Giorgio Parisi, Nobel Prize and Beyond

"Parisi is right" says the Italian Prime Minister Mario Draghi "our research funding is way lower than that of the countries around us and the government is determined to bridge this gap as much as possible, both for basic research funds and for funds toward applied research "

On October 5th 2021, the Swedish Royal Academy of Sciences announced that Giorgio Parisi is one of the Nobel Prize winners in physics. Parisi is a theoretical physicist who has been trained, has worked and works in Rome, and for the past three years he has been the President of the Accademia Nazionale dei Lincei. After a crescendo of prestigious awards, including the Boltzmann Medal and the Wolf Prize, he received the Nobel "for the discovery of the interplay of disorder and fluctuations in physical systems from atomic to planetary scales".For most of the best physicists in the world, for many international colleagues, for his collaborators and for the students of the Sapienza University who have attended his courses, Giorgio is a scientist of supreme level, one of those sitting at that imaginary table with the giants of the history of physics, from the Hellenistic period to the present day.

 

But what do physicists really do? Theorists have a fascinating job: they describe the world by grasping its essence and do it so well that with their methods one can often make accurate predictions or, some other times, one can understand why predictions are partial or impossible. From those methods derives our ability to build prodigious machines like those that drove the industrial revolution with all its branches and those that will drive the next ones. To do this, physicists use, in the Galilean tradition, that extraordinary quantitative language which is mathematics. Of course the part of the world that can be described with physics is very limited but, trying and trying again, it gradually widens as the research progresses.

This year the Nobel was awarded for the understanding of complex physical systems and, in the case of Parisi, mostly for his statistical physics theory that links disorder and fluctuations. To describe it in a colloquial way, one can draw inspiration from the same effective metaphor used in the introduction of the text by Mézard, Parisi and Virasoro: consider a system composed by many particles - imagine them as a group of people. If each of their interactions is cooperative, the system will soon find a shared agreement and the consensus reached will be certain and satisfactory for everyone. If, on the other hand, the interactions are disordered, some cooperative and others competitive, an agreement will not be found that satisfies everyone and the best compromises will be very difficult to reach. In this second case, the agreement obtained will be intrinsically fluctuating. Moreover, from the nature of those fluctuations Parisi's theory is able to obtain the main properties of those systems.

When Parisi approached these themes in the late 1970s, they were not fashionable and had no relevant practical applications. Their understanding, as the American physicist who proposed them to the community (PW Anderson, Nobel Prize for Physics in 1977) said, deserved nevertheless a thorough study for the simple fact that they represented an authentic scientific mystery.

The Nobel Prize to Parisi was therefore awarded for a theoretical result that anticipated many of the subsequent experiments, a freely invented theory as Einstein liked to say, one that the pen deposits on the white paper, and which, over time, becomes essential to understand a whole series of real phenomena. There was, however, a somehow experimental side that accompanied his work, provided by the virtual environment of computer simulations. Using his legendary familiarity with the computer as early as in the 80s, when most considered it a hellish contraption, Giorgio has always compared his theory with the data produced by the machine in a sort of meta-experiment, receiving feedback and confirmation.

More than forty years after its discovery, the conceptual scheme underlying that theory, and developed through the very rich collaborations that the author has had, has become a new paradigm of science. It represents an interpretative method with applications to the behavior of glassy materials and that of random lasers, to name just a couple. But not only that, an uncommon fact for a theory born in the bed of theoretical physics, is that the same scheme has fertilised fields distant from the hard sciences and opened new horizons such as biological systems - for the immune system, for mechanisms describing the associative memory in the brain- and ecological, economic, financial systems etc. Finally, the neural networks, which led to those marvels of modern artificial intelligence based on machine learning, are studied with ideas, methods and techniques of Parisi's theory.

