- AI model predicts whether a given quantum machine will have any quantum advantage.
- It is based on a neural network that analyzes the network structure of a quantum system and gradually learns to predict its behavior.
- This will help scientists develop new efficient quantum devices.
Quantum computing has the potential to solve various complex problems that today’s computers can’t even process. For example, it can help scientists study chemical reactions in detail and detect stable molecular structures for pharmaceutics and other fields.
However, one of the key issues in both classical and quantum computer science is computational speedup. Although quantum computers can perform way faster than classical computers, developing such machines would require a lot of time and money. Even then, no one can guarantee that these machines will exhibit quantum advantages.
Recently, a research team at Moscow Institute of Physics and Technology, ITMO University, and Valiev Institute of Physics and Technology developed a new tool that predicts whether a given quantum machine will have any quantum advantage.
This new tool is based on a neural network that analyzes the network structure of a quantum system and gradually learns to predict its behavior. It will help scientists develop new efficient quantum devices.
AI Pinpoints Candidates To Build Quantum Computers
Quantum walks have been employed in recent years to efficiently process quantum information. They are quantum counterparts of classical random walks. You can visualize this phenomenon as a particle traveling in a specific network that underlies a quantum circuit.
Unlike a classical walker’s state, the state of the quantum walker can be a coherent superposition of several positions. A device will have a quantum advantage if a particle in the device circuit exhibits quantum walk (from one network node to another) faster than its classical counterpart.
In this study, researchers used a machine learning model to identify such superior networks. The model distinguishes between networks and gradually learns to predict whether a given network will deliver any quantum advantage. This gives us the networks that can be utilized to develop an efficient quantum computer.
Illustration of AI looking out for quantum advantages
Training examples were generated by simulating the random walk dynamics of both classical and quantum particles. Each training example contained an adjacency matrix and a corresponding label (‘classical’ or ‘quantum’).
The research team also built a tool to simplify the development of computational circuits based on quantum algorithms. It could be used for conducting research in material science and biophotonics.
Quantum walks will provide a simple way (a lot simpler than architectures based on qubits and gates) to implement quantum calculations of natural phenomena. For example, they have the potential to precisely describe the excitation of photosensitive proteins like chlorophyll or rhodopsin.
Since protein is a complex biomolecule with a structure similar to that of a network, determining the quantum walk time from one network node to another may reveal what actually goes on within a molecule: where the electron will move and what sort of excitation it will cause.