Reservoir computing using networks of memristors: effects of topology and heterogeneity.

Reservoir computing (RC) has attracted significant interest as a framework for the implementation of novel neuromorphic computing architectures. Previously attention has been focussed on software-based reservoirs, where it has been demonstrated that reservoir topology plays a role in task performance, and functional advantage has been attributed to small-world and scale-free connectivity. However in hardware systems, such as electronic memristor networks, the mechanisms responsible for the reservoir dynamics are very different and the role of reservoir topology is largely unknown. Here we compare the performance of a range of memristive reservoirs in several RC tasks that are chosen to highlight different system requirements. We focus on percolating networks of nanoparticles (PNNs) which are novel self-assembled nanoscale systems that exhibit scale-free and small-world properties. We find that the performance of regular arrays of uniform memristive elements is limited by their symmetry but that this symmetry can be broken either by a heterogeneous distribution of memristor properties or a scale-free topology. The best perfomance across all tasks is observed for a scale-free network with uniform memistor properties. These results provide insight into the role of topology in neuromorphic reservoirs as well as an overview of the computational performance of scale-free networks of memristors in a range of benchmark tasks.

Reservoir computing with 3D nanowire networks.

Networks of nanowires are currently being explored for a range of applications in brain-like (or neuromorphic) computing, and especially in reservoir computing (RC). Fabrication of real-world computing devices requires that the nanowires are deposited sequentially, leading to stacking of the wires on top of each other. However, most simulations of computational tasks using these systems treat the nanowires as 1D objects lying in a perfectly 2D plane - the effect of stacking on RC performance has not yet been established. Here we use detailed simulations to compare the performance of perfectly 2D and quasi-3D (stacked) networks of nanowires in two tasks: memory capacity and nonlinear transformation. We also show that our model of the junctions between nanowires is general enough to describe a wide range of memristive networks, and consider the impact of physically realistic electrode configurations on performance. We show that the various networks and configurations have a strikingly similar performance in RC tasks, which is surprising given their radically different topologies. Our results show that networks with an experimentally achievable number of electrodes perform close to the upper bounds achievable when using the information from every wire. However, we also show important differences, in particular that the quasi-3D networks are more resilient to changes in the input parameters, generalizing better to noisy training data. Since previous literature suggests that topology plays an important role in computing performance, these results may have important implications for future applications of nanowire networks in neuromorphic computing.

Spontaneous growth of polarizing refractory metal 'nano-fins'.

Traditional polymer polarizers degrade in harsh environments and at high temperatures, reducing the polarization effect. In contrast, polarizers produced with refractory metals have vastly improved thermal stability and resistance to harsh environments but are expensive to fabricate. Here we demonstrate prototype refractory metal wire grid polarizers produced by co-sputtering molybdenum and aluminum under specific conditions. Removal of the aluminum through selective dissolution enables the nanostructure array to transmit light. The polarization spans 500-1100 nm and the extinction ratio significantly increases to >100. Possessing broadband polarization and sufficient extinction ratios, the new polarizing film has potential applications in coatings for sunglasses, windows, pyrometers, scientific instruments, and LCD panels.

Anomalously strong plasmon resonances in aluminium bronze by modification of the electronic density-of-states.

We use a combination of experimental measurements and density functional theory calculations to show that modification of the band structure of Cu by additions of Al causes an unexpected enhancement of the dielectric properties. The effect is optimized in alloys with Al contents between 10 and 15 at.% and would result in strong localized surface plasmon resonances at suitable wavelengths of light. This result is surprising as, in general, alloying of Cu increases its DC resistivity and would be expected to increase optical loss. The wavelengths for the plasmon resonances in the optimized alloy are significantly blue-shifted relative to those of pure Cu and provide a new material selection option for the range 2.2-2.8 eV.

The effect of vacancies on the optical properties of AuAl2.

AuAl2 is an intermetallic compound with a vivid purple colour attributable to a bulk plasmon energy in the visible part of the spectrum. However, the colour of as-deposited thin films is not as strong and only develops upon annealing. Density functional theory calculations of the dielectric function are presented for a variety of vacancy types and concentrations. The results support the view that the effect of annealing on colour is correlated with a reduction in concentration of Al vacancies. The effect of vacancies on the optical properties can be understood as arising from the complex interplay between interband transitions around the Fermi level and the plasmon energy.

