Search

Motivated by the success of Transformers when applied to sequences of discrete symbols, token-based world models (TBWMs) were recently proposed as sample-efficient methods. In TBWMs, the world model consumes agent experience as a language-like sequence of tokens, where each observation constitutes a sub-sequence. However, during imagination, the sequential token-by-token generation of next observations results in a severe bottleneck, leading to long training times, poor GPU utilization, and limited representations. To resolve this bottleneck, we devise a novel Parallel Observation Prediction (POP) mechanism. POP augments a Retentive Network (RetNet) with a novel forward mode tailored to our reinforcement learning setting. We incorporate POP in a novel TBWM agent named REM (Retentive Environment Model), showcasing a 15.4x faster imagination compared to prior TBWMs. REM attains superhuman performance on 12 out of 26 games of the Atari 100K benchmark, while training in less than 12 hours. Our code is available at \url{https://github.com/leor-c/REM}.

The spatial mode is an essential component of an electromagnetic field description, yet it is challenging to characterize it for optical fields with low average photon number, such as in a squeezed vacuum. We present a method for reconstruction of the spatial modes of such fields based on the homodyne measurements of their quadrature noise variance performed with a set of structured masks. We show theoretically that under certain conditions we can recover individual spatial mode distributions by using the weighted sum of the basis masks, where weights are determined using measured variance values and phases. We apply this approach to analyze the spatial structure of a squeezed vacuum field with various amount of excess thermal noise generated in Rb vapor., Comment: The following article has been accepted by AVS Quantum Science

We present an innovative concept for quantum-enhanced multiparameter optical phase sensing that can be implemented in free space, optical fiber or on-chip. Our measurable phases are in series, or cascaded, enabling measurements as a function of position with only a single input and output. We have modeled up to 20 phases, and fitting shows near-linear scaling of the power requirements for additional phases. This novel approach represents a new paradigm in multiparameter quantum metrology, and can be applied to remote sensing, communications, and geophysics., Comment: 5 pages, 4 figures. Comments are welcome

Atmospheric effects, such as turbulence and background thermal noise, inhibit the propagation of coherent light used in ON-OFF keying free-space optical communication. Here we present and experimentally validate a convolutional neural network to reduce the bit error rate of free-space optical communication in post-processing that is significantly simpler and cheaper than existing solutions based on advanced optics. Our approach consists of two neural networks, the first determining the presence of coherent bit sequences in thermal noise and turbulence and the second demodulating the coherent bit sequences. All data used for training and testing our network is obtained experimentally by generating ON-OFF keying bit streams of coherent light, combining these with thermal light, and passing the resultant light through a turbulent water tank which we have verified mimics turbulence in the air to a high degree of accuracy. Our convolutional neural network improves detection accuracy over threshold classification schemes and has the capability to be integrated with current demodulation and error correction schemes.

We combine single-pixel imaging and homodyne detection to perform full object recovery (phase and amplitude). Our method does not require any prior information about the object or the illuminating fields. As a demonstration, we reconstruct the optical properties of several semi-transparent objects and find that the reconstructed complex transmission has a phase precision of 0.02 radians and a relative amplitude precision of 0.01.

The unification of general relativity and quantum theory is one of the fascinating problems of modern physics. One leading solution is Loop Quantum Gravity (LQG). Simulating LQG may be important for providing predictions which can then be tested experimentally. However, such complex quantum simulations cannot run efficiently on classical computers, and quantum computers or simulators are needed. Here, we experimentally demonstrate quantum simulations of spinfoam amplitudes of LQG on an integrated photonics quantum processor. We simulate a basic transition of LQG and show that the derived spinfoam vertex amplitude falls within 4% error with respect to the theoretical prediction, despite experimental imperfections. We also discuss how to generalize the simulation for more complex transitions, in realistic experimental conditions, which will eventually lead to a quantum advantage demonstration as well as expand the toolbox to investigate LQG., Comment: 5 pages, 4 figures. comments are welcome

