Recent Dynamic Models Research Articles

Utilizing Inherent Bias for Memory Efficient Continual Learning: A Simple and Robust Baseline

Learning from continuously evolving data is critical in real-world applications. This type of learning, known as Continual Learning (CL), aims to assimilate new information without compromising performance on prior knowledge. However, learning new information leads to a bias in the network towards recent observations, resulting in a phenomenon known as catastrophic forgetting. The complexity increases in Online Continual Learning (OCL) scenarios where models are allowed only a single pass over data. Existing OCL approaches that rely on replaying exemplar sets are not only memory-intensive when it comes to large-scale datasets but also raise security concerns. While recent dynamic network models address memory concerns, they often present computationally demanding, over-parameterized solutions with limited generalizability. To address this longstanding problem, we propose a novel OCL approach termed “Bias Robust online Continual Learning (BRCL).” BRCL retains all intermediate models generated. These models inherently exhibit a preference for recently learned classes. To leverage this property for enhanced performance, we devise a strategy we describe as ‘utilizing bias to counteract bias.’ This method involves the development of an Inference function that capitalizes on the inherent biases of each model towards the recent tasks. Furthermore, we integrate a model consolidation technique that aligns the first layers of these models, particularly focusing on similar feature representations. This process effectively reduces the memory requirement, ensuring a low memory footprint. Despite the simplicity of the methodology to guarantee expandability to various frameworks, extensive experiments reveal a notable performance edge over leading methods on key benchmarks, getting continual learning closer to matching offline training. (Source code will be made publicly available upon the publication of this paper.)

Continuous-Time Graph Representation with Sequential Survival Process

Over the past two decades, there has been a tremendous increase in the growth of representation learning methods for graphs, with numerous applications across various fields, including bioinformatics, chemistry, and the social sciences. However, current dynamic network approaches focus on discrete-time networks or treat links in continuous-time networks as instantaneous events. Therefore, these approaches have limitations in capturing the persistence or absence of links that continuously emerge and disappear over time for particular durations. To address this, we propose a novel stochastic process relying on survival functions to model the durations of links and their absences over time. This forms a generic new likelihood specification explicitly accounting for intermittent edge-persistent networks, namely GraSSP: Graph Representation with Sequential Survival Process. We apply the developed framework to a recent continuous time dynamic latent distance model characterizing network dynamics in terms of a sequence of piecewise linear movements of nodes in latent space. We quantitatively assess the developed framework in various downstream tasks, such as link prediction and network completion, demonstrating that the developed modeling framework accounting for link persistence and absence well tracks the intrinsic trajectories of nodes in a latent space and captures the underlying characteristics of evolving network structure.

An empirical review of dynamic extreme value models for forecasting value at risk, expected shortfall and expectile

This work provides a selective review of the most recent dynamic models based on extreme value theory, in terms of their ability to forecast financial losses through different risk measures. The main characteristic of these models is that their dynamics depend only on the occurrence and magnitude of extreme events above a high threshold. Through an empirical analysis, we evaluate the predictive ability of these approaches on a set of stock market indices. In an in-sample analysis, we assess the goodness-of-fit of the different specifications. We also compare the adequacy of each model, considering how well they forecast the risk measures in the out-of-sample period. In addition, in order to identify the best-performing models, we use the model confident set procedure across different risk measures, loss functions, and score functions to identify the superior models. Finally, we identify some potential avenues for future research.

Revisiting the AA-DD Model in Zero Lower Bound

This paper proposes a simple model of a mechanism through which exchange rate can affect the link between output and government spending in zero lower bound (ZLB) periods. In our proposed model, the expected near-future interest rate is added as an endogenous variable. Unlike existing AA-DD models in ZLB, the nominal exchange rate is no longer constant. Our model predicts that the output effect of an increase in government spending in a ZLB period is deflected by an appreciation of the current exchange rate. The AA-DD model is taught in almost all economic departments. The model is also generally used by many central banks and governments. The existing AA-DD model can be misleading. Our new AA-DD model may help to update the existing model in ZLB periods. Our AA-DD model is also consistent with recent dynamic stochastic general equilibrium models in open economies in ZLB periods.

