Researchers present ‘Neural Error Mitigation’, a new method that uses neural networks to improve estimates of ground states and ground state observables obtained using VQE on short-term quantum computers


Quantum computers, which may be developed soon (in the short term), are promising and could be of great use in solving certain problems more efficiently than classical computers. Their application is in physics, chemistry and other sciences to determine the ground states of quantum systems.

Even if quantum computers perform quantum simulations efficiently, the effect of noise and limited hardware will limit these techniques. There is therefore a need to reduce the impact of noise through quantum error mitigation techniques, which the researchers have addressed.

The researchers developed a new technique called neural error mitigation that uses neural networks to improve estimates of ground states.

The researchers’ approach consists of two essential parts or phases. First, they trained a so-called ansatz NQS to represent a rough ground state created by a noisy quantum device using neural quantum state tomography (NQST). NQST is a machine learning technique that uses a small sample of empirically collected measurements to reconstruct complex quantum states. Next, they improved the current representation of the unidentified ground state using a variational Monte Carlo (VMC) approach. The Generative Machine Learning Model Transformer Architecture, which was often used to produce natural language writing and interpret images, was the NQS algorithm used by the researchers’ experiments.

The researchers tested the performance of this method on a real problem. They examined the method’s ability to recognize the wave function and ground state energy of Fermionic molecular Hamiltonians interacting with many bodies.

They used neural error mitigation to locate the ground states of H2 and LiH molecular Hamiltonians and the networked Schwinger model, created via the variational quantum solver, to show the wide range of applications of the method. Their findings demonstrate that neural error mitigation improves numerical and experimental computations of variational quantum eigensolvers to produce low-energy errors, high fidelities, and accurate estimates of more complex observables like order parameters and entanglement entropy without requiring additional quantum resources.

Future quantum simulations using short-term devices could benefit from using neural error simulation to reduce noise-related errors. This can have important ramifications for many fields of science, including chemistry, physics, and materials science, as it can lead to more accurate predictions or insightful discoveries.

This Article is written as a research summary article by Marktechpost Staff based on the research paper 'Neural Error Mitigation of Near-Term Quantum Simulations'. All Credit For This Research Goes To Researchers on This Project. Check out the paper and reference article.

Please Don't Forget To Join Our ML Subreddit

Prathvik is an ML/AI research content intern at MarktechPost, he is a 3rd year undergraduate at IIT Kharagpur. He has a keen interest in machine learning and data science. He is enthusiastic about learning more about the applications of in different fields of study.


About Author

Comments are closed.