Various algorithms for optimization and learning in adaptive systems

This thesis discusses various methods for learning and optimization in adaptive systems. Overall, it emphasizes the relationship between optimization, learning, and adaptive systems; and it illustrates the influence of underlying hardware upon the construction of efficient algorithms for learning and optimization. Chapter 1 provides a summary and an overview.

Chapter 2 discusses a method for using feed-forward neural networks to filter the noise out of noise-corrupted signals. The networks use back-propagation learning, but they use it in a way that qualifies as unsupervised learning. The networks adapt based only on the raw input data-there are no external teachers providing information on correct operation during training. The chapter contains an analysis of the learning and develops a simple expression that, based only on the geometry of the network, predicts performance.

Chapter 3 explains a simple model of the piriform cortex, an area in the brain involved in the processing of olfactory information. The model was used to explore the possible effect of acetylcholine on learning and on odor classification. According to the model, the piriform cortex can classify odors better when acetylcholine is present during learning but not present during recall. This is interesting since it suggests that learning and recall might be separate neurochemical modes (corresponding to whether or not acetylcholine is present). When acetylcholine is turned off at all times, even during learning, the model exhibits behavior somewhat similar to Alzheimer's disease, a disease associated with the degeneration of cells that distribute acetylcholine.

Chapters 4, 5, and 6 discuss algorithms appropriate for adaptive systems implemented entirely in analog hardware. The algorithms inject noise into the systems and correlate the noise with the outputs of the systems. This allows them to estimate gradients and to implement noisy versions of gradient descent, without having to calculate gradients explicitly. The methods require only noise generators, adders, multipliers, integrators, and differentiators; and the number of devices needed scales linearly with the number of adjustable parameters in the adaptive systems. With the exception of one global signal, the algorithms require only local information exchange.

Physical investigations of small particles : (I) aerosol particle charging and flux enhancement and (II) whispering gallery mode sensing

Part I

Particles are a key feature of planetary atmospheres. On Earth they represent the greatest source of uncertainty in the global energy budget. This uncertainty can be addressed by making more measurement, by improving the theoretical analysis of measurements, and by better modeling basic particle nucleation and initial particle growth within an atmosphere. This work will focus on the latter two methods of improvement.

Uncertainty in measurements is largely due to particle charging. Accurate descriptions of particle charging are challenging because one deals with particles in a gas as opposed to a vacuum, so different length scales come into play. Previous studies have considered the effects of transition between the continuum and kinetic regime and the effects of two and three body interactions within the kinetic regime. These studies, however, use questionable assumptions about the charging process which resulted in skewed observations, and bias in the proposed dynamics of aerosol particles. These assumptions affect both the ions and particles in the system. Ions are assumed to be point monopoles that have a single characteristic speed rather than follow a distribution. Particles are assumed to be perfect conductors that have up to five elementary charges on them. The effects of three body interaction, ion-molecule-particle, are also overestimated. By revising this theory so that the basic physical attributes of both ions and particles and their interactions are better represented, we are able to make more accurate predictions of particle charging in both the kinetic and continuum regimes.

The same revised theory that was used above to model ion charging can also be applied to the flux of neutral vapor phase molecules to a particle or initial cluster. Using these results we can model the vapor flux to a neutral or charged particle due to diffusion and electromagnetic interactions. In many classical theories currently applied to these models, the finite size of the molecule and the electromagnetic interaction between the molecule and particle, especially for the neutral particle case, are completely ignored, or, as is often the case for a permanent dipole vapor species, strongly underestimated. Comparing our model to these classical models we determine an “enhancement factor” to characterize how important the addition of these physical parameters and processes is to the understanding of particle nucleation and growth.

Part II

Whispering gallery mode (WGM) optical biosensors are capable of extraordinarily sensitive specific and non-specific detection of species suspended in a gas or fluid. Recent experimental results suggest that these devices may attain single-molecule sensitivity to protein solutions in the form of stepwise shifts in their resonance wavelength, \lambda_{R}, but present sensor models predict much smaller steps than were reported. This study examines the physical interaction between a WGM sensor and a molecule adsorbed to its surface, exploring assumptions made in previous efforts to model WGM sensor behavior, and describing computational schemes that model the experiments for which single protein sensitivity was reported. The resulting model is used to simulate sensor performance, within constraints imposed by the limited material property data. On this basis, we conclude that nonlinear optical effects would be needed to attain the reported sensitivity, and that, in the experiments for which extreme sensitivity was reported, a bound protein experiences optical energy fluxes too high for such effects to be ignored.

