Brain-Inspired Computer Made From Duroquinone

hasu notes that scientists at the National Institute for Materials Science at Tsukuba in Japan have created a device, consisting of 17 duroquinone molecules on a gold surface, that can in theory encode 4.3 billion outcomes. The "device" does not constitute a practical computer, since it requires both a scanning tunneling microscope and operation near absolute zero. A single duroquinone is surrounded by sixteen others, and weak chemical bonds allow a pulse to the central molecule to shift all seventeen molecules in a variety of ways. Each duroquinone has four different "settings," so a single pulse can have 4^16 possible outcomes. As a demonstration the researchers docked 8 other nano-devices to their 17-molecule computer. It is unclear how well they have characterized the inputs that result in 4.3 billion different outputs. They are working on a 3D design that would have 1,024 duroquinone molecules surrounding a central one.

Comments

Re:Elaboration Please

[psychicninja]

Could someone tell me what they mean by "operation near absolute zero."?

It means bring your coat...

Wow, 4.3 billion states?

[Kaz Kylheku]

That's like, 2^32.

Is this really a computer? Or 32 bits worth of really impractical memory? :)

Re:Wow, 4.3 billion states?

[JLF65]

:)

You noticed that as well. I replied as much to a post above. Scientists have ways of making their "discoveries" seem much more impressive than they really are. It helps keep the grant money coming in. After all, which sounds more impressive?

"I've made a 32 bit register that requires a room-size microscope and refrigerator to operate."

"I've made a molecular 'brain' that holds over FOUR BILLION states!"

Well...

[jd]

The world's first stored-program computer only stored 32 words, where one word was 40 bits in length, making this 1/40th of the capacity of Alan Turing's "Baby" (aka Manchester Mk. 1) computer. Seriously, though, this is impressive in the sense that they got the thing to work at all. Storing and recovering data from a device this small is non-trivial, especially if they've got the read to be non-destructive. At this scale, the impact of carrying out the observation is non-trivial. If they need to cool to near absolute zero, it's obviously delicate enough that they need to damp down everything to keep the system working. But precisely because almost anything can be kept constant at that temperature, I'd consider this "cheating" a little. You could probably store and recover data on almost any sufficiently uniform structure if nothing is moving.