 

But how does a Nobel winner come out? Is it a matter of nature or nurture ? A set of innate natural gifts or the intellectual environment in which he grew up? Certainly it is an artfully mixed combination of the two, according to a very secret recipe. Giorgio's natural intuition, his analytical skills and his curiosity found themselves immersed in that school of Italian physics founded by Enrico Fermi, that Amaldi had gathered in Rome after WWII. Among his mentors,  first and foremost the physicist Cabibbo, among his colleagues Altarelli, Maiani, Jona Lasinio and Francesco Guerra. To the latter we owe some of the best ideas that led to the slow metamorphosis, still in progress, of that brilliant theory of Parisi into a mathematically rigorous construct deducible from the principles of thermodynamics, a transformation that allowed mathematicians to approach the study of disordered systems. The scientific environment in Rome, in Giorgio's time, has become the country's school of excellence, the one that attracts at the same time the best students, the best researchers and the best international visitors. In fact, a remarkable and counter-trend phenomenon manifests itself at conferences on disordered systems: during discussions in small groups or coffee breaks, whether they take place in Paris, in Kyoto or in New York, many non-natives speak Italian with Italians and not infrequently among themselves as they did in Rome at Giorgio's school.

And how does the assignment of a prestigious award come about? Science is not measured in inches like heights. In a group you can always line up people and choose the highest one but talent is an infinitely richer quality.Its evaluation is essentially based on the impact of the scientific contribution but since science is also power and wealth for the host country it is natural that political as well as personal factors play a role.It is natural I say, but it is not obvious. Certainly some countries in the world are well aware of this and when it is necessary to present their candidate for a prestigious international award they silence internal tensions, choose their champion, and express themselves unanimously. A deserving scientist working in those countries where science has the consideration it deserves will have his honest chance of receiving the awards he can aspire to. But Italy does not always have the awareness of being a country, it is too busy chasing the financial bills as if they were balloons rolling down the stairs, staging and enjoying talk shows where influencers, extras and improvised politicians discuss everything and more, especially in the fields of competence of others. And so it happens that the Italians who receive awards work more often abroad after being trained in Italy, assisted by an organization that we are no longer able to express after having invented it during the Roman Empire.

A Nobel Prize that reaches an Italian who works in Italy is therefore a different phenomenon, which has followed a different path. The Nobel Prize to Parisi was a natural and unstoppable process of accumulating scientific credit, aided by the collaborations and admiration that surrounded him. The Italian scientific environment together with the international one spontaneously united around this extraordinary scientist with a kind character, alien to conflicts and generous in sharing his ideas and accompanied him to the podium. It goes without saying, as Giorgio himself often complains, that many Italian researchers have not received the recognition they deserved.

Parisi has always been concerned about the health of science in Italy, its dramatic needs, first of all the lack of investment. It is neither a mystery nor a surprise that his appeals have been systematically ignored despite promises made by politics. Prime Minister Draghi’s words and tones are therefore welcomed, perhaps it is the right time to make a change by taking the opportunity of the The National Recovery and Resilience Plan. Parisi, however, accepted with great caution the optimism of the minister of university and research on the evening of the celebrations in the great hall of the Sapienza University.

Five years ago he summarized his appeals as follows: "In a world dominated by the knowledge economy, a country that does not invest in research, development and culture has no future". Now that this has become the appeal of a Nobel prize winner, who knows if it will be enough to convince the government that a future would be useful.

 

Pierluigi Contucci, October 12th 2021.

Translated from an article published in Rivista Il Mulino

 

Accademia delle Scienze Bologna  

Intelligenza Artificiale: farsi un'idea.

 Matematica e apprendimento automatico, l'approccio meccanico statistico

Artificial Intelligence and Mathematics, CNR.

 COVID, strategie contro la terza ondata

Con Andrea Crisanti, Luca Ricolfi, Antonio Bianconi, Giampietro Ravagnan, Pierluigi Contucci (minuto 50:30), Cesare Saccani, Giuseppe Valditara, Luigi Cavanna - a cura di Laura Cuppini e Maria Serena Natale /CorriereTv

 

Colloqui Scientifici UMI, 

"Apprendimento Automatico: sfide e opportunità per la Matematica"

Abstract:

L’intelligenza artificiale sta cambiando in modo veloce e profondo il profilo tecnologico della nostra società. Eppure l’apprendimento automatico, che ora è la sua punta di diamante, non ha ancora basi scientifiche solide. Il colloquio fornirà una prospettiva sulle euristiche in cui esso si fonda e inquadrerà i principali problemi aperti. L’approccio della meccanica statistica a questi temi verrà brevemente discusso insieme a qualche risultato preliminare. La tesi di fondo è che le sfide scientifiche dell’Intelligenza Artificiale sono una grandissima opportunità per la Matematica.

