Periodic Reporting for period 2 - InfoMatter (Pathways, Memory and Information Processing in Matter)
Reporting period: 2023-03-01 to 2024-08-31
In the project infomatter, researchers are investigating whether materials can store and process information in a completely different manner than our current digital computers. This work is motivated by the realization that many of the complex phenomena seen in real materials that have vexed researchers can be understood using tools from information theory. In particular, we use materials ranging from synthetic metamaterials to crumpled sheets of paper,
to investigate if their mechanical response can be seen as performing (unusual) computions: in short, can a crumple act as a computer?
Our work is shining new light on the poorly understood physics of many of the complex materials that surround us, from granular media such as sand to glasses and crumpled sheets of paper, which in the future
may improve our handing of such complex materials - for example, handling granular matter is very energy intensive, and a better understanding of its memory effects may contribute to making industrial processes more efficient.
However, we aim for having our biggest impact in computing. Current use of enery for computing already exceeds that of airtravel, and with the exponential growth of demand for computing for, e.g. machine learning and AI, it is of paramount
importance to find new, robust, and enery efficient ways to compute. Our work on using passive materials to process information, in a manner that has key aspects in common with how our energy efficient brains work, and
circumvents the classical problems of digital computing such as the von Neumann bottleneck, will open new avenaues towards the low CO2 footprint of the future.
to investigate if their mechanical response can be seen as performing (unusual) computions: in short, can a crumple act as a computer?
Our work is shining new light on the poorly understood physics of many of the complex materials that surround us, from granular media such as sand to glasses and crumpled sheets of paper, which in the future
may improve our handing of such complex materials - for example, handling granular matter is very energy intensive, and a better understanding of its memory effects may contribute to making industrial processes more efficient.
However, we aim for having our biggest impact in computing. Current use of enery for computing already exceeds that of airtravel, and with the exponential growth of demand for computing for, e.g. machine learning and AI, it is of paramount
importance to find new, robust, and enery efficient ways to compute. Our work on using passive materials to process information, in a manner that has key aspects in common with how our energy efficient brains work, and
circumvents the classical problems of digital computing such as the von Neumann bottleneck, will open new avenaues towards the low CO2 footprint of the future.
During the first half of this project, we realized several breakthroughs that together constitute the first steps along this path. First, we realized a so-called metamaterial (a carefully shaped piece of rubber whose functions are set by its shape, not its composition) that can count signals: repeatedly compressing this material, its shape (that can be seen by the naked eye) remebers how often it is compressed. Moreover, we can extend its functionality by combining multiple of such counting materials to act as password detection. Together, this work is one of the first demonstrators of a new form of material based computing, i.e. without the need for a central processor or external memory. Second, we started exploring the computations in materials that have more than one input - for example, materials than can be squeezed at multiple locations. We are investigating both controlled metamaterials, and disordered systems such as crumples, and find that both of these naturally store information and act as unusual computer: crumpled paper can thus compute.
Finally, we developed a general framework for designing materials with integrated material bits - so called hysterons. We showed how such materials can not only count signals, but in fact can perform any sequential (step-by-step) computing that digital computers do.
So far, we have demonstrated the principles and have realized small systems by experiment.
Finally, we developed a general framework for designing materials with integrated material bits - so called hysterons. We showed how such materials can not only count signals, but in fact can perform any sequential (step-by-step) computing that digital computers do.
So far, we have demonstrated the principles and have realized small systems by experiment.
The results outline above all constitute significants steps beyond the state of the art, and are inspiring work by other groups on similar strategies for compuintg in completely different classes of materials (chemical, electronic, etc).
Our aim is to fully develop the principles of in-materia computing based on hysterons, and to develop design methods for materials that can realize these. Moreover, we aim to understand how far we can push these ideas to understand
the behavior of naturally occuring disordered materials.
Our aim is to fully develop the principles of in-materia computing based on hysterons, and to develop design methods for materials that can realize these. Moreover, we aim to understand how far we can push these ideas to understand
the behavior of naturally occuring disordered materials.