The multi-armed bandit (MAB) problem provides a convenient abstraction for many online decision problems arising in modern applications including Internet display advertising, crowdsourcing, online procurement, smart grids, etc. Several variants of the MAB problem have been proposed to extend the basic model to a variety of practical and general settings. The sleeping multi-armed bandit (SMAB) problem is one such variant where the set of available arms varies with time. This study is focused on analyzing the efficacy of the Thompson Sampling algorithm for solving the SMAB problem. Any algorithm for the classical MAB problem is expected to choose one of K available arms (actions) in each of T consecutive rounds. Each choice of an arm generates a stochastic reward from an unknown but fixed distribution. The goal of the algorithm is to maximize the expected sum of rewards over the T rounds (or equivalently minimize the expected total regret), relative to the best fixed action in hindsight. In many real-world settings, however, not all arms may be available in any given round. For example, in Internet display advertising, some advertisers might choose to stay away from the auction due to budget constraints; in crowdsourcing, some workers may not be available at a given time due to timezone difference, etc. Such situations give rise to the sleeping MAB abstraction. In the literature, several upper confidence bound (UCB)-based approaches have been proposed and investigated for the SMAB problem. Our contribution is to investigate the efficacy of a Thomp-son Sampling-based approach. Our key finding is to establish a logarithmic regret bound, which non-trivially generalizes a similar bound known for this approach in the classical MAB setting. Our bound also matches (up to constants) the best-known lower bound for the SMAB problem. Furthermore, we show via detailed simulations, that the Thompson Sampling approach in fact outperforms the known algorithms for the SMAB problem.

Orientador: Ariovaldo Verandio Garcia, Ruben Augusto Romero Lazaro / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-06T06:56:43Z (GMT). No. of bitstreams: 1 RiderFlores_MarcosJulio_D.pdf: 1021887 bytes, checksum: 6000961c2f5457b410ac691912476270 (MD5) Previous issue date: 2006 / Resumo: Neste trabalho são propostos modelos matemáticos e técnicas de solução para resolver o problema de planejamento da expansão de sistemas de transmissão através de três enfoques. a) Usando o modelo de corrente alternada do sistema de transmissão e um algoritmo heurístico construtivo especializado para resolver o problema de planejamento, e, ainda, realiza-se uma primeira tentativa de alocação de fontes de potência reativas; b) Usando o modelo de corrente contínua e técnicas de programação não-linear especializadas. Nesse caso emprega-se uma versão relaxada do problema de planejamento da expansão de sistemas de transmissão usando o modelo de corrente contínua, onde a integralidade das variáveis de investimento é desprezada. Resolve-se o problema de programação não-linear, modelado de forma matricial com um algoritmo de otimização especializado e, além disso, um algoritmo heurístico construtivo especializado é utilizado para resolver o problema de planejamento. c) Usando o modelo de corrente contínua e um algoritmo Branch and Bound (B&B) sem empregar técnicas de decomposição. Para isso foram redefinidos os chamados testes de sondagem no algoritmo B&B e em cada nó da árvore de B&B tem-se um problema de programação não-linear que são resolvidos usando a metodologia desenvolvida no item (b). Os ítens (a), (b) e (c) requerem a solução de problemas de programação não-linear diferenciados. Uma revisão das características principais da resolução iterativa dos métodos de pontos interiores é apresentada. Foi desenvolvida uma técnica baseada em uma combinação de métodos de pontos interiores de alta ordem (MPI-AO) para resolver os problemas de programação não-linear de forma rápida, eficiente e robusta. Essa combinação dos MPI-AO tem como objetivo colocar num único método as características particulares de cada um dos MPI-AO e melhorar o desempenho computacional comparado com os MPI-AO de forma individual / Abstract: In this work mathematical models and solution techniques are proposed to solve the power system transmission expansion planning problem through three approaches: a) Using the nonlinear model ofthe transmission system (AC model) and a specialized constructive heuristic algorithm to solve the problem and, yet, a first attempt to allocate reactive power sources is also considered; b) Using the direct-current (DC) model and specialized techniques of nonlinear programming. In this case a version of the