Coupled plasmonic systems and devices : applications in visible metamaterials, nanophotonic circuits, and CMOS imaging

With the size of transistors approaching the sub-nanometer scale and Si-based photonics pinned at the micrometer scale due to the diffraction limit of light, we are unable to easily integrate the high transfer speeds of this comparably bulky technology with the increasingly smaller architecture of state-of-the-art processors. However, we find that we can bridge the gap between these two technologies by directly coupling electrons to photons through the use of dispersive metals in optics. Doing so allows us to access the surface electromagnetic wave excitations that arise at a metal/dielectric interface, a feature which both confines and enhances light in subwavelength dimensions - two promising characteristics for the development of integrated chip technology. This platform is known as plasmonics, and it allows us to design a broad range of complex metal/dielectric systems, all having different nanophotonic responses, but all originating from our ability to engineer the system surface plasmon resonances and interactions. In this thesis, we demonstrate how plasmonics can be used to develop coupled metal-dielectric systems to function as tunable plasmonic hole array color filters for CMOS image sensing, visible metamaterials composed of coupled negative-index plasmonic coaxial waveguides, and programmable plasmonic waveguide network systems to serve as color routers and logic devices at telecommunication wavelengths.

Effects of international trade of food and feed and human diet shifts on food security and environmental safety: integrating scales

4 p.

Functionalization of Si(111) surfaces and the formation of mixed monolayers for the covalent attachment of molecular catalysts in photoelectrochemical devices

The functionalization of silicon surfaces with molecular catalysts for proton reduction is an important part of the development of a solar-powered, water-splitting device for solar fuel formation. The covalent attachment of these catalysts to silicon without damaging the underlying electronic properties of silicon that make it a good photocathode has proven difficult. We report the formation of mixed monolayer-functionalized surfaces that incor- porate both methyl and vinylferrocenyl or vinylbipyridyl (vbpy) moieties. The silicon was functionalized using reaction conditions analogous to those of hydrosilylation, but instead of a H-terminated Si surface, a chlorine-terminated Si precursor surface was used to produce the linked vinyl-modified functional group. The functionalized surfaces were characterized by time-resolved photoconductivity decay, X-ray photoelectron spectroscopy (XPS), electro- chemical, and photoelectrochemical measurements. The functionalized Si surfaces were well passivated, exhibited high surface coverage and few remaining reactive Si atop sites, had a very low surface recombination velocity, and displayed little initial surface oxidation. The surfaces were stable toward atmospheric and electrochemical oxidation. The surface coverage of ferrocene or bipyridine was controllably varied from 0 up to 30% of a monolayer without loss of the underlying electronic properties of the silicon. Interfacial charge transfer to the attached ferrocene group was relatively rapid, and a photovoltage of 0.4 V was generated upon illumination of functionalized n-type silicon surfaces in CH₃CN. The immobilized bipyridine ligands bound transition metal ions, and thus enabled the assembly of metal complexes on the silicon surface. XPS studies demonstrated that [Cp∗Rh(vbpy)Cl]Cl, [Cp∗Ir(vbpy)Cl]Cl, and Ru(acac)₂vbpy were assembled on the surface. For the surface prepared with iridium, x-ray absorption spectroscopy at the Ir L_III edge showed an edge energy and post-edge features virtually identical to a powder sample of [Cp∗Ir(bipy)Cl]Cl (bipy is 2,2 ́-bipyridyl). Electrochemical studies on these surfaces confirmed that the assembled complexes were electrochemically active.

Extracting material response from simple mechanical tests on hardening-softening-hardening viscoplastic solids

Compliant foams are usually characterized by a wide range of desirable mechanical properties. These properties include viscoelasticity at different temperatures, energy absorption, recoverability under cyclic loading, impact resistance, and thermal, electrical, acoustic and radiation-resistance. Some foams contain nano-sized features and are used in small-scale devices. This implies that the characteristic dimensions of foams span multiple length scales, rendering modeling their mechanical properties difficult. Continuum mechanics-based models capture some salient experimental features like the linear elastic regime, followed by non-linear plateau stress regime. However, they lack mesostructural physical details. This makes them incapable of accurately predicting local peaks in stress and strain distributions, which significantly affect the deformation paths. Atomistic methods are capable of capturing the physical origins of deformation at smaller scales, but suffer from impractical computational intensity. Capturing deformation at the so-called meso-scale, which is capable of describing the phenomenon at a continuum level, but with some physical insights, requires developing new theoretical approaches.

A fundamental question that motivates the modeling of foams is ‘how to extract the intrinsic material response from simple mechanical test data, such as stress vs. strain response?’ A 3D model was developed to simulate the mechanical response of foam-type materials. The novelty of this model includes unique features such as the hardening-softening-hardening material response, strain rate-dependence, and plastically compressible solids with plastic non-normality. Suggestive links from atomistic simulations of foams were borrowed to formulate a physically informed hardening material input function. Motivated by a model that qualitatively captured the response of foam-type vertically aligned carbon nanotube (VACNT) pillars under uniaxial compression [2011,“Analysis of Uniaxial Compression of Vertically Aligned Carbon Nanotubes,” J. Mech.Phys. Solids, 59, pp. 2227–2237, Erratum 60, 1753–1756 (2012)], the property space exploration was advanced to three types of simple mechanical tests: 1) uniaxial compression, 2) uniaxial tension, and 3) nanoindentation with a conical and a flat-punch tip. The simulations attempt to explain some of the salient features in experimental data, like
1) The initial linear elastic response.
2) One or more nonlinear instabilities, yielding, and hardening.

