26 research outputs found
Optimal high-dimensional and nonparametric distributed testing under communication constraints
No full textWe derive minimax testing errors in a distributed framework where the data is
split over multiple machines and their communication to a central machine is
limited to bits. We investigate both the - and infinite-dimensional
signal detection problem under Gaussian white noise. We also derive distributed
testing algorithms reaching the theoretical lower bounds.
Our results show that distributed testing is subject to fundamentally
different phenomena that are not observed in distributed estimation. Among our
findings, we show that testing protocols that have access to shared randomness
can perform strictly better in some regimes than those that do not. We also
observe that consistent nonparametric distributed testing is always possible,
even with as little as -bit of communication and the corresponding test
outperforms the best local test using only the information available at a
single local machine. Furthermore, we also derive adaptive nonparametric
distributed testing strategies and the corresponding theoretical lower bounds.Comment: 53 page
Optimal Distributed Composite Testing in High-Dimensional Gaussian Models With 1-Bit Communication
No full textIn this paper we study the problem of signal detection in Gaussian noise in a distributed setting where the local machines in the star topology can communicate a single bit of information. We derive a lower bound on the Euclidian norm that the signal needs to have in order to be detectable. Moreover, we exhibit optimal distributed testing strategies that attain the lower bound. </p
Mobilitetanalys baserad på mobildata
No full textThe thesis evaluates mobility based on mobile phone positions. The aim is to develop and assess different methods for travel demand estimation based on CDR data. Besides this estimation location data in cellular data is explained in more detail and a previous work based on mobile phone data and travel demand estimation is reviewed. The different methods of travel time estimation include both static and dynamic estimation. The static travel demand estimation evaluates movements in the city based on predefined time periods, whereas the dynamic estimations are based on different definitions of a trip. A trip can be defined as movements between important places, or just simply count a trip between each position, or a filtering of active states to create more accurate origin-destination matrices. The second part of the thesis includes evaluation of travel time based on CDR data before the final conclusions are drawn. The main finding of the thesis is that it is possible to assess mobility in a city based on CDR data, even if there are no validation data available
Travel demand estimation and network assignment based on cellular network data
No full textCellular networks signaling data provide means for analyzing the efficiency of an underlying transportation system and assisting the formulation of models to predict its future use. This paper describes how signaling data can be processed and used in order to act as means for generating input for traditional transportation analysis models. Specifically, we propose a tailored set of mobility metrics and a computational pipeline including trip extraction, travel demand estimation as well as route and link travel flow estimation based on Call Detail Records (CDR) from mobile phones. The results are based on the analysis of data from the Data for development "D4D" challenge and include data from Cote dlvoire and Senegal. (C) 2016 Elsevier B.V. All rights reserved.Funding Agencies|Swedish Governmental Agency for Innovation Systems (VINNOVA)</p
Numerical investigation of atomisation using a hybrid Eulerian-Lagrangian solver
No full textThis study investigates the potential of a newly released multi-phase solver to simulate atomisation in an air-blast type atomiser. The 'VOF-to-DPM' solver was used to simulate primary and secondary atomisation in an atomiser with a coaxial injector-like geometry. The solver uses a hybrid Eulerian/Eulerian-Lagrangian formulation with geometric transition criteria between the two models. In this study isothermal, non-reacting flow at room temperature was assumed. The primary focus was predicting Sauter mean diameter and droplet velocity data at a sampling plane downstream of the injector. The solver produces the expected data and predicts trends similar to those found in experimental measurements. The accuracy of the produced droplet diameters was roughly a factor 2 off compared to experiment. This is attributed primarily to mesh resolution. It was concluded that the solver has the potential to predict atomisation at a reasonable computational cost, but further study is needed to confirm its full capabilities.Accepted Author ManuscriptFluid MechanicsSpace Systems Egineerin
Numerical Investigation of Spray Formation in Air-Blast Atomizers: Numerical study of air-blast atomization using a hybrid volume of fluid/discrete phase solver
