Bidirectional Helmholtz Machines

L'entraînement sans surveillance efficace et inférence dans les modèles génératifs profonds reste un problème difficile. Une approche assez simple, la machine de Helmholtz, consiste à entraîner du haut vers le bas un modèle génératif dirigé qui sera utilisé plus tard pour l'inférence approximative. Des résultats récents suggèrent que de meilleurs modèles génératifs peuvent être obtenus par de meilleures procédures d'inférence approximatives. Au lieu d'améliorer la procédure d'inférence, nous proposons ici un nouveau modèle, la machine de Helmholtz bidirectionnelle, qui garantit qu'on peut calculer efficacement les distributions de haut-vers-bas et de bas-vers-haut. Nous y parvenons en interprétant à les modèles haut-vers-bas et bas-vers-haut en tant que distributions d'inférence approximative, puis ensuite en définissant la distribution du modèle comme étant la moyenne géométrique de ces deux distributions. Nous dérivons une borne inférieure pour la vraisemblance de ce modèle, et nous démontrons que l'optimisation de cette borne se comporte en régulisateur. Ce régularisateur sera tel que la distance de Bhattacharyya sera minisée entre les distributions approximatives haut-vers-bas et bas-vers-haut. Cette approche produit des résultats de pointe en terme de modèles génératifs qui favorisent les réseaux significativement plus profonds. Elle permet aussi une inférence approximative amérliorée par plusieurs ordres de grandeur. De plus, nous introduisons un modèle génératif profond basé sur les modèles BiHM pour l'entraînement semi-supervisé.

Properties of a method of fundamental solutions for the parabolic heat equation

We show that a set of fundamental solutions to the parabolic heat equation, with each element in the set corresponding to a point source located on a given surface with the number of source points being dense on this surface, constitute a linearly independent and dense set with respect to the standard inner product of square integrable functions, both on lateral- and time-boundaries. This result leads naturally to a method of numerically approximating solutions to the parabolic heat equation denoted a method of fundamental solutions (MFS). A discussion around convergence of such an approximation is included.

Bidirectional Helmholtz Machines

L'entraînement sans surveillance efficace et inférence dans les modèles génératifs profonds reste un problème difficile. Une approche assez simple, la machine de Helmholtz, consiste à entraîner du haut vers le bas un modèle génératif dirigé qui sera utilisé plus tard pour l'inférence approximative. Des résultats récents suggèrent que de meilleurs modèles génératifs peuvent être obtenus par de meilleures procédures d'inférence approximatives. Au lieu d'améliorer la procédure d'inférence, nous proposons ici un nouveau modèle, la machine de Helmholtz bidirectionnelle, qui garantit qu'on peut calculer efficacement les distributions de haut-vers-bas et de bas-vers-haut. Nous y parvenons en interprétant à les modèles haut-vers-bas et bas-vers-haut en tant que distributions d'inférence approximative, puis ensuite en définissant la distribution du modèle comme étant la moyenne géométrique de ces deux distributions. Nous dérivons une borne inférieure pour la vraisemblance de ce modèle, et nous démontrons que l'optimisation de cette borne se comporte en régulisateur. Ce régularisateur sera tel que la distance de Bhattacharyya sera minisée entre les distributions approximatives haut-vers-bas et bas-vers-haut. Cette approche produit des résultats de pointe en terme de modèles génératifs qui favorisent les réseaux significativement plus profonds. Elle permet aussi une inférence approximative amérliorée par plusieurs ordres de grandeur. De plus, nous introduisons un modèle génératif profond basé sur les modèles BiHM pour l'entraînement semi-supervisé.

A symbolic approach to nonlinearly perturbed heat equation

We consider a system described by the linear heat equation with adiabatic boundary conditions which is perturbed periodicaly. This perturbation is nonlinear and is characterized by a one-parameter family of quadratic maps. The system, depending on the parameters, presents very complex behaviour. We introduce a symbolic framework to analyze the system and resume its most important features.