Determination of the elastic properties of graphene by indentation and the validity of classical models of indentation.

Ab initio and empirical force field methods are used to simulate the loading of a large graphene membrane under an indenter analogous to an atomic force microscope tip. From these calculations we attempt to resolve ambiguities around determination of the elastic constants of graphene from such indentation experiments. We investigate the effect of the formation of wrinkles and more importantly the applicability of modelling the membrane as a continuous elastic sheet. By comparing empirical potential and large scale density functional theory calculations we have also assessed the performance of classical potentials in describing bending in this system. We find that the in-plane Young's modulus deduced from the indentation simulations using the classical expression for a clamped elastic membrane under a central point load is not consistent with that calculated directly from the in-plane stress-strain curve.

Transmitting hertzian optical nanoantenna with free-electron feed.

A pair of coupled gold nanorods excited by a beam of free electrons acts as a transmitting Hertzian antenna in the optical part of the spectrum. Significantly enhanced resonant emission is observed from the antenna when the electron beam is injected around the junction between the rods, where the local density of electromagnetic states is elevated.

Designing materials for plasmonic systems: the alkali-noble intermetallics.

We use electronic structure calculations based upon density functional theory to search for ideal plasmonic materials among the alkali-noble intermetallics. Importantly, we use density functional perturbation theory to calculate the electron-phonon interaction and from there use a first order solution to the Boltzmann equation to estimate the phenomenological damping frequency in the Drude dielectric function. We discuss the necessary electronic features of a plasmonic material and investigate the optical properties of the alkali-noble intermetallics in terms of some generic plasmonic system quality factors. We conclude that at low negative permittivities, KAu, with a damping frequency of 0.0224 eV and a high optical gap to bare plasma frequency ratio, outperforms gold and to some extent silver as a plasmonic material. Unfortunately, a low plasma frequency (1.54 eV) reduces its utility in modern plasmonics applications. We also discuss, briefly, the effect of local fields on the optical properties of these materials.

A review of the optical properties of alloys and intermetallics for plasmonics.

Alternative materials are required to enhance the efficacy of plasmonic devices. We discuss the optical properties of a number of alloys, doped metals, intermetallics, silicides, metallic glasses and high pressure materials. We conclude that due to the probability of low frequency interband transitions, materials with partially occupied d states perform poorly as plasmonic materials, ruling out many alloys, intermetallics and silicides as viable. The increased probability of electron-electron and electron-phonon scattering rules out many doped and glassy metals.

Optical properties of intermetallic compounds from first principles calculations: a search for the ideal plasmonic material.

First principles calculations have been used to predict the optical properties for a range of intermetallic compounds for which little or no experimental optical data are currently available. Density functional theory combined with the random phase approximation is used to calculate the dielectric functions for these compounds. The aim of this work is to investigate how the band edge and plasma frequency vary with composition in order to identify materials with promising plasmonic properties. Towards this end the intermetallic compounds chosen are composed of elements which on their own have reasonable optical properties for plasmonic applications. The position of the band edge relative to the plasma frequency is most favourable in the simple binary compounds formed from the alkali plus noble metals NaAu, KAu and KAg. In particular, for KAu the band edge and plasma frequency occur at almost the same frequency, and hence the imaginary part of the dielectric function is practically zero for frequencies below the plasma frequency. In addition, the plasma frequency in this compound is at relatively low frequency, promising a material with strong plasmon response in the infrared.

Long-term seasonal abundance patterns of Helicoverpa zea and Heliothis virescens (Lepidoptera: Noctuidae) in the Texas High Plains.