No full textThe conducted study investigates the potential of a newly released multi-phase solver to simulate atomization in liquid rocket injectors. The "VOF-to-DPM" solver was used to simulate primary and secondary atomization in an air-blast atomizer with a coaxial injector-like geometry. The solver uses a hybrid Eulerian/Eulerian-Lagrangian formulation with a geometric transition criteria between the two models. The conducted study assumed isothermal, non-reacting flow at room temperature. The primary focus was predicting Sauter Mean Diameter and droplet velocity data at a sampling plane downstream of the injection site. The results showed that the solver is able to produce the expected data and to predict trends similar to those found in experimental measurements. The accuracy of the produced droplet diameters was roughly a factor 2 off compared to experiment. This is attributed to mesh resolution. Measurements were obtained via a cooperative agreement between TU Delft and The University of Sydney. It was concluded that the solver has the potential to predict atomization at a reasonable computational cost, but further study is needed to confirm its full capabilities.Aerospace Engineerin
The Impact of Consumer Loss Aversion on Pricing
No full textWe develop a model in which a profit-maximizing monopolist with uncertain cost of production sells to loss-averse, yet rational, consumers. We first introduce (portable) techniques for analyzing the demand of such consumers, and then investigate the monopolist's pricing strategy. Compared to lower possible purchase prices, paying a higher price in the firm's pricing distribution is assessed by consumers as a loss, decreasing demand for the firm's product. We provide conditions under which a firm with continuously distributed marginal cost responds by (locally) eliminating this "comparison effect" and choosing a discrete price distribution; that is, prices are "sticky". Price stickiness is more likely to obtain when the cost distribution has high density, the price responsiveness of demand is low, or consumers are likely to purchase. Whether or not prices are sticky, the monopolist wants to at least mitigate the comparison effect, leading to countercyclical markups. On the other hand, if consumers expect to buy the product, they experience a loss if they end up not consuming it, increasing their willingness to pay for it. Thus, despite the tendency toward price stability, there are also circumstances in which a firm with unchanging cost offers random "sales" to increase customers' expectation to consume, attracting more demand at higher prices. ZUSAMMENFASSUNG - (Strategisches Preissetzungsverhalten mit verlustaversen Konsumenten) Wir analysieren das optimale Verhalten eines profitmaximierenden Monopolisten mit stochastischen Produktionskosten, der an rationale, verlustaverse Konsumenten verkauft. Hierzu entwickelt der Beitrag übertragbare Techniken, die es erlauben, die Nachfrage von verlustaversen Konsumenten herzuleiten, und bestimmt die optimale Preissetzungsstrategie des Monopolisten. Ein Konsument empfindet einen Verlust, wenn er den von ihm gezahlten Kaufpreis mit erwarteten niedrigeren Preisen des Monopolisten vergleicht. Dieser Verlust reduziert die Zahlungsbereitschaft des Konsumenten und senkt somit seine Nachfrage. Der Beitrag zeigt auf, unter welchen Bedingungen eine Firma mit kontinuierlich verteilten Grenzkosten diesen "Vergleichseffekt" (lokal) eliminiert, indem sie eine diskrete Preisverteilung wählt --- also, eine Preisverteilung mit Preisstarrheit. Diese Preisstarrheit tritt umso eher auf, je höher die Dichte der Kostenverteilung, je niedriger die Nachfrageelastizität oder je größer die Kaufwahrscheinlichkeit des Konsumenten ist. Unabhängig davon, ob die optimale Preisverteilung Preisstarrheit aufweist oder nicht, schwächt der Monopolist diesen Vergleichseffekt ab in dem er antizyklische Preisaufschläge verlangt. Auf der anderen Seite führt die Kauferwartung des Konsumenten dazu, dass er einen Verlust realisiert, wenn er das Gut nicht konsumieren kann. Eine höhere Kauferwartung führt somit zu einer höheren Zahlungsbereitschaft des Konsumenten. Daher kann es trotz der Tendenz zur Preisstarrheit auch Umstände geben, unter denen eine Unternehmung mit fixen Grenzkosten zufällige "Sonderangebote" macht, welche die Kauferwartung des Konsumenten erhöhen und somit mehr Nachfrage bei höheren Preisen generieren.Gravity Reference-dependent utility, price stickiness, monopoly pricing, kinked demand curve, countercyclical markups, sales, promotions, (seemingly) predatory pricing.
The Doctrine of Bayesian Statistics for Inverse Problems
No full textImagine you need to navigate through a completely dark cave. A well-known way of achieving this is echolocation, which works by making sounds and listening to how they are reflected back. The problem of determining the geometric shape of a space from a mixture of reflections of emitted sounds, is an example of an inverse problem. In many fields of science, there are situations where it is impossible to measure a parameter of interest directly and instead, the only available method is to measure a different object that is affected by the parameter of interest. These statistical problems can become challenging when it is difficult, or even impossible, to invert the observations directly into the parameter that one is interested in.In many cases, a statistician has a belief about the true value of the parameter before even starting the experiment. The Bayesian paradigm is an attractive method of combining the new information coming from observations with this prior belief. It gives a sound mechanism, namely the posterior distribution, to update the beliefs about the truth.Statistic
Egy kis egyetemtörténet anekdotákban
No full textTHE HISTORY OF THE UNIVERSITY IN STORIES. he writing remembers the teachers of the Reformed heological Faculty of the Debrecen University in anecdotes. he work and academic activities of the cited professors are commemorated through interesting stories and funny anecdotes. he author describes these professors amicably and amusingly as nice people who seem sometimes rigorous or eccentric, and who strived to maintain the high quality of academic work
