Condensed Matter/Statistical Mechanics Seminars
Speaker: Danilo Liarte (Cornell Univ.)
Time: Monday, August 30 at 12:00 BRT
Video (Link Here)
Title: Geometry, topology, and the complex emergent behavior of disordered materials
Abstract: Geometry and topology play a central role in physics. Their interplay and synergy lead to some of the most intriguing phenomena in condensed matter. I will discuss deep connections between the differential geometry of surfaces and the iconic defects of layered liquid crystals, and present an elegant description of their microstructure that involves Lorentz invariance and precise arrangements of conic sections. I will then explore the topology of some classes of periodic lattices to craft a model for the universal properties of disordered solids near a rigidity transition, with interest in many systems ranging from molecular glasses and granular media to biological tissues and even machine learning. Finally, I will show how we can use and extend our models of disordered elastic networks to describe a variety of phenomena appearing in colloidal suspensions, metamaterials and strange quantum liquids.
Speaker: Danilo Liarte (Cornell Univ.)
Time: Tuesday, August 31 at 15:00 BRT
Video (Link Here)
Title: Effective-medium theory and the universal behavior of disordered elastic systems
Abstract: Effective-medium theory has become one of the most powerful theoretical tools to describe the universal critical behavior of disordered elastic systems near the onset of a rigidity transition. I will discuss the approximations involved in this formalism and apply it to suitably crafted network models that exhibit “multicritical” transitions to a disordered rigid phase. I will then extract well-tested critical exponents as well as explicit formulas for the universal scaling functions governing the behavior of a large class of viscoelastic materials near the onset of rigidity.
Speaker: Jamir Marino (Univ. of Mainz)
Time: Monday, September 6 at 10:00 BRT
Video (Link Here)
Title: Spatio-temporal control of correlations with non-local dissipation
Abstract: Controlling the spread of correlations in quantum many-body systems is a key challenge at the heart of quantum science and technology. Correlations are usually destroyed by dissipation arising from coupling between a system and its environment. Here, we show that dissipation can instead be used to engineer a wide variety of spatio-temporal correlation profiles in an easily tunable manner. We describe how dissipation with any translationally-invariant spatial profile can be realized in cold atoms trapped in an optical cavity. A uniform external field and the choice of spatial profile can be used to design when and how dissipation creates or destroys correlations. We demonstrate this control by preferentially generating entanglement at a desired wavevector. We thus establish non-local dissipation as a new route towards engineering the far-from-equilibrium dynamics of quantum information, with potential applications in quantum metrology, state preparation, and transport.
Speaker: Jamir Marino (Univ. of Mainz)
Time: Tuesday, September 7 at 12:00 BRT
Video (Link Here)
Title: Criticality in driven open systems
Abstract: I will give a crash course at the level of a graduate school on non-equilibrium critical points in driven dissipative systems. With a combination of scaling arguments and elements of Keldysh field theory, I will discuss major obstructions against the existence of universality classes beyond the Halperin-Hohenberg classification, and show how these can be overcome using noise engineering. As an example, I will discuss the formation of a novel non-equilibrium fixed point in a Ising model subject to losses with power law decaying spatial profile, inspired from my first talk.
Speaker: Tiago Mendes Santos (Max Planck Institute at Dresden)
Time: Monday, September 13 at 11:00 BRT
Video (Link Here)
Title: Machine learning many-body physics
Abstract: Understanding the collective behavior of strongly correlated systems is at the center stage of dierent fields of modern physics, ranging from fundamental aspects of statistical mechanics and condensed matter to more practical aspects related to quantum simulators and quantum computing. Examples of key challenges of such fields include the development of approaches that can deal with the exponential complexity encountered in many-body systems and the characterization of novel quantum phases of matter and their transitions. This talk discusses some recent efforts in addressing such problems through the lens of machine-learning (ML) concepts and techniques. In particular, we present some applications of ML ideas to study critical phenomena and the dynamics of quantum many-body systems.
