-
Meta-Gradient Search Control: A Method for Improving the Efficiency of Dyna-style Planning
Authors:
Bradley Burega,
John D. Martin,
Luke Kapeluck,
Michael Bowling
Abstract:
We study how a Reinforcement Learning (RL) system can remain sample-efficient when learning from an imperfect model of the environment. This is particularly challenging when the learning system is resource-constrained and in continual settings, where the environment dynamics change. To address these challenges, our paper introduces an online, meta-gradient algorithm that tunes a probability with w…
▽ More
We study how a Reinforcement Learning (RL) system can remain sample-efficient when learning from an imperfect model of the environment. This is particularly challenging when the learning system is resource-constrained and in continual settings, where the environment dynamics change. To address these challenges, our paper introduces an online, meta-gradient algorithm that tunes a probability with which states are queried during Dyna-style planning. Our study compares the aggregate, empirical performance of this meta-gradient method to baselines that employ conventional sampling strategies. Results indicate that our method improves efficiency of the planning process, which, as a consequence, improves the sample-efficiency of the overall learning process. On the whole, we observe that our meta-learned solutions avoid several pathologies of conventional planning approaches, such as sampling inaccurate transitions and those that stall credit assignment. We believe these findings could prove useful, in future work, for designing model-based RL systems at scale.
△ Less
Submitted 27 June, 2024;
originally announced June 2024.
-
MOTO: Offline Pre-training to Online Fine-tuning for Model-based Robot Learning
Authors:
Rafael Rafailov,
Kyle Hatch,
Victor Kolev,
John D. Martin,
Mariano Phielipp,
Chelsea Finn
Abstract:
We study the problem of offline pre-training and online fine-tuning for reinforcement learning from high-dimensional observations in the context of realistic robot tasks. Recent offline model-free approaches successfully use online fine-tuning to either improve the performance of the agent over the data collection policy or adapt to novel tasks. At the same time, model-based RL algorithms have ach…
▽ More
We study the problem of offline pre-training and online fine-tuning for reinforcement learning from high-dimensional observations in the context of realistic robot tasks. Recent offline model-free approaches successfully use online fine-tuning to either improve the performance of the agent over the data collection policy or adapt to novel tasks. At the same time, model-based RL algorithms have achieved significant progress in sample efficiency and the complexity of the tasks they can solve, yet remain under-utilized in the fine-tuning setting. In this work, we argue that existing model-based offline RL methods are not suitable for offline-to-online fine-tuning in high-dimensional domains due to issues with distribution shifts, off-dynamics data, and non-stationary rewards. We propose an on-policy model-based method that can efficiently reuse prior data through model-based value expansion and policy regularization, while preventing model exploitation by controlling epistemic uncertainty. We find that our approach successfully solves tasks from the MetaWorld benchmark, as well as the Franka Kitchen robot manipulation environment completely from images. To the best of our knowledge, MOTO is the first method to solve this environment from pixels.
△ Less
Submitted 6 January, 2024;
originally announced January 2024.
-
Denoising total scattering data using Compressed Sensing
Authors:
James Weng,
Niklas B. Thompson,
Christopher Folmar,
James D. Martin,
Christina Hoffman
Abstract:
To obtain the best resolution for any measurement there is an ever-present challenge to achieve maximal differentiation between signal and noise over as fine of sampling dimensions as possible. In diffraction science these issues are particularly pervasive when analyzing small crystals, systems with diffuse scattering, or other systems in which the signal of interest is extremely weak and incident…
▽ More
To obtain the best resolution for any measurement there is an ever-present challenge to achieve maximal differentiation between signal and noise over as fine of sampling dimensions as possible. In diffraction science these issues are particularly pervasive when analyzing small crystals, systems with diffuse scattering, or other systems in which the signal of interest is extremely weak and incident flux and instrument time is limited. We here demonstrate that the tool of compressed sensing, which has successfully been applied to photography, facial recognition, and medical imaging, can be effectively applied to diffraction images to dramatically improve the signal-to-noise ratio (SNR) in a data-driven fashion without the need for additional measurements or modification of existing hardware. We outline a technique that leverages compressive sensing to bootstrap a single diffraction measurement into an effectively arbitrary number of virtual measurements, thereby providing a means of super-resolution imaging.
△ Less
Submitted 19 October, 2023; v1 submitted 18 October, 2023;
originally announced October 2023.
-
Thermal physics in the data age -- students judge the applicability of the equipartition theorem
Authors:
J. D. D. Martin
Abstract:
Providing students of introductory thermal physics with a plot of the heat capacities of many low density gases as a function of temperature allows them to look for systematic trends. Specifically, large amounts of heat capacity data allow students to discover the equipartition theorem, but also point to its limited applicability. Computer code to download and plot the temperature-dependent heat c…
▽ More
Providing students of introductory thermal physics with a plot of the heat capacities of many low density gases as a function of temperature allows them to look for systematic trends. Specifically, large amounts of heat capacity data allow students to discover the equipartition theorem, but also point to its limited applicability. Computer code to download and plot the temperature-dependent heat capacity data is provided.
△ Less
Submitted 4 August, 2023;
originally announced August 2023.
-
Equilibration of quantum many-body fast neutrino flavor oscillations
Authors:
Joshua D. Martin,
Duff Neill,
A. Roggero,
Huaiyu Duan,
J. Carlson
Abstract:
Neutrino gases are expected to form in high density astrophysical environments, and accurately modeling their flavor evolution is critical to understanding such environments. In this work we study a simplified model of such a dense neutrino gas in the regime for which neutrino-neutrino coherent forward scattering is the dominant mechanism contributing to the flavor evolution. We show evidence that…
▽ More
Neutrino gases are expected to form in high density astrophysical environments, and accurately modeling their flavor evolution is critical to understanding such environments. In this work we study a simplified model of such a dense neutrino gas in the regime for which neutrino-neutrino coherent forward scattering is the dominant mechanism contributing to the flavor evolution. We show evidence that the generic potential induced by this effect is non-integrable and that the statistics of its energy level spaces are in good agreement with the Wigner surmise. We also find that individual neutrinos rapidly entangle with all of the others present which results in an equilibration of the flavor content of individual neutrinos. We show that the average neutrino flavor content can be predicted utilizing a thermodynamic partition function. A random phase approximation to the evolution gives a simple picture of this equilibration. In the case of neutrinos and antineutrinos, processes like $ν_e {\barν}_e \leftrightarrows ν_μ{\barν_μ} $ yield a rapid equilibrium satisfying $n( ν_e) n({\bar ν}_e) = n( ν_μ) n({\bar ν}_μ) = n( ν_τ) n({\bar ν}_τ)$ in addition to the standard lepton number conservation in regimes where off-diagonal vacuum oscillations are small compared to $ν-ν$ interactions.
△ Less
Submitted 8 January, 2024; v1 submitted 31 July, 2023;
originally announced July 2023.
-
Comment on "Matter-wave interferometry with helium atoms in low-$l$ Rydberg states''
Authors:
D. Z. Chan,
J. D. D. Martin
Abstract:
Tommey and Hogan [Phys. Rev. A, 104, 033305 (2021)] have reported a matter-wave interference experiment using Rydberg atoms traveling through inhomogeneous electric fields at approximately 2000 m/s. Using a simplified model containing the essential physics of their experiment, we show that the phase difference measured by their observed interference fringes does not depend -- in any significant wa…
▽ More
Tommey and Hogan [Phys. Rev. A, 104, 033305 (2021)] have reported a matter-wave interference experiment using Rydberg atoms traveling through inhomogeneous electric fields at approximately 2000 m/s. Using a simplified model containing the essential physics of their experiment, we show that the phase difference measured by their observed interference fringes does not depend -- in any significant way -- on the acceleration of the Rydberg atoms, but instead simply on the uniform motion of the atoms through the inhomogeneous electric field.
