AI-enhanced tuning of quantum dot Hamiltonians toward Majorana modes

A neural network-based model has been developed to autotune quantum dot simulators for achieving Majorana modes, leveraging unsupervised learning on synthetic conductance maps. This approach utilizes a deep vision-transformer architecture trained with a physics-informed loss to correlate Hamiltonian parameters with conductance structures, allowing for efficient parameter updates that drive the system toward a topological phase. The methodology enables iterative tuning, significantly expanding the reachable parameter space for generating nontrivial zero modes, which is crucial for advancing quantum computing applications.