Levels Of Non-stationarity Research Articles

INSTABILITY OF CLOUD INFRASTRUCTURE RESOURCES AND SERVICES

Increasing non-stationarity of cloud infrastructure resources and services leads to a significant decrease in its productivity. Therefore, the purpose of the article is to identify the reasons for non-stationarity of cloud infrastructure resources and services; finding ways to reduce the level of non-stationarity . As a result of the research, the following results were obtained . Sources leading to the specified non-stationarity are identified . Existing approaches to reducing non-stationarity are analyzed . An example of basic resource allocation using standard linear programming methods is given. Variants of application of these methods for dynamic redistribution of resources are shown. Conclusion. Allocation and dynamic redistribution of resources in the cloud infrastructure can be done using standard linear programming methods. But due to the significant non-stationarity of the cloud environment, the proposed approach will reduce the productivity of cloud resources. At the same time, with an increase in the number of variables and restrictions, the computational complexity of the proposed algorithm will grow exponentially . Therefore, it is necessary to look for other approaches for the distribution and redistribution of cloud resources in conditions of significant non-stationarity .

Spatio-temporal variability in the strength, directionality, and relative importance of climate on occupancy and population densities in a philopatric mammal, the American pika (Ochotona princeps)

Species distribution models (SDMs) have been widely employed to evaluate species–environment relationships. However, when extrapolated over broad spatial scales or through time, these models decline in their predictive ability due to variation in how species respond to their environment. Many models assume species–environment relationships remain constant over space and time, hindering their ability to accurately forecast distributions. Therefore, there is growing recognition that models could be improved by accounting for spatio-temporal nonstationarity – a phenomenon wherein the factors governing ecological processes change over space or time. Here, we investigated nonstationarity in American pika (Ochotona princeps) relationships with climatic variables in the Rocky Mountains (USA). We first compared broad-scale differences in pika–climate patterns for occupancy and population density across the Southern, Central, and Northern Rockies. Next, we investigated within-ecoregion variation across four mountain ranges nested within the Northern Rockies. Lastly, we tested whether species–climate relationships changed over time within the Central Rockies ecoregion. Across all analyses, we found varying levels of nonstationarity among the climate metrics for both occupancy and density. Although we found general congruence in temperature metrics, which consistently had negative coefficients, and moisture metrics (e.g., relative humidity), which had positive coefficients, nonstationarity was greatest for summer and winter precipitation over both space and time. These results suggest that interpretations from one ecoregion should not be applied to other regions universally – especially when using precipitation metrics. The within-ecoregion analysis found much greater variation in the strength-of-relationship coefficients among the four mountain ranges, relative to the inter-regional analysis, possibly attributable to smaller sample sizes per mountain range. Lastly, the importance of several variables shifted through time from significant to insignificant in the temporal analysis. Our results collectively reveal the overall complexity underlying species–environment relationships. With rapidly shifting conditions globally, this work adds to the growing body of literature highlighting how issues of spatio-temporal nonstationarity can limit the accuracy, transferability, and reliability of models and that interpretations will likely be most robust at local to regional scales. Diagnosing, describing, and incorporating nonstationarity of species–climate relationships into models over space and time could serve as a pivotal step in creating more informative models.

Nonstationary fractionally integrated functional time series

We study a functional version of nonstationary fractionally integrated time series, covering the functional unit root as a special case. The time series taking values in an infinite-dimensional separable Hilbert space are projected onto a finite number of sub-spaces, the level of nonstationarity allowed to vary over them. Under regularity conditions, we derive a weak convergence result for the projection of the fractionally integrated functional process onto the asymptotically dominant sub-space, which retains most of the sample information carried by the original functional time series. Through the classic functional principal component analysis of the sample variance operator, we obtain the eigenvalues and eigenfunctions which span a sample version of the dominant sub-space. Furthermore, we introduce a simple ratio criterion to consistently estimate the dimension of the dominant sub-space, and use a semiparametric local Whittle method to estimate the memory parameter. Monte-Carlo simulation studies are given to examine the finite-sample performance of the developed techniques.