△ Less
Submitted 7 August, 2023; v1 submitted 11 June, 2023;
originally announced June 2023.
-
Robust Route Planning with Distributional Reinforcement Learning in a Stochastic Road Network Environment
Authors:
Xi Lin,
Paul Szenher,
John D. Martin,
Brendan Englot
Abstract:
Route planning is essential to mobile robot navigation problems. In recent years, deep reinforcement learning (DRL) has been applied to learning optimal planning policies in stochastic environments without prior knowledge. However, existing works focus on learning policies that maximize the expected return, the performance of which can vary greatly when the level of stochasticity in the environmen…
▽ More
Route planning is essential to mobile robot navigation problems. In recent years, deep reinforcement learning (DRL) has been applied to learning optimal planning policies in stochastic environments without prior knowledge. However, existing works focus on learning policies that maximize the expected return, the performance of which can vary greatly when the level of stochasticity in the environment is high. In this work, we propose a distributional reinforcement learning based framework that learns return distributions which explicitly reflect environmental stochasticity. Policies based on the second-order stochastic dominance (SSD) relation can be used to make adjustable route decisions according to user preference on performance robustness. Our proposed method is evaluated in a simulated road network environment, and experimental results show that our method is able to plan the shortest routes that minimize stochasticity in travel time when robustness is preferred, while other state-of-the-art DRL methods are agnostic to environmental stochasticity.
△ Less
Submitted 19 April, 2023;
originally announced April 2023.
-
Can a quantum mixmaster universe undergo a spontaneous inflationary phase?
Authors:
Herve Bergeron,
Jaime de Cabo Martin,
Jean-Pierre Gazeau,
Przemyslaw Malkiewicz
Abstract:
We study a semiclassical model of the mixmaster universe. We first derive the quantum model and then introduce its semiclassical approximation. We employ a general integral quantization method that respects the symmetries of the model given by the affine and the Weyl-Heisenberg groups, and can produce a wide class of quantum models. The semiclassical approximation is based on the coherent states.…
▽ More
We study a semiclassical model of the mixmaster universe. We first derive the quantum model and then introduce its semiclassical approximation. We employ a general integral quantization method that respects the symmetries of the model given by the affine and the Weyl-Heisenberg groups, and can produce a wide class of quantum models. The semiclassical approximation is based on the coherent states. The semiclassical dynamics is complex and cannot be solved by analytical methods. We focus on a key qualitative feature of the dynamics, namely, we investigate whether the primordial anisotropic universe can undergo a spontaneous inflationary phase driven by the anisotropic energy combined with semiclassical corrections. The answer to this question provides a useful perspective on the inflationary paradigm as well as on alternative bouncing models.
△ Less
Submitted 2 September, 2023; v1 submitted 14 March, 2023;
originally announced March 2023.
-
Mixmaster universe: semiclassical dynamics and inflation from bouncing
Authors:
Jaime de Cabo Martin
Abstract:
In this work we explore the quantum Bianchi type IX-model, its semi-classical features, and its relevance in early cosmology to tentatively explain inflation and production of primordial structures. We specially focus on the analytical and numerical exploration of the dynamical system derived from the phase-space portraits. Afterwards we investigate the reliability of our results with regard to in…
▽ More
In this work we explore the quantum Bianchi type IX-model, its semi-classical features, and its relevance in early cosmology to tentatively explain inflation and production of primordial structures. We specially focus on the analytical and numerical exploration of the dynamical system derived from the phase-space portraits. Afterwards we investigate the reliability of our results with regard to inflation and post inflation scenarii commonly accepted nowadays.
△ Less
Submitted 9 February, 2023; v1 submitted 2 February, 2023;
originally announced February 2023.
-
Auxiliary field diffusion Monte Carlo calculations of magnetic moments of light nuclei with chiral EFT interactions
Authors:
J. D. Martin,
S. J. Novario,
D. Lonardoni,
J. Carlson,
S. Gandolfi,
I. Tews
Abstract:
We calculate the magnetic moments of light nuclei ($A < 20$) using the auxiliary field diffusion Monte Carlo method and local two- and three-nucleon forces with electromagnetic currents from chiral effective field theory. For all nuclei under consideration, we also calculate the ground-state energies and charge radii. We generally find a good agreement with experimental values for all of these obs…
▽ More
We calculate the magnetic moments of light nuclei ($A < 20$) using the auxiliary field diffusion Monte Carlo method and local two- and three-nucleon forces with electromagnetic currents from chiral effective field theory. For all nuclei under consideration, we also calculate the ground-state energies and charge radii. We generally find a good agreement with experimental values for all of these observables. For the electromagnetic currents, we explore the impact of employing two different power countings, and study theoretical uncertainties stemming from the truncation of the chiral expansion order-by-order for select nuclei within these two approaches. We find that it is crucial to employ consistent power countings for interactions and currents to achieve a systematic order-by-order convergence.
△ Less
Submitted 25 September, 2023; v1 submitted 19 January, 2023;
originally announced January 2023.
-
Many-body neutrino flavor entanglement in a simple dynamic model
Authors:
Joshua D. Martin,
A. Roggero,
Huaiyu Duan,
J. Carlson
Abstract:
Dense neutrino gases form in extreme astrophysical sites, and the flavor content of the neutrinos likely has an important impact on the subsequent dynamical evolution of their environment. Through coherent forward scattering among neutrinos, the flavor content of the gas evolves under a time-dependent potential which can be modeled in a quantum many-body formalism as an all-to-all coupled spin-spi…
▽ More
Dense neutrino gases form in extreme astrophysical sites, and the flavor content of the neutrinos likely has an important impact on the subsequent dynamical evolution of their environment. Through coherent forward scattering among neutrinos, the flavor content of the gas evolves under a time-dependent potential which can be modeled in a quantum many-body formalism as an all-to-all coupled spin-spin interaction. This two-body potential generically introduces entanglement and greatly complicates the study of these systems. In this work we study the evolution of the quantum many-body problem as well as the typically employed mean-field approximation to it for a small number of neutrinos ($N = 16$). We consider randomly chosen one- and two-body couplings in the Hamiltonian, and the resulting evolution of several initial product states. We subsequently compare many-body and mean-field predictions for one-body observables, and we consider one- and two-body entanglement to assess under what conditions the many-body and mean-field predictions are likely to disagree. Except for a special category of prototypical initial conditions, we find that the typically employed mean-field approximation is insufficient to capture the evolution of one-body operators in the systems we consider. We also observe a loss of coherence in one- and two-body trace-reduced subsystems which suggests that the evolution may be well approximated as a classical mixture of separable states.
△ Less
Submitted 17 January, 2023;
originally announced January 2023.
-
Ambiguous power spectrum from a quantum bounce
Authors:
Jaime de Cabo Martin,
Przemysław Małkiewicz,
Patrick Peter
Abstract:
A quantum cosmological bouncing model may exhibit an ambiguity stemming from the nonclassical nature of the background evolution: two classically equivalent theories can produce two qualitatively different potentials sourcing the perturbations. We derive explicitly the quantum canonical transformation involving the quantum background to show how it leads to inequivalent theories. We identify the r…
▽ More
A quantum cosmological bouncing model may exhibit an ambiguity stemming from the nonclassical nature of the background evolution: two classically equivalent theories can produce two qualitatively different potentials sourcing the perturbations. We derive explicitly the quantum canonical transformation involving the quantum background to show how it leads to inequivalent theories. We identify the relevant quantum parameter describing the difference and expand upon the ambiguity by calculating the expected power spectra produced for initial quantum vacuum fluctuations in the contracting phase of both potentials. We find that one spectral index corresponds to all values of this parameter but one, while the other thus represents a set of measure zero.
