Anna Carbone

1. City growth positively impacts economies of scale while also bringing together undesired environmental and social degradation (Anestis and Stathakis 2024), thus requiring that the challenges pos...

Long memory processes yield important information about how complex systems evolve under positive or negative correlation among the individual processes and components. Rescaled range analysis (R/S), detrended fluctuation analysis (DFA), detrended moving average analysis (DMA) and generalized Hurst exponent (GHE) are among widely adopted methods to discriminate between correlated and anticorrelated sequences in terms of Hurst exponent. The recently proposed Kullback-Leibler cluster entropy, a measure of divergence between probability distributions of long-range correlated sequences, has the computational advantage of not relying on a linear regression of log-log data plots to estimate the Hurst exponent. In this work, the KullbackLeibler cluster entropy is illustrated on the 24 chromosomes of the T2TCHM13+Y human reference genome, including centromeric regions and short arms, i.e. gaps still present in the previously published reference assembly. The Kullback-Leibler cluster entropy can accurately detect the recently added structurally complex nucleotide regions in the 24 chromosomes, through the local variability of their correlation exponents. To test the KullbackLeibler cluster entropy performance, the results are referred to the nucleotide positions of the earlier assembly (GRCh38) and prove the algorithm accuracy in detecting the newly added strands. Biological implications are discussed, and potential uses of the Kullback-Leibler cluster divergence in machinelearning and deep-learning pipelines for genomic data are outlined.

The Kullback-Leibler cluster entropy D_{C}[P∥Q] is evaluated for the empirical and model probability distributions P and Q of the clusters formed in the realized volatility time series of five assets (S&P500, NASDAQ, DJIA, DAX, and FTSEMIB). The Kullback-Leibler functional D_{C}[P∥Q] provides complementary perspectives about the stochastic volatility process compared to the Shannon functional S_{C}[P]. While D_{C}[P∥Q] is maximum at the short timescales, S_{C}[P] is maximum at the large timescales leading to complementary optimization criteria tracing back respectively to the maximum and minimum relative entropy evolution principles. The realized volatility is modelled as a time-dependent fractional stochastic process characterized by power-law decaying distributions with positive correlation (H>1/2). As a case study, we build a multiperiod portfolio on diversity indexes derived from the Kullback-Leibler entropy measure of the realized volatility. The portfolio is robust and exhibits better performances over the horizon periods. A comparison with the portfolio built either according to the uniform distribution or in the framework of the Markowitz theory is also reported.

Digital Twin (DT) has gained significant traction over the last decade, and it has been defined many times from different perspectives by the academy and industry communities. Although DTs are widely used in e.g. manufacturing, healthcare, agriculture, their potential applications will soon extend and cover other domains. In this scenario, education is required to play a major role in building its own DT models for purposes ranging from academic to professional training. The DT technology in the education context is expected to enhance the outcomes and scopes of the standard digital e-learning approaches. A particular aspect of the research includes how the DT concept can be transferred over to e-learning and its technology adjusted for a virtual to virtual model. Furthermore, DT is a unique phenomenon, often misinterpreted, that need to be well-defined within a proper framework. In this work, we present the architecture of a Prototypical Digital Twin architecture and explain the importance of defining a framework for a DT-Learning concept.

Urban infrastructures and services—such as public transportation, innovation bodies and environmental services—are important drivers for the sustainable development of our society. How effectively citizens, institutions and enterprises interact, how quickly technological innovations are implemented and how carefully new policies are pursued, synergically determine development. In this work, data related to urban infrastructure features such as patents and recycled waste referred to 106 province areas in Italy are investigated over a period of twenty years (2001–2020). Scaling laws with exponents characterizing the above mentioned features are observed and adopted to scrutinize whether and how multiple interactions within a population have amplification effects on the recycling and innovation performance. The study shows that there is a multiplication effect of the population size on the innovation performance of territories, meaning that the dynamic interactions among the elements of the innovation eco-systems in a territory increase its innovation performance. We discuss how to use such approach and the related indexes for understanding metropolitan development policy.

An integrable Hamiltonian variant of the two species Lotka–Volterra (LV) predator–prey model, shortly referred to as geometric mean (GM) predator–prey model, has been recently introduced. Here, we perform a systematic comparison of the dynamics underlying the GM and LV models. Though the two models share several common features, the geometric mean dynamics exhibits a few peculiarities of interest. The structure of the scaled-population variables reduces to the simple harmonic oscillator with dimensionless natural time TGM varying as ωGMt with ωGM=c12c21. We found that the natural timescales of the evolution dynamics are amplified in the GM model compared to the LV one. Since the GM dynamics is ruled by the inter-species rather than the intra-species coefficients, the proposed model might be of interest when the interactions among the species, rather than the individual demography, rule the evolution of the ecosystems.