power system transmission expansion planning problem using the DC model where the integrality of the investment variables is relaxed is used. The nonlinear programming problem is solved with a specialized optimization algorithm and, moreover, a constructive heuristic algorithm is employed to solve the planning problem. c) Using the DC model and Branch and Bound (B&B) algorithm without the use of decomposition techniques. The so called fathoming tests of the B&B were redefined and at each node of the tree a nonlinear programming problem is solved using the method developed in b). Items a), b) and c) require the solution of distinct problems of nonlinear programming. A revision of the main characteristics of the iterative solution of the interior points methods is presented. An optimization technique based on a combination of the higher order interior point methods (HO-IPM) had been developed to solve the nonlinear programming problems in a fast, efficient and robust way. This combination of the HO-IPM has as objective to explore the particular characteristics of each method in a single one and to improve the comparative computational performance with the HO-IPM of individual form / Doutorado / Energia Eletrica / Doutor em Engenharia Elétrica

Esta tese estuda a adivinhação nas sete tragédias supérstites de Ésquilo entendendo-se a adivinhação não no sentido estrito de revelação de fatos futuros, mas no sentido mais amplo de um diálogo que se estabelece entre as instâncias divina e humana valendo-se de formas e recursos variados. Assim, a análise e interpretação destas tragédias priorizam os diversos aspectos deste diálogo divinatório para mostrar a adivinhação como fundamento da construção de estratégias dramáticas na tragédia esquiliana, por informar e definir tanto a peculiaridade desta poética quanto sua visão do mundo. / This thesis studies divination in the seven surviving tragedies of Aeschylus. Divination is taken here not in the strict sense of revelation of future events, but in the broader sense of a dialogue that is established between divine and human levels through a variety of forms and resources. Thus, the analysis and interpretation of these tragedies prioritises the different aspects of this divinatory dialogue, in order to show that divination is the basis for constructing the dramatic strategies in the tragedies of Aeschylus, since it informs and defines both the particular features of this poetics and its view of the world.

La plupart des travaux de recherches sur les problèmes d'ordonnancement traitent le cas des ressources renouvelables, c'est-à-dire des ressources qui sont exigées en début d'exécution de chaque tâche et sont restituées en fin d'exécution. Peu d'entre eux abordent les problèmes à ressources consommables, c'est-à-dire des ressources non restituées en fin d'exécution. Le problème de gestion de projet à contraintes de ressources (RCPSP) est le problème à ressources renouvelables le plus traité dans la littérature. Dans le cadre de cette thèse, nous nous sommes intéressés à une généralisation du problème RCPSP qui correspond au cas où les tâches sont remplacées par des événements liés par des relations de précédence étendues. Chaque événement peut produire ou consommer une quantité de ressources à sa date d'occurrence et la fonction économique reste la durée totale à minimiser. Nous avons nommé cette généralisation ERCPSP (Extended RCPSP). Nous avons élaboré des modèles de programmation linéaire pour résoudre ce problème. Nous avons proposé plusieurs bornes inférieures algorithmiques exploitant les travaux de la littérature sur les problèmes cumulatifs. Ensuite, nous avons élargi la portée des méthodes utilisées pour la mise en place de méthodes de séparation et évaluation. Nous avons traité aussi des cas particuliers par des méthodes basées sur la programmation dynamique. / This thesis investigates the Extended Resource Constrained Project Scheduling Problem (ERCPSP). ERCPSP is a general scheduling problem where the availability of a resource is depleted and replenished at the occurrence times of a set of events. It is an extension of the Resource Constrained Project Scheduling Problem (RCPSP) where activities are replaced by events, which have to be scheduled subject to generalized precedence relations. We are interested in this thesis in proposing new methodologies and approaches to solve ERCPSP. First, we study some polynomial cases of this problem and we propose a dynamic programming algorithm to solve the parallel chain case. Then, we propose lower bounds, mixed integer programming models, and a branch-and-bound method to solve ERCPSP. Finally, we develop an instance generator dedicated to this problem.

