Los sistemas software actuales son artefactos complejos cuyo comportamiento es a menudo extremadamente difícil de entender. Este hecho ha llevado al desarrollo de metodologías formales muy sofisticadas para el análisis, comprensión y depuración de programas. El análisis de trazas de ejecución consiste en la búsqueda dinámica de contenidos específicos dentro de las trazas de ejecución de un cierto programa. La búsqueda puede llevarse a cabo hacia adelante o hacia atrás. Si bien el análisis hacia adelante se traduce en una forma de análisis de impacto que identifica el alcance y las posibles consecuencias de los cambios en la entrada del programa, el análisis hacia atrás permite llevar a cabo un rastreo de la procedencia; es decir, muestra como (partes de) la salida del programa depende de (partes de) su entrada y ayuda a estimar qué dato de la entrada es necesario modificar para llevar a cabo un cambio en el resultado. En esta tesis se investiga una serie de metodologías de análisis de trazas que son especialmente adecuadas para el análisis de trazas de ejecución largas y complejas en la lógica de reescritura, que es un marco lógico y semántico especialmente adecuado para la formalización de sistemas altamente concurrentes. La primera parte de la tesis se centra en desarrollar una técnica de análisis de trazas hacia atrás que alcanza enormes reducciones en el tamaño de la traza. Esta metodología se basa en la fragmentación incremental y favorece un mejor análisis y depuración ya que la mayoría de las inspecciones, tediosas e irrelevantes, que se realizan rutinariamente en el diagnostico y la localización de errores se pueden eliminar de forma automática. Esta técnica se ilustra por medio de varios ejemplos que ejecutamos mediante el sistema iJulienne, una herramienta interactiva de fragmentación que hemos desarrollado y que implementa la técnica de análisis de trazas hacia atrás. En la segunda parte de la tesis se formaliza un sistema paramétrico, flexible y dinámico, para la exploración de computaciones en la lógica de reescritura. El esquema implementa un algoritmo de animación gen érico que permite la ejecución indeterminista de una teoría de reescritura condicional dada y que puede ser objeto de seguimiento mediante el uso de diferentes modalidades, incluyendo una ejecución gradual paso a paso y una fragmentación automática hacia adelante y/o hacia atrás, lo que reduce drásticamente el tamaño y la complejidad de las trazas bajo inspección y permite a los usuarios evaluar de forma aislada los efectos de una declaración o instrucción dada, el seguimiento de los efectos del cambio de la entrada, y obtener información sobre el comportamiento del programa (o mala conducta del mismo). Por otra parte, la fragmentación de la traza de ejecución puede identificar nuevas oportunidades de optimización del programa. Con esta metodología, un analista puede navegar, fragmentar, filtrar o buscar en la traza durante la ejecución del programa. El marco de análisis de trazas gen érico se ha implementado en el sistema Anima y describimos una profunda evaluación experimental de este que demuestra la utilidad del enfoque propuesto. / Frechina Navarro, F. (2014). A Rewriting-based, Parameterized Exploration Scheme for the Dynamic Analysis of Complex Software Systems [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/44234

Esta tesis propone una metodología de análisis dinámico que mejora el diagnóstico de programas erróneos escritos en el lenguaje Maude. La idea clave es combinar técnicas de verificación de aserciones en tiempo de ejecución con la fragmentación dinámica de trazas de ejecución para detectar automáticamente errores en tiempo de ejecución, al tiempo que se reduce el tamaño y la complejidad de las trazas a analizar. En el caso de violarse una aserción, se infiere automáticamente el criterio de fragmentación, lo que facilita al usuario identificar rápidamente la fuente del error. En primer lugar, la tesis formaliza una técnica destinada a detectar automáticamente eventuales desviaciones del comportamiento deseado del programa (síntomas de error). Esta técnica soporta dos tipos de aserciones definidas por el usuario: aserciones funcionales (que restringen llamadas a funciones deterministas) y aserciones de sistema (que especifican los invariantes de estado del sistema). La técnica de verificación dinámica propuesta es demostrablemente correcta en el sentido de que todos los errores señalados definitivamente delatan la violación de las aserciones. Tras eventuales violaciones de aserciones, se generan automáticamente trazas fragmentadas (es decir, trazas simplificadas pero igualmente precisas) que ayudan a identificar la