Reinforcement Learning in a Large Scale Photonic Network
Seminar by Daniel Brunner
from FEMTO-ST Institute/Optics Department, CNRS & University Bourgogne Franche-Comté, Besançon Cedex, France
We experimentally create a neural network via a spatial light modulator, implementing connections between 2025 in parallel based on diffractive coupling. We numerically validate the scheme for at least 34.000 photonic neurons. Based on a digital micro-mirror array we demonstrate photonic reinforcement learning and predict a chaotic time-series via our optical neural network. The prediction error efficiently converges. Finally, we give insight based on the first investigation of effects to be encountered in neural networks physically implemented in analogue substrates.
Time: 13 May 2019, 14:00
Fast volumetric light-sheet microscopy
Seminar by Omar E. Olarte
from Universidad ECCI, Colombia (and ICFO, Barcelona, Spain)
Light sheet fluorescence microscopy (LSFM) is a convenient tool for bio-imaging as it efficiently collects the generated fluorescence while at the same time minimizes photobleaching. For these reasons LSFM, being based on an intrinsic 2D illumination strategy, has been put forward as an interesting candidate for fast volumetric imaging.
We report on a LSFM microscope, combined with the use of wavefront coding (WFC) techniques, for fast volumetric imaging. This provides intrinsic 3D imaging capabilities as it extends the depth of field (DOF) of the microscope. In addition, because of the extended DOF, the light sheet can be axially scanned at fast speeds. As only the light sheet is moved, fast 3D imaging can be achieved without the need of any sample or objective movement. Since typical light scanning devices can run at KHz rates, 3D volumetric acquisition speeds will only be limited by the reading speed of the camera or the required signal to noise ratio.
WFC works at the expenses of introducing a controlled aberration to the system which blurs the resulting images. Then it requires a final deconvolution step to recover the image sharpness that would impose a limitation on the application. Here we present computational tools to perform real-time deconvolution and visualization of the images obtained with WFC-LSFM. To speed up such deconvolution processes, a routine directly developed in a GPUs has been developed. We present preliminary results on realistic computer generated images showing that real-time deconvolution and visualization is possible.
Time: 17 December 2018, 16:00
Effective Drifts in Generalized Langevin Systems
Seminar by Soon Hoe Lim
from Nordita, Stockholm, Sweden, EU
Generalized Langevin equations (GLEs) are stochastic integro-differential equations commonly used as models in non-equilibrium statistical mechanics to describe the dynamics of a particle coupled to a heat bath. From modeling point of view, it is often desirable to derive effective mathematical models, in the form of stochastic differential equations (SDEs), to capture the essential dynamics of the systems. In this talk, we consider effective SDEs describing the behavior of a large class of generalized Langevin systems in the limits when natural time scales become very small. It turns out that additional drift terms, called noise-induced drifts, appear in the effective SDEs. We discuss recent progress on the phenomena of noise-induced drift in these systems. This is joint work with Jan Wehr and Maciej Lowenstein.
Place: Soliden 3rd floor
Time: 12 December 2018, 13:00
Information Controlled Structure Formation in Artificial Microswimmer Systems
General Physics Colloquium by Frank Cichos, University of Leipzig, Germany
Abstract: Self-organization is the generation of order out of local interactions in non-equilibrium. It is deeply connected to all fields of science from physics, chemistry to biology where functional living structures self-assemble and constantly evolve all based on physical interactions. The emergence of collective animal behavior, of society or language are the result of self-organization processes as well though they involve abstract interactions arising from sensory inputs, information processing, storage and feedback resulting in collective behaviors as found, for example, in crowds of people, flocks of birds, schools of fish or swarms of bacteria.
We introduce such information based interactions to the behavior of self-thermophoretic microswimmers. A real-time feedback of swimmer positions is used as the information to control the swimming direction and speed of other swimmers. The emerging structures reveal frustrated geometries due to confinement to two dimensions. They diffuse like passive clusters of colloids, but posses internal dynamical degrees of freedom that are determined by the feedback delay and the noise in the system. As the information processing in the feedback loops can be designed almost arbitrarily complex systems with mixed feedback delays and noise will give rise to new emergent dynamics of the self-organized structures. The presented control schemes further allow the integration of machine learning algorithms to introduce an adaptive behavior of swimmers.
