Here you can find a summary of research directions I worked on recently.
The input-output balance equation is used to define rankings of constituents in the most diverse complex organizations: the very same tool that helps classify how species of an ecosystems or sectors of an economy interact with each other is useful to determine what sites of the World Wide Web - or which nodes in a social network - are the most influential. The basic principle is that constituents of a complex organization can produce outputs whose "volume" should precisely match the sum of external demand plus inputs absorbed by other constituents to function. The solution typically requires a case-by-case inversion of large matrices, which provides little to no insight on the structural features responsible for the hierarchical organization of resources. Here we show that - under very general conditions - the solution of the input-output balance equation for open systems can be described by a universal master curve, which is characterized analytically in terms of simple "mass defect" parameters - for instance, the fraction of resources wasted by each species of an ecosystem into the external environment. Our result follows from a stochastic formulation of the interaction matrix between constituents: using the replica method from the physics of disordered systems, the average (or typical) value of the rankings of a generic hierarchy can be computed, whose leading order is shown to be largely independent of the precise details of the system under scrutiny. We test our predictions on systems as diverse as the WWW PageRank, trophic levels of generative models of ecosystems, input-output tables of large economies, and centrality measures of Facebook pages.
▶︎ Bartolucci, S., Caccioli, F., Caravelli, F., & Vivo, P. (2020). Inversion-free Leontief inverse: statistical regularities in input-output analysis from partial information. ArXiv preprint arXiv:2009.06350.
▶︎ Bartolucci, S., Caccioli, F., Caravelli, F., & Vivo, P. (2020). Universal rankings in complex input-output organizations. arXiv preprint arXiv:2009.06307.
This project is founded by the UCL Centre for Blockchain Technologies and it is part of a wider collaboration with University College London, Brunel University London, Cagliari University (Italy) and a cryptocurrency trading company, BMyBit.
▶︎ Bartolucci, S., Destefanis, G., Ortu, M., Uras, N., Marchesi, M., & Tonelli, R. (2020). The butterfly “affect”: Impact of development practices on cryptocurrency prices. EPJ Data Science, 9(1), 21.
We collected and started analysing a unique dataset containing all developers’ comments about the development of the open source code of Bitcoin and Ethereum extracted from Github. In the first phase of the project, we constructed the aggregated time series of developers’ sentiment and emotions using suitably built machine learning classifiers and investigated whether they could provide insights to improve the prediction of cryptocurrency prices. We performed the Granger causality test between prices and sentiment time series and checked by using a Long-Short Term Memory Recurrent Neural Network whether including information about the time series of developers’ sentiment would increase the accuracy of price predictions.
Link to the project's presentation: https://vimeo.com/393199080
A fundamental shift in decision making processes, in the role of intermediaries (being replaced by peer-to-peer networks of not trustworthy parties), in the assets exchanged (e.g. cryptocurrencies) and the large amount of data available to businesses and consumers is profoundly changing the structure of the economy. This paradigm shift calls for new approaches to understand the “digital collective behaviour” of users and its implications.
▶︎ Bartolucci, S., & Kirilenko, A. A. (2020). A model of the optimal selection of crypto assets. Royal Society Open Science 7(8).
In this work we model the adoption of digital assets and the users’ selection process. We have created a framework to classify digital assets according to two main features, their (technological) security and their stability (in terms of governance) and then simulated a novel mechanism of assets’ selection performed by different types of users (risk prone vs. risk averse) based on the assets’ intrinsic features. We show how multiple scenarios of assets’ adoption may unfold depending on the composition of the ecosystem of assets and investors.
▶︎ Bartolucci, S., Caccioli, F., & Vivo, P. (2020). A percolation model for the emergence of the Bitcoin Lightning Network. Scientific Reports, 10(1), 1-14. TOP 100 Downloaded Physics Papers
The newly devised Lightning Network, a so-called second-layer technology built on top of the Bitcoin blockchain to provide ”off- chain” fast, frequent and cheaper payment channels between users. On both systems – traditional blockchain and Lightning Network – the users are asked to pay a fee for the submission of a new transaction whereas on the Lightning Network only the users need to lock some funds as a warranty to be able to transfer value over the network. The interplay between fees, liquidity and volume of transactions is what in principles determines the stability of the Lightning Network: within our framework based on percolation in fitness network models, we are able to determine the parameters region where such network is sustainable. The model includes parameters that could be in principle estimated from publicly available data once the evolution of the Lighting Network will have reached a stationary operable state. At the moment, a number of blockchain development companies are investing in the development and testing of this technology, which may be the solution to the limitations in scalability of the Bitcoin ecosystem compared to traditional payment system, e.g. the Visa Circuit.
In this type of networks, patterns of information can be learned and successively retrieved and recognised. As part of this research, I studied the properties of associative memories with diluted patterns, able to perform parallel retrieval of multiple information and possibly relevant for applications in artificial intelligence and machine learning. Using techniques such as Kramers-Moyal expansions for master equations, I carried out a dynamical analysis of the network evolving via Glauber sequential update. I have derived equations quantifying the evolution in time of the order parameter (magnetisation), analysing the nature and the stability of the stationary solutions in different regimes of dilution and network connectivity via linear stability analysis and Monte Carlo simulations.