Does the Climate Impact Satisfaction with Life? An Australian Spatial Study

AbstractIt is now widely acknowledged that climate change will have a considerable impact on various aspects of human existence, and this includes happiness and satisfaction with life. This study adds to the existing literature on the contribution of climate to well-being by exploring the interaction of various climate variables at the national and local levels while controlling for socioeconomic factors. Using climate data covering a 20-yr period and demographic data from the Household Income Labor Dynamics in Australia surveys, several ordinary least squares (OLS) models of interaction are developed to test the proposition that climate does influence life satisfaction. Geographically weighted regression is then applied to explore how the relationship between explanatory variables and life satisfaction varies across different regions of Australia. We find that overall rainfall, temperature, and sunshine have a small but significant effect on individual life satisfaction. The spatial analysis reveals a high level of nonstationarity in the way climate variables impact life satisfaction, suggesting that regional climate type may be an important element influencing the relationship. The understanding of this relationship may assist policy makers who develop resilience and adaptation strategies as we face the impacts of climate change.Significance StatementTo the best of our knowledge, this study is the first investigation of contributions of a wide range of climate factors to individual life satisfaction across a continent-size country that provides a novel spatial analysis of the variations in climate factor impact. The study shows that in regions with climatic conditions of high temperatures and prolonged dry periods, additional heat will adversely affect individual well-being. In view of the anticipated effects of climate change, this finding does not bode well for residents of areas that already have a hot and dry climate, as increasing temperatures and potentially longer droughts are likely to compromise their well-being. This study can inform policy making that considers adaptive climate change strategies for community well-being.

Not all long‐memory estimators are born equal: The case of nonstationary functional time series

We study a functional version of fractionally integrated time series that covers the nonstationary case when the memory parameter d is above 0.5. We project time series, with varying levels of nonstationarity, onto a finite‐dimensional subspace. We obtain the eigenvalues and eigenfunctions that span a sample version of the dominant subspace through dynamic functional principal component analysis of the sample long‐run covariance functions. Within the context of functional autoregressive fractionally integrated moving average models, we evaluate and compare finite‐sample bias and mean‐squared error among some time‐ and frequency‐domain Hurst exponent estimators via Monte Carlo simulations. We apply the estimators to Canadian female and male life‐table death counts.

Online parameter estimation under non-persistent excitations for high-rate dynamic systems

High-rate dynamic systems are defined as systems experiencing dynamic events of typical amplitudes higher than 100 gn for a duration of less than 100 ms. They are characterized by 1) large uncertainties on the external loads; 2) high levels of nonstationarity and heavy disturbance; and 3) generation of unmodeled dynamics from changes in mechanical configuration. To fully enable these systems, feedback capabilities must be developed. This includes computationally fast software and low latency hardware. This paper presents a pure time-based online parameter estimation algorithm for high-rate dynamic systems with real-time applicability. The algorithm is based on a model reference adaptive system architecture consisting of a reference system and an adaptive model. The adaptive model is built on a reduced order physical representation of the system and uncertainties are linearized. Uncertain coefficients are adapted leveraging instantaneous measurements and historical input–output data sets, termed history stack data, based on concurrent learning theory for coping with the lack of persistent excitation. The history stack is sequentially modified based on a singular value maximizing algorithm to accelerate convergence. The algorithm is numerically verified and experimentally validated on a testbed consisting of a cantilever beam with a moving cart. Numerical verifications show that the algorithm provides fast and accurate convergence when concurrent learning is used. Experimental validations show that the algorithm can successfully identify static positions of the cart, and can also track its movement relatively well, with large chattering and overshoots during travel time. The average computation speed of the algorithm per sample step, implemented in MATLAB, is 93 μs. It is envisioned that the implementation of the algorithm on an FPGA, along with refined coding, will greatly reduce computation time.

Characterization of the dynamic behavior of neural activity in Alzheimer’s disease: exploring the non-stationarity and recurrence structure of EEG resting-state activity

Objective. Mild cognitive impairment (MCI) and dementia due to Alzheimer’s disease (AD) have been shown to induce perturbations to normal neuronal behavior and disrupt neuronal networks. Recent work suggests that the dynamic properties of resting-state neuronal activity could be affected by MCI and AD-induced neurodegeneration. The aim of the study was to characterize these properties from different perspectives: (i) using the Kullback–Leibler divergence (KLD), a measure of non-stationarity derived from the continuous wavelet transform; and (ii) using the entropy of the recurrence point density () and the median of the recurrence point density (), two novel metrics based on recurrence quantification analysis. Approach. KLD, and were computed for 49 patients with dementia due to AD, 66 patients with MCI due to AD and 43 cognitively healthy controls from 60 s electroencephalographic (EEG) recordings with a 10 s sliding window with no overlap. Afterwards, we tested whether the measures reflected alterations to normal neuronal activity induced by MCI and AD. Main results. Our results showed that frequency-dependent alterations to normal dynamic behavior can be found in patients with MCI and AD, both in non-stationarity and recurrence structure. Patients with MCI showed signs of patterns of abnormal state recurrence in the theta (4–8 Hz) and beta (13–30 Hz) frequency bands that became more marked in AD. Moreover, abnormal non-stationarity patterns were found in MCI patients, but not in patients with AD in delta (1–4 Hz), alpha (8–13 Hz), and gamma (30–70 Hz). Significance. The alterations in normal levels of non-stationarity in patients with MCI suggest an initial increase in cortical activity during the development of AD. This increase could possibly be due to an impairment in neuronal inhibition that is not present during later stages. MCI and AD induce alterations to the recurrence structure of cortical activity, suggesting that normal state switching during rest may be affected by these pathologies.