Photometric Confirmation and Characterization of the Ennomos Collisional Family in the Jupiter Trojans

Collisional families offer a unique window into the interior composition of asteroid populations. Previous dynamical studies of the Jupiter Trojans have uncovered a handful of potential collisional families, two of which have been subsequently confirmed through spectral characterization. In this paper, we present new multiband photometric observations of the proposed Ennomos family and derive precise g − i colors of 75 candidate family members. While the majority of the targets have visible colors that are indistinguishable from background objects, we identify 13 objects with closely grouped dynamical properties that have significantly bluer colors. We determine that the true Ennomos collisional family is tightly confined to and , and the majority of its confirmed members have near-solar spectral slopes, including some of the bluest objects hitherto discovered in the Trojan population. The property of distinctly neutral colors that is shared by both the Ennomos family and the previously characterized Eurybates family indicates that the spectral properties of freshly exposed surfaces in the Jupiter region are markedly different than the surfaces of uncollided Trojans. This implies that the processes of ice sublimation and space weathering at 5.2 au yield a distinct regolith chemistry from the primordial environment within which the Trojans were initially accreted. It also suggests that the Trojans were emplaced in their present-day location from elsewhere sometime after the initial population formed, which is a key prediction of recent dynamical instability models of solar system evolution.

Deeper Integration or Open Regionalism in the Indian Ocean Rim Association (IORA): A CGE Approach

The Indian Ocean Rim Association (IORA) is a regional association based on the principle of open regionalism to promote liberalization through trade and investment facilitation. This study compares two main scenarios with the ingredients of a deeper regional integration compared to the IORA's present open regionalism status. Scenario I evaluates the effects of gradual tariff cuts on IORA, while Scenario II combines the impact of tariff cuts, non-tariff measures (NTM) reduction, and trade facilitation. A recent recursive dynamic model, called Global Trade Analysis Project recursive dynamic model (GTAP-RD), was used to understand the adjustment path. The result shows welfare gain for all IORA countries when a trade liberalization involving tariff cuts is complemented with NTM reduction and trade facilitation. In addition, the "NTM spillover" effect on some non-IORA members (the rest of Gulf Cooperation Council, the rest of Greater Arab Free Trade Agreement, and the United Kingdom) was observed. The results suggest that IORA has more to gain if it involves a deeper regional integration than the present form of open regionalism.

Dynamic Modeling of Trust in Automation in Human-Autonomy Teaming

Quantitative analysis of the relationship between humans and automation becomes increasingly important as the reliance of humans upon automation becomes more commonplace. We present a literature review of key factors—categorized as effects from automation, operator, and the environment—that influence trust in or reliance upon automation. Those factors are treated as parameters in a dynamical systems analysis (DSA) model whose manipulation induces phase transitions in the decision to use automation. A review of the most recent dynamical models indicates a trend toward including increasingly more parameters in increasingly sophisticated models. We review the challenges inherent in that approach and suggest an alternative with precedence in the broader DSA literature that is more parsimonious and amenable to real-time analysis.

Finite dimensional approximation to muscular response in force-fatigue dynamics using functional electrical stimulation

Recent dynamical models, based on the seminal work of V. Hill, allow to predict the muscular response to functional electrostimulation (FES), in the isometric and non-isometric cases. The physical controls are modeled as Dirac pulses and lead to a sampled-data control system, sampling corresponding to times of the stimulation, where the output is the muscular force response. Such a dynamics is suitable to compute optimized controls aiming to produce a constant force or force strengthening, but is complex for real time applications. The objective of this article is to construct a finite dimensional approximation of this response to provide fast optimizing schemes for the design of a smart electrostimulator for muscular reinforcement or rehabilitation. It is an on-going industrial project based on force-fatigue models, validated by experiments. Moreover it opens the road to application of optimal control to track a reference trajectory in the joint angular variable to produce movement in the non-isometric models.

The Nature of Low-albedo Small Bodies from 3 μm Spectroscopy: One Group that Formed within the Ammonia Snow Line and One that Formed beyond It