Bollworm, Helicoverpa zea (Boddie), and tobacco budworm, Heliothis virescens (F.), male adult (moth) activities were monitored between 1982 and 1995 by using sex pheromone traps in the Texas High Plains. Moths were monitored weekly from early March to mid-November near Lubbock and Halfway, two prominent cotton production areas in the Texas High Plains region. Based on trap captures, the bollworm-budworm complex consisted of approximately 98% bollworms and approximately 2% tobacco budworms. Seasonal activity patterns varied between location for bollworm but not for tobacco budworm. The 14-yr average (+/- SE) bollworm moth abundance (moths per trap per week) at Lubbock was significantly higher (226.5 +/- 10.4) compared with that at Halfway (153.7 +/- 8.1). Correlation analyses showed a significant positive relationship between moth abundance and average weekly temperatures, whereas a significant negative relationship was observed between moth abundance and average weekly wind velocity for both species. Analyses also showed a positive correlation between moth abundance and cumulative degree-days (> 0.0 degrees C) from 1 January. A strong positive relationship was observed between moth abundance and weekly average precipitation for both species. Average weekly abundances were positively correlated between adjacent months during most of the active cotton fruiting season (June-September). However, the relationship between populations that contributed to the overwintering generation and the following spring populations varied between species and study sites. Nevertheless, data from this study indicated that late-season moth catches could be indicative of the dynamics of the early-season moth catches the following year in the High Plains. The mean population abundance curve based on 14-yr averages showed two bollworm population peaks at Lubbock, but only one peak at Halfway. Separate degree-day-based models were developed to describe long-term seasonal abundance patterns of bollworm moths for the Lubbock and Halfway sites.

Comparison of the effects of recombinant adenovirus-mediated expression of wild-type p53 and p27Kip1 on cell cycle and apoptosis in SUDHL-1 cells derived from anaplastic large cell lymphoma.

Recombinant adenoviruses expressing wild-type p53 (AdWTp53) and p27KiP1 (Adp27) were used to compare the effects on cell cycle and apoptosis in SUDHL-1 cells derived from human anaplastic large cell lymphoma. Cells infected with AdWTp53 and Adp27 showed high level of wild-type p53 and p27KiP1 expression, respectively. The expression of these proteins resulted in G1 arrest after 24 h of infection. Although the cells persisted in G1 arrest in both cell populations after 48 and 72 h of infection, the level of apoptosis assessed by TUNEL analysis was higher in cells infected with AdWTp53. Interestingly, apoptosis was more pronounced in cells infected with Adp27 after the initial 24 h and reached a steady state at 48 and 72 h. A lower MOI of Adp27 resulted in G1 arrest associated with a low level of apoptosis in SUDHL-1 cells after 48 h of infection. This was correlated with lower expression of p27KiP1. We postulate that the time-lag and the different level of apoptosis occurring in SUDHL-1 cells infected with AdWTp53 and Adp27 are clearly related to the intrinsic biochemical pathways solicited. In this context our study provides a model to investigate these pathways and better understand the biology of this particular lymphoma. Our data also support a potential application of Adp27 for gene therapy of this lymphoma similarly to AdWTp53 as previously shown.

Biochemical differences between SUDHL-1 and KARPAS 299 cells derived from t(2;5)-positive anaplastic large cell lymphoma are responsible for the different sensitivity to the antiproliferative effect of p27(Kip1).

An inverse correlation between p27(Kip1) expression and proliferation has been recently established in tissues derived from human lymphomas. The nucleophosmin-anaplastic lymphoma kinase (NPM-ALK)/phospholipase C-gamma (PLCgamma) complex also appears to play an important role in cell proliferation and malignant transformation of anaplastic large cell lymphoma (ALCL). In this study, we report that SUDHL-1 and KARPAS 299 ALCL-derived cell lines present different sensitivity to the antiproliferative effect of recombinant adenovirus-mediated p27(Kip1) expression or to serum-starvation in culture media. The results indicate that exogenous p27(Kip1) may interact with the NPM-ALK/PLCgamma pathway in SUDHL-1 but not in KARPAS 299 cells. This interaction correlates with changes in cell cycle and cell morphology observed mainly in SUDHL-1 cells. The percentage of SUDHL-1 cells in S phase declines, whereas it is almost unchanged in KARPAS 299 cells as compared to the controls after 96 h of infection with the recombinant adenovirus. Furthermore KARPAS 299 cells are resistant to serum-starvation due to deficient p27(Kip1)-upregulation and G1 arrest, whereas SUDHL-1 cells respond with increased G1 phase and p27(Kip1)-upregulation after 48 h of serum-starvation. Both cell lines express appropriate variation of levels of cyclins E and A, and Rb-phosphorylation as expected by growing them in culture media with different FBS content. Although both cell lines express cyclin D2, SUDHL-1 cells only present high level of cyclin D3. Moreover SUDHL-1 cells express high level of PTEN and the PKB/Akt pathway is constitutively activated in both cell lines. Lastly SUDHL-1 cells show higher levels of phosphotyrosine-containing proteins that is correlated with a higher NPM-ALK-associated autophosphorylation activity compared to KARPAS 299 cells. Our study clearly identifies some of the biochemical differences that may explain the difference in sensitivity to antiproliferative stimuli shown by two cell lines derived from the same type of lymphoma.

