| Les deux révisions précédentes Révision précédente | Prochaine révisionLes deux révisions suivantes |
| mega:seminaire [2019/03/11 20:30] – [Prochaine séance] male | mega:seminaire [2019/03/11 20:35] – [Prochaine séance] male |
|---|
| ===== Prochaine séance ===== | ===== Prochaine séance ===== |
| * Vendredi **15 Mars** salle 421 | * Vendredi **15 Mars** salle 421 |
| * mini cours par **[[https://www.di.ens.fr/~lelarge/|Marc Lelarge]]** //The cavity method for the spiked Wigner model.// We consider the high-dimensional inference problem where the signal is a low-rank symmetric matrix which is corrupted by an additive Gaussian noise. Given a probabilistic model for the low-rank matrix, we compute the limit in the large dimension setting for the mutual information between the signal and the observations, as well as the matrix minimum mean squared error, while the rank of the signal remains constant. We unify and generalize a number of recent works on PCA, sparse PCA, submatrix localization or community detection by computing the information-theoretic limits for these problems in the high noise regime. \\ | * 10h30-12h00: mini cours par **[[https://www.di.ens.fr/~lelarge/|Marc Lelarge]]** //The cavity method for the spiked Wigner model.// We consider the high-dimensional inference problem where the signal is a low-rank symmetric matrix which is corrupted by an additive Gaussian noise. Given a probabilistic model for the low-rank matrix, we compute the limit in the large dimension setting for the mutual information between the signal and the observations, as well as the matrix minimum mean squared error, while the rank of the signal remains constant. We unify and generalize a number of recent works on PCA, sparse PCA, submatrix localization or community detection by computing the information-theoretic limits for these problems in the high noise regime. \\ |
| * 14h00-15h00: **[[http://math.univ-lyon1.fr/~lancien/|Cécilia Lancien]]** //Random correlation matrices: when are they with high probability classical or quantum?// Two observers performing binary outcome measurements on subsystems of a global system may obtain more strongly correlated results when they have a shared entangled quantum state than when they only have shared randomness. This well-known phenomenon of so-called Bell inequality violation can be precisely characterized mathematically. Indeed, being a classical or a quantum correlation matrix exactly corresponds to being in the unit ball of some tensor norms. In this talk, I will start with explaining all this in details. I will then look at the following problem: given a random matrix of size n, can one estimate the typical value of its "classical" and "quantum" norms, as n becomes large? For a wide class of random matrices, the answer is yes, and shows a separation between the two values. This result may be interpreted as follows: in a typical direction, the borders of the sets of classical and quantum correlations are separated away from each other. Based on joint work with Carlos Gonzalez-Guillen, Carlos Palazuelos and Ignacio Villanueva: "Random quantum correlations are generically non-classical", Ann. Henri Poincaré, 18(12):3793-3813 (2017).\\ | * 14h00-15h00: **[[http://math.univ-lyon1.fr/~lancien/|Cécilia Lancien]]** //Random correlation matrices: when are they with high probability classical or quantum?// Two observers performing binary outcome measurements on subsystems of a global system may obtain more strongly correlated results when they have a shared entangled quantum state than when they only have shared randomness. This well-known phenomenon of so-called Bell inequality violation can be precisely characterized mathematically. Indeed, being a classical or a quantum correlation matrix exactly corresponds to being in the unit ball of some tensor norms. In this talk, I will start with explaining all this in details. I will then look at the following problem: given a random matrix of size n, can one estimate the typical value of its "classical" and "quantum" norms, as n becomes large? For a wide class of random matrices, the answer is yes, and shows a separation between the two values. This result may be interpreted as follows: in a typical direction, the borders of the sets of classical and quantum correlations are separated away from each other. Based on joint work with Carlos Gonzalez-Guillen, Carlos Palazuelos and Ignacio Villanueva: "Random quantum correlations are generically non-classical", Ann. Henri Poincaré, 18(12):3793-3813 (2017).\\ |
| * 15h30-16h30: **[[http://pub.ist.ac.at/~dschroed/|Dominik Schröder]]** //Cusp Universality for Wigner-type Random Matrices.// For Wigner-type matrices, i.e. Hermitian random matrices with independent, not necessarily identically distributed entries above the diagonal, we show that at any cusp singularity of the limiting eigenvalue distribution the local eigenvalue statistics are universal and form a Pearcey process. Since the density of states typically exhibits only square root or cubic root cusp singularities, our work complements previous results on the bulk and edge universality and it thus completes the resolution of the Wigner-Dyson-Mehta universality conjecture for the last remaining universality type.\\ | * 15h30-16h30: **[[http://pub.ist.ac.at/~dschroed/|Dominik Schröder]]** //Cusp Universality for Wigner-type Random Matrices.// For Wigner-type matrices, i.e. Hermitian random matrices with independent, not necessarily identically distributed entries above the diagonal, we show that at any cusp singularity of the limiting eigenvalue distribution the local eigenvalue statistics are universal and form a Pearcey process. Since the density of states typically exhibits only square root or cubic root cusp singularities, our work complements previous results on the bulk and edge universality and it thus completes the resolution of the Wigner-Dyson-Mehta universality conjecture for the last remaining universality type.\\ |