Derivative-free optimization and neural networks for robust regression

Large outliers break down linear and nonlinear regression models. Robust regression methods allow one to filter out the outliers when building a model. By replacing the traditional least squares criterion with the least trimmed squares (LTS) criterion, in which half of data is treated as potential outliers, one can fit accurate regression models to strongly contaminated data. High-breakdown methods have become very well established in linear regression, but have started being applied for non-linear regression only recently. In this work, we examine the problem of fitting artificial neural networks (ANNs) to contaminated data using LTS criterion. We introduce a penalized LTS criterion which prevents unnecessary removal of valid data. Training of ANNs leads to a challenging non-smooth global optimization problem. We compare the efficiency of several derivative-free optimization methods in solving it, and show that our approach identifies the outliers correctly when ANNs are used for nonlinear regression.

Zeroes of Riemann's zeta function on the critical line with 20000 decimal digits accuracy

The first 12000 zeroes of Riemann's zeta function on the critical line with 20000 decimal digits accuracy. Format: the zeroes are in text file listed consecutively in decimal representation, each zero starts on a new line.

Zeroes of zeta function presented in this file were calculated on MASSIVE cluster (www.massive.org.au) using Python and packages MPmath version 0.17 and gmpy version 2.1, with a Newton based algorithm proposed by Fredrik Johansson with precision set to 20000 decimal digits. Partial recalculation with higher precision didn't show any loss of accuracy so we expect that the values are correct up to, possibly, a few last digits. We express our thanks to Fredrik Johansson for this algorithm and for development of MPmath as well.

The sampling and remix dilemma: what is the role of moral rights in the encouragement and regulation of derivative creativity

The borrowing and rearrangement of musical content, especially in the digital context, raises difficult questions for copyright law. There is significant community support for a loosening of the restrictions on the derivative (and particularly creative) use of copyright material. Law reform is called for. This paper discusses the possible introduction of a new exception to copyright infringement but notes that in the drafting of any such exception not only the economic rights but also the moral rights of the originating author need to be taken into account.

Zeroes of Riemann's zeta function on the critical line with 40000 decimal digits accuracy

The project aims at computing Riemann's zeroes with high accuracy through an analysis of large determinants using Matyasievich's Artless method. Location of Riemann's zeroes is the famous 8th Hilbert problem, and one of Clay's Institute millennium problems.

Promising pre-clinical validation of targeted radionuclide therapy using a [131I] labelled iodoquinoxaline derivative for an effective melanoma treatment

Targeted internal radionuclide therapy (TRT) would be an effective alternative to current therapies for dissemi- nated melanoma treatment. At our institution, a class of iodobenzamides has been developed as potent melanoma- seeking agents. This review described the preclinical vali- dations of a quinoxaline derivative molecule (ICF01012) as tracer for TRT application. It was selected for its high, specific and long-lasting uptake in tumour with rapid clear- ance from non-target organs providing suitable dosimetry parameters for TRT. Extended in vivo study of metabolic profiles confirmed durable tumoural concentration of the unchanged molecule form. Moreover melanin specificity of ICF01012 was determined by binding assay with syn- thetic melanin and in vivo by SIMS imaging. Then, we showed in vivo that [131I] ICF01012 treatment drastically inhibited growth of B16F0, B16Bl6 and M4Beu tumours whereas [131I] NaI or unlabelled ICF01012 treatment was without significant effect. Histological analysis showed that residual tumour cells exhibit a significant loss of aggres- siveness after treatment. This anti-tumoural effect was associated with a lengthening of the treated-mice survival time and an inhibition of lung dissemination for B16Bl6 model. Results presented here support the concept of TRT using a [131I] labelled iodoquinoxaline derivative for an effective melanoma treatment.

Effective derivative hedging and initial public offering long-run performance

We investigate the role of corporate currency risk management through the use of financial derivatives in influencing the long-run performance of a sample of Australian resources companies. We find that derivative users generally outperformed nonderivative users in the 5-year period following listing. Effective derivative users consistently outperformed the nonhedgers. Furthermore, within the population of derivative users, effective derivative users tended to perform better than ineffective hedgers. Our results indicate that effective financial risk management plays a role in long-run IPO performance.

Graphene and graphitic derivative filled polymer composites as potential sensors

Graphite and numerous graphitic-derived micro- and nano-particles have gained importance in current materials science research. These two-dimensional sheets of sp(2)-hybridized carbon atoms remarkably influence the properties of polymers. Graphene mono-layers, graphene oxides, graphite oxides, exfoliated graphite, and other related materials are derived from a parental graphite structure. In this review, we focus primarily on the role of these fillers in regulating the electrical and sensing properties of polymer composites. It has been demonstrated that the addition of an optimized mixture of graphene and or its derivatives to various polymers produces a record-high enhancement of the electrical conductivity and achieved semiconducting characteristics at small filler loading, making it suitable for sensor manufacture. Promising sensing characteristics are observed in graphite-derived composite films compared with those of micro-sized composites and the properties are explained mainly based on the filler volume fraction, nature and rate of dispersion and the filler polymer interactions at the interface. In short, this critical review aims to provide a thorough understanding of the recent advances in the area of graphitic-based polymer composites in advanced electronics. Future perspectives in this rapidly developing field are also discussed.

Estimating the term structure of volatility and fixed income derivative pricing

Estimating the parameters of the instantaneous spot interest rate process is of crucial importance for pricing fixed income derivative securities. This paper presents an estimation for the parameters of the Gaussian interest rate model for pricing fixed income derivatives based on the term structure of volatility. We estimate the term structure of volatility for US treasury rates for the period 1983 - 1995, based on a history of yield curves. We estimate both conditional and first differences term structures of volatility and subsequently estimate the implied parameters of the Gaussian model with non-linear least squares estimation. Results for bond options illustrate the effects of differing parameters in pricing.

Três métodos para o cálculo da série zeta(2n) de Riemann

Neste trabalho apresentamos três métodos distintos provando que S(n) = +1 X k=−1 (4k + 1)−n é um múltiplo racional de n para todos os inteiros n = 1, 2, 3, . . . O primeiro utiliza a teoria das função analíticas e funções geradoras. No segundo reduzimos o problema, via mudança de variável devida a E. Calabi, ao cálculo do volume de certos politopos em Rn enquanto que no terceiro usamos a teoria dos operadores integrais compactos. Cada um dos métodos tem um interesse intrínsico e está sujeito a generalizações para aplicações em novas situações.

Options listing and the volatility of the underlying asset: a study on the derivative market function

Os mercados de derivativos são vistos com muita desconfiança por inúmeras pessoas. O trabalho analisa o efeito da introdução de opções sobre ações no mercado brasileiro buscando identificar uma outra justificativa para a existência destes mercados: a alteração no nível de risco dos ativos objetos destas opções. A evidência empírica encontrada neste mercado está de acordo com os resultados obtidos em outros mercados - a introdução de opções é benéfica para o investidor posto que reduz a volatilidade do ativo objeto. Existe também uma tênue indicação de que a volatilidade se torna mais estocástica com a introdução das opções.

Toward the Optimization of an e-Tongue System Using Information Visualization: A Case Study with Perylene Tetracarboxylic Derivative Films in the Sensing Units

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Stability of multistep second derivative methods for integro-differential equations

The stability of multistep second derivative methods for integro-differential equations is examined through a test equation which allows for the construction of the associated characteristic polynomial and its region of stability (roots in the unit circle) at a proper parameter space. (c) 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.