L’étoile laser polychromatique (ELP) fournit la référence de phase à une optique adaptative (OA)pour corriger les surfaces d’onde turbulentes, y compris leur pente. L’ELP, générée dans la mésosphère parune excitation résonnante à deux photons du sodium, repose sur la déviation chromatique des images. Uneimagerie dans le visible devient possible, et est indispensable pour 80% des programmes astrophysiquesprioritaires de l'E-ELT.L’ELP requiert un écart-type des mesures de position 26 fois inférieur au cas classique. Cela m’a amené àétudier le projecteur laser interférométrique. J’ai mis au point un correcteur de base polychromatique pourégaliser la période des franges et un correcteur de phase pour compenser la réfraction atmosphérique. J’aiétudié l'optique de mesure des franges, et de séparation entre l'ELP et l’objet observé.La précision requise m’a conduit à étudier dans quelles conditions l’algorithme du maximum devraisemblance tend vers la borne de Cramér-Rao.J’ai également développé un modèle numérique de bout en bout pour simuler l’ELP depuis les lasersjusqu’à la mesure du rapport de Strehl. Je montre que pour un VLT, les rapports de Strehl sont supérieurs à40% à 500 nm sans étoile de référence, en prenant une OA qui aurait donné 50% instantané (Strehl depente : 80%). Une approche analytique valide ces résultats.Enfin, j’aborde l’application de l’ELP aux télécommunications interplanétaires et à la destruction des débrisorbitaux. / The polychromatic laser guide star (PLGS) provides adaptive optics (AO) with a phase referenceto correct corrugated wavefronts, including tip tilt. It relies on the chromatic dispersion of light returnedfrom the 2-photon resonant excitation of sodium in the mesosphere. Diffraction limited imaging in thevisible then becomes possible. This is mandatory for 80% of the prominent astrophysical cases for the EELT.A PLGS requires standard deviations of position measurements 26 times less than in classical cases. Thus Ihave studied the interferometric laser projector. I have designed a polychromatic base corrector to equalizethe fringe periods, a phase corrector to compensate atmospheric refraction and the optics for fringemeasurements and for keeping apart the PLGS from the science target images.The required accuracy leads me to study how the maximum likelihood algorithm approaches the Cramer-Rao bound.I have written an end-to-end code for numerical simulations of the PLGS, from the lasers to the Strehlmeasurement. I get for the VLT Strehl ratios larger than 40% at 500 nm if one uses an AO providing us a50% instantaneous Strehl (tip tilt Strehl : 80%). An analytical model validates these results.Finally I address the application of the PLGS to deep space communications and to space debris clearing.

On s'attache dans cette thèse à plusieurs aspects liés à la résolution de problèmes multi-objectifs, sans se limiter au cas biobjectif. Nous considérons la résolution exacte, dans le sens de la détermination de l'ensemble des points non dominés, ainsi que la résolution approchée dans laquelle on cherche une approximation de cet ensemble dont la qualité est garantie a priori.Nous nous intéressons d'abord au problème de la détermination d'une représentation explicite de la région de recherche. La région de recherche, étant donné un ensemble de points réalisables connus, exclut la partie de l'espace des objectifs que dominent ces points et constitue donc la partie de l'espace des objectifs où les efforts futurs doivent être concentrés dans la perspective de déterminer tous les points non dominés.Puis nous considérons le recours aux algorithmes de séparation et évaluation ainsi qu'aux algorithmes de ranking afin de proposer une nouvelle méthode hybride de détermination de l'ensemble des points non dominés. Nous montrons que celle-ci peut également servir à obtenir une approximation de l'ensemble des points non dominés. Cette méthode est implantée pour le problème de l'arbre couvrant de poids minimal. Les quelques propriétés de ce problème que nous passons en revue nous permettent de spécialiser certaines procédures et d'intégrer des prétraitements spécifiques. L'intérêt de cette approche est alors soutenu à l'aide de résultats expérimentaux. / This thesis deals with several aspects related to solving multi-objective problems, without restriction to the bi-objective case. We consider exact solving, which generates the nondominated set, and approximate solving, which computes an approximation of the nondominated set with a priori guarantee on the quality.We first consider the determination of an explicit representation of the search region. The search region, defined with respect to a set of known feasible points, excludes from the objective space the part which is dominated by these points. Future efforts to find all nondominated points should therefore be concentrated on the search region.Then we review branch and bound and ranking algorithms and we propose a new hybrid approach for the determination of the nondominated set. We show how the proposed method can be adapted to generate an approximation of the nondominated set. This approach is instantiated on the minimum spanning tree problem. We review several properties of this problem which enable us to specialize some procedures of the proposed approach and integrate specific preprocessing rules. This approach is finally supported through experimental results.