causa del error. Además, la técnica también sugiere una posible reparación para las reglas implicadas en la generación de los estados erróneos. La metodología propuesta se basa en (i) una notación lógica para especificar las aserciones que se imponen a la ejecución; (ii) una técnica de verificación aplicable en tiempo de ejecución que comprueba dinámicamente las aserciones; y (iii) un mecanismo basado en la generalización (ecuacional) menos general que automáticamente obtiene criterios precisos para fragmentar trazas de ejecución a partir de aserciones falsificadas. Por último, se presenta una implementación de la técnica propuesta en la herramienta de análisis dinámico basado en aserciones ABETS, que muestra cómo es posible combinar el trazado de las propiedades asertadas del programa para obtener un algoritmo preciso de análisis de trazas que resulta útil para el diagnóstico y la depuración de programas. / This thesis proposes a dynamic analysis methodology for improving the diagnosis of erroneous Maude programs. The key idea is to combine runtime assertion checking and dynamic trace slicing for automatically catching errors at runtime while reducing the size and complexity of the erroneous traces to be analyzed (i.e., those leading to states that fail to satisfy the assertions). In the event of an assertion violation, the slicing criterion is automatically inferred, which facilitates the user to rapidly pinpoint the source of the error. First, a technique is formalized that aims at automatically detecting anomalous deviations of the intended program behavior (error symptoms) by using assertions that are checked at runtime. This technique supports two types of user-defined assertions: functional assertions (which constrain deterministic function calls) and system assertions (which specify system state invariants). The proposed dynamic checking is provably sound in the sense that all errors flagged definitely signal a violation of the specifications. Then, upon eventual assertion violations, accurate trace slices (i.e., simplified yet precise execution traces) are generated automatically, which help identify the cause of the error. Moreover, the technique also suggests a possible repair for the rules involved in the generation of the erroneous states. The proposed methodology is based on (i) a logical notation for specifying assertions that are imposed on execution runs; (ii) a runtime checking technique that dynamically tests the assertions; and (iii) a mechanism based on (equational) least general generalization that automatically derives accurate criteria for slicing from falsified assertions. Finally, an implementation of the proposed technique is presented in the assertion-based, dynamic analyzer ABETS, which shows how the forward and backward tracking of asserted program properties leads to a thorough trace analysis algorithm that can be used for program diagnosis and debugging. / Esta tesi proposa una metodologia d'anàlisi dinàmica que millora el diagnòstic de programes erronis escrits en el llenguatge Maude. La idea clau és combinar tècniques de verificació d'assercions en temps d'execució amb la fragmentació dinàmica de traces d'execució per a detectar automàticament errors en temps d'execució, alhora que es reduïx la grandària i la complexitat de les traces a analitzar. En el cas de violar-se una asserció, s'inferix automàticament el criteri de fragmentació, la qual cosa facilita a l'usuari identificar ràpidament la font de l'error. En primer lloc, la tesi formalitza una tècnica destinada a detectar automàticament eventuals desviacions del comportament desitjat del programa (símptomes d'error). Esta tècnica suporta dos tipus d'assercions definides per l'usuari: assercions funcionals (que restringixen crides a funcions deterministes) i assercions de sistema (que especifiquen els invariants d'estat del sistema). La tècnica de verificació dinàmica proposta és demostrablement correcta en el sentit que tots els errors assenyalats definitivament delaten la violació de les assercions. Davant eventuals violacions d'assercions, es generen automàticament traces fragmentades (és a dir, traces simplificades però igualment precises) que ajuden a identificar la causa de l'error. A més, la tècnica també suggerix una possible reparació de les regles implicades en la generació dels estats erronis. La metodologia proposada es basa en (i) una notació lògica per a especificar les assercions que s'imposen a l'execució; (ii) una tècnica de verificació aplicable en temps d'execució que comprova dinàmicament les assercions; i (iii) un mecanisme basat en la generalització (ecuacional) menys general que automàticament obté criteris precisos per a fragmentar traces d'execució a partir d'assercions falsificades. Finalment, es presenta una implementació de la tècnica proposta en la ferramenta d'anàlisi dinàmica basat en assercions ABETS, que mostra com és possible combinar el traçat cap avant i cap arrere de les propietats assertades del programa per a obtindre un algoritme precís d'anàlisi de traces que resulta útil per al diagnòstic i la depuració de programes. / Sapiña Sanchis, J. (2017). Rewriting Logic Techniques for Program Analysis and Optimization [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/94044

