Estimating learning models from experimental data

We study the statistical properties of three estimation methods for a model of learning that is often fitted to experimental data: quadratic deviation measures without unobserved heterogeneity, and maximum likelihood withand without unobserved heterogeneity. After discussing identification issues, we show that the estimators are consistent and provide their asymptotic distribution. Using Monte Carlo simulations, we show that ignoring unobserved heterogeneity can lead to seriously biased estimations in samples which have the typical length of actual experiments. Better small sample properties areobtained if unobserved heterogeneity is introduced. That is, rather than estimating the parameters for each individual, the individual parameters are considered random variables, and the distribution of those random variables is estimated.

Nanoparticles in SiH4-Ar plasma: Modelling and comparison with experimental data

Experimental and theoretical investigations for growth of silicon nanoparticles (4 to 14 nm) in radio frequency discharge were carried out. Growth processes were performed with gas mixtures of SiH4 and Ar in a plasma chemical reactor at low pressure. A distinctive feature of presented kinetic model of generation and growth of nanoparticles (compared to our earlier model) is its ability to investigate small"critical" dimensions of clusters, determining the rate of particle production and taking into account the influence of SiH2 and Si2Hm dimer radicals. The experiments in the present study were extended to high pressure (≥20 Pa) and discharge power (≥40 W). Model calculations were compared to experimental measurements, investigating the dimension of silicon nanoparticles as a function of time, discharge power, gas mixture, total pressure, and gas flow.

Can genetic algorithms explain experimental anomalies? An application to common property resources

It is common to find in experimental data persistent oscillations in the aggregate outcomes and high levels of heterogeneity in individual behavior. Furthermore, it is not unusual to find significant deviations from aggregate Nash equilibrium predictions. In this paper, we employ an evolutionary model with boundedly rational agents to explain these findings. We use data from common property resource experiments (Casari and Plott, 2003). Instead of positing individual-specific utility functions, we model decision makers as selfish and identical. Agent interaction is simulated using an individual learning genetic algorithm, where agents have constraints in their working memory, a limited ability to maximize, and experiment with new strategies. We show that the model replicates most of the patterns that can be found in common property resource experiments.

Predicting cancer involvement of genes from heterogeneous data

Background: Systematic approaches for identifying proteins involved in different types of cancer are needed. Experimental techniques such as microarrays are being used to characterize cancer, but validating their results can be a laborious task. Computational approaches are used to prioritize between genes putatively involved in cancer, usually based on further analyzing experimental data. Results: We implemented a systematic method using the PIANA software that predicts cancer involvement of genes by integrating heterogeneous datasets. Specifically, we produced lists of genes likely to be involved in cancer by relying on: (i) protein-protein interactions; (ii) differential expression data; and (iii) structural and functional properties of cancer genes. The integrative approach that combines multiple sources of data obtained positive predictive values ranging from 23% (on a list of 811 genes) to 73% (on a list of 22 genes), outperforming the use of any of the data sources alone. We analyze a list of 20 cancer gene predictions, finding that most of them have been recently linked to cancer in literature. Conclusion: Our approach to identifying and prioritizing candidate cancer genes can be used to produce lists of genes likely to be involved in cancer. Our results suggest that differential expression studies yielding high numbers of candidate cancer genes can be filtered using protein interaction networks.

Kinematically complete analysis of the CLAS data on the proton structure Function F2 in a Regge-Dual model

The recently measured inclusive electron-proton cross section in the nucleon resonance region, performed with the CLAS detector at the Thomas Jefferson Laboratory, has provided new data for the nucleon structure function F2 with previously unavailable precision. In this paper we propose a description of these experimental data based on a Regge-dual model for F2. The basic inputs in the model are nonlinear complex Regge trajectories producing both isobar resonances and a smooth background. The model is tested against the experimental data, and the Q2 dependence of the moments is calculated. The fitted model for the structure function (inclusive cross section) is a limiting case of the more general scattering amplitude equally applicable to deeply virtual Compton scattering. The connection between the two is discussed.

Finite mixture analysis of beauty-contest data using generalised beta distributions

This paper introduces a mixture model based on the beta distribution, without preestablishedmeans and variances, to analyze a large set of Beauty-Contest data obtainedfrom diverse groups of experiments (Bosch-Domenech et al. 2002). This model gives a bettert of the experimental data, and more precision to the hypothesis that a large proportionof individuals follow a common pattern of reasoning, described as iterated best reply (degenerate),than mixture models based on the normal distribution. The analysis shows thatthe means of the distributions across the groups of experiments are pretty stable, while theproportions of choices at dierent levels of reasoning vary across groups.

