System Bits: Dec. 8

University of Michigan researchers can now prove a material that has been a puzzle for decades could open a new path to quantum computing; a new system by MIT researchers allows pattern-recognition systems to convey what they learn to humans.

popularity

Path to quantum transistors
An odd, iridescent material that’s puzzled physicists for decades turns out to be an exotic state of matter that could open a new path to quantum computers and other next-generation electronics, according to University of Michigan physicists.

The researchers have been able to confirm that several properties of the compound samarium hexaboride that raise hopes for finding the silicon of the quantum era, and believe their results are also close the case of how to classify the material—a mystery that has been investigated since the late 1960s.

The researchers provide the first direct evidence that samarium hexaboride, abbreviated SmB6, is a topological insulator. Topological insulators are, to physicists, an exciting class of solids that conduct electricity like a metal across their surface, but block the flow of current like rubber through their interior. They behave in this two-faced way despite that their chemical composition is the same throughout.

The team explained that they used a technique called torque magnetometry to observe tell-tale oscillations in the material’s response to a magnetic field that reveal how electric current moves through it. Their technique also showed that the surface of samarium hexaboride holds rare Dirac electrons, particles with the potential to help researchers overcome one of the biggest hurdles in quantum computing.

These properties are particularly enticing to scientists because SmB6 is considered a strongly correlated material. Its electrons interact more closely with one another than most solids. This helps its interior maintain electricity-blocking behavior.

Samarium hexaboride, abbreviated SmB6, is a compound made of the metal samarium and the rare metalloid boron. University of Michigan researchers have confirmed its unusual electrical properties and shown how it could advance the development of next-generation transistors for quantum computers. (Source: University of Michigan)

Samarium hexaboride, abbreviated SmB6, is a compound made of the metal samarium and the rare metalloid boron. University of Michigan researchers have confirmed its unusual electrical properties and shown how it could advance the development of next-generation transistors for quantum computers. (Source: University of Michigan)

This deeper understanding of samarium hexaboride raises the possibility that engineers might one day route the flow of electric current in quantum computers like they do on silicon in conventional electronics.

Teaching by example
While computers are good at identifying patterns in huge data sets, humans are good at inferring patterns from just a few examples. MIT researchers have created a new system that bridges these two ways of processing information, so that humans and computers can collaborate to make better decisions.

The system learns to make judgments by crunching data but distills what it learns into simple examples. In experiments, human subjects using the system were more than 20 percent better at classification tasks than those using a similar system based on existing algorithms.

The team was looking at whether they could augment a machine-learning technique so that it supported people in performing recognition-primed decision-making, which is the type of decision-making people do when they make tactical decisions. When presented with a new scenario, people don’t do search the way machines do —they try to match their current scenario with examples from their previous experience, and then think that worked in a previous scenario, and they adapt it to the new scenario.

A new machine-learning algorithm clusters data according to both a small number of shared features (circled in blue) and similarity to a representative example (far right). (Source: MIT)

A new machine-learning algorithm clusters data according to both a small number of shared features (circled in blue) and similarity to a representative example (far right). (Source: MIT)