Assignment flows for data labeling on graphs: convergence and stability

The assignment flow recently introduced in the J. Math. Imaging and Vision 58/2 (2017) constitutes a high-dimensional dynamical system that evolves on a statistical product manifold and performs contextual labeling (classification) of data given in a metric space. Vertices of an underlying corresponding graph index the data points and define a system of neighborhoods. These neighborhoods together with nonnegative weight parameters define the regularization of the evolution of label assignments to data points, through geometric averaging induced by the affine e-connection of information geometry. From the point of view of evolutionary game dynamics, the assignment flow may be characterized as a large system of replicator equations that are coupled by geometric averaging. This paper establishes conditions on the weight parameters that guarantee convergence of the continuous-time assignment flow to integral assignments (labelings), up to a negligible subset of situations that will not be encountered when working with real data in practice. Furthermore, we classify attractors of the flow and quantify corresponding basins of attraction. This provides convergence guarantees for the assignment flow which are extended to the discrete-time assignment flow that results from applying a Runge–Kutta–Munthe–Kaas scheme for the numerical geometric integration of the assignment flow. Several counter-examples illustrate that violating the conditions may entail unfavorable behavior of the assignment flow regarding contextual data classification.

Hereditary low-mode dynamo model

В данной статье рассматривается модель динамо в виде двумерной динамической системы в интегро-дифференциальной форме. В модели реализован стабилизирующий генерацию поля механизм обратной связи в виде подавления α-эффекта функционалом сверточного типа от актуальных и предыдущих значений спиральности и энергии. Наличие этого механизма подавления вводит в модель эредитарность (память). Для модели была построена численная схема ввиде совмещение разностных схем для дифференциальной и интегральной части, двухступенчатый неявный методы Рунге-Кутты и метод трапеций соотвественно. Так же были рассмотрены и графически представлены динамические режимы нашей модели. This article discusses a dynamo model in the form of a two-dimensional dynamical system in integro-differential form. The model implements a stabilizing polarization generator in the form of suppression of the a effect of convolutional type functional from current and previous helicity and energy values. The presence of this suppression mechanism introduces hereditarity (memory) into the model. For modeling, a digital scheme was constructed in the form of a combination of difference schemes for the differential and integral parts, a twostep implicit Runge-Kutta method and a trapezium method, respectively. We also reviewed and graphically presented the dynamic modes of our model.

The Cosmological Semiclassical Einstein Equation as an Infinite-Dimensional Dynamical System

We develop a comprehensive framework in which the existence of solutions to the semiclassical Einstein equation (SCE) in cosmological spacetimes is shown. Different from previous work on this subject, we do not restrict to the conformally coupled scalar field and we admit the full renormalization freedom. Based on a regularization procedure, which utilizes homogeneous distributions and is equivalent to Hadamard point splitting, we obtain a reformulation of the evolution of the quantum state as an infinite-dimensional dynamical system with mathematical features that are distinct from the standard theory of infinite-dimensional dynamical systems (e.g., unbounded evolution operators). Nevertheless, applying methods closely related to Ovsyannikov’s method, we show existence of maximal/global solutions to the SCE for vacuum-like states and of local solutions for thermal-like states. Our equations do not show the instability of the Minkowski solution described by other authors.

Ice Multiplication by Fragmentation during Quasi-Spherical Freezing of Raindrops: A Theoretical Investigation

