59 research outputs found
Who is Little Enis
No full text25 x 22 cm.Folded broadside of the poem 'Who Is Little Enis?' by Johnathan Williams, with photo by Make-Hay Studios, designed by Captain Vague & the Edgarologist, signed by author 'for David', ca. 1974. 25 x 22 cm
Splitting methods in nonsmooth convex degenerate optimization
No full textIn der vorliegenden Dissertation erforschen, entwickeln und wenden wir Splitting-Verfahren an, die sich innerhalb der konvexen degenerierten Optimierung ergeben. Mit degenerierter Optimierung bezeichnen wir den breiten theoretischen Rahmen, in dem übliche Ableitungskonzepte und der Begriff der Monotonie für mengenwertige Operatoren auch in Bezug auf Semiskalarprodukte verstanden werden können, die durch positiv semidefinite Operatoren mit potenziell großen Kernen, den sogenannten Präkonditionierern, induziert werden. In Teil I erforschen wir die Theorie der Proximalpunktverfahren mit degenerierten Präkonditionierern. Wir etablieren schwache Konvergenzergebnisse unter milden Annahmen, mit und ohne Raten, die leicht im Kontext von Splitting-Vefahren für monotone Inklusions- und konvexe Minimierungsprobleme eingesetzt werden können. Dies erlaubt uns, neue Verbindungen zwischen klassischen Splitting-Verfahren herzustellen und ebnet den Weg für die Entwicklung neuer verallgemeinerter Varianten. Zusätzlich zeigen wir, dass die Degeneriertheit des Präkonditionierers eine Reduktion der an der Iteration beteiligten Variablen ermöglicht. Unter Ausnutzung des vorgeschlagenen Rahmens stellen wir mehrere neue graphenbasierte Erweiterungen des Douglas-Rachford-Splitting-Verfahrens zur Lösung von Nullstellenproblemen für die Summe von endlich vielen maximal monotonen Operatoren vor, die nicht auf dem üblichen Produktraumtrick beruhen. Diese Verfahren können auch in leistungsfähige verteilte Algorithmen für beliebige Kommunikationsnetze überführt werden. In Teil II befassen wir uns mit der Analyse von Vorwärts-Rückwärts-Schemata in einem degenerierten Kontext und ihrer faszinierenden Verbindung zu verallgemeinerten bedingten Gradientenverfahren. Diese Analyse führt uns zur Entwicklung eines neuartigen hybriden Ansatzes, der die Lücke zwischen Proximal-Gradienten- und verallgemeinerten bedingten Gradientenverfahren schließt und die Vorteile dieser beiden klassischen Ansätze kombinieren kann. Das vorgeschlagene iterative Schema erlaubt es uns, ein neues datengetriebenes Parameter-Lernmodell, das mit Totalvariations-Entrauschen verbunden ist, effizient zu lösen, ein Problem, welches von unabhängigem Interesse ist. In Teil III stellen wir eine Fallstudie zum dynamischen Lernen von Federungssystemen bei Schienenfahrzeugen vor, eine industrielle Anwendung von präkonditionierten Splitting-Verfahren, die während einer Entsendung bei der Virtual Vehicle Research GmbH in Graz, Österreich, durchgeführt wurde.In the present dissertation, we explore, develop, and apply splitting methods that arise from the convex degenerate optimization framework. With degenerate optimization, we refer to the broad theoretical setting in which usual derivative concepts and the notion of monotonicity for multivalued operators can also be understood with respect to semiscalar products induced by positive semidefinite operators with potentially large kernels, referred to as preconditioners. In part I, we explore the theory of proximal point methods in the presence of a degenerate preconditioner. We establish weak convergence results under mild assumptions, with and without rates, that can be easily employed in the context of splitting methods for monotone inclusion and convex minimization problems. This allows us to establish new connections between classical splitting methods, and paves the way for the development of new generalized variants. Additionally, we show that the degeneracy of the preconditioner allows for a reduction of the variables involved in the iteration updates. Leveraging the proposed framework, we present several new graph-based extensions of the Douglas-Rachford Splitting method to solve root-finding problems involving the sum of finitely many maximal monotone operators, which do not rely on the usual product space trick. Notably, these can also be turned into powerful distributed schemes for any given communication network. In part II, we delve into the analysis of forward-backward schemes in a degenerate setting and their intriguing connection to generalized conditional-gradient methods. This analysis leads us to the development of a novel hybrid approach that bridges the gap between proximal-gradient and generalized conditional-gradient methods and can indeed combine the advantages of these two classical approaches. The proposed scheme allows us to efficiently tackle a new data-driven parameter learning model associated to total variation denoising, which is of independent interest. In part III, we present a case study on the dynamics learning for railway vehicle suspension systems, an industrial application of preconditioned splitting methods conducted during a secondment at Virtual Vehicle Research GmbH, in Graz, Austria.Enis ChencheneDissertation Universität Graz 2023Zusammenfassungen auf Englisch und Deutsc
The world of personalities in Selahattin Enis' novels
