The Physics of GLIMPSE

607 0

You know from A Floquet Connection that GLIMPSE stands for

Globally Linked Ising Machine Platform with Superconducting Electronics.

I have covered Ising Model and (its connection to) Integer Solutions previously.

I have mentioned Ising Machines several times, including in:

2020 Tayur Prize.

Globally Linked means All-to-All connectivity of the Spins in the Ising Model.

The Spins in GLIMPSE are mathematically equivalent to Parametric Oscillators. Parametric Oscillators were first proposed in 1831 by, wait for it, Michael Faraday, and published in – yes, you know it! – Philosophical Transactions of the Royal Society.☺️

The All-to-All connectivity is obtained not by actually connecting all the spins to each other directly. Instead, the spins are arranged on a line – connected to a resonator – and this common bus couples the spins – that is, it represents our optimization problem – by encoding the Ising parameters as the differences in the frequency of the various oscillators. This benefit of all-to-all connectivity with only a linear bus does not come free: there is an algorithmic effort that is a pre-step to get these oscillators dancing in harmony. This is called Floquet driving. GLIMPSE architecture – proposed by Onodera, Ng and McMahon at Cornell – trades-off physical complexity with algorithmic effort.

The Spins in GLIMPSE are physically realized via Josephson Parametric Oscillators.

Which brings us to Superconducting Electronics.

Superconducting was accidentally discovered in 1911 (by Onnes, who first liquified Helium, pretty much the most difficult element to be liquified, for which he received a Nobel Prize in 1913). It took until 1957 – after many false starts and failed attempts by the most capable Theoretical Physicists of that era – for Bardeen, Cooper and Schrieffer (BCS) to come up with a Quantum Mechanical explanation for Superconductivity (for which they received a Nobel Prize in 1972). In 1962, a graduate student (in his first year), Brian Jacobson, theorized that a junction made out of two superconductors separated by a thin insulator would create a current – following the logic of Quantum Mechanics and BCS Theory – without any external voltage: it was rejected not only by his thesis advisor who asked him to work on something else, but was also aggressively opposed by (the otherwise mild-mannered, soft spoken and laconic) Bardeen (who, in 1956, had won his first Nobel Prize, for Transistor). In 1963, a Bell Labs group (that included Philip Anderson, who was a well-respected Theoretical Physicist) actually built a Josephson Junction (JJ), and it behaved exactly as had been predicted! Brian Jacobson won a Nobel Prize in 1973. (For his many important theoretical contributions (unrelated to JJ), Philip Anderson was awarded a Nobel Prize in 1977).

The practical problem, however, was that superconductivity was only realized at very, very low temperatures, with valiant attempts, guided by BCS Theory and understanding of lattices and elements and compounds. Then in 1986, Georg Bednorz and Alex Mueller, from IBM (in Zurich) startled the world by exhibiting superconductivity at a much higher temperature, that cannot be explained by BCS Theory (and to date, has not found any theoretical explanation!), that led to a global race, leading to materials that we can use today (that still need cooling, but not as much cooling as previously); they received a Nobel Prize in 1987. (Yes. The Very Next Year. Not a typo!)😳

Instead of just creating one Josephson Junction (JJ), one can connect two JJs in a loop and form SQUID: Superconducting Quantum Interference Device. With SQUID, one can build a Josephson Parametric Oscillator (JPO). This is Superconducting Electronics.

Summarizing:

GLIMPSE is an All-to-All connected Ising machine, the connectivity accomplished by the Floquet driving of the various oscillators coupled through a common bus, with the oscillators themselves being made of superconducting electronics (JPO).

This is not a Turing Machine. (This is not Your Father’s Ising Machine either. ☺️)

Where is this being fabricated? At BBN and MIT Lincoln Labs.

Let me close this post with excerpts from the last chapter, Chapter 21, in The Feynman Lectures on Physics, Volume III, is A Seminar on Superconductivity, published in 1965 (the year Feynman won his Nobel Prize, for QED):

This lecture is only for entertainment…I don’t know what will come next…the transistor, the laser, and now these junctions, whose practical applications are still not known. We are really getting control of nature on a very delicate and beautiful level. I am sorry to say, gentlemen*, that to participate in this adventure it is absolutely imperative that you learn quantum mechanics as soon as possible.

And with his Epilogue:

The main purpose of my teaching has not been to prepare you for some examination…I wanted to give you some appreciation of the wonderful world…It is even possible that you may want to join in the greatest adventure that the human mind has ever begun.

(* Female undergraduates were first admitted at Caltech in 1970. Cornell was 1870. Yes, 100 years earlier.)

Leave a Reply