△ Less
Submitted 21 March, 2024; v1 submitted 23 December, 2022;
originally announced December 2022.
-
Direct interpretation of the X-ray and neutron 3D-$Δ$PDFs of yittria stabilized zirconia
Authors:
Ella M. Schmidt,
Reinhard B. Neder,
James D. Martin,
Arianna Minelli,
Marie-Hélène Lenée,
Andrew L. Goodwin
Abstract:
Three dimensional difference pair distribution functions from X-ray and neutron diffraction experiments are reported for yttria stabilized zirconia (Zr$_{0.82}$Y$_{0.18}$O$_{1.91}$). We use a quantitative analysis of the signatures in the 3D-$Δ$PDFs to establish that oxygen ions neighbouring a vacancy shift by 0.515(5)~Å along $\langle 1,0,0 \rangle$ towards the vacancy while metal ions neighbouri…
▽ More
Three dimensional difference pair distribution functions from X-ray and neutron diffraction experiments are reported for yttria stabilized zirconia (Zr$_{0.82}$Y$_{0.18}$O$_{1.91}$). We use a quantitative analysis of the signatures in the 3D-$Δ$PDFs to establish that oxygen ions neighbouring a vacancy shift by 0.515(5)~Å along $\langle 1,0,0 \rangle$ towards the vacancy while metal ions neighbouring a vacancy shift by 0.269(2)~Å along $\langle 1,1,1 \rangle$ away from the vacancy. The neutron 3D-$Δ$PDF shows a tendency for vacancies to cluster along $\langle \frac{1}{2},\frac{1}{2},\frac{1}{2}\rangle$, which results in 6-fold coordinated metal ions.
△ Less
Submitted 21 December, 2022;
originally announced December 2022.
-
Settling the Reward Hypothesis
Authors:
Michael Bowling,
John D. Martin,
David Abel,
Will Dabney
Abstract:
The reward hypothesis posits that, "all of what we mean by goals and purposes can be well thought of as maximization of the expected value of the cumulative sum of a received scalar signal (reward)." We aim to fully settle this hypothesis. This will not conclude with a simple affirmation or refutation, but rather specify completely the implicit requirements on goals and purposes under which the hy…
▽ More
The reward hypothesis posits that, "all of what we mean by goals and purposes can be well thought of as maximization of the expected value of the cumulative sum of a received scalar signal (reward)." We aim to fully settle this hypothesis. This will not conclude with a simple affirmation or refutation, but rather specify completely the implicit requirements on goals and purposes under which the hypothesis holds.
△ Less
Submitted 16 September, 2023; v1 submitted 20 December, 2022;
originally announced December 2022.
-
Should Models Be Accurate?
Authors:
Esra'a Saleh,
John D. Martin,
Anna Koop,
Arash Pourzarabi,
Michael Bowling
Abstract:
Model-based Reinforcement Learning (MBRL) holds promise for data-efficiency by planning with model-generated experience in addition to learning with experience from the environment. However, in complex or changing environments, models in MBRL will inevitably be imperfect, and their detrimental effects on learning can be difficult to mitigate. In this work, we question whether the objective of thes…
▽ More
Model-based Reinforcement Learning (MBRL) holds promise for data-efficiency by planning with model-generated experience in addition to learning with experience from the environment. However, in complex or changing environments, models in MBRL will inevitably be imperfect, and their detrimental effects on learning can be difficult to mitigate. In this work, we question whether the objective of these models should be the accurate simulation of environment dynamics at all. We focus our investigations on Dyna-style planning in a prediction setting. First, we highlight and support three motivating points: a perfectly accurate model of environment dynamics is not practically achievable, is not necessary, and is not always the most useful anyways. Second, we introduce a meta-learning algorithm for training models with a focus on their usefulness to the learner instead of their accuracy in modelling the environment. Our experiments show that in a simple non-stationary environment, our algorithm enables faster learning than even using an accurate model built with domain-specific knowledge of the non-stationarity.
△ Less
Submitted 22 May, 2022;
originally announced May 2022.
-
The primordial structure from Quantum Cosmological bouncing models
Authors:
Jaime de Cabo Martin
Abstract:
By quantizing the background as well as the perturbations in a simple one fluid cosmological model, we show that there exists an ambiguity in the choice of relevant variables, potentially leading to incompatible observational physical predictions. In a classical inflationary background, the exact same canonical transformations lead to unique predictions, so the ambiguity we put forward demands a s…
▽ More
By quantizing the background as well as the perturbations in a simple one fluid cosmological model, we show that there exists an ambiguity in the choice of relevant variables, potentially leading to incompatible observational physical predictions. In a classical inflationary background, the exact same canonical transformations lead to unique predictions, so the ambiguity we put forward demands a semiclassical background with a sufficiently strong departure from classical evolution. The latter condition is clearly satisfied by bouncing models. We propose coherent states as the tool for introducing the semiclassical universe. We solve the quantum dynamics of the perturbation modes both analytically and numerically and investigate the amplitude spectra of the perturbations. We study the underlying quantum state, the Bunch-Davies vacuum, from the point of view of late-time observers by means of the Bogolyubov transformations. In particular, we study the phase space probability distributions obtained with the standard coherent states built from instantaneous vacua. We discuss the issue of the temporal phase shift with which the modes emerge from the bounce as sine waves. Finally, we find that the model may be fitted to data and shed light on the physical universe, constraining free parameters of the bouncing universe.
△ Less
Submitted 8 March, 2022;
originally announced March 2022.
-
Classical and Quantum Evolution in a Simple Coherent Neutrino Problem
Authors:
Joshua D. Martin,
A. Roggero,
Huaiyu Duan,
J. Carlson,
V. Cirigliano
Abstract:
The extraordinary neutrino flux produced in extreme astrophysical environments like the early universe, core-collapse supernovae and neutron star mergers may produce coherent quantum neutrino oscillations on macroscopic length scales. The Hamiltonian describing this evolution can be mapped into quantum spin models with all-to-all couplings arising from neutrino-neutrino forward scattering. To date…
▽ More
The extraordinary neutrino flux produced in extreme astrophysical environments like the early universe, core-collapse supernovae and neutron star mergers may produce coherent quantum neutrino oscillations on macroscopic length scales. The Hamiltonian describing this evolution can be mapped into quantum spin models with all-to-all couplings arising from neutrino-neutrino forward scattering. To date many studies of these oscillations have been performed in a mean-field limit where the neutrinos time evolve in a product state.
In this paper we examine a simple two-beam model evolving from an initial product state and compare the mean-field and many-body evolution. The symmetries in this model allow us to solve the real-time evolution for the quantum many-body system for hundreds or thousands of spins, far beyond what would be possible in a more general case with an exponential number ($2^N$) of quantum states. We compare mean-field and many-body solutions for different initial product states and ratios of one- and two-body couplings, and find that in all cases in the limit of infinite spins the mean-field (product state) and many-body solutions coincide for simple observables. This agreement can be understood as a consequence of the fact that the typical initial condition represents a very local but dense distribution about a mean energy in the spectrum of the Hamiltonian. We explore quantum information measures like entanglement entropy and purity of the many-body solutions, finding intriguing relationships between the quantum information measures and the dynamical behavior of simple physical observables.
△ Less
Submitted 28 April, 2022; v1 submitted 23 December, 2021;
originally announced December 2021.
