Riccardo Zecchina

Timeline

1. 1988–1993

   Turin degree and PhD with Regge

   Born in Turin in 1963, he graduates in electronic engineering from the Politecnico di Torino in 1988. He earns a PhD in theoretical physics at the University of Turin in 1993, working with Tullio Regge.

2. 1997–2007

   Leading statistical physics at ICTP

   In 1997, he is appointed research scientist and head of the Statistical Physics Group at the International Centre for Theoretical Physics in Trieste, a post he holds until 2007. In 1999, Nature publishes his paper with Monasson, Kirkpatrick, Selman and Troyansky on determining computational complexity from characteristic phase transitions.

3. 2002

   Survey propagation for satisfiability

   At ICTP, he and Marc Mézard devise survey propagation, a message-passing variant of the cavity method for satisfiability problems. In 2002, he, Mézard and Giorgio Parisi publish 'Analytic and algorithmic solution of random satisfiability problems' in Science.

4. 2007–2017

   Professor at Politecnico di Torino

   In 2007, he becomes full professor of theoretical physics at the Politecnico di Torino. His European Research Council Advanced Grant runs from 2011 to 2015, for work on optimization and inference algorithms from the theory of disordered systems.

5. 2016

   The Lars Onsager Prize

   In 2016, he shares the American Physical Society's Lars Onsager Prize in Theoretical Statistical Physics with Marc Mézard and Giorgio Parisi, for applying spin-glass ideas to groups of computational problems, which yielded a new class of efficient algorithms.

6. 2016

   Dense regions and Entropy-SGD

   In 2016, he coauthors 'Unreasonable Effectiveness of Learning Neural Networks' with Carlo Baldassi, Christian Borgs, Jennifer Chayes and others, which finds rare, extremely dense and accessible regions in the weight space of networks with discrete weights. Also in 2016, he coauthors Entropy-SGD with Yann LeCun, Stefano Soatto and others, a training algorithm for deep networks that favors large flat regions of the energy landscape, for ICLR 2017.

7. 2017–2025

   Computing Sciences at Bocconi

   He moves to Bocconi University in Milan in 2017, and the Vodafone Chair in Machine Learning and Data Science, established in 2018, is held by him. In 2022, he directs the newly formed Department of Computing Sciences, and in March 2025 Alon Rosen is named its new director, with Zecchina acknowledged as a predecessor.

8. As of October 2026

   Bocconi professor and ELLIS director

   In November 2021, the four Milan universities form an ELLIS Unit, with him as one of its four scientific coordinators. He is a professor of theoretical physics at Bocconi with a chair in machine learning, Unit Director of ELLIS Unit Milan, Program Director for ELLIS's Quantum and Physics Based Machine Learning program, and an ELLIS Fellow. In October 2025, he presents work on dynamical learning in deep asymmetric recurrent neural networks at the EPFL AI Center.