Gravity Data Inversion by Adiabatic Quantum Computing

A quantum-enhanced implementation of the binary inversion method for gravity data acquisition is discussed. The subsurface structure of a single density anomaly with an assigned density contrast is calculated by using a D-Wave adiabatic quantum computer. In particular, an iterative heuristic based on quantum annealing that recovers a sharp shape of the subsurface anomaly is developed. Such a task is accomplished by collecting partial images obtained by quantum annealing processes for optimal Lagrange penalty coefficients. The results are compared with those obtained according to the same cost function minimized via genetic algorithms by conventional hardware on a realistic 2D dataset. The outcomes of this work are promising as the reconstructed model is obtained in tenths of iterations instead of the hundreds required in conventional methods. Moreover, for the part of the computation that resides in the quantum processing unit, the computational cost of the single quantum annealing descent is constant with respect to the number of degrees of freedom of the subsurface grid. The implemented method is likely to reveal its full potential on forthcoming quantum annealing devices, outperforming existing techniques. The gravity inversion method is used to map the subsurface density distributions starting from surface measurements. The article investigates a quantum-enhanced binary inversion formulation, addressing a novel application of quantum annealing to geophysics. The algorithm is tested in two different realistic use cases. When non-convexity of the associated optimization problem prevails, a hybrid classical-quantum approach provides an improved subsurface reconstruction.image