-
Unitarily inequivalent quantum cosmological bouncing models
Authors:
Jaime de Cabo Martin,
Przemysław Małkiewicz,
Patrick Peter
Abstract:
By quantising the background as well as the perturbations in a simple one fluid model, we show that there exists an ambiguity in the choice of relevant variables, potentially leading to incompatible observational physical predictions. In a classical or quantum inflationary background, the exact same canonical transformations lead to unique predictions, so the ambiguity we put forward demands a sem…
▽ More
By quantising the background as well as the perturbations in a simple one fluid model, we show that there exists an ambiguity in the choice of relevant variables, potentially leading to incompatible observational physical predictions. In a classical or quantum inflationary background, the exact same canonical transformations lead to unique predictions, so the ambiguity we put forward demands a semiclassical background with a sufficiently strong departure from classical evolution. The latter condition happens to be satisfied in bouncing scenarios, which may thus be having predictability issues. Inflationary models could evade such a problem because of the monotonic behavior of their scale factor; they do, however, initiate from a singular state which bouncing scenarios aim at solving.
△ Less
Submitted 16 November, 2021; v1 submitted 4 November, 2021;
originally announced November 2021.
-
Flavor isospin waves in one-dimensional axisymmetric neutrino gases
Authors:
Huaiyu Duan,
Joshua D. Martin,
Sivaprasad Omanakuttan
Abstract:
Flavor oscillations can occur on very short spatial and temporal scales in the dense neutrino media in a core-collapse supernova (CCSN) or binary neutron star merger (BNSM) event. Although the dispersion relations (DRs) of the fast neutrino oscillations can be obtained by linearizing the equations of motion (EoM) before the emergence of any significant flavor conversion, one largely depends on num…
▽ More
Flavor oscillations can occur on very short spatial and temporal scales in the dense neutrino media in a core-collapse supernova (CCSN) or binary neutron star merger (BNSM) event. Although the dispersion relations (DRs) of the fast neutrino oscillations can be obtained by linearizing the equations of motion (EoM) before the emergence of any significant flavor conversion, one largely depends on numerical calculations to understand this interesting phenomenon in the nonlinear regime. In this work we demonstrate that there exist nontrivial solutions to the flavor EoM that govern the fast oscillations in one-dimensional axisymmetric neutrino gases. These solutions represent a coherent flavor isospin wave similar to the magnetic spin wave in a lattice of magnetic dipoles. We also compute the DRs of such waves in some example cases which are closely related to the DRs of the fast neutrino oscillations obtained in the linear regime. This result sheds new light on the long-term behavior of fast neutrino oscillations which can have various implications for the CCSN and BNSM events.
△ Less
Submitted 16 December, 2021; v1 submitted 5 October, 2021;
originally announced October 2021.
-
Adapting the Function Approximation Architecture in Online Reinforcement Learning
Authors:
John D. Martin,
Joseph Modayil
Abstract:
The performance of a reinforcement learning (RL) system depends on the computational architecture used to approximate a value function. Deep learning methods provide both optimization techniques and architectures for approximating nonlinear functions from noisy, high-dimensional observations. However, prevailing optimization techniques are not designed for strictly-incremental online updates. Nor…
▽ More
The performance of a reinforcement learning (RL) system depends on the computational architecture used to approximate a value function. Deep learning methods provide both optimization techniques and architectures for approximating nonlinear functions from noisy, high-dimensional observations. However, prevailing optimization techniques are not designed for strictly-incremental online updates. Nor are standard architectures designed for observations with an a priori unknown structure: for example, light sensors randomly dispersed in space. This paper proposes an online RL prediction algorithm with an adaptive architecture that efficiently finds useful nonlinear features. The algorithm is evaluated in a spatial domain with high-dimensional, stochastic observations. The algorithm outperforms non-adaptive baseline architectures and approaches the performance of an architecture given side-channel information. These results are a step towards scalable RL algorithms for more general problems, where the observation structure is not available.
△ Less
Submitted 17 June, 2021;
originally announced June 2021.
-
A simple proof of Floquet's theorem for the case of unitary evolution in quantum mechanics
Authors:
J. D. D. Martin,
A. N. Poertner
Abstract:
We present a constructive proof of Floquet's theorem for the special case of unitary time evolution in quantum mechanics. The proof is straightforward and suitable for study in courses on quantum mechanics.
We present a constructive proof of Floquet's theorem for the special case of unitary time evolution in quantum mechanics. The proof is straightforward and suitable for study in courses on quantum mechanics.
△ Less
Submitted 1 March, 2021;
originally announced April 2021.
-
Fast flavor oscillations in dense neutrino media with collisions
Authors:
Joshua D. Martin,
J. Carlson,
Vincenzo Cirigliano,
Huaiyu Duan
Abstract:
We investigate the impact of the nonzero neutrino splitting and elastic neutrino-nucleon collisions on fast neutrino oscillations. Our calculations confirm that a small neutrino mass splitting and the neutrino mass hierarchy have very little effect on fast oscillation waves. We also demonstrate explicitly that fast oscillations remain largely unaffected for the time/distance scales that are much s…
▽ More
We investigate the impact of the nonzero neutrino splitting and elastic neutrino-nucleon collisions on fast neutrino oscillations. Our calculations confirm that a small neutrino mass splitting and the neutrino mass hierarchy have very little effect on fast oscillation waves. We also demonstrate explicitly that fast oscillations remain largely unaffected for the time/distance scales that are much smaller than the neutrino mean free path but are damped on larger scales. This dam** originates from both the direct modification of the dispersion relation of the oscillation waves in the neutrino medium and the flattening of the neutrino angular distributions over time. Our work suggests that fast neutrino oscillation waves produced near the neutrino sphere can propagate essentially unimpeded which may have ramifications in various aspects of the supernova physics.
△ Less
Submitted 1 March, 2021; v1 submitted 4 January, 2021;
originally announced January 2021.
-
Variational Filtering with Copula Models for SLAM
Authors:
John D. Martin,
Kevin Doherty,
Caralyn Cyr,
Brendan Englot,
John Leonard
Abstract:
The ability to infer map variables and estimate pose is crucial to the operation of autonomous mobile robots. In most cases the shared dependency between these variables is modeled through a multivariate Gaussian distribution, but there are many situations where that assumption is unrealistic. Our paper shows how it is possible to relax this assumption and perform simultaneous localization and map…
▽ More
The ability to infer map variables and estimate pose is crucial to the operation of autonomous mobile robots. In most cases the shared dependency between these variables is modeled through a multivariate Gaussian distribution, but there are many situations where that assumption is unrealistic. Our paper shows how it is possible to relax this assumption and perform simultaneous localization and map** (SLAM) with a larger class of distributions, whose multivariate dependency is represented with a copula model. We integrate the distribution model with copulas into a Sequential Monte Carlo estimator and show how unknown model parameters can be learned through gradient-based optimization. We demonstrate our approach is effective in settings where Gaussian assumptions are clearly violated, such as environments with uncertain data association and nonlinear transition models.
△ Less
Submitted 2 August, 2020;
originally announced August 2020.
