You are a robot and you live in a Markov decision process (MDP) with a finite or an infinite number of transitions from state-action to next states. You got brains and so you plan before you act. Luckily, your roboparents equipped you with a generative model to do some Monte-Carlo planning. The world is waiting for you and you have no time to waste. You want your planning to be efficient. Sample-efficient. Indeed, you want to exploit the possible structure of the MDP by exploring only a subset of states reachable by following near-optimal policies. You want guarantees on sample complexity that depend on a measure of the quantity of near-optimal states. You want something, that is an extension of Monte-Carlo sampling (for estimating an expectation) to problems that alternate maximization (over actions) and expectation (over next states). But you do not want to StOP with exponential running time, you want something simple to implement and computationally efficient. You want it all and you want it now. You want TrailBlazer.
Twisted bilayer photonic crystals introduce a twist between two stacked photonic crystal slabs, enabling strong modulation of their electromagnetic properties. The change in the twist angle strongly influences the resonant frequencies and available propagating diffraction orders with applications including sensing, lasing, slow light or wavefront engineering. In this work, we design and analyze twisted bilayer crystals capable of steering light in a direction controlled by the twist angle. In order to achieve beam steering, the device efficiently routes input power into a single, twist-dependent, transmitted diffraction order. The outgoing light then follows the orientation of this diffraction order, externally controlled by the twist angle. The optimization is performed using high-efficiency heuristic optimization method which enabled a data-oriented approach to further understand the design operation. The optimized device demonstrates an efficiency above 90% across twist angles ranging from 0 to 30 degrees for both TE and TM polarizations. Extending the optimization to include left- and right-handed polarizations yields overall accuracy nearing 90% when averaged across the entire
The Quantum Approximate Optimization Algorithm (QAOA) is one of the most promising candidates for achieving quantum advantage over classical computers. However, existing compilers lack specialized methods for optimizing QAOA circuits. There are circuit patterns inside the QAOA circuits, and current quantum hardware has specific qubit connectivity topologies. Therefore, we propose Coqa to optimize QAOA circuit compilation tailored to different types of quantum hardware. Our method integrates a linear nearest-neighbor (LNN) topology and efficiently map the patterns of QAOA circuits to the LNN topology by heuristically checking the interaction based on the weight of problem Hamiltonian. This approach allows us to reduce the number of SWAP gates during compilation, which directly impacts the circuit depth and overall fidelity of the quantum computation. By leveraging the inherent patterns in QAOA circuits, our approach achieves more efficient compilation compared to general-purpose compilers. With our proposed method, we are able to achieve an average of 30% reduction in gate count and a 39x acceleration in compilation time across our benchmarks.
In recent years, computer-aided automatic polyp segmentation and neoplasm detection have been an emerging topic in medical image analysis, providing valuable support to colonoscopy procedures. Attentions have been paid to improving the accuracy of polyp detection and segmentation. However, not much focus has been given to latency and throughput for performing these tasks on dedicated devices, which can be crucial for practical applications. This paper introduces a novel deep neural network architecture called BlazeNeo, for the task of polyp segmentation and neoplasm detection with an emphasis on compactness and speed while maintaining high accuracy. The model leverages the highly efficient HarDNet backbone alongside lightweight Receptive Field Blocks for computational efficiency, and an auxiliary training mechanism to take full advantage of the training data for the segmentation quality. Our experiments on a challenging dataset show that BlazeNeo achieves improvements in latency and model size while maintaining comparable accuracy against state-of-the-art methods. When deploying on the Jetson AGX Xavier edge device in INT8 precision, our BlazeNeo achieves over 155 fps while yieldin
Old and recent puzzles of GRBs and SGRs find a solution with a model based on the fast blazing of very collimated thin gamma Jets. Damped oscillating afterglows in GRB030329 find a natural explanation assuming a very thin Jet whose persistent activity and different angle of view maybe combined at once with the Supernovae power and the apparent huge GRBs output. The same thin beaming offer an understanding of the apparent SGR-Pulsar power connection. A thin collimated precessing Gamma Jet model for both GRBs and SGRs, at their different scaled luminosity (10^{44} - 10^{38}erg s^-1), explains the existence of few identical energy spectra and time evolution of these sources leading to a unified model. Their similarity with the huge precessing Jets in AGN, QSRs and Radio-Galaxies inspires this smaller scale SGR-GRB model. The spinning-precessing Jet explains the rare mysterious X-Ray precursors in GRBs and SGRs events. Any large Gamma Jet off-axis beaming to the observer might lead to the X-Flash events without any GRB signals, as the most recent XRF030723. Its possible re-brightening would confirm the evidence of the variable pointing of the jet in or off line towards the observer. In
Machine learning tools have illustrated their potential in many significant sectors such as healthcare and finance, to aide in deriving useful inferences. The sensitive and confidential nature of the data, in such sectors, raise natural concerns for the privacy of data. This motivated the area of Privacy-preserving Machine Learning (PPML) where privacy of the data is guaranteed. Typically, ML techniques require large computing power, which leads clients with limited infrastructure to rely on the method of Secure Outsourced Computation (SOC). In SOC setting, the computation is outsourced to a set of specialized and powerful cloud servers and the service is availed on a pay-per-use basis. In this work, we explore PPML techniques in the SOC setting for widely used ML algorithms-- Linear Regression, Logistic Regression, and Neural Networks. We propose BLAZE, a blazing fast PPML framework in the three server setting tolerating one malicious corruption over a ring (\Z{\ell}). BLAZE achieves the stronger security guarantee of fairness (all honest servers get the output whenever the corrupt server obtains the same). Leveraging an input-independent preprocessing phase, BLAZE has a fast inpu