Nonstationarity and Abandonment in Markovian Queues with Application to Call Centers

Both large and small firms maintain call centers to establish contact between themselves and their customers. A call center is staffed by Customer Service Representatives (CSRs). In addition to CSRs, a call center needs computers and telecommunication equipment such as an Automatic Call Distributor (ACD). If calls arrive according to a Poisson arrival process and the service times have an exponential distribution, the call center can be modeled as an M/M/s queue where s is the number of CSRs. Typical calculations include finding the number of CSRs required and finding the number of trunks lines required. However, if call center models ignore the abandonment behavior of customers, they distort information that is relevant to management. Typically, ignoring abandonment would lead to overstaffing, as fewer CSRs are actually needed, because some of the callers abandon the system. Also, nonstationarity of arrivals is highly prevalent in call centers. Green, Kolesar and Whitt plot hourly arrival rates for a financial services call center. There is significant variation in arrivals by time of day. In this paper, we model call centers as multiserver Markovian queues with both nonstationarity and abandonment. Nonstationarity is modeled by having a Poisson arrival stream with time dependent mean, which varies according to a sinusoid. Abandonment is defined in terms of an exponential patience random variable, which is the amount of time the caller would be on hold before abandoning the call. We numerically study the performance of this nonstationary M(t)/M/s+M queue with abandonment and compare its performance measures with those of the stationary M/M/s+M queue. We find that approximating a nonstationary system by a stationary system, even at low levels of nonstationarity, can lead to significant errors. Similar results in a system without abandonment, have been obtained by Green, Kolesar and Svoronos. Additionally, we find that abandonment dampens the effect of nonstationarity. Since abandonment and nonstationarity are both present together in real call centers, a real call center with a high level of abandonment behaves closer to an ideal stationary system.

Exploring non-stationarity patterns in schizophrenia: neural reorganization abnormalities in the alpha band

Objective. The aim of this paper was to characterize brain non-stationarity during an auditory oddball task in schizophrenia (SCH). The level of non-stationarity was measured in the baseline and response windows of relevant tones in SCH patients and healthy controls. Approach. Event-related potentials were recorded from 28 SCH patients and 51 controls. Non-stationarity was estimated in the conventional electroencephalography frequency bands by means of Kullback–Leibler divergence (KLD). Relative power (RP) was also computed to assess a possible complementarity with KLD. Main results. Results showed a widespread statistically significant increase in the level of non-stationarity from baseline to response in all frequency bands for both groups. Statistically significant differences in non-stationarity were found between SCH patients and controls in beta-2 and in the alpha band. SCH patients showed more non-stationarity in the left parieto-occipital region during the baseline window in the beta-2 band. A leave-one-out cross validation classification study with feature selection based on binary stepwise logistic regression to discriminate between SCH patients and controls provided a positive predictive value of 72.73% and negative predictive value of 78.95%. Significance. KLD can characterize transient neural reorganization during an attentional task in response to novelty and relevance. Our findings suggest anomalous reorganization of neural dynamics in SCH during an oddball task. The abnormal frequency-dependent modulation found in SCH patients during relevant tones is in agreement with the hypothesis of aberrant salience detection in SCH. The increase in non-stationarity in the alpha band during the active task supports the notion that this band is involved in top-down processing. The baseline differences in the beta-2 band suggest that hyperactivation of the default mode network during attention tasks may be related to SCH symptoms. Furthermore, the classification improved when features from both KLD and RP were used, supporting the idea that these measures can be complementary.

