Conditioned quantum-assisted deep generative surrogate for particle-binary vector indicating thecalorimeter...

Area of Science:

• High Energy Physics
• Computational Physics
• Quantum Computing

Background:

• Particle collision simulations are crucial for High Luminosity Large Hadron Collider (HL-LHC) experiments like ATLAS and CMS.
• Current simulation methods, such as Geant4, incur substantial computational costs, requiring millions of CPU-years for the HL-LHC.
• Calorimeter simulations are particularly computationally intensive.

Purpose of the Study:

• To develop a computationally efficient method for simulating particle collisions.
• To reduce the significant computational burden of High Luminosity Large Hadron Collider (HL-LHC) simulations.
• To explore the application of quantum computing in high-energy physics simulations.

Main Methods:

• A conditioned quantum-assisted generative model was proposed, integrating a conditioned variational autoencoder (VAE) and a conditioned restricted Boltzmann machine (RBM).
• The RBM architecture was optimized for D-Wave's Pegasus quantum annealer, utilizing flux bias for conditioning.
• An adaptive method for estimating effective inverse temperature was introduced.

Main Results:

• The proposed model combines the universal approximation capabilities of classical RBMs with the speed and scalability of quantum annealing.
• The framework was validated using Dataset 2 of the CaloChallenge, demonstrating its effectiveness.
• This approach offers a significant reduction in simulation time and computational resources.

Conclusions:

• The conditioned quantum-assisted generative model presents a viable solution to the computational challenges in high-energy physics simulations.
• This hybrid quantum-classical approach shows promise for accelerating scientific discovery at future colliders.
• Further development and application of this method could revolutionize experimental design and data analysis in particle physics.