The model-inherent relationships between the material properties and the overall stress-strain behavior were validated against the available experimental data. The material properties include the gradient in stiffness along the height, plastic and elastic compressibility, and hardening. Each of these tests was evaluated in terms of their efficiency in extracting material properties. The uniaxial simulation results proved to be a combination of structural and material influences. Out of all deformation paths, flat-punch indentation proved to be superior since it is the most sensitive in capturing the material properties.

Cell-targeted regulation of gene expression through synthetic RNA devices

The ability to interface with and program cellular function remains a challenging research frontier in biotechnology. Although the emerging field of synthetic biology has recently generated a variety of gene-regulatory strategies based on synthetic RNA molecules, few strategies exist through which to control such regulatory effects in response to specific exogenous or endogenous molecular signals. Here, we present the development of an engineered RNA-based device platform to detect and act on endogenous protein signals, linking these signals to the regulation of genes and thus cellular function.

We describe efforts to develop an RNA-based device framework for regulating endogenous genes in human cells. Previously developed RNA control devices have demonstrated programmable ligand-responsive genetic regulation in diverse cell types, and we attempted to adapt this class of cis-acting control elements to function in trans. We divided the device into two strands that reconstitute activity upon hybridization. Device function was optimized using an in vivo model system, and we found that device sequence is not as flexible as previously reported. After verifying the in vitro activity of our optimized design, we attempted to establish gene regulation in a human cell line using additional elements to direct device stability, structure, and localization. The significant limitations of our platform prevented endogenous gene regulation.

We next describe the development of a protein-responsive RNA-based regulatory platform. Employing various design strategies, we demonstrated functional devices that both up- and downregulate gene expression in response to a heterologous protein in a human cell line. The activity of our platform exceeded that of a similar, small-molecule-responsive platform. We demonstrated the ability of our devices to respond to both cytoplasmic- and nuclear-localized protein, providing insight into the mechanism of action and distinguishing our platform from previously described devices with more restrictive ligand localization requirements. Finally, we demonstrated the versatility of our device platform by developing a regulatory device that responds to an endogenous signaling protein.

The foundational tool we present here possesses unique advantages over previously described RNA-based gene-regulatory platforms. This genetically encoded technology may find future applications in the development of more effective diagnostic tools and targeted molecular therapy strategies.

New optoelectronic devices using GaAs-GaAlAs epitaxy

Three subjects related to epitaxial GaAs-GaAlAs optoelectronic devices are discussed in this thesis. They are:

1. Embedded Epitaxy

This is a technique of selective multilayer growth of GaAs- Ga_1-xAl_xAs single crystal structures through stripe openings in masking layers on GaAs substrates. This technique results in prismatic layers of GaAs and Ga_1-xAl_xAs "embedded" in each other and leads to controllable uniform structures terminated by crystal faces. The dependence of the growth habit on the orientation of the stripe openings has been studied. Room temperature embedded double heterostructure lasers have been fabricated using this technique. Threshold current densities as low as 1.5 KA/cm² have been achieved.

2. Barrier Controlled PNPN Laser Diode

It is found that the I-V characteristics of a PNPN device can be controlled by using potential barriers in the base regions. Based on this principle, GaAs-GaAlAs heterostructure PNPN laser diodes have been fabricated. GaAlAs potential barriers in the bases control not only the electrical but also the optical properties of the device. PNPN lasers with low threshold currents and high breakover voltage have been achieved. Numerical calculations of this barrier controlled structure are presented in the ranges where the total current is below the holding point and near the lasing threshold.

3. Injection Lasers on Semi-Insulating Substrates

GaAs-GaAlAs heterostructure lasers fabricated on semi-insulating substrates have been studied. Two different laser structures achieved are: (1) Crowding effect lasers, (2) Lateral injection lasers. Experimental results and the working principles underlying the operation of these lasers are presented. The gain induced guiding mechanism is used to explain the lasers' far field radiation patterns. It is found that Zn diffusion in Ga_1-xAl_xAs depends on the Al content x, and that GaAs can be used as the diffusion mask for Zn diffusion in Ga_1-xAl_xAs. Lasers having very low threshold currents and operating in a stable single mode have been achieved. Because these lasers are fabricated on semi-insulating substrates, it is possible to integrate them with other electronic devices on the same substrate. An integrated device, which consists of a crowding effect laser and a Gunn oscillator on a common semi-insulating GaAs substrate, has been achieved.