m
• D

F Nous contacter

0

Documents : Post-edited  Conférences Vidéo Chapitrées | enregistrements trouvés : 200

O

P Q

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Network-guided feature selection in high-dimensional genomic data Azencott, Chloé (Auteur de la Conférence) | CIRM (Editeur )

Differences in disease predisposition or response to treatment can be explained in great part by genomic differences between individuals. This has given birth to precision medicine, where treatment is tailored to the genome of patients. This field depends on collecting considerable amounts of molecular data for large numbers of individuals, which is being enabled by thriving developments in genome sequencing and other high-throughput experimental technologies.
Unfortunately, we still lack effective methods to reliably detect, from this data, which of the genomic features determine a phenotype such as disease predisposition or response to treatment. One of the major issues is that the number of features that can be measured is large (easily reaching tens of millions) with respect to the number of samples for which they can be collected (more usually of the order of hundreds or thousands), posing both computational and statistical difficulties.
In my talk I will discuss how to use biological networks, which allow us to understand mutations in their genomic context, to address these issues. All the methods I will present share the common hypotheses that genomic regions that are involved in a given phenotype are more likely to be connected on a given biological network than not.
Differences in disease predisposition or response to treatment can be explained in great part by genomic differences between individuals. This has given birth to precision medicine, where treatment is tailored to the genome of patients. This field depends on collecting considerable amounts of molecular data for large numbers of individuals, which is being enabled by thriving developments in genome sequencing and other high-throughput ex...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Learning with differentiable perturbed optimizers Berthet, Quentin (Auteur de la Conférence) | CIRM (Editeur )

Machine learning pipelines often rely on optimization procedures to make discrete decisions (e.g. sorting, picking closest neighbors, finding shortest paths or optimal matchings). Although these discrete decisions are easily computed in a forward manner, they cannot be used to modify model parameters using first-order optimization techniques because they break the back-propagation of computational graphs. In order to expand the scope of learning problems that can be solved in an end-to-end fashion, we propose a systematic method to transform a block that outputs an optimal discrete decision into a differentiable operation. Our approach relies on stochastic perturbations of these parameters, and can be used readily within existing solvers without the need for ad hoc regularization or smoothing. These perturbed optimizers yield solutions that are differentiable and never locally constant. The amount of smoothness can be tuned via the chosen noise amplitude, whose impact we analyze. The derivatives of these perturbed solvers can be evaluated eciently. We also show how this framework can be connected to a family of losses developed in structured prediction, and describe how these can be used in unsupervised and supervised learning, with theoretical guarantees.
We demonstrate the performance of our approach on several machine learning tasks in experiments on synthetic and real data.
Machine learning pipelines often rely on optimization procedures to make discrete decisions (e.g. sorting, picking closest neighbors, finding shortest paths or optimal matchings). Although these discrete decisions are easily computed in a forward manner, they cannot be used to modify model parameters using first-order optimization techniques because they break the back-propagation of computational graphs. In order to expand the scope of learning ...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Signaling gradients in surface dynamics as basis for regeneration in flatworms Stevens, Angela (Auteur de la Conférence) | CIRM (Editeur )

We introduce and analyze a mathematical model for the regeneration of planarian flatworms. This system of differential equations incorporates dynamics of head and tail cells which express positional control genes that in turn translate into localized signals that guide stem cell differentiation. Orientation and positional information is encoded in the dynamics of a long range wnt-related signaling gradient.
We motivate our model in relation to experimental data and demonstrate how it correctly reproduces cut and graft experiments. In particular, our system improves on previous models by preserving polarity in regeneration, over orders of magnitude in body size during cutting experiments and growth phases. Our model relies on tristability in cell density dynamics, between head, trunk, and tail. In addition, key to polarity preservation in regeneration, our system includes sensitivity of cell differentiation to gradients of wnt-related signals measured relative to the tissue surface. This process is particularly relevant in a small tissue layer close to wounds during their healing, and modeled here in a robust fashion through dynamic boundary conditions.
We introduce and analyze a mathematical model for the regeneration of planarian flatworms. This system of differential equations incorporates dynamics of head and tail cells which express positional control genes that in turn translate into localized signals that guide stem cell differentiation. Orientation and positional information is encoded in the dynamics of a long range wnt-related signaling gradient.
We motivate our model in relation to ...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Estimating parameters in the initial phase of an epidemic Scalia Tomba, Gianpaolo (Auteur de la Conférence) | CIRM (Editeur )