Treatment of musculoskeletal pain with traditional Chinese herbal medicine.

Traditional Chinese herbal medicine (TCHM) uses naturally occurring plant, animal, and mineral substances to treat or assist in the treatment of the full spectrum of human disease. With the rise in popularity of alternative or complementary medicine, all physicians will encounter patients using TCHMs. TCHM should be taken under the supervision of a provider whose herbal training and competence is commensurate with the intensity of the herbal regimen and the severity of the clinical condition. TCHM can be valuable in the treatment of all kinds of pain: either as primary or adjunctive therapy depending on the clinical situation. TCHM therapies are prescribed in the framework of a unique diagnostic approach, and are highly specific with regard to type of pain and location of pain.

Gross and microscopic findings in the human esophagus after esophageal variceal band ligation: a postmortem analysis.

Endoscopic sclerotherapy has been the therapy of choice for controlling acute variceal hemorrhage. Recently, endoscopic band ligation has been utilized for the management of esophageal varices with fewer complications and greater efficacy. We report the pathologic findings in the human esophagus 5 days after esophageal variceal band ligation.

Relationships of structural dimensions to bench press strength in college males.

The purpose of this study was to determine the relationships between structural dimensions and bench press performance in college males. Members of required fitness classes (n = 170) were measured after 14 weeks of strength and aerobic endurance training. Anthropometric dimensions included upper arm and chest circumferences, upper and lower arm lengths, shoulder and hip widths, %fat, and height. Arm muscle cross-sectional area (CSA) was calculated from upper arm circumference corrected for triceps skinfold. Drop distance was measured from the bar to the pectoral muscles. Multiple regression analysis selected upper arm CSA, %fat, and chest circumference as the best items to predict bench press strength (R = 0.83; SEE = 11.6 kg). Cross-validation of the prediction equation on a similar sample (n = 89) produced an r = 0.74 between predicted and actual bench press (t = 0.53, p greater than 0.50). In a second cross validation sample (n = 57) who had trained more extensively with weights, the correlation between predicted and actual bench press was r = 0.57 (p less than 0.05). The prediction equations significantly (t = 6.59, p less than 0.01) underestimated bench press performance in the more extensively weight trained subjects. The results of this study suggest that bench press performance is related to structural dimensions in males and that extensive strength training may alter the relationship between size and strength.

Factors affecting anaerobic power output in the Margaria-Kalamen test.

The purpose of this study was to determine the effect of three approach distances and two vertical ascent methods on approach velocity, vertical velocity and power output of the Margaria-Kalamen test. Male (n = 43) and female (n = 53) university students were tested using a traditional and a modified (ramp) Margaria-Kalamen test with approaches of 2, 6 and 10m. The average of five trials for each approach distance was used to calculate anaerobic power. Males averaged 14.0% greater approach velocity, 28.8% greater vertical velocity and 41.1% greater power output than females, regardless of approach distance or vertical ascent method. Anaerobic power output was not significantly different between the sexes for any method when the effects of the percentage of fat and lean body mass were removed by the covariance technique. The 2m approach produced significantly slower approach velocity, vertical velocity and power output than the other approaches. The ramp method resulted in significantly greater approach velocity (11.5%), vertical velocity (9.6%) and anaerobic power output (9.4%) than the stair method. Males averaged 8.4% higher power output using the ramp, while females averaged 11.0% higher than the conventional stair method. Females appeared to benefit more than males from using a ramp in the anaerobic power test. Anaerobic power measurement from the Margaria-Kalamen method can be maximized using a ramp method.