Florent Krzakala


From Statistical physics to computer science and back...

Originally, my main research focus has been the statistical physics of disordered systems and the study of the collective behavior of complex assemblies of simple elementary components. This includes glassy systems such as spin glasses, random polymers or a jammed ensemble of colloids, quantum problems but also simpler models in a out-of-equilibrium situation, for instance ferromagnets quenched at their critical points.

It turns out that this problematic naturally arises in other fields than physics, and in particular in computer science problems as diverse as combinatorial optimization, statistical inference, machine learning and signal processing. This has become over the last few years a rapidly evolving and exciting field at the interface of many disciplines, which we study together with the SPHINX group.



SPARCS

My main project over the last few years has been the SPARCS ones. This page contains a list of papers and conferences where the ERC Sparcs project has been featured. The ASPICS PAGE contains codes, papers and data on the algorithms we have been developing using statistical physics for inference in interdisciplinary applications, with a special focus on compressed sensing.

Publications

These are links to my complete list of publications, classified into different area. Please note that this classification is highly subjective and that the distinction is sometime subtle. I worked on:

* Constraint Satisfaction Problems

* Spin Glasses and Random Fields

* Glass and Jamming Transitions

* Quantum Phase Transitions

* Out-of-Equilibrium Dynamics

* RNA Folding

* Statistical inference on graphs

* Inference, Reconstruction and Learning

* Error Correcting Codes

* Computational optics


All my publications are also available on the arxiv, on Google Scholar and on citebase.
My Ph.D Thesis (in french) is available thanks to the CNRS web site.
You can also download my Habilitation a diriger des recherches. I also keep a complete list of my collaborators.


Selected talks on-line

* A serie of three lectures in ICTS Bangalore in India, in December 2016 on Inference and statistical phyaics.


* Invited talk at the "journée Claude Shannon" in Paris; INRIA in July; 2016 on "Shannon Capacity and Spatially-Coupled Superposition Codes"


* Invited talk at the Simons Institue (Berkeley) in May 2016 on Low Rank Matrix factorization:


* Keynote talk at the GRETSI 25 in Lyon 2015 (in french) on Statistical Physics and signal processing:


* Invited talk at the Big Data conference in Harvard (2015) on Optimal compressed sensing with spatial coupling and message passing:


* Invited talk at the Simons Institue (Berkeley) in February 2015 on Sparse Superposition Codes for the Error-Correction over the AWGN Channel:


* Paris: Workshop on Community Detection at Institut Henri Poincare, February 2015The Bethe Hessian Operator for graph clustering.
* Philips Research (Suresnes) April 2012: January 2011: a seminar (in pdf) on our new approach to compressed sensing.
* ENS (France) February 2012: a seminar (in pdf) for student on the use of statistical physics outside physics.
* Bardonecchia (Italy) February 2010: a seminar (in pdf) on Bayesian Inference in community detection called Hard, easy and impossible inference.
* Harvard University (Cambridge, USA) December 2010: A seminar (in pdf) on the glass transition as a melting process.
* International Center of Theoritical Physics (Trieste, Italy) May 2010: A seminar (in pdf) on Quantum computation from a spin glass perspective.
* International Center of Theoritical Physics (Trieste, Italy) November 2009: A seminar (in pdf) on Adiabatic dynamics in glassy systems.
* Center of Research in Mathematics (Barcelona, Spain) October 2009: My seminar (in pdf) in the workshop on "Techniques and Challenges from Statistical Physics". I spoke about Where the really hard problems are ? .
* University of Chicago (IL, USA) October 2009: A seminar (in pdf) on Jamming transition in mean field models .
* Santa Fe (NM, USA) Auguest 2009 : My seminar in the workshop "Physics of Algorithm '09" on simulating impossible to-simulate models .
* Princeton University (NJ USA). A seminar (in html) on The Quantum cavity Method.
* Los Alamos (NM USA), LANL. A seminar (in flash) on Spin Glass Quantum transition.
* Zurich, ETH Theory Seminar, April 2007 on the coloring of random graphs.
* Paris, Summer School SPIN-GLASS 2007. See my lecture on 3d spin glasses.
* Kyoto, "International Workshop on Statistical-Mechanical Informatics 2007", September 2007, invited talk on Phase Transition In Random Constraint Satisfaction Problems .


Contact

Professor Florent Krzakala

University Pierre et Marie Curie
Lab. de Physique Statistique
ENS and UMR CNRS 8550
Tel: +33 (0)1 44 32 25 82
Fax: +33 (0)1 40 79 47 31

Ecole Normale Superieure
24 Rue Lhomond
75231 PARIS Cedex 05
florent.krzakala@ens.fr

To come and see me
ENS is in the center of Paris, close to the Pantheon. My office is L284 in the Physics Departement (second floor).