-
Autonomous Exploration Under Uncertainty via Deep Reinforcement Learning on Graphs
Authors:
Fanfei Chen,
John D. Martin,
Yewei Huang,
**kun Wang,
Brendan Englot
Abstract:
We consider an autonomous exploration problem in which a range-sensing mobile robot is tasked with accurately map** the landmarks in an a priori unknown environment efficiently in real-time; it must choose sensing actions that both curb localization uncertainty and achieve information gain. For this problem, belief space planning methods that forward-simulate robot sensing and estimation may oft…
▽ More
We consider an autonomous exploration problem in which a range-sensing mobile robot is tasked with accurately map** the landmarks in an a priori unknown environment efficiently in real-time; it must choose sensing actions that both curb localization uncertainty and achieve information gain. For this problem, belief space planning methods that forward-simulate robot sensing and estimation may often fail in real-time implementation, scaling poorly with increasing size of the state, belief and action spaces. We propose a novel approach that uses graph neural networks (GNNs) in conjunction with deep reinforcement learning (DRL), enabling decision-making over graphs containing exploration information to predict a robot's optimal sensing action in belief space. The policy, which is trained in different random environments without human intervention, offers a real-time, scalable decision-making process whose high-performance exploratory sensing actions yield accurate maps and high rates of information gain.
△ Less
Submitted 24 July, 2020;
originally announced July 2020.
-
Fusing Concurrent Orthogonal Wide-aperture Sonar Images for Dense Underwater 3D Reconstruction
Authors:
John McConnell,
John D. Martin,
Brendan Englot
Abstract:
We propose a novel approach to handling the ambiguity in elevation angle associated with the observations of a forward looking multi-beam imaging sonar, and the challenges it poses for performing an accurate 3D reconstruction. We utilize a pair of sonars with orthogonal axes of uncertainty to independently observe the same points in the environment from two different perspectives, and associate th…
▽ More
We propose a novel approach to handling the ambiguity in elevation angle associated with the observations of a forward looking multi-beam imaging sonar, and the challenges it poses for performing an accurate 3D reconstruction. We utilize a pair of sonars with orthogonal axes of uncertainty to independently observe the same points in the environment from two different perspectives, and associate these observations. Using these concurrent observations, we can create a dense, fully defined point cloud at every time-step to aid in reconstructing the 3D geometry of underwater scenes. We will evaluate our method in the context of the current state of the art, for which strong assumptions on object geometry limit applicability to generalized 3D scenes. We will discuss results from laboratory tests that quantitatively benchmark our algorithm's reconstruction capabilities, and results from a real-world, tidal river basin which qualitatively demonstrate our ability to reconstruct a cluttered field of underwater objects.
△ Less
Submitted 20 July, 2020;
originally announced July 2020.
-
On Catastrophic Interference in Atari 2600 Games
Authors:
William Fedus,
Dibya Ghosh,
John D. Martin,
Marc G. Bellemare,
Yoshua Bengio,
Hugo Larochelle
Abstract:
Model-free deep reinforcement learning is sample inefficient. One hypothesis -- speculated, but not confirmed -- is that catastrophic interference within an environment inhibits learning. We test this hypothesis through a large-scale empirical study in the Arcade Learning Environment (ALE) and, indeed, find supporting evidence. We show that interference causes performance to plateau; the network c…
▽ More
Model-free deep reinforcement learning is sample inefficient. One hypothesis -- speculated, but not confirmed -- is that catastrophic interference within an environment inhibits learning. We test this hypothesis through a large-scale empirical study in the Arcade Learning Environment (ALE) and, indeed, find supporting evidence. We show that interference causes performance to plateau; the network cannot train on segments beyond the plateau without degrading the policy used to reach there. By synthetically controlling for interference, we demonstrate performance boosts across architectures, learning algorithms and environments. A more refined analysis shows that learning one segment of a game often increases prediction errors elsewhere. Our study provides a clear empirical link between catastrophic interference and sample efficiency in reinforcement learning.
△ Less
Submitted 9 June, 2020; v1 submitted 27 February, 2020;
originally announced February 2020.
-
On the validity of many-mode Floquet theory with commensurate frequencies
Authors:
A. N. Poertner,
J. D. D. Martin
Abstract:
Many-mode Floquet theory [T.-S. Ho, S.-I. Chu, and J. V. Tietz, Chem. Phys. Lett., v. 96, 464 (1983)] is a technique for solving the time-dependent Schrödinger equation in the special case of multiple periodic fields, but its limitations are not well understood. We show that for a Hamiltonian consisting of two time-periodic couplings of commensurate frequencies (integer multiples of a common frequ…
▽ More
Many-mode Floquet theory [T.-S. Ho, S.-I. Chu, and J. V. Tietz, Chem. Phys. Lett., v. 96, 464 (1983)] is a technique for solving the time-dependent Schrödinger equation in the special case of multiple periodic fields, but its limitations are not well understood. We show that for a Hamiltonian consisting of two time-periodic couplings of commensurate frequencies (integer multiples of a common frequency), many-mode Floquet theory provides a correct expression for unitary time evolution. However, caution must be taken in the interpretation of the eigenvalues and eigenvectors of the corresponding many-mode Floquet Hamiltonian, as only part of its spectrum is directly relevant to time evolution. We give a physical interpretation for the remainder of the spectrum of the Hamiltonian. These results are relevant to the engineering of quantum systems using multiple controllable periodic fields.
△ Less
Submitted 5 December, 2019;
originally announced December 2019.
-
Spectral swaps in a two-dimensional neutrino ring model
Authors:
Joshua D. Martin,
J. Carlson,
Huaiyu Duan
Abstract:
Neutrinos emitted deep within a supernova explosion experience a self-induced index of refraction. In the stationary, one-dimensional (1D) supernova "bulb model", this self-induced refraction can lead to a collective flavor transformation which is coherent among different neutrino momentum modes. Such collective oscillations can produce partial swaps of the energy spectra of different neutrino spe…
▽ More
Neutrinos emitted deep within a supernova explosion experience a self-induced index of refraction. In the stationary, one-dimensional (1D) supernova "bulb model", this self-induced refraction can lead to a collective flavor transformation which is coherent among different neutrino momentum modes. Such collective oscillations can produce partial swaps of the energy spectra of different neutrino species as the neutrinos stream away from the proto-neutron star. However, it has been demonstrated that the spatial symmetries (such as the spherical symmetry in the bulb model) can be broken spontaneously by collective neutrino oscillations in multi-dimensional models. Using a stationary, 2D neutrino ring model we demonstrate that there exist two limiting scenarios where collective oscillations may occur. In one limit, the collective flavor transformation begins at a radius with relatively high neutrino densities and develops small-scale flavor structures. The loss of the spatial correlation in the neutrino flavor field results in similar (average) energy spectra for the anti-neutrinos of almost all energies and the neutrinos of relatively high energies. In the other limit, the flavor transformation starts at a radius where the neutrino densities are smaller (e.g., due to the suppression of the high matter density near the proto-neutron star). Although the spatial symmetry is broken initially, it is restored as the neutrino densities decrease, and the neutrinos of different flavors partially swap their energy spectra as in the 1D bulb model. This finding may have interesting ramifications in other aspects of supernova physics.
△ Less
Submitted 15 January, 2020; v1 submitted 21 November, 2019;
originally announced November 2019.
-
Dynamic fast flavor oscillation waves in dense neutrino gases
Authors:
Joshua D. Martin,
Changhao Yi,
Huaiyu Duan
Abstract:
The flavor transformation in a dense neutrino gas can have a significant impact on the physical and chemical evolution of its surroundings. In this work we demonstrate that a dynamic, fast flavor oscillation wave can develop spontaneously in a one-dimensional (1D) neutrino gas when the angular distributions of the electron neutrino and antineutrino cross each other. Unlike the 2D stationary models…
▽ More
The flavor transformation in a dense neutrino gas can have a significant impact on the physical and chemical evolution of its surroundings. In this work we demonstrate that a dynamic, fast flavor oscillation wave can develop spontaneously in a one-dimensional (1D) neutrino gas when the angular distributions of the electron neutrino and antineutrino cross each other. Unlike the 2D stationary models which are plagued with small-scale flavor structures, the fast flavor oscillation waves remain coherent in the dynamic 1D model in both the position and momentum spaces of the neutrino. The electron lepton number is redistributed and transported in space as the flavor oscillation wave propagates, although the total lepton number remains constant. This result may have interesting implications in the neutrino emission in and the evolution of the compact objects such as core-collapse supernovae.
