Our recent study published in eLife showed that computer modeling can help identify the role (that) MRSA infections from the community play in hospital outbreaks and test ways to control them.
Nature Communication Editors’ highlights. The the study shows that not only were fake news sources mentioned as frequently as traditional outlets, but fake or biased news moved in significantly different directions, depending on the political slant. In fact, to the researchers’ surprise, they found that right-wing voters tended to influence the output of people producing fake news tweets, not the other way around.
The paper ‘The k-core as a predictor of structural collapse in mutualistic ecosystems’, which can be found here, illustrates how the the collapse of a network representing an ecosystem can be predicted by the k-core, a topological invariant of the system. The article, published in Nature Physics, presents an analytical approach, supported by numerical solutions, to the study of the dynamical equations that find the dependence of the tipping point of a system with the parameters of the network of the interacting species. The theory then, identifies a technique with which one can monitor mutualistic ecosystems and prevent their extinction.
Our work on the use of optimal percolation to find influentual nodes in brain networks has been selected by the editors of Nature Communications for inclusion in an anthology of new research on complex systems. The paper, “Finding influential nodes for integration in brain networks using optimal percolation theory,” can be found in the collection here. Optimal percolation has emerged as one of the most effective ways to find influencers in a wide range of networks from social networks to the brain, and our work specifically can be employed in the prognosis of diseases, the modulation of brain function to mitigate certain health conditions, and identifying areas that must be avoided during brain tumor surgery in order to preserve the maximum amount of function in patients.
Furthermore, CCNY has highlighted our paper and research in the following press release.
G. Del Ferraro, A. Moreno, B. Min, F. Morone, Ú. Pérez-Ramírez, L. Pérez-Cervera, L. C. Parra, A. Holodny, S. Canals, H. A. Makse. Finding influential nodes for integration in brain networks using optimal percolation theory and supporting information, Nature Comm. 9, 2274 (2018).
Current events have brought the spread of false information via social media to the forefront of national consciousness. With Mark Zuckerberg’s ongoing testimony on Capitol Hill regarding Facebook’s role in the proliferation of fake news giving us insight into what happened, it is important to identify the how, i.e., the mechanism by which misinformation goes viral. Using machine learning, complex network theory, and methods such as Granger causality on a massive Twitter dataset numbering 171 million relevant Tweets, members of our lab uncovered patterns in how various sources of fake news influenced– and were influenced by– their followers.
CCNY physicist tracks influence of fake news on Presidential election CCNY News (2018), Rebecca Rivera
Our research regarding the influence of fake news spread via Twitter on the 2016 U.S. Presidential Election has been featured in CCNY Communications. Patricia Reilly writes:
Volume 355 of Science (February 03, 2017) focuses on the use of scientific methods to predict population-level trends. The work of Alexandre Bovet, Flaviano Morone, and Hernán A. Makse using analysis of Twitter to predict the outcome of the 2016 US Presidential Election (see previous post) is featured. Technology and understanding of human motivations and current circumstances are key to such prediction. Although there is still work to be done before a completely accurate method of prediction is attained, the tools we are developing at present are a step firmly in the right direction. For further reading:
Prediction and its limits Science 355 (2017), Barbara R. Jasny and Richard Stone
The Pulse of the People Science 355 (2017), John Bohannon
The recently-concluded United States Presidential election was the end to one of the most divisive and vitriolic campaign seasons in recent memory. With one of the candidates being active on Twitter, there was a good opportunity to study whether or not the election results could be predicted from patterns in Twitter’s network of users. Percolation theory, statistical physics of complex networks, natural language processing, and machine learning are combined in an analysis of a network of election-related interactions between Twitter users created from retweets, replies, quotes, and hashtags. The analytics pioneered by Alexandre Bovet, Flaviano Morone, and Hernán Makse were able to anticipate the New York Times polling average by up to 13 days, providing a more instantaneous snapshot of general public opinion than classic polling techniques were able to. See the paper by Alexandre Bovet, Flaviano Morone, and Hernán Makse here: Predicting election trends with Twitter: Hillary Clinton versus Donald Trump.
Nature 524, 38–39 (06 August 2015)
In complex networks, some nodes are more important than others. The most important nodes are those whose elimination induces the network’s collapse, and identifying them is crucial in many circumstances, for example, if searching for the most effective way to stop a disease from spreading. But this is a hard task, and most methods available for the purpose are essentially based on trial-and-error. Here, Flaviano Morone and Hernán Makse devise a rigorous method to determine the most influential nodes in random networks by mapping the problem onto optimal percolation and solving the optimization problem with an algorithm that the authors call ‘collective influence’. They find that the number of optimal influencers is much smaller, and that low-degree nodes can play a much more important role in the network than previously thought.
The retweet data in Twitter that we used in this paper can be downloaded here.
