Evolutionary algorithm to design high-cooperativity optical cavities

Abstract

Using an evolutionary algorithm combined with a gradient descent (GD) method we design optical cavities with significantly enhanced strong coupling rates between cavity photons and a single quantum emitter. Our approach allows us to find specially designed non-spherical mirrors which lead to high-finesse cavity eigenmodes with large field enhancement at the center of the cavity. The method is based on adding consecutive perturbations to an initial spherical mirror shape using the GD method for optimization. We present mirror profiles which demonstrate higher cavity cooperativity than any spherical cavity of the same size. Finally, we demonstrate numerically how such a cavity enhances the operation frequency and purity of coupling a Ca+ ion to an optical fiber photon.

Convolutional neural networks for mode on-demand high finesse optical resonator design

“…We consider cavities with cylindrical rotational symmetry consisting of two identical mirrors with a spherical profile perturbed by a harmonic modulation given by By limiting the ranges of the period P and the amplitude of the perturbation A , we can choose parameters which are achievable by modern fabrication tools, see “ Fabrication considerations ”. While choosing such a 2-parameter family of mirror shapes may look restrictive, we have shown in previous work 19 , 20 , 31 that already very simple mirror deviation shapes from spherical can significantly enhance the cooperativity by generating eigenmodes that are superpositions of several Laguerre-Gaussian modes. We will show below that we can also achieve fairly complex mode shapes within the parameter family of Eq.…”

“…Traditionally, only cavities with spherical mirrors supporting Laguerre–Gaussian modes have been considered for these applications, but they provide only a limited design space that may not fulfill the requirements of specific applications. Allowing for non-spherical cavity mirrors provides a much wider range of design options and can generate cavity eigenmodes that have superior properties compared to standard Laguerre-Gaussian modes 19 , 20 . However, the large design space then makes inverse design problems very difficult to solve.…”

“…However, the large design space then makes inverse design problems very difficult to solve. In our previous work 20 we exploited a gradient descent method for this task. Because of the frequent occurrence of local minima in the design optimization, we had to combine the gradient descent method with an evolutionary algorithm, which required a large number of forward-design calculations.…”

Retrieving space-dependent polarization transformations via near-optimal quantum process tomography

“…In the first case, we devise a genetic search of the minimum of the cost function, emulating the principle of Darwinian natural selection [36][37][38][39]. GAs have been largely employed in the contexts of quantum simulation [40], quantum annealing [41], as well as for diverse optical applications [42][43][44][45][46][47][48][49][50]. The second approach relies on training a neural network (NN) to predict the process generating a limited number of experimental outcomes.…”