Un graphe métrique G(X;D) est un graphe dont l’ensemble des sommets est l’ensemble X des points d’un espace métrique (X; d), et dont les arêtes relient les paires fx; yg de sommets telles que d(x; y) 2 D. Dans cette thèse, nous considérons deux problèmes qui peuvent être interprétés comme des problèmes de graphes métriques dans Rn. Premièrement, nous nous intéressons au célèbre problème d’empilements de sphères, relié au graphe métrique G(Rn; ]0; 2r[) pour un rayon de sphère r donné. Récemment, Venkatesh a amélioré d’un facteur log log n la meilleure borne inférieure connue pour un empilement de sphères donné par un réseau, pour une suite infinie de dimensions n. Ici nous prouvons une version effective de ce résultat, dans le sens où l’on exhibe, pour la même suite de dimensions, des familles finies de réseaux qui contiennent un réseaux dont la densité atteint la borne de Venkatesh. Notre construction met en jeu des codes construits sur des corps cyclotomiques, relevés en réseaux grâce à un analogue de la Construction A. Nous prouvons aussi un résultat similaire pour des familles de réseaux symplectiques. Deuxièmement, nous considérons le graphe distance-unité G associé à une norme k_k. Le nombre m1 (Rn; k _ k) est défini comme le supremum des densités réalisées par les stables de G. Si la boule unité associée à k _ k pave Rn par translation, alors il est aisé de voir que m1 (Rn; k _ k) > 1 2n . C. Bachoc et S. Robins ont conjecturé qu’il y a égalité. On montre que cette conjecture est vraie pour n = 2 ainsi que pour des régions de Voronoï de plusieurs types de réseaux en dimension supérieure, ceci en se ramenant à la résolution de problèmes d’empilement dans des graphes discrets. / A distance graph G(X;D) is a graph whose set of vertices is the set of points X of a metric space (X; d), and whose edges connect the pairs fx; yg such that d(x; y) 2 D. In this thesis, we consider two problems that may be interpreted in terms of distance graphs in Rn. First, we study the famous sphere packing problem, in relation with thedistance graph G(Rn; (0; 2r)) for a given sphere radius r. Recently, Venkatesh improved the best known lower bound for lattice sphere packings by a factor log log n for infinitely many dimensions n. We prove an effective version of this result, in the sense that we exhibit, for the same set of dimensions, finite families of lattices containing a lattice reaching this bound. Our construction uses codes over cyclotomic fields, lifted to lattices via Construction A. We also prove a similar result for families of symplectic lattices. Second, we consider the unit distance graph G associated with a norm k _ k. The number m1 (Rn; k _ k) is defined as the supremum of the densities achieved by independent sets in G. If the unit ball corresponding with k _ k tiles Rn by translation, then it is easy to see that m1 (Rn; k _ k) > 1 2n . C. Bachoc and S. Robins conjectured that the equality always holds. We show that this conjecture is true for n = 2 and for several Voronoï cells of lattices in higher dimensions, by solving packing problems in discrete graphs.

Weak intermolecular interactions have very strong impact on the structures and properties of life giving molecules like H2O, DNA, RNA etc. These interactions are responsible for many biological phenomena. The directional preference of some of these interactions is used for designing different synthetic approaches in the supramolecular chemistry. The work reported in this Thesis comprises of investigations of weak intermolecular interactions in gas phase using home-built Pulsed Nozzle Fourier Transform Microwave (PN-FTMW) spectrometer as an experimental tool and ab-initio and Atoms in Molecules (AIM) theory as theoretical tools. The spectrometer which is coupled with a pulsed nozzle is used to record pure rotational spectra of the molecular clusters in a jet cooled molecular beam. In the molecular beam molecules/complexes are free from interactions with other molecules/complexes and thus, spectroscopy in the molecular beams provides information about the 'isolated' molecule/complex under investigation. The rotational spectra of the molecules/complexes in the molecular beam provide their geometry in the ground vibrational states. These experimental geometries can be used to test the performance and accuracy of theoretical models like ab-initio theory, when applied to the weakly bound complexes. Further the AIM theory can be used to gain insights into the nature and strength of the intermolecular interactions present in the system under investigation. Chapter I of this Thesis gives a brief introduction of intermolecular interactions. Other than hydrogen bonding, which is considered as the most important intermolecular interaction, many other intermolecular interactions involving different atoms have been observed in past few decades. The chapter summarizes all these interactions. The chapter also gives a brief introduction to the experimental and theoretical methods used to probe these interactions. In Chapter II, the experimental and theoretical methods used in this work are summarized. Details of our home-built PN-FTMW spectrometer are given in this chapter. The chapter also discusses briefly the theoretical methods like ab-initio, AIM and Natural bond orbital (NBO) analysis. We have made few changes in the mode of control of one of our delay generators which have also been described. Chapter III and