Connecting complex networks is known to exacerbate perturbations and lead to cascading failures, but natural networks of networks like the brain are surprisingly stable. A theory now proposes that network structure holds the key to understanding this paradox. See paper Saulo D. S. Reis, Yanqing Hu, Andres Babino, Jose S. Andrade Jr, Santiago Canals, Mariano Sigman, Hernan A. Makse, “Avoiding catastrophic failure in correlated networks of networks”, Nature Physics 10, 762-767 (October 2014). Published pdf version. Supplementary Information, SI. News & Views Multilayer Networks: Dangerous Liaisons? by Ginestra Bianconi, pdf. CUNY press release. CSIC press release. Tendencias cientificas. Sociedade Brasileira de Fisica, SBF. Why natural networks are more stable than man-made networks, phys.org. Tus neuronas mejoran las redes que mueven al mundo, El Pais. A estabilidade do cerebro, Pesquisa FAPESP. Disenian redes tecnologicas mediante experimentos con el cerebro humano, abc.es.
From MIT Technology Review. Nobody has figured out how to spot the most influential spreaders of information in a real-world network. Now that looks set to change with important implications, not least for the superspreaders themselves. Who are the most influential spreaders of information on a network? That’s a question that marketers, bloggers, news services and even governments would like answered. Not least because the answer could provide ways to promote products quickly, to boost the popularity of political parties above their rivals and to seed the rapid spread of news and opinions. The question of how to find the superspreaders remains open. That looks set to change thanks to the work of our group after performing the first study of superspreaders on real networks. See paper in Scientific Reports 4, 5547 (2014). Published PDF version. PDF version in arxiv.org.
by Adrian Baule and Hernan Makse. Random packings of objects of a particular shape are ubiquitous in science and engineering. However, such jammed matter states have eluded any systematic theoretical treatment due to the strong positional and orientational correlations involved. In recent years progress on a fundamental description of jammed matter could be made by starting from a constant volume ensemble in the spirit of conventional statistical mechanics. Recent work has shown that this approach, first introduced by S. F. Edwards more than two decades ago, can be cast into a predictive framework to calculate the packing fractions of both spherical and non-spherical particles. Soft Matter 10, 4423-4429 (2014); DOI: 10.1039/C3SM52783B. See cover pdf
. Work done in collaboration with Jose S. Andrade from Universidade Federal de Ceara, Brazil. Large cities are more productive than small ones so it shouldn’t come as a surprise that they produce more CO2 as well, say physicists. See press coverage at MIT Technology Review and Arizona News. Original article: Oliveira, E.A., Andrade, J.S. & Makse, H.A. Large cities are less green. Sci. Rep. 4, 4235; DOI:10.1038/srep04235 (2014), pdf. At the same time, the science of cites shows that bigger cities have fewer suicides per capita than smaller ones. Our new study shows that residents of bigger cities are less likely to commit suicide, suggesting the lonely find solace among increased opportunity for social interaction. Computational anthropologists Melo, Moreira, Makse and Soares suggest that the kind of emotional intensity associated with suicide might dissipate more easily in big cities, where there are more people to shoulder the burden, an idea known as emotional epidemics. Put another way, suicides are essentially a social phenomenon. See full article in the Arxiv and press coverage at MIT Technology Review. See also Medical Daily, The Guardian and the piece of Olga Khazan in The Atlantic: Hell might be other people, but they might just save you from yourself. Article at nextcity.org by Rebecca Tuhus-Dubrow dueling on the different definitions of cities including our CCA algoritms.
by Adrian Baule, Romain Mari, Lin Bo, Louis Portal, Hernan A. Makse. Finding the densest random packing of particles with a non-spherical shape is a long standing mathematical problem. Here, the authors develop a method based on a mean-field estimation of the Voronoi volume which can predict densest random packings in good agreement with empirical results. Date 23 Jul 2013 doi: 10.1038/ncomms3194. Nature Communications 4, Article number: 2194 doi:10.1038/ncomms3194. pdf
March 21, 2013. Paper in PLOS ONE. System-wide networks of proteins are indispensable for organisms. Function and evolution of these networks are among the most fascinating research questions in biology. Bioinformatician Thomas Rattei, University of Vienna, and physicist Hernan Makse, City University New York (CUNY), have reconstructed ancestral protein networks. The results are of high interest not only for evolutionary research but also for the interpretation of genome sequence data. Read more at: Phys.org. Full dataset of reconstructed ancestral protein interaction networks available here.