A novel two-stage discrete crack method based on the screened Poisson equation and local mesh refinement

We propose an alternative crack propagation algo- rithm which effectively circumvents the variable transfer procedure adopted with classical mesh adaptation algo- rithms. The present alternative consists of two stages: a mesh-creation stage where a local damage model is employed with the objective of defining a crack-conforming mesh and a subsequent analysis stage with a localization limiter in the form of a modified screened Poisson equation which is exempt of crack path calculations. In the second stage, the crack naturally occurs within the refined region. A staggered scheme for standard equilibrium and screened Poisson equa- tions is used in this second stage. Element subdivision is based on edge split operations using a constitutive quantity (damage). To assess the robustness and accuracy of this algo- rithm, we use five quasi-brittle benchmarks, all successfully solved.

Damage and fracture algorithm using the screened Poisson equation and local remeshing

We propose a crack propagation algorithm which is independent of particular constitutive laws and specific element technology. It consists of a localization limiter in the form of the screened Poisson equation with local mesh refinement. This combination allows the cap- turing of strain localization with good resolution, even in the absence of a sufficiently fine initial mesh. In addition, crack paths are implicitly defined from the localized region, cir- cumventing the need for a specific direction criterion. Observed phenomena such as mul- tiple crack growth and shielding emerge naturally from the algorithm. In contrast with alternative regularization algorithms, curved cracks are correctly represented. A staggered scheme for standard equilibrium and screened equations is used. Element subdivision is based on edge split operations using a given constitutive quantity (either damage or void fraction). To assess the robustness and accuracy of this algorithm, we use both quasi-brittle benchmarks and ductile tests.

How to get an equation named after you (part I)

Abstract|IT Nelle lauree triennali le pubblicazioni scientifiche non vengono studiate. Per quanto possa rivelarsi un piccolo intervento, lo scopo di questa tesi è invece quello di rendere più accessibili ai pochi interessati alcune vecchie e polverose pubblicazioni. Inoltre, la presente tesi non dovrebbe essere trattata come un lavoro a se stante, ma come il primo mattone di un progetto che comprende tutte le maggiori pubblicazioni della storia. How to get an equation named after you (Part I) in particolare discute la serie di pubblicazioni del 1926 di Schrödinger "Quantisierung als Eigenwertproblem" e l’articolo del 1928 di Dirac "The Quantum Theory of the Electron", ovvero quei lavori dove le equazioni di Schrödinger e Dirac vennero per prime derivate. La serie di articoli del 1926 sommano ad un totale di oltre 100 pagine. Inizialmente Schrödinger dimostra come ciò su cui è basata la sua teoria possa spiegare correttamente fenomeni conosciuti come l’atomo di idrogeno e l’effetto Stark, per poi derivare la famosa equazione d’onda complessa del secondo ordine. I procedimenti matematici, in questi articoli, sono complicati e molte delle dimostrazioni non vengono mostrate oppure risultano inutilmente lunghe. La pubblicazione di Dirac invece ha principalmente a che fare con la derivazione dell’equazione, la sua generalizzazione e l’invarianza relativistica. Dimostra inoltre che tale equazione è compatibile con passate teorie. La lettura di Dirac è molto più sistematica, dato il largo utilizzo di dimostrazioni matematiche laddove Schrödinger avrebbe usato parole.

Dynamical models in neuroscience: the delay FitzHugh-Nagumo equation

Il primo modello matematico in grado di descrivere il prototipo di un sistema eccitabile assimilabile ad un neurone fu sviluppato da R. FitzHugh e J. Nagumo nel 1961. Tale modello, per quanto schematico, rappresenta un importante punto di partenza per la ricerca nell'ambito neuroscientifico delle dinamiche neuronali, ed è infatti capostipite di una serie di lavori che hanno puntato a migliorare l’accuratezza e la predicibilità dei modelli matematici per le scienze. L’elevato grado di complessità nello studio dei neuroni e delle dinamiche inter-neuronali comporta, tuttavia, che molte delle caratteristiche e delle potenzialità dell’ambito non siano ancora state comprese appieno. In questo lavoro verrà approfondito un modello ispirato al lavoro originale di FitzHugh e Nagumo. Tale modello presenta l’introduzione di un termine di self-coupling con ritardo temporale nel sistema di equazioni differenziali, diventa dunque rappresentativo di modelli di campo medio in grado di descrivere gli stati macroscopici di un ensemble di neuroni. L'introduzione del ritardo è funzionale ad una descrizione più realistica dei sistemi neuronali, e produce una dinamica più ricca e complessa rispetto a quella presente nella versione originale del modello. Sarà mostrata l'esistenza di una soluzione a ciclo limite nel modello che comprende il termine di ritardo temporale, ove tale soluzione non può essere interpretata nell’ambito delle biforcazioni di Hopf. Allo scopo di esplorare alcune delle caratteristiche basilari della modellizzazione del neurone, verrà principalmente utilizzata l’impostazione della teoria dei sistemi dinamici, integrando dove necessario con alcune nozioni provenienti dall’ambito fisiologico. In conclusione sarà riportata una sezione di approfondimento sulla integrazione numerica delle equazioni differenziali con ritardo.