In this work, we theoretically and experimentally demonstrate the possibility to create an image of an opaque object using a few-photon thermal optical field. We utilize the Quadrature-Noise Shadow Imaging (QSI) technique that detects the changes in the quadrature-noise statistics of the probe beam after its interaction with an object. We show that such thermal QSI scheme has an advantage over the classical differential imaging when the effect of dark counts is considered. At the same time, the easy availability of thermal sources for any wavelength makes the method practical for broad range of applications, not accessible with, e.g. quantum squeezed light. As a proof of principle, we implement this scheme by two different light sources: a pseudo-thermal beam generated by rotating ground glass (RGG) method and a thermal beam generated by Four-Wave Mixing (FWM) method. The RGG method shows simplicity and robustness of QSI scheme while the FWM method validates theoretical signal-to-noise ratio predictions. Finally, we demonstrate low-light imaging abilities with QSI by imaging a biological specimen on a CCD camera, detecting as low as 0.03 photons on average per pixel per 1.7 microseconds exposure., Comment: 9 pages, 7 figures

Quantum information theory sets the ultimate limits for any information-processing task. In rangefinding and LIDAR, the presence or absence of a target can be tested by detecting different states at the receiver. In this Letter, we use quantum hypothesis testing for an unknown coherent-state return signal in order to derive the limits of symmetric and asymmetric error probabilities of single-shot ranging experiments. We engineer a single measurement independent of the range, which in some cases saturates the quantum bound and for others is presumably the best measurement to approach it. In addition, we verify the theoretical predictions by performing numerical simulations. This work bridges the gap between quantum information and quantum sensing and engineering and will contribute to devising better ranging sensors, as well as setting the path for finding practical limits for other quantum tasks., Comment: 7 pages, 7 figures. comments are welcome

Quantum entanglement is a crucial resource for a wide variety of quantum technologies. However, the current state-of-art methods to generate quantum entanglement in optomechanical systems are not as efficient as all-optical methods utilizing nonlinear crystals. This article proposes a new scheme in which two single-mode squeezed light fields are injected into an optomechanical cavity. We demonstrate through our numerical simulations that the quantum entanglement can be substantially enhanced with the careful selection of squeezing strength and squeezing angle of the two quadrature squeezed light fields. Our results represent a significant improvement in output bipartite photon-photon entanglement over the previously demonstrated schemes using two coherent light fields as inputs. These simulations predict a maximum increase in bipartite optical entanglement by a factor of about 6, as well as increases in the quantum noise of the output light. A perceived loss of quantum information at certain squeezing angles is attributed to tripartite entanglement between the two optical fields and the optomechanical oscillator (OMO). At particular squeezing angles, the bipartite (or tripartite) entanglement can be increased, thus introducing a method of optically controlling the intracavity entanglement. These mechanics can benefit various optical quantum technologies utilizing optomechanical entanglement and continuous variable quantum optics.

We experimentally demonstrate an imaging technique based on quantum noise modification after interaction with an opaque object. By using a homodyne-like detection scheme, we eliminate the detrimental effect of the camera's dark noise, making this approach particularly attractive for imaging scenarios that require weak illumination. Here, we reconstruct the image of an object illuminated with a squeezed vacuum using a total of 800 photons, utilizing less than one photon per frame on average.

Quantum entanglement is the key to many applications like quantum key distribution, quantum teleportation, and quantum sensing. However, reliably generating quantum entanglement in macroscopic systems has proved to be a challenge. Here, we present a detailed analysis of ponderomotive entanglement generation which utilizes optomechanical interactions to create quantum correlations. We numerically calculate an entanglement measure -- the logarithmic negativity -- for the quantitative assessment of the entanglement. Experimental limitations, including thermal noise and optical loss, from measurements of an existing experiment were included in the calculation, which is intractable to solve analytically. This work will play an important role in the development of ponderomotive entanglement devices.