△ Less
Submitted 15 November, 2019; v1 submitted 11 September, 2019;
originally announced September 2019.
-
Stochastically Dominant Distributional Reinforcement Learning
Authors:
John D. Martin,
Michal Lyskawinski,
Xiaohu Li,
Brendan Englot
Abstract:
We describe a new approach for managing aleatoric uncertainty in the Reinforcement Learning (RL) paradigm. Instead of selecting actions according to a single statistic, we propose a distributional method based on the second-order stochastic dominance (SSD) relation. This compares the inherent dispersion of random returns induced by actions, producing a more comprehensive and robust evaluation of t…
▽ More
We describe a new approach for managing aleatoric uncertainty in the Reinforcement Learning (RL) paradigm. Instead of selecting actions according to a single statistic, we propose a distributional method based on the second-order stochastic dominance (SSD) relation. This compares the inherent dispersion of random returns induced by actions, producing a more comprehensive and robust evaluation of the environment's uncertainty. The necessary conditions for SSD require estimators to predict accurate second moments. To accommodate this, we map the distributional RL problem to a Wasserstein gradient flow, treating the distributional Bellman residual as a potential energy functional. We propose a particle-based algorithm for which we prove optimality and convergence. Our experiments characterize the algorithm performance and demonstrate how uncertainty and performance are better balanced using an \textsc{ssd} policy than with other risk measures.
△ Less
Submitted 7 October, 2020; v1 submitted 17 May, 2019;
originally announced May 2019.
-
Nonlinear flavor development of a two-dimensional neutrino gas
Authors:
Joshua D. Martin,
Sajad Abbar,
Huaiyu Duan
Abstract:
We present a numerical survey of the nonlinear flavor development of dense neutrino gases. This study is based on the stationary, two-dimensional ($x$ and $z$), two-beam, monochromatic neutrino line model with a periodic boundary condition along the $x$ direction. Similar to a previous work, we find that small-scale flavor structures can develop in a neutrino gas even if the physical conditions ar…
▽ More
We present a numerical survey of the nonlinear flavor development of dense neutrino gases. This study is based on the stationary, two-dimensional ($x$ and $z$), two-beam, monochromatic neutrino line model with a periodic boundary condition along the $x$ direction. Similar to a previous work, we find that small-scale flavor structures can develop in a neutrino gas even if the physical conditions are nearly homogeneous along the $x$ axis initially. The power diffusion from the large-scale to small-scale structures increases with the neutrino density and helps to establish a semi-exponential dependence of the magnitudes of the Fourier moments on the corresponding wave numbers. The overall flavor conversion probabilities in the neutrino gases with small initial sinusoidal perturbations reach certain equilibrium values at large distances which are mainly determined by the neutrino-antineutrino asymmetry. Similar phenomena also exist in a neutrino gas with a localized initial perturbation, albeit only inside an expanding flavor conversion region. Our work suggests that a statistical treatment may be possible for the collective flavor oscillations of a dense neutrino gas in a multi-dimensional environment.
△ Less
Submitted 26 August, 2019; v1 submitted 18 April, 2019;
originally announced April 2019.
-
The dispersion relation of the fast neutrino oscillation wave
Authors:
Changhao Yi,
Lei Ma,
Joshua D. Martin,
Huaiyu Duan
Abstract:
A dense neutrino medium can support flavor oscillation waves which are coherent among different momentum modes of the neutrinos. The dispersion relation (DR) branches of such a wave with complex frequencies and/or wave numbers can lead to the exponential growth of the wave amplitude which in turn will engender a collective flavor transformation in the neutrino medium. In this work we propose that…
▽ More
A dense neutrino medium can support flavor oscillation waves which are coherent among different momentum modes of the neutrinos. The dispersion relation (DR) branches of such a wave with complex frequencies and/or wave numbers can lead to the exponential growth of the wave amplitude which in turn will engender a collective flavor transformation in the neutrino medium. In this work we propose that the complex DR branches of the neutrino oscillation wave should be bound by the critical points of the DR. We demonstrate how this theory can be applied to the neutrino medium with an (approximate) axial symmetry about the propagation direction of the neutrino oscillation wave. We also show how the flavor instabilities in this medium can be identified by tracing the critical points of the DR as the electron lepton number distribution of the neutrino medium is changed continuously.
△ Less
Submitted 12 March, 2019; v1 submitted 6 January, 2019;
originally announced January 2019.
-
Bioinformatics and Medicine in the Era of Deep Learning
Authors:
Davide Bacciu,
Paulo J. G. Lisboa,
José D. Martín,
Ruxandra Stoean,
Alfredo Vellido
Abstract:
Many of the current scientific advances in the life sciences have their origin in the intensive use of data for knowledge discovery. In no area this is so clear as in bioinformatics, led by technological breakthroughs in data acquisition technologies. It has been argued that bioinformatics could quickly become the field of research generating the largest data repositories, beating other data-inten…
▽ More
Many of the current scientific advances in the life sciences have their origin in the intensive use of data for knowledge discovery. In no area this is so clear as in bioinformatics, led by technological breakthroughs in data acquisition technologies. It has been argued that bioinformatics could quickly become the field of research generating the largest data repositories, beating other data-intensive areas such as high-energy physics or astroinformatics. Over the last decade, deep learning has become a disruptive advance in machine learning, giving new live to the long-standing connectionist paradigm in artificial intelligence. Deep learning methods are ideally suited to large-scale data and, therefore, they should be ideally suited to knowledge discovery in bioinformatics and biomedicine at large. In this brief paper, we review key aspects of the application of deep learning in bioinformatics and medicine, drawing from the themes covered by the contributions to an ESANN 2018 special session devoted to this topic.
△ Less
Submitted 27 February, 2018;
originally announced February 2018.
-
Reduction of the dc electric field sensitivity of circular Rydberg states using non-resonant dressing fields
Authors:
Y. Ni,
P. Xu,
J. D. D. Martin
Abstract:
Non-resonant dressing fields can make the transition frequency between two circular Rydberg states insensitive to second order variations in the dc electric field. Perturbation theory can be used to establish the required dressing field amplitude and frequency. The same perturbative approach may be used to understand removal of the first order dependence of the transition frequency on electric fie…
▽ More
Non-resonant dressing fields can make the transition frequency between two circular Rydberg states insensitive to second order variations in the dc electric field. Perturbation theory can be used to establish the required dressing field amplitude and frequency. The same perturbative approach may be used to understand removal of the first order dependence of the transition frequency on electric field about a bias dc electric field [Hyafil et al. Phys. Rev. Lett., v. 93, 103001 (2004)]. The directional alignment of the dressing and dc fields is critical in determining the electric field sensitivity of the dressed transition frequencies. This sensitivity is significantly larger for circular Rydberg states compared to low-angular momentum Rydberg states of Rb.
△ Less
Submitted 21 September, 2015;
originally announced September 2015.
