Analyseur de spectre | EPFL Graph Search

Un analyseur de spectre est un instrument de mesure destiné a afficher un signal dans le domaine frequentiel contrairement à un oscilloscope qui affiche le signal dans le domaine temporel. Les signaux peuvent être de natures diverses : électrique, optique, sonore, radioélectrique. Classification Classification par type de mesure Analyseur de spectre électrique Un analyseur de spectre électrique permet la mesure de la tension de signaux électriques dans le domaine fréquentiel. Les mesures peuvent aller de quelques dixièmes de Hz à plusieurs centaines de GHz. Analyseur de réseau Un analyseur de réseau est un analyseur de spectre électrique équipé d'un générateur de signaux, permettant ainsi l'analyse fréquentielle des lignes de transmission. Cependant il ne permet pas d'analyser les réseaux sans fils. Analyseur de champ électromagnétique Un analyseur de champ électromagnétique permet la mesure des champs magnétique et él

Hierarchical Micromold Structuring via 2PP Micro-Stereolithography

Ho-Yun Lee

Inspired by nature, a common way to create hydrophobic surfaces is by introducing hierarchical and fractal microstructures on surface. While plenty of studies have demonstratedcreating PDC (polymer derived ceramic) for several applications include SiC-fiber for composite materials and cantilever in ceramic MEMs, no research from the author’s knowledge is devoted to creating hydrophobic PDC surface by fabricate micro size array-like structure on PDC. Therefore, fabricate hydrophobic PDC surface is the main goal of the project. To achieve the idea, a novel approach is introduced and presented in the report. The approach combines micro molding of PDC and 3D printing hydrophobic surface. By exploiting micro-stereolithography base on Two Photon Polymerization, the project first printed various 3D structure including fractal Sierpinski tetrahedron, pyramid, biomimic arrays and QR codes. The printed photoresists show very well define structures by using OM characterization and the hydrophobicity of structure were verified by drop shape analyser. The designed micro molds were printed on the KOH etched mold for PDC molding. In the end, using this novel method, first ever micro-featured PDC were successfully replicated. Although the hydrophobicity for pyramid arrays on PDC is not yet verified, several improvements are provided to optimize future experiment.

2020

Effects of resonator geometry and substrate stiffness on the tunability of a deformable microwave metasurface

Romain Christophe Rémy Fleury, Taha Goudarzi, Rayehe Karimi Mahabadi

Tunable metasurfaces can shift their resonant frequency through different approaches, one of which is applying mechanical deformations. Here, we show the effects of two key factors on the tunability of deformable metasurfaces; the resonator geometry and substrate stiffness. To show the effects, we compared the tunability of unit cells with three resonator geometries and three common substrates at microwave frequencies from 1 GHz to 10 GHz under a given mechanical deformation. We showed that the resonator geometry affects the deformation field, as a consequence, causes different resonant frequency shifts. Moreover, it affects the stress field in the metasurface which in turn limits the maximum allowable deformation before mechanical failure. In addition, increasing the substrate stiffness has controversial effects on the tunability of the metasurface. It can either increase or decrease the tuning spectrum by transferring more deformation to the resonators and constraining the rigid body motions. On the other hand, increasing the substrate stiffness increases the force required for tuning which is a limiting factor for applications. Accordingly, a trade-off exists between the desired tuning spectrum and the required amount of force for changing the metasurface response. Both of these factors limit the functionality of tunable metasurfaces.

2022

Rigorous optimization recipes for sparse and low rank inverse problems with applications in data sciences

Anastasios Kyrillidis

Many natural and man-made signals can be described as having a few degrees of freedom relative to their size due to natural parameterizations or constraints; examples include bandlimited signals, collections of signals observed from multiple viewpoints in a network-of-sensors, and per-flow traffic measurements of the Internet. Low-dimensional models (LDMs) mathematically capture the inherent structure of such signals via combinatorial and geometric data models, such as sparsity, unions-of-subspaces, low-rankness, manifolds, and mixtures of factor analyzers, and are emerging to revolutionize the way we treat inverse problems (e.g., signal recovery, parameter estimation, or structure learning) from dimensionality-reduced or incomplete data. Assuming our problem resides in a LDM space, in this thesis we investigate how to integrate such models in convex and non-convex optimization algorithms for significant gains in computational complexity. We mostly focus on two LDMs:

(i)

sparsity and

(ii)

low-rankness. We study trade-offs and their implications to develop efficient and provable optimization algorithms, and--more importantly--to exploit convex and combinatorial optimization that can enable cross-pollination of decades of research in both.

EPFL2014

Rigorous optimization recipes for sparse and low rank inverse problems with applications in data sciences

Anastasios Kyrillidis

(i)

sparsity and

(ii)

EPFL2014