses by examining their microstructure. Thanks to recent advances in deep learning, it is possible to obtain accurate fracture predictions even for highly disordered solids such as glasses. Unfortunately, however, the huge number of parameters used by artificial neural networks often makes a physical interpretation of the results impossible. This problem affects not only fracture prediction but is found in multiple applications of artificial intelligence. Researchers at the Center for Complexity and Biosystems applied a method to identify the areas of the microstructural image most used by the neural network for fracture prediction. This provided insight into what characteristics make a material more susceptible to fracture. “Neural networks are black boxes”-explains Stefano Zapperi, professor of theoretical matter physics and coordinator of the research-“and this is an important limitation in scientific research where the main purpose is to explain the origin of a phenomenon.  Thanks to the method we used, it was possible to better understand what are the relevant aspects that determine the fracture of the material and thus obtain not only a prediction but also a greater fundamental understanding of the mechanics of glasses.” “The strategy we developed lends itself to further applications”-adds Roberto Guerra, associate professor in the physics department and co-author of the paper-“such as to design disordered materials with better fracture toughness properties.”  F. Font-Clos et al. Predicting the failure of two-dimensional silica glasses, Nature Communications (2022)

High density electroencephalography (hdEEG) provides an accessible but indirect tool to record brain activity in real time. While the recorded signals encode a large amount of information, it is not straightforward to interpret them. This would be extremely useful for several potential applications such as disease diagnosis and monitoring. Researchers at the Center for Complexity and Biosystemts of the University of Milan, coordinated by Caterina La Porta from the Department of Environmental Science and Policy, have devised a new method to visualize and analyze hdEEG recordings based on a multilayer network representation. The work was just published in the journal Frontiers in Network Physiology. A network representation provides an intuitive picture of the spatial connectivity underlying an hdEEG recording. In order to minimize the information lost in the network projection, CCamp;B researchers have constructed an algorithm that creates a network while maximizing the information content. The algorithm was then tested on hdEEG signals recorded during sleep in individuals with mental health issues by researchers at the Department of Health Sciences of University of Milan who collaborated to this research. The work also involved researchers at the Physics Department “Aldo Pontremoli” and at the Department of Biomedical and Clinical Sciences ‘Luigi Sacco’ of the University of Milan. “By computing a set of statistical indicators of the network topology, we were able to reveal significant differences between patients with mood disorder and healthy subjects,” explains Stefano Zapperi who coauthored the study. “The analysis also indicates that patients display a highly correlated activity in some regions of the brain”, adds Caterina La Porta, “a very important result, because we could use our algorithm to identify patients directly from non-invasive hdEEG recordings with potential applications of the method also to other pathological conditions.” Read the paper: Font-Clos F, Spelta B, D’Agostino A, Donati F, Sarasso S, Canevini MP, Zapperi S and La Porta CAM (2021) Information Optimized Multilayer Network Representation of High Density Electroencephalogram Recordings. Front. Netw. Physiol. 1:746118. doi: 10.3389/fnetp.2021.746118https://www.frontiersin.org/articles/10.3389/fnetp.2021.746118/full

Mechanical metamaterials actuators achieve pre-determined input-output operations exploiting architectural features encoded within a single 3D printed element, thus removing the need of assembling different structural components. Actuators are currently used in many applications from machines to robots. A team of researchers from the Center for Complexity and Biosystems of the University of Milan used artificial intelligence (AI) to design metamaterial actuators with high efficiency, surpassing traditional human design. Their work was published in Nature Communications. Designing materials with tailored mechanical properties and functionality still remains a great scientific and technological challenge, with huge potential for engineering applications and societal gains. A revolutionary approach of recent years has been to focus on materials that are structured on the macroscale producing mechanical metamaterials, a novel class of artificial materials engineered to have exceptional properties and responses that are difficult to find in conventional materials. These peculiar properties find natural applications in industrial design, as architectural motifs or reinforcement patterns for textiles, beams and other structures. The increased focus on metamaterials is stimulated by the recent advances in digital manufacturing technologies, such as 3D printing, which enable an easier design of such material structures with the removal of many of the constraints in scale and geometry at low  cost. In metamaterials actuators constituent cells work together in a well-defined order to obtain a desired macroscopic movement. Current design strategies for metamaterial structures and machines are essentially based on manual operations, but CC&B researchers showed that AI can improve considerably the design process. “We combined different computational algorithms to be able to optimize the response of the actuator and then compared the efficiency  with human designed actuators. Machine design always beats human design!” explains Silvia Bonfanti, postdoctoral fellow at the Department of Physics and first author of the paper, “We have also tested that the predicted structures could be 3D printed and confirmed their high efficiency” adds Roberto Guerra, researcher at the Department of Physics and co-author of the paper. “The algorithm we devised has  practical applications for human-machine interactions as interactive/responsive components and we are currently pursuing a commercialization strategy thanks to the support of the European Research Council” concludes Stefano Zapperi, professor of theoretical physics of matter  the Department of Physics, who coordinated the study. The work is supported by the proof of concept grant METADESIGN from the European Research Council. Link to the paper:

Prof. Zapperi is spending the current academic year in Germany as a visiting professor at FAU Erlangen-Nürnberg and LMU München supported by the Alexander von Humboldt Foundation thanks to the Humboldt Research Award. He is hosted in Fürth by Prof. Zaiser at the FAU Department of Materials Science and in München by Prof. Frey at the LMU Department of Physics and the Arnold Sommerfeld Center for Theoretical Physics. Read below the news release from LMU and an interview from FAU.

For some, making friends isn’t easy. Being social isn’t easy. Some of us are more introverted than extroverted. Myself, I have driven both Uber and Lyft, and I’ve found that one suited me better because I tend to be more introverted.

For some, making friends isn’t easy. Being social isn’t easy. Some of us are more introverted than extroverted. Myself, I have driven both Uber and Lyft, and I’ve found that one suited me better because I tend to be more introverted.

For some, making friends isn’t easy. Being social isn’t easy. Some of us are more introverted than extroverted. Myself, I have driven both Uber and Lyft, and I’ve found that one suited me better because I tend to be more introverted.

For some, making friends isn’t easy. Being social isn’t easy. Some of us are more introverted than extroverted. Myself, I have driven both Uber and Lyft, and I’ve found that one suited me better because I tend to be more introverted.

For some, making friends isn’t easy. Being social isn’t easy. Some of us are more introverted than extroverted. Myself, I have driven both Uber and Lyft, and I’ve found that one suited me better because I tend to be more introverted.

For some, making friends isn’t easy. Being social isn’t easy. Some of us are more introverted than extroverted. Myself, I have driven both Uber and Lyft, and I’ve found that one suited me better because I tend to be more introverted.