arXiv subjects

Luis Santos

Publications and source records attributed to Luis Santos.

At least 55 records · Page 3Linked to original sources

arXiv2021

PettingZoo: Gym for Multi-Agent Reinforcement Learning

This paper introduces the PettingZoo library and the accompanying Agent Environment Cycle ("AEC") games model. PettingZoo is a library of diverse sets of multi-agent environments with a universal, elegant Python API. PettingZoo was developed with the goal of accelerating research in Multi-Agent Reinforcement Learning ("MARL"), by making work more interchangeable, accessible and reproducible akin to what OpenAI's Gym library did for single-agent reinforcement learning. PettingZoo's API, while inheriting many features of Gym, is unique amongst MARL APIs in that it's based around the novel AEC games model. We argue, in part through case studies on major problems in popular MARL environments, that the popular game models are poor conceptual models of games commonly used in MARL and accordingly can promote confusing bugs that are hard to detect, and that the AEC games model addresses these problems.

cs.LG↗

arXiv2021

May I Take Your Order? On the Interplay Between Time and Order in Process Mining

Process mining starts from event data. The ordering of events is vital for the discovery of process models. However, the timestamps of events may be unreliable or imprecise. To further complicate matters, also causally unrelated events may be ordered in time. The fact that one event is followed by another does not imply that the former causes the latter. This paper explores the relationship between time and order. Moreover, it describes an approach to preprocess event data having timestamp-related problems. This approach avoids using accidental or unreliable orders and timestamps, creates partial orders to capture uncertainty, and allows for exploiting domain knowledge to (re)order events. Optionally, the approach also generates interleavings to be able to use existing process mining techniques that cannot handle partially ordered event data. The approach has been implemented using ProM and can be applied to any event log.

cs.DB↗

arXiv2021

Agent Environment Cycle Games

Partially Observable Stochastic Games (POSGs) are the most general and common model of games used in Multi-Agent Reinforcement Learning (MARL). We argue that the POSG model is conceptually ill suited to software MARL environments, and offer case studies from the literature where this mismatch has led to severely unexpected behavior. In response to this, we introduce the Agent Environment Cycle Games (AEC Games) model, which is more representative of software implementation. We then prove it's as an equivalent model to POSGs. The AEC games model is also uniquely useful in that it can elegantly represent both all forms of MARL environments, whereas for example POSGs cannot elegantly represent strictly turn based games like chess.

cs.LG↗

arXiv2021

Two-dimensional supersolidity in a dipolar quantum gas

Supersolidity -- a quantum-mechanical phenomenon characterized by the presence of both superfluidity and crystalline order -- was initially envisioned in the context of bulk solid helium, as a possible answer to the question of whether a solid could have superfluid properties. While supersolidity has not been observed in solid helium (despite much effort), ultracold atomic gases have provided a fundamentally new approach, recently enabling the observation and study of supersolids with dipolar atoms. However, unlike the proposed phenomena in helium, these gaseous systems have so far only shown supersolidity along a single direction. By crossing a structural phase transition similar to those occurring in ionic chains, quantum wires, and theoretically in chains of individual dipolar particles, we demonstrate the extension of supersolid properties into two dimensions, providing an important step closer to the bulk situation envisioned in helium. This opens the possibility of studying rich excitation properties, including vortex formation, as well as ground-state phases with varied geometrical structure in a highly flexible and controllable system.

cond-mat.quant-gas↗

arXiv2021

RadVR: A 6DOF Virtual Reality Daylighting Analysis Tool

This work introduces RadVR, a virtual reality tool for daylighting analysis that simultaneously combines qualitative assessments through immersive real-time renderings with quantitative physically correct daylighting simulations in a 6DOF virtual environment. By taking a 3D building model with material properties as input, RadVR allows users to (1) perform physically-based daylighting simulations via Radiance, (2) study sunlight in different hours-of-the-year, (3) interact with a 9-point-in-time matrix for the most representative times of the year, and (4) visualize, compare, and analyze daylighting simulation results. With an end-to-end workflow, RadVR integrates with 3D modeling software that is commonly used by building designers. Additionally, by conducting user experiments we compare the proposed system with DIVA for Rhino, a Radiance-based tool that uses conventional 2D-displays. The results show that RadVR can provide promising assistance in spatial understanding tasks, navigation, and sun position analysis in virtual reality.

cs.GR↗

arXiv2021

Multiplayer Support for the Arcade Learning Environment

The Arcade Learning Environment ("ALE") is a widely used library in the reinforcement learning community that allows easy programmatic interfacing with Atari 2600 games, via the Stella emulator. We introduce a publicly available extension to the ALE that extends its support to multiplayer games and game modes. This interface is additionally integrated with PettingZoo to allow for a simple Gym-like interface in Python to interact with these games. We additionally introduce experimental baselines for all environments included.