Comparison of a deterministic and a data driven model to describe MBR fouling

Membrane bioreactors (MBRs) are a combination of activated sludge bioreactors and membrane filtration, enabling high quality effluent with a small footprint. However, they can be beset by fouling, which causes an increase in transmembrane pressure (TMP). Modelling and simulation of changes in TMP could be useful to describe fouling through the identification of the most relevant operating conditions. Using experimental data from a MBR pilot plant operated for 462days, two different models were developed: a deterministic model using activated sludge model n°2d (ASM2d) for the biological component and a resistance in-series model for the filtration component as well as a data-driven model based on multivariable regressions. Once validated, these models were used to describe membrane fouling (as changes in TMP over time) under different operating conditions. The deterministic model performed better at higher temperatures (>20°C), constant operating conditions (DO set-point, membrane air-flow, pH and ORP), and high mixed liquor suspended solids (>6.9gL-1) and flux changes. At low pH (<7) or periods with higher pH changes, the data-driven model was more accurate. Changes in the DO set-point of the aerobic reactor that affected the TMP were also better described by the data-driven model. By combining the use of both models, a better description of fouling can be achieved under different operating conditions

Memòria justificativa de l'estada de recerca

Estudi elaborat a partir d’una estada a l’Associação para o Desenvolvimento da Aerodinânica Industrial (ADAI) de la Universitat de Coimbra, Portugal, entre març i juliol de 2006. Aquesta disposa d'un laboratori d'assaigs i té medis suficients per a cremar de manera controlada parcel•les prèviament delimitades en terreny forestal. Això permet observar el fenomen dels incendis forestals a dues escales de treball diferents. L’objectiu ha estat l’obtenció de dades experimentals sobre la propagació de fronts de flames que avancen sobre combustible tractat amb retardants sota l’efecte del pendent o el vent. S’ha participat en proves experimentals de camp i se n’han realitzat dues en instal•lacions de laboratori en què l’efecte del pendent o de la velocitat del vent podia ser variat. Degut a l’elevat nombre de variables que entren en joc l’anàlisi acurada de les dades encara està en procés.

Model based fault diagnosis with application to an automotive engine

Report for the scientific sojourn at the University of Linköping between April to July 2007. Monitoring of the air intake system of an automotive engine is important to meet emission related legislative diagnosis requirements. During the research the problem of fault detection in the air intake system was stated as a constraint satisfaction problem over continuous domains with a big number of variables and constraints. This problem was solved using Interval-based Consistency Techniques. Interval-based consistency techniques are shown to be particularly efficient for checking the consistency of the Analytical Redundancy Relations (ARRs), dealing with uncertain measurements and parameters, and using experimental data. All experiments were performed on a four-cylinder turbo-charged spark-ignited SAAB engine located in the research laboratory at Vehicular System Group - University of Linköping.

Experimentació amb un receptor software GPS-Galileo d'alta sensibilitat per a entorns interiors

Este proyecto se centra en el análisis de señales GPS, utilizando un receptor software desarrollado con Matlab en un proyecto de investigación para la Agencia Espacial Europea (ESA), llevado a cabo por parte del departamento de Telecomunicaciones e Ingeniería de Sistemas de la ETSE. Este software utiliza técnicas de procesado de señal de alta sensibilidad (HS-GNSS) que permite al usuario determinar su posición en entornos de difícil propagación como puede ser el caso de los escenarios interiores. Los datos experimentales se analizan en función del nivel de multipath que afecta a la señal de cada uno de los satélites, y la degradación que los escenarios interiores provocan en las señales, a causa del mobiliario, paredes, personas, etc. Para analizar los datos experimentales, se ha utilizado una métrica presentada en el congreso internacional EuCAP 2009, con la que es posible caracterizar las señales en función del nivel de multipath.

Contact melting of a three-dimensional phase change material on a flat substrate

In this paper a model is developed to describe the three dimensional contact melting process of a cuboid on a heated surface. The mathematical description involves two heat equations (one in the solid and one in the melt), the Navier-Stokes equations for the flow in the melt, a Stefan condition at the phase change interface and a force balance between the weight of the solid and the countering pressure in the melt. In the solid an optimised heat balance integral method is used to approximate the temperature. In the liquid the small aspect ratio allows the Navier-Stokes and heat equations to be simplified considerably so that the liquid pressure may be determined using an igenfunction expansion and finally the problem is reduced to solving three first order ordinary differential equations. Results are presented showing the evolution of the melting process. Further reductions to the system are made to provide simple guidelines concerning the process. Comparison of the solutions with experimental data on the melting of n-octadecane shows excellent agreement.

