arXiv subjects

Amir Gat

Publications and source records attributed to Amir Gat.

4 recordsLinked to original sources

arXiv2024

Multistable Physical Neural Networks

Artificial neural networks (ANNs), which are inspired by the brain, are a central pillar in the ongoing breakthrough in artificial intelligence. In recent years, researchers have examined mechanical implementations of ANNs, denoted as Physical Neural Networks (PNNs). PNNs offer the opportunity to view common materials and physical phenomena as networks, and to associate computational power with them. In this work, we incorporated mechanical bistability into PNNs, enabling memory and a direct link between computation and physical action. To achieve this, we consider an interconnected network of bistable liquid-filled chambers. We first map all possible equilibrium configurations or steady states, and then examine their stability. Building on these maps, both global and local algorithms for training multistable PNNs are implemented. These algorithms enable us to systematically examine the network's capability to achieve stable output states and thus the network's ability to perform computational tasks. By incorporating PNNs and multistability, we can design structures that mechanically perform tasks typically associated with electronic neural networks, while directly obtaining physical actuation. The insights gained from our study pave the way for the implementation of intelligent structures in smart tech, metamaterials, medical devices, soft robotics, and other fields.

cs.NE↗

arXiv2023

Dynamics of fixed-volume pinned film -- dealing with a non-self-adjoint thin film problem

The use of thin liquid films has expanded beyond lubrication and coatings, and into applications in actuators and adaptive optical elements. In contrast to their predecessors, whose dynamics can be typically captured by modelling infinite or periodic films, these applications are characterized by a finite amount of liquid in an impermeable domain. The global mass conservation constraint, together with common boundary conditions (e.g., pinning) create quantitatively and qualitatively different dynamics than those of infinite films. Mathematically, this manifests itself as a non-self-adjoint problem. This work presents a combined theoretical and experimental study for this problem. We provide a time-dependent closed-form analytical solution for the linearized non-self-adjoint system that arises from these boundary conditions. We highlight that, in contrast to self-adjoint problems, here special care should be given to deriving the adjoint problem to reconstruct the solution based on the eigenfunctions properly. We compare these solutions with those obtained for permeable and periodic boundary conditions, representing common models for self-adjoint thin-film problems. We show that while the initial dynamics are nearly identical, the boundary conditions eventually affect the film deformation as well as its response time. To experimentally illustrate the dynamics and to validate the theoretical model, we fabricated an experimental setup that subjects a thin liquid film to a prescribed normal force distribution through dielectrophoresis, and used high-frame-rate digital holography to measure the film deformation in real-time. The experiments agree well with the model and confirm that confined films exhibit different behaviour which could not be predicted by existing models.

physics.flu-dyn↗

arXiv2018

Elastic deformations driven by non-uniform lubrication flows

The ability to create dynamic deformations of micron-sized structures is relevant to a wide variety of applications such as adaptable optics, soft robotics, and reconfigurable microfluidic devices. In this work we examine non-uniform lubrication flow as a mechanism to create complex deformation fields in an elastic plate. We consider a Kirchoff-Love elasticity model for the plate and Hele-Shaw flow in a narrow gap between the plate and a parallel rigid surface. Based on linearization of the Reynolds equation, we obtain a governing equation which relates elastic deformations to gradients in non-homogenous physical properties of the fluid (e.g. body forces, viscosity, and slip velocity). We then focus on a specific case of non-uniform Helmholtz-Smoluchowski electroosmotic slip velocity, and provide a method for determining the zeta-potential distribution necessary to generate arbitrary static and quasi-static deformations of the elastic plate. Extending the problem to time-dependent solutions, we analyze transient effects on asymptotically static solutions, and finally provide a closed form solution for a Green's function for time periodic actuations.

physics.flu-dyn↗

arXiv2014

Solid-liquid composite structures: elastic beams with embedded liquid-filled parallel-channel networks

Deformation due to embedded fluidic networks is currently studied in the context of soft-actuators and soft-robotics. Expanding on this concept, beams can be designed so that the pressure in the channel-network is created directly from external forces acting on the beam, and thus can be viewed as passive solid-liquid composite structure. We obtain a continuous function relating the network geometry to the deformation. This enables design of networks creating arbitrary steady and time varying deformation-fields as well as to eliminate deformation created by external forces.

cond-mat.soft↗