A quantum computer that uses giant atoms can mimic the human brain

A quantum computer that uses giant atoms can mimic the human brain

Because of the quantum properties of matter, in particular superpositions and complexity, quantum computers promise to offer unprecedented computing opportunities that will solve serious problems in several areas of fundamental research. However, these computers are still very sensitive to external disturbances and are still NEXQ.

In 2018, American physicist John Prescill explained in Quantum magazine that we had entered the NISQ era.

This restriction prevents the effective use of quantum machines for quantum machine learning. Therefore, research is aimed at developing quantum valves that are increasingly accurate and resistant to disturbance. Recently, quantum machine learning models inspired by brain dynamics have emerged as a way to circumvent the hardware limitations of NEXQ devices. In this context, the Harvard University team shows that a quantum computer built from giant atoms can theoretically simulate certain brain functions.

Atom-based quantum neural network

The giant atoms considered are the so-called Ridberg atoms, the atoms in the excruciated state, in which case the ruby atoms, whose main quantum number is very large; their huge electronic orbits include large dipole-dipole interactions that allow confusion; these atoms are large because some of their electrons rotate away from the core; they are very light-sensitive and can be controlled by lasers; they are often used in experiments of quantum decoherence, a theory that tries to link quantum physics rules to classical physics.

Using computer simulations, Rodrigo Araz Bravo and his colleagues at Harvard University have shown that these atoms can be used to build a new type of quantum computer, in particular, they have discovered that Reedberg's six atoms can be manipulated.

In this quantum system, neural networks are more complex than classical computers, so in theory they can perform even more complex tasks in less time.

Challenges to memory and decision-making

In particular, Bravo and his team have demonstrated that their quantum RNN is capable of replicating several cognitive tasks, such as multi-tasking, decision-making and long-term memory. In order to come to this conclusion, they have modelled the bombing of ruby atoms with two different laser pulses, and then trained the neural network to select a more intensive pulse to develop its decision-making capacity.

In order to remember, the team repeated the same simulation, but between the two laser pulses there was a delay of one tenth microsecond. In other words, qRNN had to learn to remember the first pulse as soon as it received the second. Researchers note that in regular RNNs such as decision-making and working memory, all neurons require a connection. Since the connection is limited by physical limitations, they chose a special architecture to prevent the isolation of neurons from each other.

Our brain is probably the most efficient "machine" in terms of information processing and energy consumption. Emulation of some of its capabilities using quantum systems would be a breakthrough. Bravo and his colleagues are already working on creating this computer.