AI, una risorsa per l’Italia. Strategie per il futuro - AI, a resource for Italy. Strategies for the future

Estratto del mio contributo al Webinar "Intelligenza Artificiale e Sistema Paese" (27 giugno 2020), organizzato da Lettera 150. Testo bilingue italiano e inglese.

%%%

AI, una risorsa per l’Italia. Strategie per il futuro

Pierluigi Contucci – Professore Ordinario di Fisica Matematica – Università di Bologna

Quella dell’Intelligenza Artificiale è una rivoluzione in atto e in questi anni sta iniziando una fase per molti versi simile a quella della rivoluzione industriale che ha trasformato radicalmente la società moderna. Allora i primi motori a vapore mostravano che era possibile generare energia dalla combustione e con essa azionare macchine che non necessitavano di forza umana o animale. Oggi le nuove tecniche di apprendimento automatico sintetizzano parte dell’informazione contenuta nelle grandi basi di dati e generano una conoscenza che, pur piccola e rudimentale, non proviene da cervello biologico umano o animale. L’esempio lampante è la classificazione delle immagini: dopo aver visto alcune (molte!) foto di una persona la macchina ne riconosce di nuove e sa distinguerle da quelle di altre persone. Diversi istituti di ricerca stimano l’impatto di questa nuova tecnologia nel PIL europeo del prossimo decennio superiore mediamente al 16%, il doppio della crescita che abbiamo avuto nel decennio successivo alla seconda guerra mondiale. E’ evidente quindi che occorre governare in modo virtuoso la tecnologia emergente, la scienza di cui abbisogna e la sua interazione con l’umano che deve essere sempre al centro dello sviluppo. Queste premesse da sole fanno comprendere la complessità del problema di governo che ci troviamo ad affrontare e la necessità di coordinare in modo efficiente la sua organizzazione. Se da un lato infatti la realizzazione di una tale crescita va realizzata con solidi strumenti tecnico-scientifici che dovranno essere coltivati e promossi dall’altro il problema del se e come questa crescita verrà incanalata nel benessere dei cittadini è tutt’altro che scontato e merita la alta attenzione e priorità. Gli studi degli ultimi due decenni sul benessere che va ben oltre il prodotto interno lordo, su come questo si misuri in maniera non arbitraria, potranno servire da guida.
Una strategia per il paese va quindi pensata in ottica continentale e nel nostro caso europea. Nel panorama internazionale infatti Stati Uniti e Cina sono gli attori principali e nessun paese europeo può, da solo, entrare nel vivo della competizione. Serve invece una forte alleanza in EU che punti sulla condivisione di risorse finanziarie, culturali e umane oltre che su comuni direzioni di sviluppo concordate tra i paesi membri.
Il caso italiano merita una serie di considerazioni a parte essendo frutto di un ambiente molto peculiare. Da oltre mezzo secolo circola infatti una narrazione secondo cui le nostre forze produttive, economiche, accademiche e culturali sono allo sbando e prive di impulso verso il futuro. Sono periodiche ma piuttosto inascoltate le smentite, dati alla mano, di queste persistente vulgata anche perché molto meno godibili nel gossip del passaparola. Il paese è ricchissimo di ingegno che sarà la materia prima della crescita che sperabilmente sarà capace di raggiungere. Quel che manca è un apparato organizzativo e istituzionale in grado di promuovere l’ingegno, metterlo in condizione di operare in modo fluido invece che imbrigliarlo e umiliarlo di continuo con normative scritte di fretta e magari con buone intenzioni ma che inevitabilmente si trasformano in cavilli vessatori. Il lavoro all’interno della commissione strategica del MIUR, la collaborazione e l’intesa con l’omonimo francese su AI, ha mostrato chiaramente le direzioni verso cui muoversi e i rischi da scongiurare. L’intrinseca trasversalità di AI, ottima dote in fase di alta velocità, è al contempo causa di grande viscosità nelle fasi inziali dello sviluppo. Per fare un esempio a livello di paese non c’è dicastero che non sia coinvolto o interessato, direttamente o meno, da questa nuova tecnologia e allo stato attuale ognuno tenderà a tirare ed egemonizzare le parti di sua competenza. Scendendo a livello accademico: sono moltissimi i settori scientifico disciplinari che credono di essere i naturali depositari di questa scienza che, per inciso, scienza non lo è ancora. Per eliminare la frizioni presenti ad ogni livello è assolutamente indispensabile individuare strutture di raccordo e coordinamento. Partendo dalle fondamenta (ma non necessariamente in ordine cronologico) serve un centro di ricerca in AI, magari Europeo, che funzioni analogamente al CERN. Serve poi un’agenzia per l’AI con una impostazione non dissimile a quella dell’Agenzia Spaziale Italiana che raccordi mondo accademico e industriale e canalizzi le risorse nazionali e produttive. Infine, e questo magari dopo aver avviato i centri precedenti, servirà prima o poi un ministero in AI che funga da collettore politico finanziario. E’ impensabile gestire una rivoluzione di tipo industriale senza un dicastero dedicato.