We present evidence, via a large survey of 191 new spectra along with previously published spectra, of a divide in the 3 μm spectral properties of the low-albedo asteroid population. One group (“sharp types,” or STs, with band centers <3 μm) has a spectral shape consistent with carbonaceous chondrite meteorites, while the other group (“not sharp types,” or NSTs, with bands centered >3 μm) is not represented in the meteorite literature but is as abundant as the STs among large objects. Both groups are present in most low-albedo asteroid taxonomic classes, and, except in limited cases, taxonomic classifications based on 0.5–2.5 μm data alone cannot predict whether an asteroid is an ST or NST. Statistical tests show that the STs and NSTs differ in average band depth, semimajor axis, and perihelion at confidence levels ≥98% while not showing significant differences in albedo. We also show that many NSTs have a 3 μm absorption band shape like comet 67P and likely represent an important small-body composition throughout the solar system. A simple explanation for the origin of these groups is formation on opposite sides of the ammonia snow line, with the NST group accreting H2O and NH3 and the ST group only accreting H2O, with subsequent thermal and chemical evolution resulting in the minerals seen today. Such an explanation is consistent with recent dynamical modeling of planetesimal formation and delivery and suggests that much more outer solar system material was delivered to the main asteroid belt than would be thought based on the number of D-class asteroids found today.

Banking crises and financial instability: Empirical and historical lessons

The paper examines the importance of financial instability for the development of four Norwegian banking crises. The crises are the Post First World War Crisis during the early 1920s, the mid 1920s Monetary Crisis, the Great Depression in the 1930s, and the Scandinavian Banking Crisis of 1987–1993. The paper first offers a description of the financial instability hypothesis applied by Minsky and Kindleberger, and in a recent dynamic financial crisis model. Financial instability is defined as a lack of financial markets and institutions that provide capital and liquidity at a sustainable level under stress. Financial instability basically evolves during times of overheating, overspending and extended credit granting. This is most common during significant booms. The process has devastating effects after markets have turned into a state of negative development.The paper tests the validity of the financial instability hypothesis using a quantitative structural time series model. It reveals upheaval of 10 financial and macroeconomic indicators prior to all the four crises, resulting in a state of economic overheating and asset bubble creation. This is basically explained by huge growth in debts. The overheating caused the following banking crises. Finally, the paper discusses the four crises qualitatively. Again, the conclusion is that a significant increase in money supply and debt caused overheating, asset bubbles, and thereafter, financial and banking crises, which in turn spread to other markets and industries and caused huge slumps in the real economy.

A Hysteresis Dynamic Mathematical Model Approach to Parametric Estimation System

The main contribution of this paper is the proposal of a recent hysteresis dynamic model which is successfully employed within a posited signal modulator. The modulation of signals is a commonly required stage in many engineering applications, such as telecommunications, power electronics, and control, among others. In this paper, the effectiveness of a signal modulator based on the well-known Delta modulator when it contains a dynamic hysteresis system within its main structure is presented. To do that, it is resorted to an application of the granted Hysteresis-Delta Modulator. This application consists of including the modulator within an adaptive scheme, since it is well known that the persistent excitation condition is required, for instance, in parameter estimation tasks. Hence, the main functional property of the modulator with hysteresis is its ability of producing a modulated signal with uniform high-frequency content even when its input is not a permanent persistent excitation signal. To highlight the main contribution of this paper, a numerical experiment of a parameter estimation system is developed to compare the performance of the modulator with the proposed hysteresis model and two other previously reported hysteresis systems. That is, three different scenarios have been tested in the parameter estimation of a nonminimum phase system. Finally, the numerical experiments confirm that the proposed hysteresis model along with the modulator provides the best performance as expected.

Measurement of fragment-mass distributions from fission of At214 following the Be9+Tl205 reaction

Background: The recent surprising experimental discovery of asymmetric fission in the nucleus $^{180}\mathrm{Hg}$ at low excitation energy has renewed interest concerning our understanding of the role of nuclear shell effects in the dynamics of the fission process. New theoretical models have been developed that claim to successfully explain the observed asymmetric fission of $^{180}\mathrm{Hg}$ and also predict the nature of mass distributions of several other preactinide nuclei for which experimental data are scarce.Purpose: We investigated fragment-mass distributions in the fission of the preactinide nucleus $^{214}\mathrm{At}$ at different excitation energies near the Coulomb barrier. The present results were compared with the predictions from a recent state-of-the-art theoretical calculation, as well as with our previous measurement for the nearby nucleus $^{210}\mathrm{Po}$.Methods: A pulsed heavy-ion beam was used in the experiment. Fission fragments were detected with two large-area position-sensitive multiwire proportional counters placed at the folding angle. Mass distributions were extracted from time-of-flight differences and position information ($\ensuremath{\theta},\ensuremath{\phi}$) for the fission fragments.Results: An asymmetry in fragment-mass distribution in the fission of $^{214}\mathrm{At}$ was observed at an excitation energy of $\ensuremath{\approx}31$ MeV, where a shell effect is found to influence the potential energy surface. At higher excitation energies, mass distributions are predominantly symmetric indicating the absence of shell effects.Conclusion: This finding is consistent with the prediction of a recent macroscopic-microscopic dynamical model. However, for the preactinides, static models with nascent fragment shell structure are likely to produce similar results.