In recent years, new pandemic threats have become more and more frequent (SARS, bird flu, swine flu, Ebola, MERS, nCoV...) and analyses of data from the early spread more and more common and rapid. Particular interest is usually focused on the estimation of $R_{0}$ and various methods, essentially based estimates of exponential growth rate and generation time distribution, have been proposed. Other parameters, such as fatality rate, are also of interest. In this talk, various sources of bias arising because observations are made in the early phase of spread will be discussed and also possible remedies proposed.
In recent years, new pandemic threats have become more and more frequent (SARS, bird flu, swine flu, Ebola, MERS, nCoV...) and analyses of data from the early spread more and more common and rapid. Particular interest is usually focused on the estimation of $R_{0}$ and various methods, essentially based estimates of exponential growth rate and generation time distribution, have been proposed. Other parameters, such as fatality rate, are also of ...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Evolutionary rescue over a phenotype fitness landscape: across the mutation rate spectrum Martin, Guillaume (Auteur de la Conférence) | CIRM (Editeur )

Evolutionary rescue (ER) is the process by which a population, initially destined to extinction due to environmental stress, avoids extinction via adaptive evolution. One of the widely observed pattern of ER (especially in the study of antibiotic resistance) is that it is more likely to occur in mild than in strong stress. This may be due either to purely demographic effects (extinction is faster in strong stress) or to evolutionary effects (adaptation is harder in strong stress). Disentangling the two and predicting the likelihood of ER has important medical or agronomic implications, but also has a strong potential for empirical testing of eco-evolutionary theory, as ER experiments are widespread (at least in microbial systems) and fairly rapid to perform.
Here, I will present results from three recent articles [1-3] where we considered the probability of ER, and the distribution of extinction times, in a classic phenotype-fitness landscape: Fisher’s geometric model (FGM). In our (classic) version of the FGM, fitness is a quadratic function of traits, with an optimum that depends on the environment. This model has received some empirical support with respects to its ability to reproduce or even predict patterns of context dependence in mutation effects on fitness (be it environmental or genetic context).
In our FGM-ER scenario, a population is initially adapted to the current optimum (either a clone or at mutation selection balance). The environment shifts abruptly and the optimum position, plus possibly peak height and width are modified. We follow the evolutionary and demographic response to this change, assuming a density-independent demography (which we approximate by continuous branching process CB process or Feller process).
In spite of its simplicity, the FGM displays fairly distinct behaviors depending on the relative strength of selection and mutation: this yields different approaches to deal with the FGM-ER scenario. I will thus present the different approaches we have used so far: from the strong selection, weak mutation regime to the weak mutation strong selection regime, and discuss possible extensions at the transition between these regimes.
Evolutionary rescue (ER) is the process by which a population, initially destined to extinction due to environmental stress, avoids extinction via adaptive evolution. One of the widely observed pattern of ER (especially in the study of antibiotic resistance) is that it is more likely to occur in mild than in strong stress. This may be due either to purely demographic effects (extinction is faster in strong stress) or to evolutionary effects ...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Unique ergodicity for foliations on compact Kähler surfaces Sibony, Nessim (Auteur de la Conférence) | CIRM (Editeur )

How to study the dynamics of a holomorphic polynomial vector field in $\mathbb{C}^{2}$? What is the replacement of invariant measure? I will survey some surprising rigidity results concerning the behavior of these dynamical system. It is helpful to consider the extension of this dynamical system to the projective plane.
Consider a foliation in the projective plane admitting a unique invariant algebraic curve. Assume that the foliation is generic in the sense that its singular points are hyperbolic. With T.-C. Dinh, we showed that there is a unique positive $dd^{c}$-closed (1, 1)-current of mass 1 which is directed by the foliation. This is the current of integration on the invariant curve. A unique ergodicity theorem for the distribution of leaves follows: for any leaf $L$, appropriate averages on $L$ converge to the current of integration on the invariant curve (although generically the leaves are dense). The result uses our theory of densities for currents. It extends to Foliations on Kähler surfaces.
I will describe a recent result, with T.-C. Dinh and V.-A. Nguyen, dealing with foliations on compact Kähler surfaces. If the foliation, has only hyperbolic singularities and does not admit a transverse measure, in particular no invariant compact curve, then there exists a unique positive $dd^{c}$-closed (1, 1)-current of mass 1 which is directed by the foliation( it’s like uniqueness of invariant measure for discrete dynamical systems). This improves on previous results, with J.-E. Fornæss, for foliations (without invariant algebraic curves) on the projective plane. The proof uses a theory of densities for positive $dd^{c}$-closed currents (an intersection theory).
How to study the dynamics of a holomorphic polynomial vector field in $\mathbb{C}^{2}$? What is the replacement of invariant measure? I will survey some surprising rigidity results concerning the behavior of these dynamical system. It is helpful to consider the extension of this dynamical system to the projective plane.
Consider a foliation in the projective plane admitting a unique invariant algebraic curve. Assume that the foliation is ...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  How to compute using quantum walks Kendon, Vivien (Auteur de la Conférence) | CIRM (Editeur )