[pt] Esta dissertação tem como objeto de estudo uma adaptação da peça Hamlet, de William Shakespeare, intitulada srsly Hamlet (falando sério, Hamlet, em tradução livre), de autoria de Courtney Carbone. A proposta da pesquisa é discutir, por meio da análise da adaptação, alguns procedimentos que, usados no processo de reescrita, visam não só manter alguns dos elementos do que se pode chamar de universo shakespeariano, mas também garantir que a peça do Bardo seja apreciada pelo público infantil e juvenil contemporâneo. A obra de Carbone, publicada em 2015 pela Random House Books for Young Readers, é o primeiro título da Série OMG Shakespeare. A série é composta por quatro adaptações de peças de Shakespeare e busca levar a dramaturgia do poeta inglês para os leitores jovens de uma forma lúdica, sendo o seu principal diferencial em relação a outras adaptações voltadas para os jovens a apresentação dos textos no formato de mensagens de WhatsApp. Signos visuais e sonoros se sobrepõem nessas reescritas inovadoras das peças shakespearianas. A partir de uma perspectiva descritivista, este estudo averigua como a natureza multimodal e intermidiática de srsly Hamlet contribui para o processo de construção de sentidos e para a criação de imagens contemporâneas de Shakespeare e de Hamlet. Através da pesquisa das estratégias de adaptação empregadas (linguísticas e culturais), examina-se a influência das expectativas da cultura-alvo na manipulação da trama, da narrativa e da linguagem. Embora a adaptação de Carbone esteja, em princípio, voltada para o mercado literário e público anglo-americanos, espera-se que esta investigação sobre o processo de recriação de uma obra considerada canônica possa (i) contribuir para uma discussão mais ampla sobre os elementos que caracterizam, nos tempos atuais, as reescritas voltadas para a literatura infantil e juvenil impressa; e (ii) trazer novas possibilidades à reflexão sobre como o entrelaçamento entre qualidades intermidiáticas e a narrativa multimodal em uma obra afeta não só a criação de significados, mas também a experiência vivenciada pelos leitores. / [en] The research object of this master s thesis is an adaptation of William Shakespeare s Hamlet entitled srsly Hamlet, written by Courtney Carbone. The purpose of this study is to discuss, through the analysis of the adaptation, some procedures used in the rewriting process that aim not only to retain some elements of what may be called Shakespearean universe, but also to ensure that the Bard s Hamlet will be enjoyed by contemporary juvenile audiences. Carbone s work, published in 2015 under the Random House Books for Young Readers, is the first title of the OMG Shakespeare Series. The series is made up of four adaptations of Shakespeare s plays and aims at bringing the English playwright s work to contemporary young readers in a playful way. One may say that the main gap between OMG books and other written adaptations aimed at youths is the presentation of the text in WhatsApp-message format. Visual and aural signs overlap in these innovative rewritings of the Bard s plays. This research takes a descriptive perspective to go over how the intermedial and multimodal nature of Carbone s srsly Hamlet plays a part in the process of creating meaning and in the shape of contemporary images of Shakespeare and his Hamlet. Also, through the analysis of the adaptation strategies employed (linguistic and cultural), the influence of the target culture s expectations on the manipulation of plot, narrative and language is examined. Although Carbone s srsly Hamlet originally targets Anglo-American literary markets and audiences, this investigation of the process of recreation of a canonical-standing work may hopefully (i) contribute to a broader discussion of what characterizes adaptations directed at print children s literature in contemporary times; and (ii) add new possibilities to the reflections on how the intertwinement of intermedial qualities and multimodal narratives in literary works affects not only the construction of meaning, but also readers experience.