%%%

AI, a resource for Italy. Strategies for the future

Pierluigi Contucci – Full Professor in Mathematical Physics, University of Bologna

Artificial Intelligence is a revolution in progress. In recent years it is going through a phase similar to that of the industrial revolution that has radically transformed the human society and accompanied it into the modern era. At that time the first steam engines showed that it was possible to generate energy from combustion and with it to operate machines that did not require human or animal power. Today the new machine learning techniques synthesise part of the information contained in the large databases and generate knowledge that, although small and rudimentary, does not come from the human or animal biological brain. The striking example is the classification of images: after seeing some (many!) photos of a person, the machine recognizes new ones and can distinguish them from those of other people. Several research institutes estimate the impact of this new technology on the European GDP for the next decade by an average above 16%, twice as much the growth that we had in the decade following the Second World War. It is therefore clear that such emerging technology must be governed in a virtuous way, as much as the science it needs and its interaction with the human who must always be at the center of it. These premises alone give a glimpse of the complexity of the governance problem we are facing and the need to efficiently coordinate its organization. If on the one hand the realization of such growth must be achieved with solid technical-scientific tools that must be cultivated and promoted on the other, the problem of whether and how this growth will be channeled into the well-being of citizens is far from obvious and deserves high attention and priority. The studies of the past two decades on well-being that goes far beyond the gross domestic product, on how this is measured in a non-arbitrary way, can serve as a guide.
A strategy for the country must therefore be thought of from a continental , European, perspective. In the international panorama, in fact, the United States and China are the main actors and no European country can, on its own, enter the heart of the competition. Instead, a strong alliance in the EU is needed which focuses on the sharing of financial, cultural and human resources as well as on common development directions agreed between the member countries.
The Italian case deserves a series of separate considerations as it is the result of a very peculiar environment. For more than half a century, a narrative has circulated according to which our productive, economic, academic and cultural forces are in disarray and have no impulse towards the future. The denials of these persistent vulgata are periodic but rather unheard, also because they are much less enjoyable in gossip based word of mouth. The country is fully endowed of inventive talent which will be the raw material of the growth that it will hopefully be able to achieve. What is missing is an organizational and institutional apparatus capable of promoting that talent, putting it in a position to operate in a fluid way rather than harnessing it and continuously humiliating it with regulations written in a hurry and perhaps with good intentions but which inevitably turn into vexing quibbles. The work within the MIUR strategic commission, the collaboration and understanding with the French side on AI, clearly showed the directions in which to move and the risks to be carefully avoided. The intrinsic transversality of AI, an excellent gift at high speed, is at the same time a cause of great viscosity in the initial stages of development. To give an example at the country level there is no dicastery that is not involved or interested, directly or not, by this new technology and at the time being everyone will tend to pull and dominate the parts of his own competence. Going down to the academic level: there are many scientific disciplinary sectors that believe they are the natural depositories of this science which, incidentally, is not a mature science yet. To eliminate the frictions present at each level it is absolutely essential to identify and create institutional and coordination structures. Starting from the foundations (but not necessarily in chronological order) we need a research center in AI, perhaps European, which would work similarly to CERN. Then we need an agency for AI with functions similar to the Italian Space Agency that connects academia and industry and channels national and productive resources. Finally, this step to be done after starting the previous centers, sooner or later we will need an AI ministry that will serve as a political and financial center. It is unthinkable to manage an industrial revolution without a dedicated dicastery.

 The potential for AI in Industry

Engisoft Newsletter interview (page 6)