Chapter V of this Thesis are dedicated to the propargyl alcohol complexes. Propargyl alcohol (PA) is a molecule of astrophysical interest. It is also important in combustion chemistry since propargyl radical is considered as the precursor in soot formation. Moreover, PA is a multifunctional molecule, having a hydroxyl (-OH) and an acetylenic (-C≡C-H) group. Both of the groups can individually act as hydrogen bond acceptor as well as donor and thus PA provides an exciting possibility of studying many different types of weak interactions. Due to internal motion of -OH group, PA monomer can exist in gauche as well as trans form. However, rotational spectra of PA-monomer show the presence of only gauche conformer. In Chapter III, rotational spectra of Ar•••PA complex are discussed. The pure rotational spectra of the parent Ar•••PA complex and its two deuterated isotopologues, Ar•••PA-D (OD species) and Ar•••PA-D (CD species), could be observed and fitted within experimental uncertainty. The structural fitting confirmed a structure in which PA is present as gauche conformer and argon interacts with both the O-H group and the acetylenic group leading to Ar•••H-O and Ar•••π interactions respectively. Presence of these interactions was further confirmed by AIM theoretical analysis. In all the three isotopologues c-type rotational transitions showed significant splitting. Splitting patterns in the three isotopologues suggest that it originates mainly due to the large amplitude motion of the hydroxyl group and the motion is weakly coupled with the carbon chain bending motion. No evidence for the complex with trans conformer of PA was found. Although, we could not observe Ar•••trans-PA complex experimentally, we decided to perform ab-initio and AIM theoretical calculations on this complex as well. AIM calculations suggested the presence of Ar•••H-O and a unique Ar•••C interaction in this complex which was later found to be present in the Ar•••methanol complex as well. This prompted us to explore different possible interactions in methanol, other than the well known O-H•••O hydrogen bonding interactions, and eventually led us to an interesting interaction which we termed as carbon bond. Chapter IV discusses carbon bonding interaction in different complexes. Electrostatic potential (ESP) calculations show that tetrahedral face of methane is electron-rich and thus can act as hydrogen/halogen bond acceptor. This has already been observed in many complexes, e.g. CH4•••H2O/HF/HCl/ClF etc., both experimentally and theoretically. However, substitution of one of the hydrogens of methane with -OH leads to complete reversal of the properties of the CH3 tetrahedral face and this face in methanol is electron-deficient. We found that CH3 face in methanol interacts with electron rich sites of HnY molecules and leads to the formation of complexes stabilized by Y•••C-X interactions. This interaction was also found to be present in the complexes of many different CH3X (X=OH/F/Cl/Br/NO2/NF2 etc.) molecules. AIM, NBO and C-X frequency shift analyses suggest that this interaction could be termed as "carbon bond". The carbon bonding interactions could be important in understanding hydrophobic interactions and thus could play an important role in biological phenomena like protein folding. The carbon bonding interaction could also play a significant role in the stabilization of the transition state in SN2 reactions. In Chapter V of this Thesis rotational spectra of propargyl alcohol dimer are discussed. Rotational spectra of the parent dimer and its three deuterated (O-D) isotopologues (two mono-substituted and one bi-substituted) could be recorded and fitted within experimental uncertainty. The fitted rotational constants are close to one of the ab-initio predicted structure. In the dimer also propargyl alcohol exists in the gauche form. Atoms in molecules analysis suggests that the experimentally observed dimer is bound by O-H•••O, O-H•••π and C-H•••π interactions. Chapter VI of the thesis explores the 'electrophore concept'. To observe the rotational spectra of any species and determine its rotational constant by microwave spectroscopy, the species should have a permanent dipole moment. Can we obtain rotational constants of a species having no dipole moment via microwave spectroscopy? Electrophore concept can be used for this purpose. An electrophore is an atom or molecule which could combine with another molecule having no dipole moment thereby forming a complex with a dipole moment, e.g. Argon atom is an electrophore in Ar•••C6H6 complex. The microwave spectra of Ar•••13CC5H6 and Ar•••C6H5D complexes were recorded and fitted. The A rotational constant of these complexes was found to be equal to the C rotational constant of 13CC5H6 and C6H5D molecules respectively and thus we could determine the C rotational constant of microwave 'inactive' 13CC5H6. This concept could be used to obtain the rotational spectra of parallel displaced benzene-dimer if it exists. We recently showed that the square pyramidal Fe(CO)5 can act as hydrogen bond acceptor. Appendix I summarizes the extension of this work and discusses interactions of trigonal bipyramidal Fe(CO)5 with HF, HCl, HBr and ClF. Our initial attempts on generating a chirped pulse to be used in a new broadband spectrometer are summarized in Appendix II. Preliminary investigations on the propargyl•••water complex are summarized in Appendix III.