Programs have precise semantics, so we can use mathematical proof to establish their properties. These proofs are often too large to validate with the usual "social process" of mathematics, so instead we create and check them with theorem-proving software. This software must be advanced enough to make the proof process tractable, but this very sophistication casts doubt upon the whole enterprise: who verifies the verifier? We begin with a simple proof checker, Level 1, that only accepts proofs composed of the most primitive steps, like Instantiation and Cut. This program is so straightforward the ordinary, social process can establish its soundness and the consistency of the logical theory it implements (so we know theorems are "always true"). Next, we develop a series of increasingly capable proof checkers, Level 2, Level 3, etc. Each new proof checker accepts new kinds of proof steps which were not accepted in the previous levels. By taking advantage of these new proof steps, higher-level proofs can be written more concisely than lower-level proofs, and can take less time to construct and check. Our highest-level proof checker, Level 11, can be thought of as a simplified version of the ACL2 or NQTHM theorem provers. One contribution of this work is to show how such systems can be verified. To establish that the Level 11 proof checker can be trusted, we first use it, without trusting it, to prove the fidelity of every Level n to Level 1: whenever Level n accepts a proof of some phi, there exists a Level 1 proof of phi. We then mechanically translate the Level 11 proof for each Level n into a Level n - 1 proof---that is, we create a Level 1 proof of Level 2's fidelity, a Level 2 proof of Level 3's fidelity, and so on. This layering shows that each level can be trusted, and allows us to manage the sizes of these proofs. In this way, our system proves its own fidelity, and trusting Level 11 only requires us to trust Level 1. / text

Ce mémoire examine l’ethos d’historien de François Du Creux dans ses Historiæ canadensis publiées en 1664. Cette œuvre relativement peu connue, qui consiste, pour une bonne part, en une réécriture latine des Relations jésuites de Nouvelle-France, pose des problèmes de statut et de finalité découlant notamment des transformations que connaissait l’écriture de l’histoire à l’époque de Du Creux. L’étude de l’ethos est ici envisagée comme un moyen d’aborder les grands enjeux de l’œuvre en précisant le projet auctorial lié au statut de l’historien. À partir des occurrences de la première personne dans les pièces liminaires et dans le corps du texte, nous dégageons d’abord les trois statuts énonciatifs de la figure auctoriale, en tant que Français, Jésuite et historien. Ces statuts évoquent des questions religieuses et politiques, souvent polémiques, qui sont associées à la Compagnie de Jésus au xviie siècle. L’auteur les convoque pour établir son autorité en se référant à sa tâche d’historien, en défendant ses positions sur les questions qui concernent son ordre et en faisant valoir sa familiarité avec les acteurs et les faits historiques. Nous montrons enfin comment les figures adoptées par l’auteur orientent la lecture des événements racontés et affinent la compréhension de la visée de l’œuvre. / This thesis examines the ethos François Du Creux as a historian in his Historiæ canadensis published in 1664. This relatively unknown work, which mostly consists of a rewriting of the New France Jesuits’ Relations, presents problems of status and purpose largely derived from the transformations that were characteristic of the writing of history in Du Creux’s time. The study of the ethos is used here as a way to address the key issues of the book by clarifying the authorial work associated with the status of the historian. First, we have drawn from occurrences of verbs in the first person in the introductory pieces and the main text three descriptive statuses of the authorial figure: Frenchman, Jesuit, and historian. Theses statuses are linked to religious and politic matters, often controversial, that were associated with the Society of Jesus during the 17th century. Du Creux invokes them to establish his authority by citing his historian function, by defending his positions on matters that concern his order, and by asserting his familiarity with the historical players and facts. Lastly, we demonstrate how the figures used by Du Creux guide the reader’s understanding of the events that are described and deepen the comprehension of the work’s aim.

This work is about diagrammatic languages, how they can be represented, and what they in turn can be used to represent. More specifically, it focuses on representations and applications of string diagrams. String diagrams are used to represent a collection of processes, depicted as "boxes" with multiple (typed) inputs and outputs, depicted as "wires". If we allow plugging input and output wires together, we can intuitively represent complex compositions of processes, formalised as morphisms in a monoidal category. While string diagrams are very intuitive, existing methods for defining them rigorously rely on topological notions that do not extend naturally to automated computation. The first major contribution of this dissertation is the introduction of a discretised version of a string diagram called a string graph. String graphs form a partial adhesive category, so they can be manipulated using double-pushout graph rewriting. Furthermore, we show how string graphs modulo a rewrite system can be used to construct free symmetric traced and compact closed categories on a monoidal signature. The second contribution is in the application of graphical languages to quantum information theory. We use a mixture of diagrammatic and algebraic techniques to prove a new classification result for strongly complementary observables. Namely, maximal sets of strongly complementary observables of dimension D must be of size no larger than 2, and are in 1-to-1 correspondence with the Abelian groups of order D. We also introduce a graphical language for multipartite entanglement and illustrate a simple graphical axiom that distinguishes the two maximally-entangled tripartite qubit states: GHZ and W. Notably, we illustrate how the algebraic structures induced by these operations correspond to the (partial) arithmetic operations of addition and multiplication on the complex projective line. The third contribution is a description of two software tools developed in part by the author to implement much of the theoretical content described here. The first tool is Quantomatic, a desktop application for building string graphs and graphical theories, as well as performing automated graph rewriting visually. The second is QuantoCoSy, which performs fully automated, model-driven theory creation using a procedure called conjecture synthesis.

