SPICE Workshop on Quantum Spinoptics, June 18th - June 20th 2024
Jonathan Keeling
I will discuss the prospects, challenges, and open questions associated with using cold atoms in a multimode (confocal) cavity to realise associative memory [1] and spin glass [2,3] phases.
The associative memory is realised by a confocal cavity QED neural network, with the cavity modes serving as synapses, connecting a network of superradiant atomic spin ensembles, which serve as the neurons. Memories are encoded in the connectivity matrix between the spin ensembles and can be accessed through the input and output of patterns of light. I will discuss how this particular physical system introduces specific forms of the connectivity matrix, and particular forms of dynamics of the system. Remarkably, we find these dynamics perform better than widely-used approaches for the Hopfield model.
The same system can also realise a spin glass, when the distribution of atom positions is large enough such that the connectivity matrix becomes uncorrelated. I will discuss how replica symmetry breaking can be observed in this glassy phase, both using theoretical quantum trajectory calculations [2] and in experiments [3].
[1] Enhancing Associative Memory Recall and Storage Capacity Using Confocal Cavity QED. B. P. Marsh, Y. Guo, R. M. Kroeze, S. Gopalakrishnan, S. Ganguli, J. Keeling, and B. L. Lev. Phys. Rev. X 11 021048 (2021)[2] Entanglement and Replica Symmetry Breaking in a Driven-Dissipative Quantum Spin Glass. B. P. Marsh, R. M. Kroeze, S. Ganguli, S. Gopalakrishnan, J. Keeling, and B. L. Lev. Phys. Rev. X 14 011026 (2024)
[3] Replica symmetry breaking in a quantum-optical vector spin glass. R. M. Kroeze, B. P. Marsh, D. Atri Schuller, H. S. Hunt, S. Gopalakrishnan, J. Keeling, B. L. Lev.arXiv:2311.04216