Effects of early intense bombardment on megaregolith evolution and on lunar (and planetary) surface samples

AbstractImpact rates in the first 500 Myr of the solar system are critical to an understanding of lunar geological history, but they have been controversial. The widely accepted, post‐Apollo paradigm of early lunar impact cratering (about 1975–2014) proposed very low or negligible impact cratering in the period from accretion (&gt;4.4 Ga) to ~4.0 Ga ago, followed by an ~170 million year long spike of cataclysmic cratering, during which most prominent multi‐ring impact basins formed at age ~3.9 Ga. More recent dynamical models suggest very early intense impact rates, declining throughout the period from accretion until an age of ~3.0 Ga. These models remove the basin‐forming spike. This shift has important consequences vis‐à‐vis megaregolith evolution and properties of rock samples that can be collected on the lunar surface today. We adopt the Morbidelli et al. (2018) “accretion tail” model of early intense bombardment, declining as a function of time. We find effects differing from the previous models: early crater saturation and supersaturation; disturbance of magma ocean solidification; deep early megaregolith; and erosive destruction of the earliest multi‐ring basins, their impact melts, and their ejecta blankets. Our results explain observations such as differences in numbers of early lunar impact melts versus numbers of early igneous crustal rocks, highland breccias containing impact melts as old as 4.35 Ga, absence of a 170 Myr long spike in impact melt ages at 3.9 Ga among lunar and asteroidal meteorites, and GRAIL observations of lunar crustal structure.

Evidence of a Supershear Transition Across a Fault Stepover

AbstractSupershear earthquakes, propagating faster than the Earth's shear wave velocity, can generate strong ground motion at distances far from the ruptured fault. Despite the hazards associated with these earthquakes, the exact fault properties that govern their occurrence are not well constrained. Although early studies associated supershear ruptures with simple fault geometries, recent dynamic rupture models have revealed a supershear transition mechanism over complex fault geometries such as fault stepovers. Here we present the first observation of a supershear transition on a fault stepover system during the 2017 Mw 7.7 Komandorsky Islands earthquake. Using a high‐resolution back‐projection technique, we find that the earthquake's rupture velocity accelerates from 2.1 to 5.0 km/s between two offset faults, demonstrating the viability of a new supershear transition mechanism occurring in nature. Given the fault complexity of the Earth's transform plate boundaries, this result may improve our understanding of supershear rupture processes and their associated hazards.

Rising inequality and de-democratization

ABSTRACTThe literature on economic determinants of democratization has identified most importantly the positive effect of economic development and the negative effect of income inequality. We confirm these results using more recent data and dynamic panel models. In this regard, the 2018 World Inequality Report has noted that inequality is highest in the Middle East, where it has stayed stable at that high level for the past several decades.

Data-Driven Predictive Modeling of Neuronal Dynamics Using Long Short-Term Memory

Modeling brain dynamics to better understand and control complex behaviors underlying various cognitive brain functions have been of interest to engineers, mathematicians and physicists over the last several decades. With the motivation of developing computationally efficient models of brain dynamics to use in designing control-theoretic neurostimulation strategies, we have developed a novel data-driven approach in a long short-term memory (LSTM) neural network architecture to predict the temporal dynamics of complex systems over an extended long time-horizon in future. In contrast to recent LSTM-based dynamical modeling approaches that make use of multi-layer perceptrons or linear combination layers as output layers, our architecture uses a single fully connected output layer and reversed-order sequence-to-sequence mapping to improve short time-horizon prediction accuracy and to make multi-timestep predictions of dynamical behaviors. We demonstrate the efficacy of our approach in reconstructing the regular spiking to bursting dynamics exhibited by an experimentally-validated 9-dimensional Hodgkin-Huxley model of hippocampal CA1 pyramidal neurons. Through simulations, we show that our LSTM neural network can predict the multi-time scale temporal dynamics underlying various spiking patterns with reasonable accuracy. Moreover, our results show that the predictions improve with increasing predictive time-horizon in the multi-timestep deep LSTM neural network.