De légendaires, les grandes figures féminines des mythes anciens sont devenues, au fil du XIXe siècle, emblématiques. Le mouvement s’amplifie vers la fin du siècle et l’imaginaire « féminin » se nourrit alors d’un discours social qui contribue à construire la féminité en termes de menace et de dépravation. Les figures mythiques prêteront leurs traits à celle de la femme fatale, devenue le symbole de la dégénérescence de la société française. Engrangeant dans son corps représenté tous les vices du siècle, la figure féminine nous est apparue éminemment révélatrice quant à la compréhension d’une époque. Or, la figure de la femme fatale s’avère fondamentalement ambivalente et Lilith, pouvant à la fois incarner l’amour et la destruction, affiche ce double visage de la féminité. Nous démontrons qu’il existe une relation étroite entre la profonde ambivalence du mythe de Lilith et les représentations de la femme fatale et pour ce faire, procédons à une analyse comparative de l’œuvre de Rachilde et Octave Mirbeau qui, dans La Jongleuse et Le Jardin des supplices, réécrivent le mythe de Lilith. De la comparaison des deux Lilith, ressortent deux représentations extrêmement contrastées de la femme fatale : alors que Rachilde dresse toute droite son héroïne dans son désir ascensionnel, Mirbeau construit une Clara toute en mollesse et assoiffée de chair. Par l’analyse des rapports qui s’articulent entre deux écritures, nous démontrons que la dualité inhérente au mythe de Lilith répond à l’instabilité d’une société aux prises avec de multiples angoisses en matière d’identité sexuelle. Cette comparaison nous amène aussi à nous interroger quant aux traces d’une certaine sexuation dans la voix littéraire. / Over the course of the nineteenth century the legendary female figures of ancient myth had become emblematic of the female sex. This association grew stronger toward the end of the century and the “feminine imaginary” fed itself on a social discourse that contributed to the construction of femininity in terms of menace and depravity. The mythical figures that lent their faces to representations of the femme fatale became symbols of the degeneration of French society. With all of the vices of the century gathered into her body, this female figure appears to us as eminently revealing as to the understanding of an era. However, the figure of the femme fatale is fundamentally ambivalent, and the mythical figure of Lilith, which can embody love as well as destruction, represents the two opposing aspects of nineteenth century representations of femininity. This study shows that there is a direct relationship between the profound ambivalence that characterizes the myth of Lilith and representations of the femme fatale. To this end, we undertake a comparative analysis of the works of Rachilde and Octave Mirbeau, who rewrite the myth of Lilith in The Juggler and The Torture Garden. Two extremely contrasting representations emerge from the comparison between the two “Liliths”: as Rachilde portrays her upstanding heroine’s desire to transcend her body, Mirbeau constructs his Clara as soft and mired in the body, thirsty for carnal pleasure. By examining the relationships that become apparent between these works, we demonstrate that the duality inherent in the myth of Lilith responds to the instability of a society grappling with multiple anxieties regarding sexual identity. This comparison, therefore, allows us to interrogate the traces of a specific mode of sexuation in the literary voice.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal.