Place: PJ Lecture Hall
Ripples in Thin Films
Seminar by Mazi Jalaal
from the Physics of Fluids laboratory
at the University of Twente, the Netherlands, EU
We present experimental observations of capillary ripples at the contact line of a droplet, spreading on a pre-wetted surface.
We use Digital Holographic Microscopy to measure the micro-scale undulation of the thin film. By raising the capillary number, the amplitude of the undulations increases at first and subsequently decreases.
At critical values of the capillary number, the ripples disappear. Using linear stability analysis, we further provide theoretical counterparts for the experimental observations, explaining the non-monotonic dependency on the capillary number
Place: PJ Lecture Hall
Time: 9 October, 2018, 11:00
Cell differentiation and pattern formation in the transition to multicellularity: lessons from the microbial world
Seminar by Mariana Benitez Keinrad
from the Laboratorio Nacional de Ciencias de la Sostenibilidad,
Universidad Nacional Autónoma de México (UNAM), Mexico.
Multicellular development occurs in plants, animals and other lineages, and involves the complex interaction among biochemical, physical and ecological factors. Our group has focused on the study of microbial multicellular organisms, which have been considered useful models to study the evolutionary transition to multicelullarity. I present some of our theoretical and experimental work, and discuss the physical and chemical processes that, in coordination with molecular regulatory networks, appear to be relevant for cell differentiation, patterning and morphogenesis in microbial aggregates.
Place: Soliden 3rd floor
Time: 8 October, 2018, 12:15
Physics of active soft matter
General Physics Colloquium by Hartmut Löwen, Heinrich-Heine Universität Düsseldorf, Germany
Abstract: Ordinary materials are “passive” in the sense that their constituents are typically made by inert particles which are subjected to thermal fluctuations, internal interactions and external fields but do not move on their own. Living systems, like schools of fish, swarms of birds, pedestrians and swimming microbes are called “active matter” since they are composed of self-propelled constituents. Active matter is intrinsically in nonequilibrium and exhibits a plethora of novel phenomena as revealed by a recent combined effort
of statistical theory, computer simulation and real-space experiments. The colloquium talk provides an introduction into the physics of active matter focussing on biological and artificial microswimmers as key examples of active soft matter . A number of single-particle and collective phenomena in active matter will be adressed ranging from the circle swimming to inertial delay effects.
 For a review, see: C. Bechinger, R. di Leonardo, H. Löwen, C. Reichhardt, G. Volpe, G. Volpe, Active particles in complex and crowded environments, Reviews of Modern Physics 88, 045006 (2016).
Place: PJ Lecture Hall
Alejandro V. Arzola is a Visiting Professor from the Universidad Nacional Autónoma de México in Mexico City. His visiting position is financed through the Linnaeus Palme International Exchange Programme.
Alejandro was born in Oaxaca in the south of Mexico. He studied for a PhD at the Universidad Nacional Autónoma de México (UNAM) in Mexico City, worked as a posdoctoral researcher at the Institutte of Scientific Instruments in Brno, Czech Republic, and at UNAM. Since 2014 he joined the group of Optical Micromanipulation at the Institute of Physics in UNAM.
He is interested in optical micromanipulation and related research fields. His latest research deals with the transport of Brownian particles in optical landscapes under breaking space-time symmetries, a system which is known in the literature as ratchets. He is also interested in the behavior of microscopic particles in structured light fields with spin and orbital angular momentum.
Qiwei Li is a bachelor student at the University of Würzburg. He will do his summer internship at the Soft Matter Lab from July 17 to September 20, 2018, with a grant from DAAD (Deutscher Akademischer Austauschdienst). He will work on nonlinear phenomena occurring in critical mixtures.