cs.LG↗

arXiv2020

Cluster dynamics in two-dimensional lattice gases with inter-site interactions

Sufficiently strong inter-site interactions in extended-Hubbard and XXZ spin models result in dynamically-bound clusters at neighboring sites. We show that the dynamics of these clusters in two-dimensional lattices is remarkably different and richer than that of repulsively-bound on-site clusters in gases without inter-site interactions. Whereas on-site pairs move in the same lattice as individual particles, nearest-neighbor dimers perform an interacting quantum walk in a different lattice geometry, leading to a peculiar dynamics characterized by more than one time scale. The latter is general for any lattice geometry, but it is especially relevant in triangular and diamond lattices, where dimers move resonantly in an effective kagome and Lieb lattice, respectively. As a result, dimers experience partial quasi-localization due to an effective flat band, and may move slower than longer clusters. This surprising link between anomalously slow quantum walk dynamics in these models and flat-band physics may be readily observed in experiments with lanthanide atoms.

cond-mat.quant-gas↗

arXiv2020

Excited-state quantum phase transitions in spinor Bose-Einstein condensates

Excited-state quantum phase transitions (ESQPTs) extend the notion of quantum phase transitions beyond the ground state. They are characterized by closing energy gaps amid the spectrum. Identifying order parameters for ESQPTs poses however a major challenge. We introduce spinor Bose-Einstein condensates as a versatile platform for studies of ESQPTs. Based on the mean-field dynamics, we define a topological order parameter that distinguishes between excited-state phases, and discuss how to interferometrically access the order parameter in current experiments. Our work opens the way for the experimental characterization of excited-state quantum phases in atomic many-body systems.

cond-mat.quant-gas↗

arXiv2020

Topological inheritance in half-SSH Hubbard models

The interplay between interparticle interactions and topological features may result in unusual phenomena. Interestingly, interactions may induce topological features in an originally trivial system, as we illustrate for the case of a one-dimensional two-component Hubbard model in which one component is subjected to Su-Schrieffer-Heeger(SSH) dimerization, whereas the other one is not. We show that due to inter-component interactions the topological properties of one component are induced in the originally trivial one. Although for large interactions topological inheritance may be readily explained by on-site pairing, we show that the threshold for full inheritance occurs at weak interactions, for which the components are not yet paired. We illustrate this inheritance by discussing both bulk and edge properties, as well as dynamical observables as mean chiral displacement and charge pumping.

cond-mat.quant-gas↗

arXiv2020

Universal algebraic growth of entanglement entropy in many-body localized systems with power-law interactions

Power-law interactions play a key role in a large variety of physical systems. In the presence of disorder, these systems may undergo many-body localization for a sufficiently large disorder. Within the many-body localized phase the system presents in time an algebraic growth of entanglement entropy, $S_{vN}(t)\propto t^γ$. Whereas the critical disorder for many-body localization depends on the system parameters, we find by extensive numerical calculations that the exponent $γ$ acquires a universal value $γ_c\simeq 0.33$ at the many-body localization transition, for different lattice models and decay powers. Moreover, our results suggest an intriguing relation between $γ_c$ and the critical minimal decay power of interactions necessary for the observation of many-body localization.

cond-mat.dis-nn↗

arXiv2020

Self-bound Doubly-Dipolar Bose-Einstein condensates

We analyze the physics of self-bound droplets in a doubly dipolar Bose-Einstein condensate (DDBEC) composed by particles with both electric and magnetic dipole moments. Using the particularly relevant case of dysprosium, we show that the anisotropy of the doubly-dipolar interaction potential is highly versatile and nontrivial, depending critically on the relative orientation and strength between the two dipole moments. This opens novel possibilities for exploring intriguing quantum many-body physics. Interestingly, by varying the angle between the two dipoles we find a dimensional crossover from quasi one-dimensional to quasi two-dimensional self-bound droplets. This opens a so far unique scenario in condensate physics, in which a dimensional crossover is solely driven by interactions in the absence of any confinement.

cond-mat.quant-gas↗

arXiv2019

Deconfining disordered phase in two-dimensional quantum link models

We explore the ground-state physics of two-dimensional spin-$1/2$ $U(1)$ quantum link models, one of the simplest non-trivial lattice gauge theories with fermionic matter within experimental reach for quantum simulations. Whereas in the large mass limit we observe Neél-like vortex-antivortex and striped crystalline phases, for small masses there is a transition from the striped phases into a disordered phase whose properties resemble those at the Rokhsar-Kivelson point of the quantum dimer model. This phase is characterized on ladders by boundary Haldane-like properties, such as vanishing parity and finite string ordering. Moreover, from studies of the string tension between gauge charges, we find that whereas the stripe phases are confined, the novel disordered phase present clear indications of being deconfined. Our results open exciting perspectives of studying highly non-trivial physics in quantum simulators, such as spin-liquid behavior and confinement-deconfinement transitions, without the need of explicitly engineering plaquette terms.

cond-mat.quant-gas↗

arXiv2019

Quantum Zeno-based Detection and State Engineering of Ultracold Polar Molecules

We present and analyze a toolbox for the controlled manipulation of ultracold polar molecules, consisting of detection of molecules, atom-molecule entanglement, and engineering of dissipative dynamics. Our setup is based on fast chemical reactions between molecules and atoms leading to a quantum Zeno-based collisional blockade in the system. We demonstrate that the experimental parameters for achieving high fidelities can be found using a straightforward numerical optimization. We exemplify our approach for a system comprised of NaK molecules and Na atoms and we discuss the consequences of residual imperfections such as a finite strength of the quantum Zeno blockade.