Ancien activiste d’extrême gauche, ouvrier, alpiniste, Erri De Luca (1950-) est un écrivain « non-croyant » hanté par le texte biblique. Refusant toute autorité (qu’elle soit étatique, divine ou littéraire), il s’inscrit « contre » tous, dans un « à-côté », dans un hors norme traductif et exégétique biblique qui est pourtant norme de l’herméneutique rabbinique. Notre travail s’attache à montrer comment De Luca propose à un lecteur protéiforme une mise en scène des modalités d’une lecture qui s’écrit. Dans une écriture du non-dit, l’auteur s’inspire de sa lecture, de sa traduction et de son commentaire du texte biblique pour mieux questionner les genres. Il crée ainsi, dans une œuvre babélienne où les origines sont horizon, dans une zone mouvante qui échappe à toute catégorisation, un « entre-deux ». Sa position scripturale est duale, entre culture innée (d’origine chrétienne) et culture inventée (d’origine juive). Sa re-traduction n’est pas écriture seconde ou sous-écriture ; elle naît comme écriture. Décontextualisant, littérarisant, resémantisant la méthode exégétique rabbinique du midrash, De Luca propose une forme de réception personnelle du texte fondateur : il lorgne dans les interstices du texte biblique afin de lui conférer une « autre possibilité ». / Erri De Luca (Naples, 1950-), former far-left activist, workman, climber, is a “non-believer” who is haunted by the Bible. Refusing any authority (let it be from a State, a God, or a literary rule), he places himself “against” everyone, on the “sideline”, outside the standards of translation, outside the standards of biblical exegetic compendiums. Yet, De Luca's method fits well in the standards of rabbinical hermeneutical methods. Our work's goal is to show how De Luca offers to a proteiform reader an arrangement of his writing. While De Luca's writing signifies by what is kept quiet, his readings, his translations, and its biblical compendiums serve his inspiration to better question the genre. He therefore creates, in a babelical masterpiece where origins are horizon, in a shifting area that refuses any classification, a place “in-between” every other places. His writing position is dual, in-between an inborn culture (of Christian origin) and a created culture (of Jewish origin). His re-translation is not yet “another writing” nor a “cheap writing”: his translation erects itself as writing. With his very personal style, outside of any context, but flourishing with literary references and by reusing the rabbinical hermeneutical method of the midrash, De Luca proposes a personal reading of the Bible. He squeezes within the Bible's interstices in order to give it an “other possibility”.

Proust est un lecteur régulier et exigeant de Baudelaire, auquel il consacre des pages critiques qui ont compté dans la réception du poète au cours du XXe siècle. Toutefois, les liens qui unissent les deux auteurs sont loin de se réduire à cette dimension métatextuelle et permettent d’interroger plus avant la notion même d’intertextualité. La principale difficulté dans l’identification de la présence de Baudelaire tient à son côté diffus qui s’inscrit dans les linéaments de l’écriture. Loin d’être explicites, les renvois à l’œuvre du poète se font au mieux sur le mode de l’allusion, sinon au gré d’un filtrage intertextuel et culturel qui gomme presque totalement les traces de ce souvenir. Nous proposons de revenir à une définition de l’intertextualité entendue dans un sens large, associant l’étude du texte à l’histoire et à l’anthropologie culturelles, soulignant ainsi que l’écriture est autant mémoire qu’oubli, et, partant, que la mémoire du lecteur n’est pas moins ambivalente. Atteindre le Baudelaire de Proust suppose ainsi de comprendre quelle idée le romancier se faisait du poète, mais aussi par quelles médiations l’œuvre de Baudelaire, passée au filtre des sensations de Proust, de son histoire, de son époque s’est imprimée dans la Recherche. En d’autres termes, il convient tout à la fois d’établir la sociologie d’une lecture et celle d’un lecteur.Apparaissent ainsi deux Baudelaire : l’un est explicite et rapidement usée par la conversation et l’autre innerve la Recherche. On voit qu’un Baudelaire chasse l’autre, et que les réseaux intertextuels ne peuvent être que souterrains. L’œuvre de Baudelaire chez Proust est ainsi lue au prisme de la mémoire et de l’oubli. / A voracious and meticulous reader of Baudelaire, Proust’s non-fiction writing pays substantial attention to the poet, in criticism which would influence the reception of Baudelaire’s work throughout the twentieth century.However, the imbrication of these two writers extends far beyond the issue of metatextuality. Rather, by highlighting the pervasive presence of Baudelaire in Proust’s oeuvre, the object of scrutiny becomes the very notion of intertextuality itself. The challenge of identifying Baudelairean references in the Recherche arises from their diffusion. Seldom explicit, Proust’s allusions to the poet’s work are subtle, if not near camouflaged by the cultural and intertextual filters. This thesis argues for a return to intertextual analysis in the broadest sense of the term, namely a study that devotes attention to the text’s historical and cultural-anthropological facets. In so doing, the thesis underlines how the act of writing involves memory as much as forgetfulness, and that the reader’s memory is as ambivalent as the author’s. But also, and more importantly, my thesis undertakes to demonstrate the ways in which Baudelaire’s poetry is transformed through the Proustian prism: that is, the narrative and mood of the Recherche, being written in a particular period, with a particular aesthetic sensibility. The social conditions of a reading, and the social determinism of a reader, are placed here in parallel. There are thus two Baudelaire in the Recherche. One is referenced explicitly in banal exchanges; the other is referenced implicitly a Baudelaire diffuse who vitalises the Recherche. And so, each Baudelaire chases the other, creating intertextual networks.