Improved Selective Cholesterol Adsorption By Molecularly Imprinted Poly(methacrylic Acid)/silica (pmaa-sio₂) Hybrid Material Synthesized With Different Molar Ratios.

The present paper describes the synthesis of molecularly imprinted polymer - poly(methacrylic acid)/silica and reports its performance feasibility with desired adsorption capacity and selectivity for cholesterol extraction. Two imprinted hybrid materials were synthesized at different methacrylic acid (MAA)/tetraethoxysilane (TEOS) molar ratios (6:1 and 1:5) and characterized by FT-IR, TGA, SEM and textural data. Cholesterol adsorption on hybrid materials took place preferably in apolar solvent medium, especially in chloroform. From the kinetic data, the equilibrium time was reached quickly, being 12 and 20 min for the polymers synthesized at MAA/TEOS molar ratio of 6:1 and 1:5, respectively. The pseudo-second-order model provided the best fit for cholesterol adsorption on polymers, confirming the chemical nature of the adsorption process, while the dual-site Langmuir-Freundlich equation presented the best fit to the experimental data, suggesting the existence of two kinds of adsorption sites on both polymers. The maximum adsorption capacities obtained for the polymers synthesized at MAA/TEOS molar ratios of 6:1 and 1:5 were found to be 214.8 and 166.4 mg g(-1), respectively. The results from isotherm data also indicated higher adsorption capacity for both imprinted polymers regarding to corresponding non-imprinted polymers. Nevertheless, taking into account the retention parameters and selectivity of cholesterol in the presence of structurally analogue compounds (5-α-cholestane and 7-dehydrocholesterol), it was observed that the polymer synthesized at the MAA/TEOS molar ratio of 6:1 was much more selective for cholesterol than the one prepared at the ratio of 1:5, thus suggesting that selective binding sites ascribed to the carboxyl group from MAA play a central role in the imprinting effect created on MIP.

Graded Meshes In Bio-thermal Problems With Transmission-line Modeling Method.

In this study, the transmission-line modeling (TLM) applied to bio-thermal problems was improved by incorporating several novel computational techniques, which include application of graded meshes which resulted in 9 times faster in computational time and uses only a fraction (16%) of the computational resources used by regular meshes in analyzing heat flow through heterogeneous media. Graded meshes, unlike regular meshes, allow heat sources to be modeled in all segments of the mesh. A new boundary condition that considers thermal properties and thus resulting in a more realistic modeling of complex problems is introduced. Also, a new way of calculating an error parameter is introduced. The calculated temperatures between nodes were compared against the results obtained from the literature and agreed within less than 1% difference. It is reasonable, therefore, to conclude that the improved TLM model described herein has great potential in heat transfer of biological systems.

Relationship Between Aerobic And Anaerobic Parameters From 3-minute All-out Tethered Swimming And 400-m Maximal Front Crawl Effort.