No full textSelâhattin Enis [Atabeyoğlu] (1892-1942), edebiyat anlayışı ve dünya görüşüyle kimilerince “Türkiye'nin Zola'sı” olarak anılmıştır. Bastırdığı bir hikâye kitabı ve dördü gazetelerde tefrika hâlinde yayımlanmış sekiz romanı bulunan Selâhattin Enis, ilk eseri Neriman'dan başlayarak tüm romanlarında toplum sorunlarını, başta Batılılaşmanın yanlış anlaşılmasına dayanan ahlâkî bozulmaların bireyleri götürdüğü mutsuzluğu, keskin bir gerçekçilikle göstermekten çekinmemiştir. Sanatçının Meşrutiyet dönemi eseri olmasına rağmen Servet-i Fünûn döneminin izlerini taşıyan ilk romanı Neriman, dar bir şahıs kadrosuna sahiptir. Bu ilk romanı takip eden diğer romanlarında sanatçı giderek şahıs kadrosunu genişletir, her bir romanında vermek istediği daha fazla mesaja mukabil eserlerine daha fazla şahıs dâhil eder. 1910- 1930 tarihleri arasında yazdığı sekiz romana olabildiğince fazla mesaj sığdırma çabası içinde olan sanatçı, son dört romanında şahıs kadrosunu öylesine genişletir ki kimi zaman kahramanın adını karıştırır, kimi zaman bir önceki romanında yer alan kahramanını bir sonraki romanında ufak tefek değişikliklerle yeniden kullanır. Sanatçının şahıs kadrosunun büyük çoğunluğu tiplemelerden ibaret olan romanları; şahısların okura önceden yazar tarafından her yönleri ile tanıtılması, yazarın taraf olması, kahramanlarına karşı beslediği olumlu ve olumsuz hislerini okura açıkça yansıtması gibi yönleri ile Tanzimat dönemi romanlarının izinden gider. Bilhassa tefrika romanlarında okurun ilgisini çekme çabası ile birbirinden renkli tiplemelere yer verirken meddahvari bir tutum sergileyen sanatçıyı natüralist olma çabası, bilhassa yozlaşmış şahısların görünüşü, kokusu, eşyası, yaşadığı mekânı kullanımının tasvirinde ifrada sürükler. Selâhattin Enis’in romanları bu hâlleri ile Tanzimat’tan Cumhuriyet’e her dönemin kimi özelliklerini yansıtan; konak sakinlerinden eski yangın yeri yaşayanlarına, köşklerdeki hanımefendilerden kaldırım kadınlarına kadar rengârenk bir şahıs kadrosu teşkil eder. Anahtar Sözcükler: Selâhattin Enis, roman, şahıslar, Meşrutiyet dönemi, Cumhuriyet dönemi, natüralizmSelâhattin Enis [Atabeyoğlu] (1892-1942) is, based on his perception of literature and philosophy, considered as “Turkey’s Zola” by some persons. Selâhattin Enis, with his one published and eight novels among which four were published as serials in the newspapers, in all his novels starting from his first novel named “Neriman” showed no hesitation to demonstrate, with absolute realism, social problems and misery the individuals were led by the moral corruptions based on misunderstanding of Westernisation. Neriman, the first novel of the author which although being a work of Constitutional Monarchy Period bore the traces of Servet-i Fünun (The Wealth of Sciences) Period contained a narrow range of characters. In his novels following this first, the author gradually widened the range of his characters and in each novel included more characters against the message he sought to give. The author striving to include as more messages as possible in his eight novels he wrote between 1910-1930 so expanded the range of characters in his last four novels that he sometimes confused the names of his characters and sometimes integrated some of his characters of a previous novel to the next with some slight changes. The novels of the author, many of the characters of which consisted of typecasting, follow the lead of novels of the Constitutional Monarchy Period in terms of introducing the characters in all aspects to readers beforehand, the author taking a side and clearly reflecting the reader his positive or negative feelings he bore for his characters. The fervour to be a naturalist of the artist who showed an encomiastic manner while including in his novels characters one more vivacious than the other in an effort to grip the readers particularly in serials drags the artist into exaggeration in depicting the appearance, odour, belongings and dwellings of particularly the corrupted persons. The novels of Selâhattin Enis, as they are, consist of a variety of characters from mansion dwellers to old fireground residents, ladies of pavilions to women of sidewalks reflecting some features of each period from Constitutional Monarchy Period to the Republic Period. Key Words: Selâhattin Enis, novel, characters, Constitutional Monarchy Period, Republican period, naturalism
Wasserstein medians: robustness, PDE characterization and numerics
No full text38 pages, 6 figuresWe investigate the notion of Wasserstein median as an alternative to the Wasserstein barycenter, which has become popular but may be sensitive to outliers. In terms of robustness to corrupted data, we indeed show that Wasserstein medians have a breakdown point of approximately 1 2. We give explicit constructions of Wasserstein medians in dimension one which enable us to obtain L p estimates (which do not hold in higher dimensions). We also address dual and multimarginal reformulations. In convex subsets of R d , we connect Wasserstein medians to a minimal (multi) flow problem à la Beckmann and a system of PDEs of Monge-Kantorovich-type, for which we propose a p-Laplacian approximation. Our analysis eventually leads to a new numerical method to compute Wasserstein medians, which is based on a Douglas-Rachford scheme applied to the minimal flow formulation of the problem
Wasserstein medians: robustness, PDE characterization and numerics
