Smart materials that are actually smart 1 . Combinatorial design of textured mechanical metamaterials

As I type this commentary on my computer, I am reminded that the revolutionary advances in electronics over the past sixty years has made computing ubiquitous in our lives. Yet, the desktop computer on which I am typing can draw up to 200 Watts of power while the laptop I will probably switch to as I continue writing later will draw 60 Watts of power. In contrast, the human brain, which is not too shabby as a computing device, uses only an estimated 20 Watts of power [1]. Beyond this, there is a growing sense that many of the computers of the future will be distributed throughout our environment as soft, flexible, wearable devices and in the form of smart materials. Unlike the smart materials of today, future smart materials will be “smart” in the sense that they will be autonomously sense their environment, decide how to respond, and actuate to perform a task according to the environmental cues. As examples, one might think of clotting agents that detect and heal pipeline ruptures, or health and infrastructure monitoring devices. To remain flexible and reduce weight, these future materials must perform under low or even intermittent power, perhaps even harnessing ambient noise, rather than drawing on bulky batteries. On the other hand, the types of processing that is required need not be nearly as complex as what is required to compile a LTEXdocument, suggesting that, perhaps, smart materials need not perform their computations at the speed or complexity we might desire from our cell phones. What would a soft computer look like? One possibility is based on the self-assembly of DNA tiles with specifically-tailored interactions in some kind of “smart soup” [2]. Yet, even as far back as the earlier 19th century, Charles Babbage proposed another computing paradigm, the difference engine [3]. The difference engine was a hypothetic computing device (since built and demonstrated [3]) which, through the turning of gears and motion of rigid other elements, performs computations entirely mechanically. Of course, this device is not soft, is not small, and essentially reproduces the kind of computation already performed