Ice multiplication by fragmentation during collision–freezing of supercooled rain or drizzle is investigated. A zero–dimensional dynamical system describes the time evolution of number densities of supercooled drops and ice crystals in a mixed–phase cloud. The characteristic time–scale for this collision–freezing ice fragmentation is controlled by the collision efficiency, the number of ice fragments per freezing event, and the available number concentration of supercooled drops. The rate of the process is proportional to the number of supercooled drops available. Thus, ice may multiply extensively, even when the fragmentation number per freezing event is relatively small. The ratio of total numbers of ice particles to those from the first ice, namely the ‘ice–enhancement factor’, is controlled both by the number of fragments per freezing event and the available number concentration of supercooled drops in a similar manner. Especially, when ice fragmentation by freezing of supercooled drops is considered in isolation, the number of originally–existing supercooled drops multiplied by the fragmentation number per freezing event yields the eventual number of ice crystals. When supercooled drops are continuously generated by coalescence, ice crystals from freezing fragmentation also continuously increase asymptotically at a rate equal to the generation rate of supercooled drops multiplied by the fragmentation number per freezing event. All these results are expressed by simple analytical forms, thanks to the simplicity of the theoretical model. These parameters can practically be used as a means for characterizing observed mixed–phase clouds.

Stability in a Two-Dimensional Dynamical System of Endogenous Growth with Public Capital

The aim of this study is to present a stability in a two-dimensional dynamical system of endogenous growth with public capital. We assume the simple model of the economic growth, in which both private and public capital can influence on the rate of growth of knowledge. The public capital is rival but non excludable goods, i.e. there is a congestion in use of public capital. The model of growth is formulated as a two-dimensional dynamical system. Using mathematical methods of dynamical systems, we analyze growth paths as well as the stationary states of the system and their stability.

Locating and Stabilizing Unstable Periodic Orbits Embedded in the Horseshoe Map

Based on the theory of symbolic dynamical systems, we propose a novel computation method to locate and stabilize the unstable periodic points (UPPs) in a two-dimensional dynamical system with a Smale horseshoe. This method directly implies a new framework for controlling chaos. By introducing the subset based correspondence between a planar dynamical system and a symbolic dynamical system, we locate regions sectioned by stable and unstable manifolds comprehensively and identify the specified region containing a UPP with the particular period. Then Newton’s method compensates the accurate location of the UPP with the regional information as an initial estimation. On the other hand, the external force control (EFC) is known as an effective method to stabilize the UPPs. By applying the EFC to the located UPPs, robust controlling chaos is realized. In this framework, we never use ad hoc approaches to find target UPPs in the given chaotic set. Moreover, the method can stabilize UPPs with the specified period regardless of the situation where the targeted chaotic set is attractive. As illustrative numerical experiments, we locate and stabilize UPPs and the corresponding unstable periodic orbits in a horseshoe structure of the Duffing equation. In spite of the strong instability of UPPs, the controlled orbit is robust and the control input retains being tiny in magnitude.

Shaping Dynamics With Multiple Populations in Low-Rank Recurrent Networks

An emerging paradigm proposes that neural computations can be understood at the level of dynamic systems that govern low-dimensional trajectories of collective neural activity. How the connectivity structure of a network determines the emergent dynamical system, however, remains to be clarified. Here we consider a novel class of models, gaussian-mixture, low-rank recurrent networks in which the rank of the connectivity matrix and the number of statistically defined populations are independent hyperparameters. We show that the resulting collective dynamics form a dynamical system, where the rank sets the dimensionality and the population structure shapes the dynamics. In particular, the collective dynamics can be described in terms of a simplified effective circuit of interacting latent variables. While having a single global population strongly restricts the possible dynamics, we demonstrate that if the number of populations is large enough, a rank R network can approximate any R-dimensional dynamical system.

Asymptotic (statistical) periodicity in two-dimensional maps

<p style='text-indent:20px;'>In this paper we give a new sufficient condition for the existence of asymptotic periodicity of Frobenius–Perron operators corresponding to two–dimensional maps. Asymptotic periodicity for strictly expanding systems, that is, all eigenvalues of the system are greater than one, in a high-dimensional dynamical system was already known. Our new result enables one to deal with systems having an eigenvalue smaller than one. The key idea for the proof is to use a function of bounded variation defined by line integration. Finally, we introduce a new two-dimensional dynamical system numerically exhibiting asymptotic periodicity with different periods depending on parameter values, and discuss the application of our theorem to the example.</p>