No full text38 pages, 6 figuresWe investigate the notion of Wasserstein median as an alternative to the Wasserstein barycenter, which has become popular but may be sensitive to outliers. In terms of robustness to corrupted data, we indeed show that Wasserstein medians have a breakdown point of approximately 1 2. We give explicit constructions of Wasserstein medians in dimension one which enable us to obtain L p estimates (which do not hold in higher dimensions). We also address dual and multimarginal reformulations. In convex subsets of R d , we connect Wasserstein medians to a minimal (multi) flow problem à la Beckmann and a system of PDEs of Monge-Kantorovich-type, for which we propose a p-Laplacian approximation. Our analysis eventually leads to a new numerical method to compute Wasserstein medians, which is based on a Douglas-Rachford scheme applied to the minimal flow formulation of the problem
Wasserstein medians: robustness, PDE characterization and numerics
No full text38 pages, 6 figuresWe investigate the notion of Wasserstein median as an alternative to the Wasserstein barycenter, which has become popular but may be sensitive to outliers. In terms of robustness to corrupted data, we indeed show that Wasserstein medians have a breakdown point of approximately 1 2. We give explicit constructions of Wasserstein medians in dimension one which enable us to obtain L p estimates (which do not hold in higher dimensions). We also address dual and multimarginal reformulations. In convex subsets of R d , we connect Wasserstein medians to a minimal (multi) flow problem à la Beckmann and a system of PDEs of Monge-Kantorovich-type, for which we propose a p-Laplacian approximation. Our analysis eventually leads to a new numerical method to compute Wasserstein medians, which is based on a Douglas-Rachford scheme applied to the minimal flow formulation of the problem
Wasserstein medians: robustness, PDE characterization and numerics
No full textWe investigate the notion of Wasserstein median as an alternative to the
Wasserstein barycenter, which has become popular but may be sensitive to
outliers. In terms of robustness to corrupted data, we indeed show that
Wasserstein medians have a breakdown point of approximately . We
give explicit constructions of Wasserstein medians in dimension one which
enable us to obtain estimates (which do not hold in higher dimensions).
We also address dual and multimarginal reformulations. In convex subsets of
, we connect Wasserstein medians to a minimal (multi) flow
problem \`a la Beckmann and a system of PDEs of Monge-Kantorovich-type, for
which we propose a -Laplacian approximation. Our analysis eventually leads
to a new numerical method to compute Wasserstein medians, which is based on a
Douglas-Rachford scheme applied to the minimal flow formulation of the problem.Comment: 38 pages, 6 figure
A consensus-based algorithm for non-convex multiplayer games
No full textIn this paper, we present a novel consensus-based zeroth-order algorithm tailored for non-convex multiplayer games. The proposed method leverages a metaheuristic approach using concepts from swarm intelligence to reliably identify global Nash equilibria. We utilize a group of interacting particles, each agreeing on a specific consensus point, asymptotically converging to the corresponding optimal strategy. This paradigm permits a passage to the mean-field limit, allowing us to establish convergence guarantees under appropriate assumptions regarding initialization and objective functions. Finally, we conduct a series of numerical experiments to unveil the dependency of the proposed method on its parameters and apply it to solve a nonlinear Cournot oligopoly game involving multiple goods
Graph and distributed extensions of the Douglas-Rachford method
No full textIn this paper, we propose several graph-based extensions of the
Douglas-Rachford splitting (DRS) method to solve monotone inclusion problems
involving the sum of maximal monotone operators. Our construction is based
on a two-layer architecture that we refer to as bilevel graphs, to which we
associate a generalization of the DRS algorithm that presents the prescribed
structure. The resulting schemes can be understood as unconditionally stable
frugal resolvent splitting methods with a minimal lifting in the sense of Ryu
[Math Program 182(1):233-273, 2020], as well as instances of the (degenerate)
Preconditioned Proximal Point method, which provides robust convergence
guarantees. We further describe how the graph-based extensions of the DRS
method can be leveraged to design new fully distributed protocols. Applications
to a congested optimal transport problem and to distributed Support Vector
Machines show interesting connections with the underlying graph topology and
highly competitive performances with state-of-the-art distributed optimization
approaches.Comment: 23 pages, 4 figure
A hybrid proximal generalized conditional gradient method and application to total variation parameter learning
No full textIn this paper we present a new method for solving optimization problems
involving the sum of two proper, convex, lower semicontinuous functions, one of
which has Lipschitz continuous gradient. The proposed method has a hybrid
nature that combines the usual forward-backward and the generalized conditional
gradient method. We establish a convergence rate of under mild
assumptions with a specific step-size rule and show an application to a total
variation parameter learning problem, which demonstrates its benefits in the
context of nonsmooth convex optimization.Comment: 6 pages, 3 figures, 1 tabl
- …