Quantum walks are widely and successfully used to model diverse physical processes. This leads to computation of the models, to explore their properties. Quantum walks have also been shown to be universal for quantum computing. This is a more subtle result than is often appreciated, since it applies to computations run on qubit-based quantum computers in the single walker case, and physical quantum walkers in the multi-walker case (quantum cellular automata). Nonetheless, quantum walks are powerful tools for quantum computing when correctly applied. I will explain the relationship between quantum walks as models and quantum walks as computational tools, and give some examples of their application in both contexts.
Quantum walks are widely and successfully used to model diverse physical processes. This leads to computation of the models, to explore their properties. Quantum walks have also been shown to be universal for quantum computing. This is a more subtle result than is often appreciated, since it applies to computations run on qubit-based quantum computers in the single walker case, and physical quantum walkers in the multi-walker case (quantum ...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  On determinants of random matrices Zeitouni, Ofer (Auteur de la Conférence) | CIRM (Editeur )

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Emergence of wandering stable components Berger, Pierre (Auteur de la Conférence) | CIRM (Editeur )

In a joint work with Sebastien Biebler, we show the existence of a locally dense set of real polynomial automorphisms of $\mathbb{C}^{2}$ displaying a stable wandering Fatou component; in particular this solves the problem of their existence, reported by Bedford and Smillie in 1991. These wandering Fatou components have non-empty real trace and their statistical behavior is historical with high emergence. The proof follows from a real geometrical model which enables us to show the existence of an open and dense set of $C^{r}$ families of surface diffeomorphisms in the Newhouse domain, each of which displaying a historical, high emergent, wandering domain at a dense set of parameters, for every $2\leq r\leq \infty$ and $r=\omega$. Hence, this also complements the recent work of Kiriki and Soma, by proving the last Taken's problem in the $C^{\infty }$ and $C^{\omega }$-case.
In a joint work with Sebastien Biebler, we show the existence of a locally dense set of real polynomial automorphisms of $\mathbb{C}^{2}$ displaying a stable wandering Fatou component; in particular this solves the problem of their existence, reported by Bedford and Smillie in 1991. These wandering Fatou components have non-empty real trace and their statistical behavior is historical with high emergence. The proof follows from a real g...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Averages of Zagier L-functions Balkanova, Olga (Auteur de la Conférence) | CIRM (Editeur )

In 1976, Zagier established a functional equation for the generalized Dirichlet L-functions that are part of the Fourier-Whittaker expansion of halfintegral weight Eisenstein series. The special values of these L-functions at 1/2 and at 1 are of particular interest because of the connection with the Selberg trace formula, with moments of symmetric square L-functions and with the prime geodesic theorem. In this talk, we describe various properties of Zagier L-functions and consider several problems related to the asymptotic evaluation of averages of special L-values.
In 1976, Zagier established a functional equation for the generalized Dirichlet L-functions that are part of the Fourier-Whittaker expansion of halfintegral weight Eisenstein series. The special values of these L-functions at 1/2 and at 1 are of particular interest because of the connection with the Selberg trace formula, with moments of symmetric square L-functions and with the prime geodesic theorem. In this talk, we describe various ...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Failure of the Brauer-Manin obstruction for a simply connected fourfold, and an orbifold version of the Mordell theorem Kebekus, Stefan (Auteur de la Conférence) | CIRM (Editeur )

