arXiv subjects

Matthew Filipovich

Publications and source records attributed to Matthew Filipovich.

3 recordsLinked to original sources

arXiv2025

Direct space-time modeling of mechanically dressed dipole-dipole interactions with electromagnetically-coupled oscillating dipoles

We study the radiative dynamics of coupled electric dipoles, modelled as Lorentz oscillators (LOs), in the presence of real-time mechanical oscillations. The dipoles are treated in a self-consistent way through a direct electromagnetic simulation approach that fully includes the dynamical movement of the charges, accounting for radiation reaction, emission and absorption. This allows for a powerful numerical solution of optomechanical resonances without any perturbative approximations for the mechanical motion. The scaled population (excitation) dynamics of the LOs are investigated as well as the emitted radiation and electromagnetic spectra, which demonstrates how the usual dipole-dipole resonances couple to the underlying Floquet states, yielding multiple spectral peaks that are separated from the superradiant and subradiant states by an integer number of the mechanical oscillation frequency. Moreover, we observe that when the mechanical amplitude and frequency are sufficiently large, these additional spectral peaks undergo further modification, including spectral splitting, spectral squeezing, or shifting. These observations are fully corroborated by a theoretical Floquet analysis conducted on two coupled harmonic oscillators.

quant-ph↗

arXiv2024

Streamlined optical training of large-scale modern deep learning architectures with direct feedback alignment

Modern deep learning relies nearly exclusively on dedicated electronic hardware accelerators. Photonic approaches, with low consumption and high operation speed, are increasingly considered for inference but, to date, remain mostly limited to relatively basic tasks. Simultaneously, the problem of training deep and complex neural networks, overwhelmingly performed through backpropagation, remains a significant limitation to the size and, consequently, the performance of current architectures and a major compute and energy bottleneck. Here, we experimentally implement a versatile and scalable training algorithm, called direct feedback alignment, on a hybrid electronic-photonic platform. An optical processing unit performs large-scale random matrix multiplications, which is the central operation of this algorithm, at speeds up to 1500 TeraOPS under 30 Watts of power. We perform optical training of modern deep learning architectures, including Transformers, with more than 1B parameters, and obtain good performances on language, vision, and diffusion-based generative tasks. We study the scaling of the training time, and demonstrate a potential advantage of our hybrid opto-electronic approach for ultra-deep and wide neural networks, thus opening a promising route to sustain the exponential growth of modern artificial intelligence beyond traditional von Neumann approaches.

cs.ET↗

arXiv2023

Consistent, multidimensional differential histogramming and summary statistics with YODA 2

Histogramming is often taken for granted, but the power and compactness of partially aggregated, multidimensional summary statistics, and their fundamental connection to differential and integral calculus make them formidable statistical objects, especially when very large data volumes are involved. But expressing these concepts robustly and efficiently in high-dimensional parameter spaces and for large data samples is a highly non-trivial challenge -- doubly so if the resulting library is to remain usable by scientists as opposed to software engineers. In this paper we summarise the core principles required for consistent generalised histogramming, and use them to motivate the design principles and implementation mechanics of the re-engineered YODA histogramming library, a key component of physics data-model comparison and statistical interpretation in collider physics.

hep-ph↗