RESUMO
We investigate the dynamics of particulate matter, nitrogen oxides, and ozone concentrations in Hong Kong. Using fluctuation functions as a measure for their variability, we develop several simple data models and test their predictive power. We discuss two relevant dynamical properties, namely, the scaling of fluctuations, which is associated with long memory, and the deviations from the Gaussian distribution. While the scaling of fluctuations can be shown to be an artifact of a relatively regular seasonal cycle, the process does not follow a normal distribution even when corrected for correlations and non-stationarity due to random (Poissonian) spikes. We compare predictability and other fitted model parameters between stations and pollutants.
RESUMO
Frequency measurements indicate the state of a power grid. In fact, deviations from the nominal frequency determine whether the grid is stable or in a critical situation. We aim to understand the fluctuations of the frequency on multiple time scales with a recently proposed method based on detrended fluctuation analysis. It enables us to infer characteristic time scales and generate stochastic models. We capture and quantify known features of the fluctuations like periodicity due to the trading market, response to variations by control systems, and stability of the long time average. We discuss similarities and differences between the British grid and the continental European grid.
RESUMO
Autocorrelations in stationary systems are time-symmetric, irrespective of the signal's properties. Linear dynamics is usually associated with signals which are statistically time inversion invariant. This is known to be broken for non-Gaussian models. In this paper, we develop a theoretical framework of time reversibility of linear models based on noncontinuous driving. We identify the inverse decay exponent of the autocorrelation function as either a characteristic time for the building-up of extreme events in the time series or of relaxations after these extreme events. If the characteristic time is known, the dynamics can be inverted in both directions in time and the residuals can be compared, which gives a criterion for the type of time inversion asymmetry. The method is applied to two time series from atmospheric science with different behavior.
RESUMO
Deviations from Brownian motion leading to anomalous diffusion are found in transport dynamics from quantum physics to life sciences. The characterization of anomalous diffusion from the measurement of an individual trajectory is a challenging task, which traditionally relies on calculating the trajectory mean squared displacement. However, this approach breaks down for cases of practical interest, e.g., short or noisy trajectories, heterogeneous behaviour, or non-ergodic processes. Recently, several new approaches have been proposed, mostly building on the ongoing machine-learning revolution. To perform an objective comparison of methods, we gathered the community and organized an open competition, the Anomalous Diffusion challenge (AnDi). Participating teams applied their algorithms to a commonly-defined dataset including diverse conditions. Although no single method performed best across all scenarios, machine-learning-based approaches achieved superior performance for all tasks. The discussion of the challenge results provides practical advice for users and a benchmark for developers.