Magic Dust Computing – New Kind of Computing
Russian professor Natalia Berloff, now at the University of Cambridge in the United Kingdom, has a history of important discoveries, such as quasiparticles that make quantum mechanics visible to the naked eye, as well as helping to elucidate the mystery of unidentified electron objects.
But a few years ago when she and her colleagues at the Skolkovo Institute of Science and Technology in Russia tried to demonstrate that quantum particles could be used to create a new type of computing, scientific journals found the revolutionary idea too much and simply refused to publish his ideas, until then only theoretical.
One reviewer said, ‘Who would be crazy enough to try to implement this ?!’ So we had to do it ourselves, and now we prove our proposal with experimental data, “she said.
Finding the Global Minimum
Berloff proposes to do computing using what she calls “magic dust,” quasiparticles that combine light and matter, known as polaritons, and which is from an emerging technology called plasmonics.
The idea is to use the polaritons, so they fill empty spaces pointing directly to the simplest solution to the more complex problems. Predictions indicate that this type of computing will outperform any current supercomputer and even future quantum computers in speed and complexity.
Virtually all the computing we do today-from modeling protein folding to financial market behavior, designing new materials and sending fully automated missions to deep space-depends on our ability to find the ideal solution for formulation math of a problem: the absolute minimum number of steps it takes to solve this problem.
What Professor Berloff realized is that the search for an ideal mathematical solution is analogous to the search for the lowest point on a mountainous terrain with many valleys, moats, and wells. A wanderer may descend a hill and think that he has reached the lowest point of the whole landscape, but there may be a deeper valley just behind the next mountain.
If this search already seems a bit scary on a natural terrain – “Who would be crazy enough to try this?”, Remember? – imagine complexity when you consider a space of several dimensions. “This is exactly the problem we have to deal with when the function we want to minimize is a real-life problem with many unknowns, parameters, and constraints,” explained Berloff.
Even a quantum computer, when constructed, will at best offer the quadratic acceleration for the quest for the global minimum through a “brute force” approach. Modern supercomputers can only deal with a small subset of these problems when the size of the function to be minimized is small or when the underlying structure of the problem allows finding the ideal solution quickly even for a large dimensionality function.
Computing with magic dust
What if Berloff and his colleagues faced this problem from an unexpected angle: What if, instead of walking all over the mountainous terrain in search of the lowest point, you fill the landscape with a magical dust that has the property of only shining at the deepest level? Soon, the solution will appear on time very quickly.
They created the magic dust of polaritons firing a laser over stacked layers of selected atoms, such as gallium, arsenic, indium, and aluminum. The electrons in these layers emit and absorb light of a specific color. The mass of the polaritons is ten thousand times smaller than that of the electrons and can reach enough densities to form a state of matter known as Bose-Einstein condensate. In this condensate the quantum phases of the polaritons synchronize, creating a single macroscopic quantum object that can be detected by photoluminescence measurements.
The essential step then was missing: to create a landscape that corresponds to the function to be minimized and to force the magic dust to condense at its lowest point.
To do this, the group focused on a particular type of optimization problem, but one that is generic enough that any other difficult problem can be simulated. This problem is called the minimization of the XY model, which is one of the most fundamental models of statistical mechanics.
The team then created polaritons at the vertices of an arbitrary graph. As the polaritons condense, their quantum phases are arranged in a configuration that corresponds to the absolute minimum of the function.
With this practical demonstration, the reviewer thought the team was not that crazy, and the scientific article was taken seriously and published.
“We are beginning to explore the potential of polariton graphs to solve complex problems,” said Professor Pavlos Lagoudakis, co-author of the paper and responsible for the experiments. “We are now expanding our device to hundreds of nodes as we test its fundamental computing power. The ultimate goal is to build a quantum simulator in a microchip that works under ambient conditions.”