Almost one decade ago, Poonen constructed the first examples of algebraic varieties over global fields for which Skorobogatov’s etale Brauer-Manin obstruction does not explain the failure of the Hasse principle. By now, several constructions are known, but they all share common geometric features such as large fundamental groups.
This talk discusses a construction of simply connected fourfolds over global fields of positive characteristic for which the Brauer-Manin machinery fails. Contrary to earlier work in this direction, our construction does not rely on major conjectures. Instead, we establish a new diophantine result of independent interest: a Mordell-type theorem for Campana’s "geometric orbifolds" over function fields of positive characteristic. Along the way, we also construct the first example of simply connected surface of general type over a global field with a non-empty, but non-Zariski dense set of rational points.
Joint work with Pereira and Smeets.
Almost one decade ago, Poonen constructed the first examples of algebraic varieties over global fields for which Skorobogatov’s etale Brauer-Manin obstruction does not explain the failure of the Hasse principle. By now, several constructions are known, but they all share common geometric features such as large fundamental groups.
This talk discusses a construction of simply connected fourfolds over global fields of positive characteristic for ...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Spectral decompositions and an extension of a theorem of Atzmon: a couple leading to spectral subspaces for Bishop operators Gallardo-Gutiérrez, Eva A. (Auteur de la Conférence) | CIRM (Editeur )

Bishop’s operator arose in the fifties as possible candidates for being counterexamples to the Invariant Subspace Problem. Several authors addressed the problem of finding invariant subspaces for some of these operators; but still the general problem is open. In this talk, we shall discuss about recent results on the existence of invariant subspaces which are indeed spectral subspaces for Bishop operators, by providing an extension of a Theorem of Atzmon (Joint work with M. Monsalve-Lopez).
Bishop’s operator arose in the fifties as possible candidates for being counterexamples to the Invariant Subspace Problem. Several authors addressed the problem of finding invariant subspaces for some of these operators; but still the general problem is open. In this talk, we shall discuss about recent results on the existence of invariant subspaces which are indeed spectral subspaces for Bishop operators, by providing an extension of a Theorem ...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Triality Elduque, Alberto (Auteur de la Conférence) | CIRM (Editeur )

Duality in projective geometry is a well-known phenomenon in any dimension. On the other hand, geometric triality deals with points and spaces of two different kinds in a sevendimensional projective space. It goes back to Study (1913) and Cartan (1925), and was soon realizedthat this phenomenon is tightly related to the algebra of octonions, and the order 3 outer automorphisms of the spin group in dimension 8.
Tits observed, in 1959, the existence of two different types of geometric triality. One of them is related to the octonions, but the other one is better explained in terms of a class of nonunital composition algebras discovered by the physicist Okubo (1978) inside 3x3-matrices, and which has led to the definition of the so called symmetric composition algebras.
This talk will review the history, classification, and their connections with the phenomenon of triality, of the symmetric composition algebras.
Duality in projective geometry is a well-known phenomenon in any dimension. On the other hand, geometric triality deals with points and spaces of two different kinds in a sevendimensional projective space. It goes back to Study (1913) and Cartan (1925), and was soon realizedthat this phenomenon is tightly related to the algebra of octonions, and the order 3 outer automorphisms of the spin group in dimension 8.
Tits observed, in 1959, the ...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  On ellipsephic integers Dartyge, Cécile (Auteur de la Conférence) | CIRM (Editeur )

The term " ellipsephic " was proposed by Christian Mauduit to denote the integers with missing digits in a given basis. This talk is a survey on several results on the multiplicative properties of these integers.

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  $L^{r}$-Helmholtz-Weyl decomposition in 3D exterior domains Kozono, Hideo (Auteur de la Conférence) | CIRM (Editeur )