-
Coherent manipulation of cold Rydberg atoms near the surface of an atom chip
Authors:
J. D. Carter,
J. D. D. Martin
Abstract:
Coherent superpositions of the 49s and 48s Rydberg states of cold Rb atoms were studied near the surface of an atom chip. The superpositions were created and manipulated using microwaves resonant with the two-photon 49s-48s transition. Coherent behavior was observed using Rabi flop**, Ramsey sequences, spin-echo and spin-locking. These results are discussed in the context of Rydberg atoms as ele…
▽ More
Coherent superpositions of the 49s and 48s Rydberg states of cold Rb atoms were studied near the surface of an atom chip. The superpositions were created and manipulated using microwaves resonant with the two-photon 49s-48s transition. Coherent behavior was observed using Rabi flop**, Ramsey sequences, spin-echo and spin-locking. These results are discussed in the context of Rydberg atoms as electric field noise sensors. We consider the coherence of systems quadratically coupled to noise fields with 1/f^k power spectral densities (k \approx 1).
△ Less
Submitted 6 October, 2013; v1 submitted 8 August, 2013;
originally announced August 2013.
-
Rydberg atoms with a reduced sensitivity to dc and low-frequency electric fields
Authors:
L. A. Jones,
J. D. Carter,
J. D. D. Martin
Abstract:
A non-resonant microwave dressing field at 38.465 GHz was used to eliminate the static electric dipole moment difference between the $49s_{1/2}$ and $48s_{1/2}$ Rydberg states of $^{87}$Rb in dc fields of approximately 1 V/cm. The reduced susceptibility to electric field fluctuations was measured using 2-photon microwave spectroscopy. An anomalous spectral doublet is attributed to polarization ell…
▽ More
A non-resonant microwave dressing field at 38.465 GHz was used to eliminate the static electric dipole moment difference between the $49s_{1/2}$ and $48s_{1/2}$ Rydberg states of $^{87}$Rb in dc fields of approximately 1 V/cm. The reduced susceptibility to electric field fluctuations was measured using 2-photon microwave spectroscopy. An anomalous spectral doublet is attributed to polarization ellipticity in the dressing field. The demonstrated ability to inhibit static dipole moment differences --- while retaining sensitivity to high frequency fields --- is applicable to sensors and/or quantum devices using Rydberg atoms.
△ Less
Submitted 17 January, 2013;
originally announced January 2013.
-
Electric field sensing near the surface microstructure of an atom chip using cold Rydberg atoms
Authors:
J. D. Carter,
O. Cherry,
J. D. D. Martin
Abstract:
The electric fields near the heterogeneous metal/dielectric surface of an atom chip were measured using cold atoms. The atomic sensitivity to electric fields was enhanced by exciting the atoms to Rydberg states that are 10^8 times more polarizable than the ground state. We attribute the measured fields to charging of the insulators between the atom chip wires. Surprisingly, it is observed that the…
▽ More
The electric fields near the heterogeneous metal/dielectric surface of an atom chip were measured using cold atoms. The atomic sensitivity to electric fields was enhanced by exciting the atoms to Rydberg states that are 10^8 times more polarizable than the ground state. We attribute the measured fields to charging of the insulators between the atom chip wires. Surprisingly, it is observed that these fields may be dramatically lowered with appropriate voltage biasing, suggesting configurations for the future development of hybrid quantum systems.
△ Less
Submitted 29 June, 2012;
originally announced June 2012.
-
Injection-locked diode laser current modulation for Pound-Drever-Hall frequency stabilization using transfer cavities
Authors:
C. E. Liekhus-Schmaltz,
R. Mantifel,
M. Torabifard,
I. B. Burgess,
J. D. D. Martin
Abstract:
A phase modulated RF current source is applied to an injection locked diode laser operating at $780\unit{nm}$. This produces tunable phase modulated sidebands of the laser suitable for stabilizing the length of an optical transfer cavity using the Pound-Drever-Hall technique. The Pound-Drever-Hall signal is anti-symmetric about the lock point, despite the presence of significant diode laser amplit…
▽ More
A phase modulated RF current source is applied to an injection locked diode laser operating at $780\unit{nm}$. This produces tunable phase modulated sidebands of the laser suitable for stabilizing the length of an optical transfer cavity using the Pound-Drever-Hall technique. The Pound-Drever-Hall signal is anti-symmetric about the lock point, despite the presence of significant diode laser amplitude modulation. The stabilized optical transfer cavity is used to frequency stabilize a 776 nm external cavity diode laser. The stability and tunability of this transfer cavity locked laser is established by observation of the hyperfine components of the $^{87}$Rb $5P_{3/2}-5D_{5/2}$ transition in a vapor cell.
△ Less
Submitted 24 December, 2011; v1 submitted 1 September, 2011;
originally announced September 2011.
-
Understanding Pound-Drever-Hall locking using voltage controlled radio-frequency oscillators: An undergraduate experiment
Authors:
C. E. Liekhus-Schmaltz,
J. D. D. Martin
Abstract:
We have developed a senior undergraduate experiment that illustrates frequency stabilization techniques using radio-frequency electronics. The primary objective is to frequency stabilize a voltage controlled oscillator to a cavity resonance at 800 MHz using the Pound-Drever-Hall method. This technique is commonly applied to stabilize lasers at optical frequencies. By using only radio-frequency equ…
▽ More
We have developed a senior undergraduate experiment that illustrates frequency stabilization techniques using radio-frequency electronics. The primary objective is to frequency stabilize a voltage controlled oscillator to a cavity resonance at 800 MHz using the Pound-Drever-Hall method. This technique is commonly applied to stabilize lasers at optical frequencies. By using only radio-frequency equipment it is possible to systematically study aspects of the technique more thoroughly, inexpensively, and free from eye hazards. Students also learn about modular radio-frequency electronics and basic feedback control loops. By varying the temperature of the resonator, students can determine the thermal expansion coefficients of copper, aluminum, and super invar.
△ Less
Submitted 3 December, 2011; v1 submitted 3 August, 2011;
originally announced August 2011.
-
Energy shifts of Rydberg atoms due to patch fields near metal surfaces
Authors:
J. D. Carter,
J. D. D. Martin
Abstract:
The statistical properties of patch electric fields due to a polycrystalline metal surface are calculated. The fluctuations in the electric field scale like 1/z^2, when z >> w, where z is the distance to the surface, and w is the characteristic length scale of the surface patches. For typical thermally evaporated gold surfaces these field fluctuations are comparable to the image field of an elemen…
▽ More
The statistical properties of patch electric fields due to a polycrystalline metal surface are calculated. The fluctuations in the electric field scale like 1/z^2, when z >> w, where z is the distance to the surface, and w is the characteristic length scale of the surface patches. For typical thermally evaporated gold surfaces these field fluctuations are comparable to the image field of an elementary charge, and scale in the same way with distance to the surface. Expressions for calculating the statistics of the inhomogeneous broadening of Rydberg atom energies due to patch electric fields are presented. Spatial variations in the patch fields over the Rydberg orbit are found to be insignificant.
△ Less
Submitted 7 March, 2011; v1 submitted 9 August, 2010;
originally announced August 2010.
-
ac electric-field-induced resonant energy transfer between cold Rydberg atoms
Authors:
J. A. Petrus,
P. Bohlouli-Zanjani,
J. D. D. Martin
Abstract:
An oscillating electric field at 1.356 GHz was used to promote the resonant energy transfer process: $43d_{5/2}+43d_{5/2} \to 45p_{3/2}+41f$ between translationally cold $^{85}$Rb Rydberg atoms. The ac Stark shifts due to this dressing field created degeneracies between the initial and final two-atom states of this process. The ac field strength was scanned to collect spectra which are analogous…
▽ More
An oscillating electric field at 1.356 GHz was used to promote the resonant energy transfer process: $43d_{5/2}+43d_{5/2} \to 45p_{3/2}+41f$ between translationally cold $^{85}$Rb Rydberg atoms. The ac Stark shifts due to this dressing field created degeneracies between the initial and final two-atom states of this process. The ac field strength was scanned to collect spectra which are analogous to dc electric-field-induced resonant energy transfer spectra. Different resonances were observed for different magnetic sublevels involved in the process. Compared to earlier work performed at higher frequencies, the choice of dressing frequency and structure of the spectra may be intuitively understood, by analogy with the dc field case.