The main aim of this investigation was to verify the relationship of the variables measured during a 3-minute all-out test with aerobic (i.e., peak oxygen uptake [(Equation is included in full-text article.)] and intensity corresponding to the lactate minimum [LMI]) and anaerobic parameters (i.e., anaerobic work) measured during a 400-m maximal performance. To measure force continually and to avoid the possible influences caused by turns, the 3-minute all-out effort was performed in tethered swimming. Thirty swimmers performed the following tests: (a) a 3-minute all-out tethered swimming test to determine the final force (equivalent to critical force: CF3-MIN) and the work performed above CF3-MIN (W'3-MIN), (b) a LMI protocol to determine the LMI during front crawl swimming, and (c) a 400-m maximal test to determine the (Equation is included in full-text article.)and total anaerobic contribution (WANA). Correlations between the variables were tested using the Pearson's correlation test (p ≤ 0.05). CF3-MIN (73.9 ± 13.2 N) presented a high correlation with the LMI (1.33 ± 0.08 m·s; p = 0.01) and (Equation is included in full-text article.)(4.5 ± 1.2 L·min; p = 0.01). However, the W'3-MIN (1,943.2 ± 719.2 N·s) was only moderately correlated with LMI (p = 0.02) and (Equation is included in full-text article.)(p = 0.01). In summary, CF3-MIN determined during the 3-minute all-out effort is associated with oxidative metabolism and can be used to estimate the aerobic capacity of swimmers. In contrast, the anaerobic component of this model (W'3-MIN) is not correlated with WANA.

Modeling The Effects Of Temperature And Ph On The Resistance Of Alicyclobacillus Acidoterrestris In Conventional Heat-treated Fruit Beverages Through A Meta-analysis Approach.

In this work, all publicly-accessible published findings on Alicyclobacillus acidoterrestris heat resistance in fruit beverages as affected by temperature and pH were compiled. Then, study characteristics (protocols, fruit and variety, °Brix, pH, temperature, heating medium, culture medium, inactivation method, strains, etc.) were extracted from the primary studies, and some of them incorporated to a meta-analysis mixed-effects linear model based on the basic Bigelow equation describing the heat resistance parameters of this bacterium. The model estimated mean D* values (time needed for one log reduction at a temperature of 95 °C and a pH of 3.5) of Alicyclobacillus in beverages of different fruits, two different concentration types, with and without bacteriocins, and with and without clarification. The zT (temperature change needed to cause one log reduction in D-values) estimated by the meta-analysis model were compared to those ('observed' zT values) reported in the primary studies, and in all cases they were within the confidence intervals of the model. The model was capable of predicting the heat resistance parameters of Alicyclobacillus in fruit beverages beyond the types available in the meta-analytical data. It is expected that the compilation of the thermal resistance of Alicyclobacillus in fruit beverages, carried out in this study, will be of utility to food quality managers in the determination or validation of the lethality of their current heat treatment processes.

Supercritical CO2 recovery of caffeine from green coffee oil: new experimental solubility data and modeling

The caffeine solubility in supercritical CO2 was studied by assessing the effects of pressure and temperature on the extraction of green coffee oil (GCO). The Peng-Robinson¹ equation of state was used to correlate the solubility of caffeine with a thermodynamic model and two mixing rules were evaluated: the classical mixing rule of van der Waals with two adjustable parameters (PR-VDW) and a density dependent one, proposed by Mohamed and Holder² with two (PR-MH, two parameters adjusted to the attractive term) and three (PR-MH3 two parameters adjusted to the attractive and one to the repulsive term) adjustable parameters. The best results were obtained with the mixing rule of Mohamed and Holder² with three parameters.

A segunda lei da termodinâmica

Considering intrinsic characteristics of the system exclusively, both statistical and information theory interpretations of the second law are used to provide more comprehensive meanings for the concepts of entropy, temperature, and Helmholtz and Gibbs energies. The coherence of Clausius inequality to these concepts is emphasized. The aim of this work is to re-discuss the second law of thermodynamics in accordance to homogeneous processes thermodynamics, a temporal science which is the very special oversimplification of continuum mechanics for spatially constant intensive properties.

Resolvendo a equação de Schrödinger utilizando procedimentos numéricos fundamentais

A combination of the variational principle, expectation value and Quantum Monte Carlo method is used to solve the Schrödinger equation for some simple systems. The results are accurate and the simplicity of this version of the Variational Quantum Monte Carlo method provides a powerful tool to teach alternative procedures and fundamental concepts in quantum chemistry courses. Some numerical procedures are described in order to control accuracy and computational efficiency. The method was applied to the ground state energies and a first attempt to obtain excited states is described.