It is known that in 3D exterior domains Ω with the compact smooth boundary $\partial \Omega$, two spaces $X^{r}_{har}\left ( \Omega \right )$ and $V^{r}_{har}\left ( \Omega \right )$ of $L^{r}$-harmonic vector fields $h$ with $h\cdot v\mid _{\partial \Omega }= 0$ and $h\times v\mid _{\partial \Omega }= 0$ are both of finite dimensions, where $v$ denotes the unit outward normal to $\partial \Omega$. We prove that for every $L^{r}$-vector field $u$, there exist $h\in X^{r}_{har}\left ( \Omega \right )$, $w\in H^{1,r}\left ( \Omega \right )^{3}$ with div $w= 0$ and $p\in H^{1,r}\left ( \Omega \right )$ such that $u$ is uniquely decomposed as $u= h$ + rot $w$ + $\bigtriangledown p$.
On the other hand, if for the given $L^{r}$-vector field $u$ we choose its harmonic part $h$ from $V^{r}_{har}\left ( \Omega \right )$, then we have a similar decomposition to above, while the unique expression of $u$ holds only for $1< r< 3$. Furthermore, the choice of $p$ in $H^{1,r}\left ( \Omega \right )$ is determined in accordance with the threshold $r= 3/2$.
Our result is based on the joint work with Matthias Hieber, Anton Seyferd (TU Darmstadt), Senjo Shimizu (Kyoto Univ.) and Taku Yanagisawa (Nara Women Univ.).
It is known that in 3D exterior domains Ω with the compact smooth boundary $\partial \Omega$, two spaces $X^{r}_{har}\left ( \Omega \right )$ and $V^{r}_{har}\left ( \Omega \right )$ of $L^{r}$-harmonic vector fields $h$ with $h\cdot v\mid _{\partial \Omega }= 0$ and $h\times v\mid _{\partial \Omega }= 0$ are both of finite dimensions, where $v$ denotes the unit outward normal to $\partial \Omega$. We prove that for every $L^{r}$-vector field $u... Déposez votre fichier ici pour le déplacer vers cet enregistrement. Post-edited The local density approximation in density functional theory Seiringer, Robert (Auteur de la Conférence) | CIRM (Editeur ) We present a mathematically rigorous justification of the Local Density Approximation in density functional theory. We provide a quantitative estimate on the difference between the grand-canonical Levy-Lieb energy of a given density (the lowest possible energy of all quantum states having this density) and the integral over the Uniform Electron Gas energy of this density. The error involves gradient terms and justifies the use of the Local Density Approximation in situations where the density is very flat on sufficiently large regions in space. (Joint work with Mathieu Lewin and Elliott Lieb) We present a mathematically rigorous justification of the Local Density Approximation in density functional theory. We provide a quantitative estimate on the difference between the grand-canonical Levy-Lieb energy of a given density (the lowest possible energy of all quantum states having this density) and the integral over the Uniform Electron Gas energy of this density. The error involves gradient terms and justifies the use of the ... Déposez votre fichier ici pour le déplacer vers cet enregistrement. Post-edited Rationality of Fano 3-folds over non-closed fields Kuznetsov, Alexander (Auteur de la Conférence) | CIRM (Editeur ) In the talk I will discuss rationality criteria for Fano 3-folds of geometric Picard number 1 over a non-closed field$k$of characteristic 0. Among these there are 8 types of geometrically rational varieties. We prove that in one of these cases any variety of this type is k-rational, in four cases the criterion of rationality is the existence of a$k$-rational point, and in the last three cases the criterion is the existence of a$k$-rational point and a k rational curve of genus 0 and degree 1, 2, and 3 respectively. The last result is based on recent results of Benoist-Wittenberg. This is a joint work with Yuri Prokhorov. In the talk I will discuss rationality criteria for Fano 3-folds of geometric Picard number 1 over a non-closed field$k$of characteristic 0. Among these there are 8 types of geometrically rational varieties. We prove that in one of these cases any variety of this type is k-rational, in four cases the criterion of rationality is the existence of a$k$-rational point, and in the last three cases the criterion is the existence of a$k$-rational ... Déposez votre fichier ici pour le déplacer vers cet enregistrement. Post-edited Chain conditions, unbounded colorings and the$C$-sequence spectrum Rinot, Assaf (Auteur de la Conférence) | CIRM (Editeur ) The productivity of the$\kappa $-chain condition, where$\kappa $is a regular, uncountable cardinal, has been the focus of a great deal of set-theoretic research. In the 1970’s, consistent examples of$kappa-cc$posets whose squares are not$\kappa-cc$were constructed by Laver, Galvin, Roitman and Fleissner. Later, ZFC examples were constructed by Todorcevic, Shelah, and others. The most difficult case, that in which$\kappa = \aleph{_2}$, was resolved by Shelah in 1997. In the first part of this talk, we shall present analogous results regarding the infinite productivity of chain conditions stronger than$\kappa-cc$. In particular, for any successor cardinal$\kappa$, we produce a ZFC example of a poset with precaliber$\kappa$whose$\omega ^{th}$power is not$\kappa-cc$. To do so, we introduce and study the principle$U(\kappa , \mu , \theta , \chi )$asserting the existence of a coloring$c:\left [ \kappa \right ]^{2}\rightarrow \theta $satisfying a strong unboundedness condition. In the second part of this talk, we shall introduce and study a new cardinal invariant$\chi \left ( \kappa \right )$for a regular uncountable cardinal$\kappa$. For inaccessible$\kappa$,$\chi \left ( \kappa \right )$may be seen as a measure of how far away$\kappa$is from being weakly compact. We shall prove that if$\chi \left ( \kappa \right )> 1$, then$\chi \left ( \kappa \right )=max(Cspec(\kappa ))$, where: (1) Cspec$(\kappa)$:= {$\chi (\vec{C})\mid \vec{C}$is a sequence over$\kappa$}$\setminus \omega$, and (2)$\chi \left ( \vec{C} \right )$is the least cardinal$\chi \leq \kappa $such that there exist$\Delta\in\left [ \kappa \right ]^{\kappa }$and b :$\kappa \rightarrow \left [ \kappa \right ]^{\chi }$with$\Delta \cap \alpha \subseteq \cup _{\beta \in b(\alpha )}C_{\beta }$for every$\alpha < \kappa$. We shall also prove that if$\chi (\kappa )=1$, then$\kappa$is greatly Mahlo, prove the consistency (modulo the existence of a supercompact) of$\chi (\aleph_{\omega +1})=\aleph_{0}$, and carry a systematic study of the effect of square principles on the$C$-sequence spectrum. In the last part of this talk, we shall unveil an unexpected connection between the two principles discussed in the previous parts, proving that, for infinite regular cardinals$\theta< \kappa ,\theta \in Cspec(\kappa )$if there is a closed witness to$U_{(\kappa ,\kappa ,\theta ,\theta )}$. This is joint work with Chris Lambie-Hanson. The productivity of the$\kappa $-chain condition, where$\kappa $is a regular, uncountable cardinal, has been the focus of a great deal of set-theoretic research. In the 1970’s, consistent examples of$kappa-cc$posets whose squares are not$\kappa-cc$were constructed by Laver, Galvin, Roitman and Fleissner. Later, ZFC examples were constructed by Todorcevic, Shelah, and others. The most difficult case, that in which$\kappa = \aleph{_2}\$, ...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Challenges in achieving scalable and robust linear solvers Grigori, Laura (Auteur de la Conférence) | CIRM (Editeur )