△ Less
Submitted 11 September, 2009;
originally announced September 2009.
-
Enhancement of Rydberg atom interactions using ac Stark shifts
Authors:
P. Bohlouli-Zanjani,
J. A. Petrus,
J. D. D. Martin
Abstract:
The ac Stark effect was used to induce resonant energy transfer between translationally cold Rydberg atoms. The Rb Rydberg atoms were obtained by laser excitation of cold atoms from a magneto-optical trap. Using two-photon microwave spectroscopy of the 49s_{1/2}-50s_{1/2} transition, the field strength of a 28.5 GHz dressing field was calibrated. When this dressing field was set at specific fiel…
▽ More
The ac Stark effect was used to induce resonant energy transfer between translationally cold Rydberg atoms. The Rb Rydberg atoms were obtained by laser excitation of cold atoms from a magneto-optical trap. Using two-photon microwave spectroscopy of the 49s_{1/2}-50s_{1/2} transition, the field strength of a 28.5 GHz dressing field was calibrated. When this dressing field was set at specific field strengths, the two-atom dipole-dipole process 43d_{5/2} + 43d_{5/2} -> 45p_{3/2} + 41f was dramatically enhanced, due to induced degeneracy of the initial and final states. A series of observed resonant field strengths correspond to different magnetic sublevel possibilities for the initial and final states. These agree well with calculated resonance fields based on a perturbative ac Stark shift formula. This method for enhancing interactions is complementary to dc electric-field-induced resonant energy transfer, but has more flexibility due to the possibility of varying the applied frequency.
△ Less
Submitted 23 December, 2006;
originally announced December 2006.
-
Determination of the Rb ng-series quantum defect by electric-field-induced resonant energy transfer between cold Rydberg atoms
Authors:
K. Afrousheh,
P. Bohlouli-Zanjani,
J. A. Petrus,
J. D. D. Martin
Abstract:
Resonant energy transfer between cold Rydberg atoms was used to determine Rydberg atom energy levels, at precisions approaching those obtainable in microwave spectroscopy. Laser cooled Rb atoms from a magneto-optical trap were optically excited to 32d Rydberg states. The two-atom process 32d(j=5/2) + 32d(j=5/2) -> 34p(j=3/2) + 30g is resonant at an electric field of approximately 0.3 V/cm. This…
▽ More
Resonant energy transfer between cold Rydberg atoms was used to determine Rydberg atom energy levels, at precisions approaching those obtainable in microwave spectroscopy. Laser cooled Rb atoms from a magneto-optical trap were optically excited to 32d Rydberg states. The two-atom process 32d(j=5/2) + 32d(j=5/2) -> 34p(j=3/2) + 30g is resonant at an electric field of approximately 0.3 V/cm. This process is driven by the electric dipole-dipole interaction, which is allowed due to the partial f character that the g state acquires in an electric field. The experimentally observed resonant field, together with the Stark map calculation is used to make a determination of the Rb ng-series quantum defect: delta_g (n=30) = 0.00405(6).
△ Less
Submitted 19 December, 2006; v1 submitted 18 August, 2006;
originally announced August 2006.
-
Optical transfer cavity stabilization using current-modulated injection-locked diode lasers
Authors:
P. Bohlouli-Zanjani,
K. Afrousheh,
J. D. D. Martin
Abstract:
It is demonstrated that RF current modulation of a frequency stabilized injection-locked diode laser allows the stabilization of an optical cavity to adjustable lengths, by variation of the RF frequency. This transfer cavity may be used to stabilize another laser at an arbitrary wavelength, in the absence of atomic or molecular transitions suitable for stabilization. Implementation involves equi…
▽ More
It is demonstrated that RF current modulation of a frequency stabilized injection-locked diode laser allows the stabilization of an optical cavity to adjustable lengths, by variation of the RF frequency. This transfer cavity may be used to stabilize another laser at an arbitrary wavelength, in the absence of atomic or molecular transitions suitable for stabilization. Implementation involves equipment and techniques commonly used in laser cooling and trap** laboratories, and does not require electro- or acousto-optic modulators. With this technique we stabilize a transfer cavity using a RF current-modulated diode laser which is injection locked to a 780 nm reference diode laser. The reference laser is stabilized using polarization spectroscopy in a Rb cell. A Ti:sapphire ring laser at 960 nm is locked to this transfer cavity and may be precisely scanned by varying the RF modulation frequency. We demonstrate the suitability of this system for the excitation of laser cooled Rb atoms to Rydberg states.
△ Less
Submitted 7 June, 2006;
originally announced June 2006.
-
Resonant electric dipole-dipole interactions between cold Rydberg atoms in a magnetic field
Authors:
K. Afrousheh,
P. Bohlouli-Zanjani,
J. D. Carter,
A. Mugford,
J. D. D. Martin
Abstract:
Laser cooled Rb atoms were optically excited to 46d Rydberg states. A microwave pulse transferred a fraction of the atoms to the 47p Rydberg state. The resonant electric dipole-dipole interactions between atoms in these two states were probed using the linewidth of the two-photon microwave transition 46d-47d. The presence of a weak magnetic field (approximately 0.5 G) reduced the observed line b…
▽ More
Laser cooled Rb atoms were optically excited to 46d Rydberg states. A microwave pulse transferred a fraction of the atoms to the 47p Rydberg state. The resonant electric dipole-dipole interactions between atoms in these two states were probed using the linewidth of the two-photon microwave transition 46d-47d. The presence of a weak magnetic field (approximately 0.5 G) reduced the observed line broadening, indicating that the interaction is suppressed by the field. The field removes some of the energy degeneracies responsible for the resonant interaction, and this is the basis for a quantitative model of the resulting suppression. A technique for the calibration of magnetic field strengths using the 34s-34p one-photon transition is also presented.
△ Less
Submitted 21 April, 2006;
originally announced April 2006.
-
Spectroscopic observation of resonant electric dipole-dipole interactions between cold Rydberg atoms
Authors:
K. Afrousheh,
P. Z. Bohlouli,
D. Vagale,
A. Mugford,
M. Fedorov,
J. D. D. Martin
Abstract:
Resonant electric dipole-dipole interactions between cold Rydberg atoms were observed using microwave spectroscopy. Laser-cooled Rb atoms in a magneto-optical trap were optically excited to 45d Rydberg states using a pulsed laser. A microwave pulse transferred a fraction of these Rydberg atoms to the 46p state. A second microwave pulse then drove atoms in the 45d state to the 46d state, and was…
▽ More
Resonant electric dipole-dipole interactions between cold Rydberg atoms were observed using microwave spectroscopy. Laser-cooled Rb atoms in a magneto-optical trap were optically excited to 45d Rydberg states using a pulsed laser. A microwave pulse transferred a fraction of these Rydberg atoms to the 46p state. A second microwave pulse then drove atoms in the 45d state to the 46d state, and was used as a probe of interatomic interactions. The spectral width of this two-photon probe transition was found to depend on the presence of the 46p atoms, and is due to the resonant electric dipole-dipole interaction between 45d and 46p Rydberg atoms.
△ Less
Submitted 7 November, 2004;
originally announced November 2004.