Parallel Nonstationary Direct Policy Search for Risk-Averse Stochastic Optimization

This paper presents an algorithmic strategy to nonstationary policy search for finite-horizon, discrete-time Markovian decision problems with large state spaces, constrained action sets, and a risk-sensitive optimality criterion. The methodology relies on modeling time-variant policy parameters by a nonparametric response surface model for an indirect parametrized policy motivated by Bellman’s equation. The policy structure is heuristic when the optimization of the risk-sensitive criterion does not admit a dynamic programming reformulation. Through the interpolating approximation, the level of nonstationarity of the policy, and consequently, the size of the resulting search problem can be adjusted. The computational tractability and the generality of the approach follow from a nested parallel implementation of derivative-free optimization in conjunction with Monte Carlo simulation. We demonstrate the efficiency of the approach on an optimal energy storage charging problem, and illustrate the effect of the risk functional on the improvement achieved by allowing a higher complexity in time variation for the policy. The online supplement is available at https://doi.org/10.1287/ijoc.2016.0733 .

Traffic Signal Prediction Using Elman Neural Network and Particle Swarm Optimization

Prediction of traffic is very crucial for its management. Because of human involvement in the generation of this phenomenon, traffic signal is normally accompanied by noise and high levels of non-stationarity. Therefore, traffic signal prediction as one of the important subjects of study has attracted researchers’ interests. In this study, a combinatorial approach is proposed for traffic signal prediction, based on Neural Networks and Particle Swarm Optimization algorithm. Elman Neural Network is chosen from amongst many types of Neural Networks due to its feedbacked structure. To this purpose, Particle Swarm optimization algorithm is utilized for adequate training of the Neural Network, instead of common gradient descent based methods. In this work, wavelet transform is employed as a part of the preprocessing stage, for the elimination of transient phenomena as well as for more efficient training of the Neural Network. Simulations are carried out to verify performance of the proposed method, and the results demonstrate good performance in comparison to other methods.

Estimating Buoyancy Heat Flux Using the Surface Renewal Technique over Four Amazonian Forest Sites in Brazil

The surface renewal (SR) method was applied for the first time to measurements of air temperature over four Amazonian forest sites and different seasons in order to obtain estimates of buoyancy heat flux. The required calibration of this method against eddy covariance resulted in a value for a specific parameter that is close to the range reported in other studies, contributing to the generalization of the SR method to different kinds of canopies. The comparison with fluxes obtained using the eddy covariance technique revealed a good match between the two methods for different sites, heights and seasons. Sites with high levels of non-stationarity in the signals of temperature and wind speed presented higher scatter in the regression with fluxes from eddy covariance. For a particular site with previously reported influence of low-frequency motions, the regression was only satisfactory, i.e., slope parameter close to unity and small offset, when oscillations with periods longer than $$\approx $$ 13 min were filtered out. The SR method has a great potential due to the simplicity of the instrumentation required. However, care should be taken when measuring under the influence of mesoscale motions, which can lead to high levels of non-stationarity, compromising the fundamental concepts of the SR theory.

Quantifying the Non-stationarity of Vehicle Vibrations with the Run Test

This paper investigates the applicability of the run test as a tool to quantify the statistical non-stationarity of road vehicle vibrations. The run test was applied to the moving root-mean-square (RMS) time history of a number of vibration records measured from a variety of vehicle types, routes and vehicle speeds. The paper discusses the limitations associated with calculating the moving RMS of random signals especially with respect to the window width. When applied to the set of vibration records (with segments of inactivity removed), the run test indicates that every record is non-stationary. A run ratio parameter was introduced to quantify the level of non-stationarity, which shows that the result is sometimes dependent on the RMS window width. Further analysis correlating the run test results with the statistical distribution or the RMS indicates that the run ratio parameter has some merit in quantifying the level of non-stationarity in road vehicle vibrations. Copyright © 2013 John Wiley & Sons, Ltd.

The slope of the psychometric function and non-stationarity of thresholds in spatiotemporal contrast vision

The slope of the two-interval, forced-choice psychometric function (e.g. the Weibull parameter, β) provides valuable information about the relationship between contrast sensitivity and signal strength. However, little is known about how or whether β varies with stimulus parameters such as spatiotemporal frequency and stimulus size and shape. A second unresolved issue concerns the best way to estimate the slope of the psychometric function. For example, if an observer is non-stationary (e.g. their threshold drifts between experimental sessions), β will be underestimated if curve fitting is performed after collapsing the data across experimental sessions. We measured psychometric functions for 2 experienced observers for 14 different spatiotemporal configurations of pulsed or flickering grating patches and bars on each of 8days. We found β≈3 to be fairly constant across almost all conditions, consistent with a fixed nonlinear contrast transducer and/or a constant level of intrinsic stimulus uncertainty (e.g. a square law transducer and a low level of intrinsic uncertainty). Our analysis showed that estimating a single β from results averaged over several experimental sessions was slightly more accurate than averaging multiple estimates from several experimental sessions. However, the small levels of non-stationarity (SD≈0.8dB) meant that the difference between the estimates was, in practice, negligible.