The introduction of negative interest rates, as an additional tool of unconventional monetary policy in recent years, leads to many, yet largely polarized debates about the implications of the unconventional monetary policy of central banks. The main reason for their introduction in some economies, namely the threat of deflation, is also questioned, as well as the negative side effects of this instrument on financial stability and economic growth, and finally, their limited scope is under scrutiny. Due to the still insufficient research in this field, the aim of the diploma thesis is to evaluate the concerns and the main long-term risks and impacts of the monetary policy using negative interest rates in the context of deflation threats. The analysis of available data confirms both the concerns about the negative effects of negative interest rates on the banking sector, the recovery process of public finances, the possible rise of a price bubble on the financial markets and real estate markets, or on the possibility of excessive risk taking, and on the other hand draws attention to the direct dependence between negative impacts and the duration of monetary policy using negative interest rates.

L’objectif de ce travail est de comprendre les effets d’un changement de composition des modèles fromagers sur la mobilité, la libération et la perception de molécules de la flaveur (sel, composés d’arôme). Six modèles fromagers aromatisés ont été formulés (3 ratios lipides/protéines (L/P) et 2 teneurs en sel). La microstructure et les propriétés rhéologiques des modèles fromagers ont été caractérisées respectivement par microscopie confocale et par compression uniaxiale. La mobilité des ions sodium a été analysée par RMN 23Na. La cinétique de libération des ions sodium a été suivie dans l’eau, puis dans la salive, en situation de consommation. La libération rétronasale des composés d’arôme a été suivie par nose-space APCI-MS, simultanément au suivi des déglutitions, et de la mastication par électromyographie. Enfin, les propriétés sensorielles des modèles fromagers (intensité salée, arôme, texture) ont été étudiées.Une diminution du ratio L/P et une diminution de la teneur en sel diminuent la taille des gouttelettes lipidiques et augmentent la fermeté. Cela conduit à une diminution de la mobilité des ions sodium, qui se traduit par une diminution de la quantité de sodium libéré dans la salive et une diminution de la perception salée. De plus, le maximum de libération d’arôme est atteint plus tard et la perception aromatique est diminuée. Ces effets peuvent être expliqués par la répartition lipides/protéines observée en microscopie, par une déglutition plus tardive et une plus grande activité masticatoire / The aim of this work is to understand the effects of changes in the composition of model cheeses on mobility, release and perception of flavor molecules (salt, aroma compounds). Six flavoured model cheeses were formulated (3 lipid/protein (L/P) ratios and 2 salt contents). The microstructure and the rheological properties of the model cheeses were characterized respectively by confocal microscopy and by uniaxial compression test. The mobility of sodium ions was analyzed by 23Na NMR. The kinetics of sodium release was followed in water and then in saliva during consumption of the model cheeses. The retronasal release of aroma compounds was followed by nose-space APCI-MS, simultaneously with the study of swallowing and chewing by electromyography. Then, the sensory properties of the model cheeses (saltiness, aroma, texture) were studied.A decrease in the L/P ratio and a decrease in salt content reduce the fat droplet size and increase the hardness. This leads to a decrease in sodium ion mobility, which results in a decrease in sodium release in saliva and a decrease in salty perception. Moreover, the maximum intensity of aroma release is reached later and the aroma perception is decreased. These effects can be explained by the observed lipid/protein distribution in microscopy, by the later swallowing and by the higher chewing activity