A regression algorithm for accelerated lattice QCD that exploits sparse inference on the D-Wave quantum annealer.

Area of Science:

• Quantum Computing
• Machine Learning
• Computational Physics

Background:

• Sparse inference is crucial for analyzing complex datasets.
• Quantum annealers offer potential for solving challenging optimization problems.

Purpose of the Study:

• To propose and evaluate a novel regression algorithm leveraging quantum annealing for sparse inference.
• To demonstrate the algorithm's performance on lattice quantum chromodynamics (LQC) simulation data.

Main Methods:

• A regression algorithm was developed using a learned dictionary optimized for sparse inference.
• Independent and dependent variables were concatenated and encoded into a dictionary for sparse reconstruction.
• A D-Wave quantum annealer was used to solve the non-convex sparse coding optimization problem.
• The algorithm was tested on LQC data, initializing the dependent variable to its average value.

Main Results:

• The quantum regression algorithm successfully shifted the dependent variable closer to its true value.
• Good prediction performance was achieved on the LQC simulation data.
• Testing with 20 to 64 logical qubits showed that increased qubit count improved prediction accuracy.

Conclusions:

• Quantum annealing provides a viable approach for solving sparse coding optimization in regression tasks.
• The proposed algorithm demonstrates effective prediction capabilities for scientific simulations.
• Scaling results suggest that larger quantum annealers will yield enhanced accuracy.