This talk focuses on challenges that we address when designing linear solvers that aim at achieving scalability on large scale computers, while also preserving numerical robustness. We will consider preconditioned Krylov subspace solvers. Getting scalability relies on reducing global synchronizations between processors, while also increasing the arithmetic intensity on one processor. Achieving robustness relies on ensuring that the condition number of the preconditioned matrix is bounded. We will discuss two different approaches for this. The first approach relies on enlarged Krylov subspace methods that aim at computing an enlarged subspace and obtain a faster convergence of the iterative method. The second approach relies on a multilevel Schwarz preconditioner, a multilevel extension of the GenEO preconditioner, that is basedon constructing robustly a hierarchy of coarse spaces. Numerical results on large scale computers, in particular for linear systems arising from solving linear elasticity problems, will discuss the efficiency of the proposed methods.
This talk focuses on challenges that we address when designing linear solvers that aim at achieving scalability on large scale computers, while also preserving numerical robustness. We will consider preconditioned Krylov subspace solvers. Getting scalability relies on reducing global synchronizations between processors, while also increasing the arithmetic intensity on one processor. Achieving robustness relies on ensuring that the condition ...

Déposez votre fichier ici pour le déplacer vers cet enregistrement.

Post-edited  Twistor theory for LQG Eastwood, Michael (Auteur de la Conférence) | CIRM (Editeur )

Twistor Theory was proposed in the late 1960s by Roger Penrose as a potential geometric unification of general relativity and quantum mechanics. During the past 50 years, there have been many mathematical advances and achievements in twistor theory. In physics, however, there are aspirations yet to be realised. Twistor Theory and Loop Quantum Gravity (LQG) share a common background. Their aims are very much related. Is there more to it? This talk will sketch the geometry and symmetry behind twistor theory with the hope that links with LQG can be usefully strengthened. We believe there is something significant going on here: what could it be?
Twistor Theory was proposed in the late 1960s by Roger Penrose as a potential geometric unification of general relativity and quantum mechanics. During the past 50 years, there have been many mathematical advances and achievements in twistor theory. In physics, however, there are aspirations yet to be realised. Twistor Theory and Loop Quantum Gravity (LQG) share a common background. Their aims are very much related. Is there more to it? This ...

Z