Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
QI: Fachverband Quanteninformation
QI 18: Poster II
QI 18.19: Poster
Mittwoch, 20. März 2024, 11:00–14:30, Poster A
Learning the Ground Energy Profile of H2 Using Variational Quantum Circuits — Sergio Cotrino and •Carlos Viviescas — Departamento de Física, Universidad Nacional de Colombia, Bogotá, Colombia
Leveraging Machine Learning techniques with quantum data enables both information processing and learning on quantum systems. We applied Meta-Variational Quantum Eigensolver (meta-VQE) to learn a molecule’s ground energy profile using a set of training points. We trained an ansatz quantum circuit using a non-linear Gaussian encoding for circuit parameters, with the interatomic distance as a free variable. This approach delivers a reliable approximation of the ground energy across a specific interatomic distance range. Furthermore, it generates effective initial parameters for standard VQE training, yielding superior results (opt-meta-VQE). We implemented Meta-VQE using both analytic noise-free simulations and 10,000-shots simulations in PennyLane’s quantum computing framework. The analytic simulation accurately models the potential energy surface for an H2 molecule within chemical accuracy, employing a hardware-inspired ansatz and the ADAM optimizer. The 10,000-shots simulation approximates the energy profile but is less precise due to sample variability. Meta-VQE introduces an innovative method for information extraction and production by learning from quantum data within variational quantum circuits.
Keywords: quantum computing; quantum chemistry; quantum circuits; Variational Quantum Eigensolver