First, we consider the problem of extracting physical properties from minimally processed data sets that are generated by Monte Carlo simulations of lattice models or experiments with quantum simulators. In particular, we consider how the minimum number of variables needed to accurately describe the essential features of a data set – which is known as the intrinsic dimension (ID) – behaves in the vicinity of dierent types of phase transitions. We show that the ID allows characterizing (classical and quantum) critical points and reveals universal properties as the critical exponent ν (associated with the correlation length’s divergence). In the second part of the talk, we discuss an approach for solving many-body problems based on the ecient compression of quantum states with Artificial Neural Network (ANN), termed Neural Quantum State (NQS). In particular, we present generalizations of the NQS framework to compute the dynamical and spectral properties of quantum many-body models relevant to experiments with quantum simulators.
Speaker: Tiago Mendes Santos (Max Planck Institute at Dresden)
Time: Tuesday, September 14 at 11:00 BRT
Video (Link Here)
Title: Quantum entanglement and unsupervised learning in many-body physics
Abstract: Finding proper observables to characterize quantum phases of matter and critical points remains a key challenge in many-body physics. Over the last twenty years, entanglement has emerged as a fundamental tool for this task. For example, entanglement measures provide access to universal quantities related to critical points and phases of matter. Similarly, new approaches developed in machine-learning (ML) concepts have emerged as a new potential tool to identify universal properties of many-body systems from minimally processed physical data sets. This talk aims to discuss aspects related to entanglement and unsupervised ML in such context. Despite the central role of entanglement as a diagnostic tool for low-energy properties of many-body Hamiltonians, its measurement has so far been elusive both from an experimental and beyond one dimension numerical point of view. In the first part of this talk, we discuss an approach to access ground-state entanglement measures of spin models. The method is based on the thermodynamic study of lattice entanglement Hamiltonian of ground states obtained via eld theoretical insights, and it is implemented with quantum Monte Carlo simulations. In the second part of this talk, we discuss how certain features of minimally processed physical data sets can be used to reveal universal properties related to classical and quantum critical points. In particular, we consider the behavior of the the minimum number of variables needed to accurately describe the important features of a data set – the intrinsic dimension (ID) – in the vicinity of different phase transitions. We show that the ID uniquely characterizes the critical regime of different types of phase transitions. Our work reveals how raw data sets display unique signatures of universal behavior and suggest direct parallelism between conventional order parameters in real space and the ID in the data space.
Speaker: Victor Quito (Iowa State Univ.)
Time: Thursday, September 16 at 13:00 BRT
Video (Link Here)
Title: Floquet-tuning correlated systems
Abstract: Floquet-tuning consists of coupling quantum systems to time-periodic potentials. It provides a versatile platform for inducing fascinating quantum phases, some of them rare or even impossible to find in equilibrium. In this talk, I will present the basic features of Floquet theory and highlight the diverse phenomena it can induce, from time-crystals to spin liquids. I will also show how to accommodate unpolarized quasi-monochromatic light in Floquet theory, allowing for tuning without breaking any underlying symmetry. Then, I will describe how coupling to unpolarized light provides a flexible way of driving strongly correlated phases, focusing on frustrated magnetism and heavy-fermion physics.
Speaker: Victor Quito (Iowa State Univ.)
Time: Tuesday, September 21 at 11:00 BRT
Video (Link Here)
Title: Emergent symmetries in one-dimensional strongly disordered interacting systems
Abstract: The almost century-old concept of symmetry breaking has been immensely fruitful for understanding various phases in condensed matter systems and high-energy physics. A different, and in a sense, complementary question that has been the focus of increasing attention is that of emergent symmetries: phases that, at low energies, show higher symmetry than in their microscopic description. The general mechanism for symmetry enhancement is not known, and examples remain few and far between. In this talk, I will introduce a generic route in which strongly disordered interacting chains present phases with emergent symmetries. I will show that chains with explicit SO(N) symmetry accommodate two different phases with emergent SU(N) symmetries, one with the ground state formed of singlets of pairs of SO(N) spins (a ‘mesonic’ phase) and another with singlets made of multiples of N SO(N) spins (a ‘baryonic’ phase).