Modelització de la producció de melanina a partir de polifenol oxidasa d'Agaricus bisporus

Les melanines són un grup heterogeni de polímers producte de reaccions enzimàtiques en els teixits vegetals que contenen compostos fenòlics o polifenòlics. Estudis recents han descobert algunes propietats benèfiques de les melanines sobre la salut, tals com antioxidants, antiinflamatòries, immunològiques i propietats anti-tumorals. Així, no només la seva eliminació ha de ser examinada, sinó que també podria considerar-se la seva addició a aliments funcionals de nova creació. D’aquesta manera, es requereix conèixer el mecanisme cinètic de la lanogènesi abans de la seva possible utilització industrial. S’ha desenvolupat un model cinètic per explicar la formació de melanina a partir de L-tirosina utilitzant polifenol oxidasa d’Agaricus bisporus i monitoritzant l'absorbància de la solució. Aquesta expressió permet descriure la formació de melanina en funció del temps de reacció i obtenir alguns paràmetres importants que defineixen el producte, com el coeficient d'extinció. L’absorbància comença a créixer després d'un període de latència en què es produeixen productes intermedis incolors. El coeficient d'extinció dels productes resultants no és un valor constant, perquè depèn de les condicions de cada experiment. La tirosinasa tingué un menor efecte catalitzador sobre la L-tirosina (primera reacció que catalitza), que sobre L-DOPA (segona reacció).

Analytical expressions for the flame front speed in the downward combustion of thin solid fuels and comparison to experiments

We derive analytical expressions for the propagation speed of downward combustion fronts of thin solid fuels with a background flow initially at rest. The classical combustion model for thin solid fuels that consists of five coupled reaction-convection-diffusion equations is here reduced into a single equation with the gas temperature as the single variable. For doing so we apply a two-zone combustion model that divides the system into a preheating region and a pyrolyzing region. The speed of the combustion front is obtained after matching the temperature and its derivative at the location that separates both regions.We also derive a simplified version of this analytical expression expected to be valid for a wide range of cases. Flame front velocities predicted by our analyticalexpressions agree well with experimental data found in the literature for a large variety of cases and substantially improve the results obtained from a previous well-known analytical expression

Reaction-diffusion wave fronts: Multigeneration biological species under climate change

A generalization of reaction-diffusion models to multigeneration biological species is presented. It is based on more complex random walks than those in previous approaches. The new model is developed analytically up to infinite order. Our predictions for the speed agree to experimental data for several butterfly species better than existing models. The predicted dependence for the speed on the number of generations per year allows us to explain the change in speed observed for a specific invasion

Ab initio benchmark study for the oxidative addition of CH4 to Pd: importance of basis-set flexibility and polarization

To obtain a state-of-the-art benchmark potential energy surface (PES) for the archetypal oxidative addition of the methane C-H bond to the palladium atom, we have explored this PES using a hierarchical series of ab initio methods (Hartree-Fock, second-order Møller-Plesset perturbation theory, fourth-order Møller-Plesset perturbation theory with single, double and quadruple excitations, coupled cluster theory with single and double excitations (CCSD), and with triple excitations treated perturbatively [CCSD(T)]) and hybrid density functional theory using the B3LYP functional, in combination with a hierarchical series of ten Gaussian-type basis sets, up to g polarization. Relativistic effects are taken into account either through a relativistic effective core potential for palladium or through a full four-component all-electron approach. Counterpoise corrected relative energies of stationary points are converged to within 0.1-0.2 kcal/mol as a function of the basis-set size. Our best estimate of kinetic and thermodynamic parameters is -8.1 (-8.3) kcal/mol for the formation of the reactant complex, 5.8 (3.1) kcal/mol for the activation energy relative to the separate reactants, and 0.8 (-1.2) kcal/mol for the reaction energy (zero-point vibrational energy-corrected values in parentheses). This agrees well with available experimental data. Our work highlights the importance of sufficient higher angular momentum polarization functions, f and g, for correctly describing metal-d-electron correlation and, thus, for obtaining reliable relative energies. We show that standard basis sets, such as LANL2DZ+ 1f for palladium, are not sufficiently polarized for this purpose and lead to erroneous CCSD(T) results. B3LYP is associated with smaller basis set superposition errors and shows faster convergence with basis-set size but yields relative energies (in particular, a reaction barrier) that are ca. 3.5 kcal/mol higher than the corresponding CCSD(T) values