Gaussian process volatility model

The prediction of time-changing variances is an important task in the modeling of financial data. Standard econometric models are often limited as they assume rigid functional relationships for the evolution of the variance. Moreover, functional parameters are usually learned by maximum likelihood, which can lead to over-fitting. To address these problems we introduce GP-Vol, a novel non-parametric model for time-changing variances based on Gaussian Processes. This new model can capture highly flexible functional relationships for the variances. Furthermore, we introduce a new online algorithm for fast inference in GP-Vol. This method is much faster than current offline inference procedures and it avoids overfitting problems by following a fully Bayesian approach. Experiments with financial data show that GP-Vol performs significantly better than current standard alternatives.

High-frequency programmable acoustic wave device realized through ferroelectric domain engineering

Surface acoustic wave devices are extensively used in contemporary wireless communication devices. We used atomic force microscopy to form periodic macroscopic ferroelectric domains in sol-gel deposited lead zirconate titanate, where each ferroelectric domain is composed of many crystallites, each of which contains many microscopic ferroelastic domains. We examined the electro-acoustic characteristics of the apparatus and found a resonator behavior similar to that of an equivalent surface or bulk acoustic wave device. We show that the operational frequency of the device can be tailored by altering the periodicity of the engineered domains and demonstrate high-frequency filter behavior (>8GHz), allowing low-cost programmable high-frequency resonators. © 2014 AIP Publishing LLC.

Graphene-based optical modulator realized in metamaterial split-ring resonators operating in the THz frequency range

The integration of quantum cascade lasers with devices capable of efficiently manipulating terahertz light, represents a fundamental step for many different applications. Split-ring resonators, sub-wavelength metamaterial elements exhibiting broad resonances that are easily tuned lithographically, represent the ideal route to achieve such optical control of the incident light. We have realized a design based on the interplay between metallic split rings and the electronic properties of a graphene monolayer integrated into a single device. By acting on the doping level of graphene, an active modulation of the optical intensity was achieved in the frequency range between 2.2 THz and 3.1 THz, with a maximum modulation depth of 18%.

High speed liquid crystal over silicon display based on the flexoelectro-optic effect.

One of the key technologies to evolve in the displays market in recent years is liquid crystal over silicon (LCOS) microdisplays. Traditional LCOS devices and applications such as rear projection televisions, have been based on intensity modulation electro-optical effects, however, recent developments have shown that multi-level phase modulation from these devices is extremely sought after for applications such as holographic projectors, optical correlators and adaptive optics. Here, we propose alternative device geometry based on the flexoelectric-optic effect in a chiral nematic liquid crystal. This device is capable of delivering a multilevel phase shift at response times less than 100 microsec which has been verified by phase shift interferometry using an LCOS test device. The flexoelectric on silicon device, due to its remarkable characteristics, enables the next generation of holographic devices to be realized.

Switching performance of epitaxially grown normally-off 4H-SiC JFET

Static and dynamic behavior of the epitaxially grown dual gate trench 4H-SiC junction field effect transistor (JFET) is investigated. Typical on-state resistance Ron was 6-10mΩcm2 at VGS = 2.5V and the breakdown voltage between the range of 1.5-1.8kV was realized at VGS = -5V for normally-off like JFETs. It was found that the turn-on energy delivers the biggest part of the switching losses. The dependence of switching losses from gate resistor is nearly linear, suggesting that changing the gate resistor, a way similar to Si-IGBT technology, can easily control di/dt and dv/dt. Turn-on losses at 200°C are lower compared to those at 25°C, which indicates the influence of the high internal p-type gate layer resistance. Inductive switching numerical analysis suggested the strong influence of channel doping conditions on the turn-on switching performance. The fast switching normally-off JFET devices require heavily doped narrow JFET channel design. © (2009) Trans Tech Publications, Switzerland.

Effects of mechanical and electrical coupling on the parametric sensitivity of mode localized sensors

We compare and contrast the effects of two distinctly different mechanisms of coupling (mechanical and electrical) on the parametric sensitivity of micromechanical sensors utilizing mode localization for sensor applications. For the first time, the strong correlation between mode localization and the phenomenon of 'eigenvalue loci-veering' is exploited for accurate quantification of the strength of internal coupling in mode localized sensors. The effects of capacitive coupling-spring tuning on the parametric sensitivity of electrically coupled resonators utilizing this sensing paradigm is also investigated and a mass sensor with sensitivity tunable by over 400% is realized. ©2009 IEEE.

Uniform stability of a particle approximation of the optimal filter derivative

Sequential Monte Carlo methods, also known as particle methods, are a widely used set of computational tools for inference in non-linear non-Gaussian state-space models. In many applications it may be necessary to compute the sensitivity, or derivative, of the optimal filter with respect to the static parameters of the state-space model; for instance, in order to obtain maximum likelihood model parameters of interest, or to compute the optimal controller in an optimal control problem. In Poyiadjis et al. [2011] an original particle algorithm to compute the filter derivative was proposed and it was shown using numerical examples that the particle estimate was numerically stable in the sense that it did not deteriorate over time. In this paper we substantiate this claim with a detailed theoretical study. Lp bounds and a central limit theorem for this particle approximation of the filter derivative are presented. It is further shown that under mixing conditions these Lp bounds and the asymptotic variance characterized by the central limit theorem are uniformly bounded with respect to the time index. We demon- strate the performance predicted by theory with several numerical examples. We also use the particle approximation of the filter derivative to perform online maximum likelihood parameter estimation for a stochastic volatility model.