Collisionless relaxation of ion distributions downstream of laminar quasi‐perpendicular shocks

Directed flow of incident ions provides the free energy which is redistributed in a shock among heated ions and electrons, accelerated particles, and magnetic compression. In low Mach number laminar shock the main channel of conversion is into downstream gyrating ions. Just behind the shock transition the ion distribution is substantially nongyrotropic, which results in spatially periodic variations of the ion pressure and, consequently, in time stationary downstream oscillations of the magnetic field. In the absence of significant level of nonstationarity, gyrotropization is due to the gyrophase mixing and slow. Theoretical analysis of the phenomenon and supporting hybrid simulations are presented. It is shown that these oscillations are more likely to be observed at low Mach number low β shocks, while at higher Mach numbers or higher β they may be obscured by waves crossing the shocks.

Coherence and short-term synchronization are insensitive to motor unit spike train nonstationarity

Motor units discharge at variable rates and thereby produce spike trains that are inherently nonstationary. For signal analysis techniques that assume data stationarity, such as those for detecting and quantifying coherence and short-term synchronization (STS), nonstationarity presents a potential source of inaccuracy. If coherence and STS indices are sensitive to the level of nonstationarity, broad application of these techniques to nonstationary spike trains could diminish the validity of previous and future motor unit synchronization studies. To address this concern, spike trains taken from intramuscular EMG recorded in two hand muscles during a 2-min, submaximal isometric task were checked for significant nonstationarity using the Mann–Kendall test. Linear regressions were then performed for coherence and STS indices against the mean spike train nonstationarity strength for all physiological spike train pairs ( n = 92). Peak coherence, coherence index, k′ − 1, and CIS were used to quantify synchronization, whereas two indices (StIn and StatAv) were used to quantify nonstationarity. To extend the scope and power of this study, an integrate-to-fire model was used to create a much larger pool of simulated spike train pairs ( n = 2520) with predetermined synchronization and stationarity strengths and discharge properties that closely approximated those of the physiological pool. Regressions for both physiological and simulated pairs revealed that coherence, STS, and their indices were insensitive to stationarity. Therefore, motor unit spike train coherence and STS measurements should be robust across all synchronization and nonstationarity levels typical for steady submaximal tasks. Additionally, this finding should hold for all spike trains with similar levels of synchronization and nonstationarity, regardless of origin.

Augmenting the lot sizing order quantity when demand is probabilistic

Abstract In this paper we consider a single item, discrete time, lot sizing situation where demand is random and its parameters (e.g., mean and standard deviation) can change with time. For the appealing criterion of minimizing expected total relevant costs per unit time until the moment of the next replenishment we develop two heuristic ways of selecting an appropriate augmentation quantity beyond the expected total demand through to the planned (deterministic) time of the next replenishment. The results of a set of numerical experiments show that augmentation is important, particularly when orders occur frequently (i.e., the fixed cost of a replenishment is low relative to the costs of carrying one period of demand in stock) and the coefficient of variability of demand is relatively low, but also under other specified circumstances. The heuristic procedures are also shown to perform very favourably against a hindsight, baseline ( s , S ) policy, especially for larger levels of non-stationarity.

Developments and Comparisons on the Definition of an Instantaneous Damping Estimator for Structures under Natural Excitation

Accurate identification of damping constitutes a relevant issue for structural health monitoring. In fact, high modal or localized damping ratios are often associated with the presence of structural damage. This question may become even more challenging when identifying a structure from its natural vibration (due to wind, traffic etc.). This article describes the results of the latest studies, conducted by this research group, focusing on the definition of an instantaneous estimator of damping for the identification of structures under unknown non-stationary excitation. A method is presented, in particular, which is based on instantaneous curve-fitting applied directly to time-frequency representations of dynamic response signals. One of the strong points of this technique is its flexibility: from each signal, in fact, a time function of damping is extracted, providing punctual information on estimate stability. A series of numerical applications to Multiple Degree of Freedom (M-DOF) systems is presented; the performance of the method is examined under increasing levels of non stationarity of the excitation. Finally, the results are compared with those obtained, on the same data, through the use of three known time-domain identification methods: Direct System Parameter Identification (DSPI), Eigensystem Realisation Algorithm (ERA), Polyreference Time Domain (PRTD). The latter two were used with the Random Decrement function.