We present a technique for squeezed light detection based on direct imaging of the displaced-squeezed-vacuum state using a CCD camera. We show that the squeezing parameter can be accurately estimated using only the first two moments of the recorded pixel-to-pixel photon fluctuation statistics, with accuracy that rivals that of the standard squeezing detection methods such as a balanced homodyne detection. Finally, we numerically simulate the camera operation, reproducing the noisy experimental results with low signal samplings and confirming the theory with high signal samplings., Comment: 5 pages, 4 figures, supplemental information included, comments are very welcome

The problem of simulating complex quantum processes on classical computers gave rise to the field of quantum simulations. Quantum simulators solve problems, such as Boson sampling, where classical counterparts fail. In another field of physics, the unification of general relativity and quantum theory is one of the greatest challenges of our time. One leading approach is Loop Quantum Gravity (LQG). Here, we connect these two fields and design a linear-optical simulator such that the evolution of the optical quantum gates simulates the spinfoam amplitudes of LQG. It has been shown that computing transition amplitudes in simple quantum field theories falls into the class BQP -- which strongly suggests that computing transition amplitudes of LQG are classically intractable. Therefore, these amplitudes are efficiently computable with universal quantum computers which are, alas, possibly decades away. We propose here an alternative special-purpose linear-optical quantum computer, which can be implemented using current technologies. This machine is capable of efficiently computing these quantities. This work opens a new way to relate quantum gravity to quantum information and will expand our understanding of the theory., Comment: 5 pages, 3 figures, comments are welcome

We present a technique that improves the signal-to-noise-ratio (SNR) of range-finding, sensing, and other light-detection applications. The technique filters out low photon numbers using photon-number-resolving detectors (PNRDs). This technique has no classical analog and cannot be done with classical detectors. We investigate the properties of our technique and show under what conditions the scheme surpasses the classical SNR. Finally, we simulate the operation of a rangefinder, showing improvement with a low number of signal samplings and confirming the theory with a high number of signal samplings., Comment: 5 pages, 4 figures, comments are welcome

A single photon has many physical degrees of freedom (DOF) that can carry the state of a high-dimensional quantum system. Nevertheless, only a single DOF is usually used in any specific demonstration. Furthermore, when more DOF are being used, they are analyzed and measured one at a time. We introduce a two-qubit information system, realized by two degrees of freedom of a single photon: polarization and time. The photon arrival time is divided into two time-bins representing a qubit, while its polarization state represents a second qubit. The time difference between the two time-bins is created without an interferometer at the picosecond scale, which is much smaller than the detector's response time. The two physically different DOF are analyzed simultaneously by photon bunching between the analyzed photon and an ancilla photon. Full two-qubit states encoded in single photons were reconstructed using quantum state tomography, both when the two DOF were entangled and when they were not, with fidelities higher than 96%.

LiDAR (laser based radar) systems are a major part of many new real-world interactive systems, one of the most notable being autonomous cars. The current market LiDAR systems are limited by detector sensitivity: when output power is at eye-safe levels, the range is limited. Long range operation also slows image acquisition as flight-time increases. We present an approach that combines a high sensitivity photon number resolving diode with machine learning and a micro-mechanical digital mirror device to achieve safe and fast long range 3D scanning., Comment: 7 pages, 14 figures, SPIE Photonics west 2018

Single-photon detectors are widely used in modern quantum optics experiments and applications. Like all detectors, it is important for these devices to be accurately calibrated. A single-photon detector is calibrated by determining its detection efficiency; the standard method to measure this quantity requires comparison to another detector. Here, we suggest a method to measure the detection efficiency of a single photon detector without requiring an external reference detector. Our method is valid for individual single-photon detectors as well as multiplexed detectors, which are known to be photon number resolving. The method exploits the photon-number correlations of a nonlinear source, as well as the nonlinear loss of a single photon detector that occurs when multiple photons are detected simultaneously. We have analytically modeled multiplexed detectors and used the results to experimentally demonstrate calibration of a single photon detector without the need for an external reference detector., Comment: 3 figures, 2 tables, comments are welcome

Entangled coherent states are shown to emerge, with high fidelity, when mixing coherent and squeezed vacuum states of light on a beam-splitter. These maximally entangled states, where photons bunch at the exit of a beamsplitter, are measured experimentally by Fock-state projections. Entanglement is examined theoretically using a Bell-type nonlocality test and compared with ideal entangled coherent states. We experimentally show nearly perfect similarity with entangled coherent states for an optimal ratio of coherent and squeezed vacuum light. In our scheme, entangled coherent states are generated deterministically with small amplitudes, which could be beneficial, for example, in deterministic distribution of entanglement over long distances., Comment: 6 pages, 6 figures, comments are welcome