Migration of D-type asteroids from the outer Solar System inferred from carbonate in meteorites

Recent dynamical models of Solar System evolution and isotope studies of rock-forming elements in meteorites have suggested that volatile-rich asteroids formed in the outer Solar System beyond Jupiter's orbit, despite being currently located in the main asteroid belt. The ambient temperature under which asteroids formed is a crucial diagnostic to pinpoint the original location of asteroids and is potentially determined by the abundance of volatiles they contain. In particular, abundances and 13C/12C ratios of carbonates in meteorites record the abundances of carbon-bearing volatile species in their parent asteroids. However, the sources of carbon for these carbonates remain poorly understood. Here we show that the Tagish Lake meteorite contains abundant carbonates with consistently high 13C/12C ratios. The high abundance of 13C-rich carbonates in Tagish Lake excludes organic matter as their main carbon source. Therefore, the Tagish Lake parent body, presumably a D-type asteroid, must have accreted a large amount of 13C-rich CO2 ice. The estimated 13C/12C and CO2/H2O ratios of ice in Tagish Lake are similar to those of cometary ice. Thus, we infer that at least some D-type asteroids formed in the cold outer Solar System and were subsequently transported into the inner Solar System owing to an orbital instability of the giant planets.

Geophysical evidence that Saturn’s Moon Phoebe originated from a C-type asteroid reservoir

ABSTRACT Saturn’s Moon Phoebe has been suggested to originate from the Kuiper Belt. However, its density is twice that of Kuiper Belt objects (KBOs) in the same size class, which challenges that relationship. Since the internal evolution of mid-sized planetesimals (100–300 km in diameter) is primarily driven by the amount of accreted short-lived radioisotopes, it is possible to constrain the relative times of formation of these bodies based on their bulk porosity content, hence their densities. From modelling the thermal evolution of KBOs, we infer a difference in formation timing between these bodies and Phoebe. This confirms prior suggestions for a delayed accretion timeframe with increasing distance from the Sun. This geophysical finding combined with spectral observations suggests Phoebe formed in the same region as C-type asteroids and support recent dynamical models for a C-type body reservoir between the orbits of the giant planets. On the other hand, the similarly low densities of mid-sized D-type asteroids, Trojan asteroids, and KBOs add to the growing evidence that these objects shared a common reservoir near or beyond the orbit of Neptune and were heat starved overall.

Revisiting the AA-DD Model in Zero Lower Bound

This paper proposes a simple model of a mechanism through which exchange rate can affect the link between output and government spending in zero lower bound (ZLB) periods. In our proposed model, the expected near-future interest rate is added as an endogenous variable. Unlike existing AA-DD models in ZLB, the nominal exchange rate is no longer constant. Our model predicts that the output effect of an increase in government spending in a ZLB period is deflected by an appreciation of the current exchange rate. The AA-DD model is taught in almost all economic departments. The model is also generally used by many central banks and governments. The existing AA-DD model can be misleading. Our new AA-DD model may help to update the existing model in ZLB periods. Our AA-DD model is also consistent with recent dynamic stochastic general equilibrium models in open economies in ZLB periods.

Dynamic analysis of slender rotor of vertically suspended centrifugal pumps due to various hydraulic design factors

If a rotor is excited at one of its natural frequencies, a resonance occurs. The resonance can lead to large amplitude oscillations and subsequently failures. Field experience proves that the excessive vibration of the rotor of centrifugal pumps shows itself in wear and fatigue failure of shafts, bearings, wear rings, seals and impellers. So, modal analysis of a pump rotor is of great concern to researchers, international standard organizations and industries. In this research, dynamic behavior of a nonlinear finite element model of slender rotor of vertically suspended centrifugal pumps is studied. The dynamic model of the rotor is discretized into an appropriate number of three-dimensional Timoshenko beam elements. In addition to the main nonlinear terms and particularly the geometric stiffening effect, the gyroscopic effect has been also taken into account. The full-order equations of the motion, which are obtained by the Lagrangian approach and the finite element method, are integrated into a computational plan. The ability of the recent nonlinear dynamic model, which is intended to furnish a basic model for further development of a more comprehensive model, is examined to track the coupling between different types of vibrations. The comprehensive modal analysis of the rotor due to various hydraulic design factors is presented.