We report on the design and characterization of exact and heuristic algorithms to solve atom reconfiguration problems. These algorithms can be used to prepare deterministic configurations of atoms in two-dimensional arrays of optical traps, as well as to realize quantum many-body systems with dynamic connectivity graphs and time-varying interactions. We numerically quantify the operational performance of our algorithms using realistic experimental parameters. Our results indicate that implementing the redistribution-reconfiguration (red-rec) algorithm would enable the assembly of arrays of 256 and 1024 atoms using 512 and 2048 optical traps with a success probability of 91.3(2)$~\%$ and 21(1)$~\%$, respectively. Further rejecting configurations of atoms containing less atoms than a given threshold results in greater mean success probability.
We further report on the design and characterization of a low-latency reconfiguration system to perform feedback control experiments on atomic systems. The system exploits low-latency communication protocols among hardware devices and parallel processing on a CPU or a GPU to speed up the process of acquiring and processing images, generating control sequences, and synthesizing and streaming waveforms actuating active diffractive optical elements. We benchmark and optimize the computational runtime for different reconfiguration problem sizes, identifying the regime for which the GPU outperforms the CPU. This system can readily be deployed for real-time operation to actuate dynamic arrays of optical traps, realizing adaptive, variational, and error correction protocols, as well as synthesizing and streaming optimal control pulses.
|Presenter name||Alexandre Cooper-Roy|
|How will you attend ICAP-27?||I am planning on in-person attendance|