Quasi-one-dimensional quantum models are ideal for theoretically exploring the physical phenomena associated with strong correlations. In this thesis we study three examples where strong correlations play an important role in the static or dynamic properties of the system. Firstly, we examine the behaviour of a doped fermionic two-leg ladder in which umklapp interactions are present. Such interactions arise at special band fillings and can be induced by the formation of charge density wave order in an array of two-leg ladders with long-range (three-dimensional) interactions. For the umklapp which arises from the half-filling of one of the bands, we show that the low-energy theory has a number of phases, including a strong coupling regime in which the dominant fluctuations are superconducting in nature. These superconducting fluctuations carry a finite wave vector – they are the one-dimensional analogue of Fulde-Ferrell-Larkin-Ovchinnikov superconductivity. In a second example, we consider a quantum spin model which captures the essential one-dimensional physics of CoNb₂O₆, a quasi-one-dimensional Ising ferromagnet. Motivated by high-resolution inelastic neutron scattering experiments, we calculate the dynamical structure in the paramagnetic phase and show that a small misalignment of the transverse field can lead to quasi-particle breakdown – a surprising broadening in the single particle mode observed in experiment. Finally, we study the out-of-equilibrium dynamics of a model with tuneable integrability breaking. When integrability is broken by the presence of weak interactions, we show that the system relaxes to a non-thermal state on intermediate time scales, the so-called “prethermalization plateau”. We describe the approximately stationary behaviour in this regime by constructing a generalised Gibbs ensemble with charges deformed to leading order in perturbation theory. Expectation values of these charges are time-independent, but interestingly the charges do not commute with the Hamiltonian to leading order in perturbation theory. Increasing the strength of the integrability breaking interactions leads to behaviour compatible with thermalisation. In each case we use a combination of perturbative analytical calculations and non-perturbative numerical computations to study the problem at hand.

One of the most important results emerging from string theory is the gauge gravity duality (AdS/CFT correspondence) which tells us that certain problems in particular gravitational backgrounds can be exactly mapped to a particular dual gauge theory a quantum theory very similar to the one explaining the interactions between fundamental subatomic particles. The chief merit of the duality is that a difficult problem in one theory can be mapped to a simpler and solvable problem in the other theory. The duality can be used both ways. Most of the current theoretical framework is suited to study equilibrium systems, or systems where time dependence is at most adiabatic. However in the real world, systems are almost always out of equilibrium. Generically these scenarios are described by quenches, where a parameter of the theory is made time dependent. In this dissertation I describe some of the work done in the context of studying quantum quench using the AdS/CFT correspondence. We recover certain universal scaling type of behavior as the quenching is done through a quantum critical point. Another question that has been explored in the dissertation is time dependence of the gravity theory. Present cosmological observations indicate that our universe is accelerating and is described by a spacetime called de-Sitter(dS). In 2011 there had been a speculation over a possible duality between de-Sitter gravity and a particular field theory (Euclidean SP(N) CFT). However a concrete realization of this proposition was still lacking. Here we explicitly derive the dS/CFT duality using well known methods in field theory. We discovered that the time dimension emerges naturally in the derivation. We also describe further applications and extensions of dS/CFT.

Cette thèse porte sur l'étude de systèmes quantiques à N-corps à température nulle, où le comportement du système n'est alors soumis qu'aux effets quantiques. Je vais présenter ici une approche variationnelle développée avec Tommaso Roscilde, mon directeur de thèse, et Fabio Mezzacapo, mon co-encadrant de thèse, pour étudier ces systèmes.Cette approche se base sur une parametrisation de l’état quantique (dit Ansatz) à laquelle on applique une procédure d’optimisation variationnelle lui permettant de reproduire l'évolution d'un système soumis à l'équation de Schrödinger, tout en limitant le nombre de variables considérées. En considérant une évolution en temps imaginaire, il est possible d'étudier l'état fondamental d'un système. Je me suis ainsi intéressé à un modèle de chaîne XX de spins 1/2, dont les corrélations à longue portée rendent l'étude difficile, et adapté ainsi notre approche pour reproduire au mieux les corrélations et l'intrication du système. Je me suis ensuite intéressé au modèle J1-J2 dont la structure de signe non positive des coefficients de l’état quantique pose un défi important pour les approches Monte Carlo; et dans laquelle la frustration magnétique induit une transition de phase quantique (d’un état aux corrélations à longue porté vers un état non magnétique avec formation d’un cristal de lien de valence). Je me suis enfin intéressé à l'évolution temporelle d'un système à N-corps à partir d'un état non stationnaire. J'ai pu étudier la capacité de notre approche à reproduire la croissance linéaire de l’intrication dans le temps, ce qui est un obstacle fondamental pour les approches alternatives telles que le groupe de renormalisation de la matrice densité. / This thesis presents a study of quantum many-body systems at zero temperature, where the behavior of the system is purely driven by the quantum effects. I will introduce a variationnal approach developped with Tommaso Roscilde, my PhD supervisor, and Fabio Mezzacapo, my co-supervisor, in order to study these systems.This approach is based on a parametrisation of the quantum state (named Ansatz) on which we apply a variational optimisation, allowing us reproduce the system's evolution under Schrödinger's equation with a limited number of variables.By considering an imaginary-time evolution, it is possible to reconstruct the system's ground state. I focused on S=1/2 XX spin chain, where the long-range quantum correlations complicate a variational study; and I have specifically targeted our Ansatz in order to reproduce the correlations and the entanglement of the ground state. Moreover I considered the antiferromagnetic S=1/2 J1-J2 spin chain, where the non-trivial sign structure of the coefficients of the quantum state introduces an important challenge for the quantum Monte Carlo approach; and where the magnetic frustration induces a quantum phase transition (from a state with long range correlations to a non-magnetic state in the form of a valence-bond crystal).Finally I focused on the time evolution of a quantum many-body system starting from a non-stationary state. I studied the ability of our approach to reproduce the linear increase of the entanglement during time, which is a fondamental obstacle for other approaches such as the density-matrix renormalization group.