Dr. Hernan Makse’s APS citation reads:
The overarching theme of Dr. Makse’s research is the theoretical understanding of complexity. Dr. Makse’s original area of interest is the study of jammed matter, spanning from granular materials, colloidal suspensions, dense emulsions to glasses, in search of unifying theoretical frameworks. Under his 2003 NSF CAREER award, he studied statistical mechanics of particulate systems far from equilibrium. He is, however, continually coming up with new applications for the laws of physical systems, and by 2005, he was studying the dynamics of social networks under NSF auspices. Dr. Makse continues his ground-breaking work on granular matter, and, increasingly, he is applying the principles of statistical mechanics to the organization of complex networks from biological systems, to urban economics and social networks. This interdisciplinary work is at the interface of physics and disciplines such as neuroscience, biology and sociology. In recent papers he has addressed the function and evolution of protein networks, the environmental factors which may affect the spread of obesity, what makes the best spreaders of information in a social network, and a new way to define cities based on clustering algorithms from percolation theory. Dr. Makse travels the world in search of collaborators willing to take the same intellectual risks he does, and his lab at CCNY’s Benjamin Levich Institute is home to graduate students from China, Brazil, Argentina, Chile, France, Italy and the like.
What are our motivations in choosing our online friends? In our recent paper in PRX we study high-quality data and show how we can estimate the influence of different social drivers. In simpler words, how probable is it that we reply to friend requests or how often do we connect to popular people? Although we may not realize this, our connections evolve and we may find ourselves in an environment different from our choices. Physics Review X 2, 031014 (2012). pdf
Paper in Scientific Reports. Press release: An international team of researchers’ study of the spatial patterns of the spread of obesity suggests America’s bulging waistlines may have more to do with collective behavior than genetics or individual choices. The team, led by City College of New York physicist HernaÃ¡n Makse, found correlations between the epidemic’s geography and food marketing and distribution patterns. Talk at the Wolfram Data Summit 2012. Press Releases: CCNY. NSF Highlights. Science Daily. Medicalxpress. Science Blog. Supermarket or not Supermarket (Care2.com). The visible embryo. The Atlantic cities. Jeff Nesbit’s “On the Edge” blog in US News. Dataset available here.
Who are the best spreaders of information in a social network? Best connected individuals may not be the most influential spreaders. Instead, location in the network, as defined by the k-shell, determines influence. Click here for more information and dataset 1 and dataset 2: Twitter, Facebook, LiveJournal and full APS collaboration network. Paper: pdf or cond-mat. Press releases: Technology review, Science Daily, Fast company, Science for SEO, Emedia, NSF. F1000 prime article. Collaboration between Bar-Ilan University, Boston University, Stockholm University, NYU, and CCNY. Nature Physics 2010. (High resolution image and cover, created with the lanet-vi tool).
We investigate the network of human cell differentiation from the fertilized egg up to a crying baby. PNAS, 2010. Dataset of cell types. The full network is published in the Supporting Information in PNAS.
The City Clustering Algorithm, CCA, allows for a test of Zipf’s law for cities of all sizes. We find (ta,tan,ta,tan..) that Zipf’s law is surprisingly valid up to small cities of a few hundred inhabitants. Collaboration with Xavier Gabaix, Stern, NYU. Paper on Zipf’s law for all cities. Published in American Economic Review, August 2011. Below is an image of all the population clusters identified by the CCA in the USA and the CCA cluster around London superimposed with a Google maps.
We find scaling laws in human communication patterns (PNAS). A recent paper presents a new way to define cities based on clustering algorithms from percolation theory. We find that the growth rate of cities and its standard deviation follow (surprise, surprise..) power-laws with the city size, in contradiction to Gibrat’s law.
The small world-fractal transition and information flow. See recent paper in Phys. Rev. Lett. 2010.
Paper in the Nature issue of May 29, 2008, and Supplementary Materials. News & Views editorial by Zamponi. Press release. Nature Physics: Research Highlights, p435. Physics World. Science Daily. Physorg.com. Genetic Engineering & Biotechnology News.The following follow up papers are in Physica A and in cond-mat: Jamming I: A Hamiltonian for jammed matter. Jamming II: A phase diagram for jammed matter. Jamming III: Characterizing Randomness via the Entropy of Jammed Matter. Jamming IV: A distribution of volumes and coordination number in jammed matter: mesoscopic approximation. Jamming V: Jamming in two dimensions.
Can you improve the box-covering of a network? Download the algorithms and Databases of complex networks used in our studies to calculate the fractal dimension of a complex network. Including our PNAS paper on Scaling Theory of Transport in Complex Biological Networks in the May 2007 issue (pdf, supplementary information).
Explanation of self-similarity of Complex Networks. What is the relation between these Romanesque broccoli and the protein-protein interaction network of E.coli? From the January 2005 issue of Nature.
News and Views Editorial by Strogatz on Self-similar Complex Networks in the January 27th, 2005 issue of Nature.
Jamming is even cooler than you thought. A review article by H. A. Makse, J. Brujic and S. F. Edwards.
See a recent News Feature in the October 19th, 2003 issue of Nature.
A News and Views Nature Editorial by Bob Behringer in Nature.
Hernan Makse’s first scientific paper (well, second, after a small paper on dyslexic behavior in neural networks published in 1992). It was published in Nature on October 19, 1995. The study models the morphologies of cities in terms of correlated percolation and proposes a new way to look at cities, in the words of Michael Batty in the attached commentary in Nature published in New and View.