cond-mat.quant-gas↗

arXiv2019

Observation of Roton Mode Population in a Dipolar Quantum Gas

The concept of a roton, a special kind of elementary excitation, forming a minimum of energy at finite momentum, has been essential to understand the properties of superfluid $^4$He. In quantum liquids, rotons arise from the strong interparticle interactions, whose microscopic description remains debated. In the realm of highly-controllable quantum gases, a roton mode has been predicted to emerge due to magnetic dipole-dipole interactions despite of their weakly-interacting character. This prospect has raised considerable interest; yet roton modes in dipolar quantum gases have remained elusive to observations. Here we report experimental and theoretical studies of the momentum distribution in Bose-Einstein condensates of highly-magnetic erbium atoms, revealing the existence of the long-sought roton mode. By quenching the interactions, we observe the roton appearance of peaks at well-defined momentum. The roton momentum follows the predicted geometrical scaling with the inverse of the confinement length along the magnetisation axis. From the growth of the roton population, we probe the roton softening of the excitation spectrum in time and extract the corresponding imaginary roton gap. Our results provide a further step in the quest towards supersolidity in dipolar quantum gases.

cond-mat.quant-gas↗

arXiv2019

Observation of a dipolar quantum gas with metastable supersolid properties

The competition of dipole-dipole and contact interactions leads to exciting new physics in dipolar gases, well-illustrated by the recent observation of quantum droplets and rotons in dipolar condensates. We show that the combination of the roton instability and quantum stabilization leads under proper conditions to a novel regime that presents supersolid properties, due to the coexistence of stripe modulation and phase coherence. In a combined experimental and theoretical analysis, we determine the parameter regime for the formation of coherent stripes, whose lifetime of a few tens of milliseconds is limited by the eventual destruction of the stripe pattern due to three-body losses. Our results open intriguing prospects for the development of long-lived dipolar supersolids.

cond-mat.quant-gas↗

arXiv2018

Dimensional crossover for the beyond-mean-field correction in Bose gases

We present a detailed beyond-mean-field analysis of a weakly interacting Bose gas in the crossover from three to low dimensions. We find an analytical solution for the energy and provide a clear qualitative picture of the crossover in the case of a box potential with periodic boundary conditions. We show that the leading contribution of the confinement-induced resonance is of beyond-mean-field order and calculate the leading corrections in the three- and low-dimensional limits. We also characterize the crossover for harmonic potentials in a model system with particularly chosen short- and long-range interactions and show the limitations of the local-density approximation. Our analysis is applicable to Bose-Bose mixtures and gives a starting point for developing the beyond-mean-field theory in inhomogeneous systems with long-range interactions such as dipolar particles or Rydberg-dressed atoms.

cond-mat.quant-gas↗

arXiv2018

Characterizing Deep-Learning I/O Workloads in TensorFlow

The performance of Deep-Learning (DL) computing frameworks rely on the performance of data ingestion and checkpointing. In fact, during the training, a considerable high number of relatively small files are first loaded and pre-processed on CPUs and then moved to accelerator for computation. In addition, checkpointing and restart operations are carried out to allow DL computing frameworks to restart quickly from a checkpoint. Because of this, I/O affects the performance of DL applications. In this work, we characterize the I/O performance and scaling of TensorFlow, an open-source programming framework developed by Google and specifically designed for solving DL problems. To measure TensorFlow I/O performance, we first design a micro-benchmark to measure TensorFlow reads, and then use a TensorFlow mini-application based on AlexNet to measure the performance cost of I/O and checkpointing in TensorFlow. To improve the checkpointing performance, we design and implement a burst buffer. We find that increasing the number of threads increases TensorFlow bandwidth by a maximum of 2.3x and 7.8x on our benchmark environments. The use of the tensorFlow prefetcher results in a complete overlap of computation on accelerator and input pipeline on CPU eliminating the effective cost of I/O on the overall performance. The use of a burst buffer to checkpoint to a fast small capacity storage and copy asynchronously the checkpoints to a slower large capacity storage resulted in a performance improvement of 2.6x with respect to checkpointing directly to slower storage on our benchmark environment.

cs.DC↗

arXiv2018

Probing the exchange statistics of one-dimensional anyon models

We propose feasible scenarios for revealing the modified exchange statistics in one-dimensional anyon models in optical lattices based on an extension of the multicolor lattice-depth modulation scheme introduced in [{Phys. Rev. A 94, 023615 (2016)}]. We show that the fast modulation of a two-component fermionic lattice gas in the presence a magnetic field gradient, in combination with additional resonant microwave fields, allows for the quantum simulation of hardcore anyon models with periodic boundary conditions. Such a semi-synthetic ring set-up allows for realizing an interferometric arrangement sensitive to the anyonic statistics. Moreover, we show as well that simple expansion experiments may reveal the formation of anomalously bound pairs resulting from the anyonic exchange.

cond-mat.quant-gas↗