The apparently huge energy budget of the gamma ray burst GRB 990123 led to the final collapse of the isotropic fireball model, forcing even the most skeptical to consider a beamed Jet emission correlated to a supernova (SN) explosion. Similarly the surprising giant flare from the soft gamma repeater SGR 1806-20 that occurred on 2004 December 27, may induce the crisis of the magnetar model. If the apparently huge energy associated to this flare has been radiated isotropically, the magnetar should have consumed at once most of (if not all) the energy stored in the magnetic field. On the contrary we think that a thin collimated precessing jet, blazing on-axis, may be the source of such apparently huge bursts with a moderate output power. We discuss the possible role of the synchrotron emission and electromagnetic showering of PeV electron pairs from muon bundles. A jet made of muons may play a key role in avoiding the opacity of the SN-GRB radiation field. We propose a similar mechanism to explain the emission of SGRs. In this case we also examine the possibility of a primary hadronic jet that would produce ultra relativistic e+ e- (1 - 10 PeV) from pion- muon or neutron decay. Such e
Gamma Ray Burst and Soft Gamma Repeaters are neither standard candle nor isotropic explosions. Our model explain them as strong blazing of a light-house, spinning and precessing gamma jet. Such jets at maximal output (as GRBs in Supernova like sources at cosmic edges) or at late lower power stages (SGRs in nearer planetary nebulae in galactic halo) may blaze the observer by extreme beaming $(Ω< 10^{-8})$ and apparent huge luminosity. \keywords{GRB, Jet, Inverse Compton, SGR}
NASA’s Swift Observatory observed a supermassive black hole ripping apart a star more than 30,000 light-years from the center of a distant galaxy。 The extraordinary flare briefly outshone its entire host galaxy in ultraviolet light and revealed a black hole about a million times the Sun’s mass
High Energy Cosmic Rays (C.R.) versus Neutrino and Neutralino induced Air-Shower maybe tested at Horizons by their muons, gamma and Cerenkov blazing signals. Inclined and Horizontal C.R. Showers (70-90 zenith angle) produce secondary (gamma, e+, e-) mostly suppressed by high column atmosphere depth. Earliest shower Cherenkov photons are diluted by large distances and by air opacity, while secondary penetrating muons and their successive decay into electrons and gamma, may revive additional Cerenkov lights. GeVs gamma telescopes at the top of the mountains or in Space may detect at horizons PeVs up to EeV C.R. and their secondaries. Details on arrival angle and column depth, shower shape, timing signature of photon flash intensity, may inform us on the altitude interaction and primary UHECR composition. Below the horizons, at zenith angle among copious single albedo muons, rare up-going showers traced by muon bundles would give evidence of rare tau Earth-Skimming neutrinos, at EeVs energies. Their rate may be comparable with 6.3 PeVs anti-neutrino electron induced air-shower (mostly hadronic) originated above and also below horizons, in interposed atmosphere by W resonance at Glasho
We discuss the dynamics of microquasar jets in the interstellar medium, with specific focus on the effects of the X-ray binaries' space velocity with respect to the local Galactic standard of rest. We argue that, during late stages in the evolution of large scale radio nebulae around microquasars, the ram pressure of the interstellar medium due to the microquasar's space velocity becomes important and that microquasars with high velocities form the Galactic equivalent of extragalactic head-tail sources, i.e., that they leave behind trails of stripped radio plasma. Because of their higher space velocities, low-mass X-ray binaries are more likely to leave trails than high-mass X-ray binaries. We show that the volume of radio plasma released by microquasars over the history of the Galaxy is comparable to the disk volume and argue that a fraction of a few percent of the radio plasma left behind by the X-ray binary is likely mixed with the neutral phases of the ISM before the plasma is removed from the disk by buoyancy. Because the formation of microquasars is an unavoidable by-product of star formation, and because they can travel far from their birth places, their activity likely has
Correlation functions are widely used in extra-galactic astrophysics to extract insights into how galaxies occupy dark matter halos and in cosmology to place stringent constraints on cosmological parameters. A correlation function fundamentally requires computing pair-wise separations between two sets of points and then computing a histogram of the separations. Corrfunc is an existing open-source, high-performance software package for efficiently computing a multitude of correlation functions. In this paper, we will discuss the SIMD AVX512F kernels within Corrfunc, capable of processing 16 floats or 8 doubles at a time. The latest manually implemented Corrfunc AVX512F kernels show a speedup of up to $\sim 4\times$ relative to compiler-generated code for double-precision calculations. The AVX512F kernels show $\sim 1.6\times$ speedup relative to the AVX kernels and compare favorably to a theoretical maximum of $2\times$. In addition, by pruning pairs with too large of a minimum possible separation, we achieve a $\sim 5-10\%$ speedup across all the SIMD kernels. Such speedups highlight the importance of programming explicitly with SIMD vector intrinsics for complex calculations that
A list of questions regarding Gamma Ray Bursts (GRBs) and Soft Gamma Repeaters (SGRs) remain unanswered within the Fireball-cone and Magnetar explosive scenarios. A persistent, thin (less than micron-sr solid angle) precessing and spinning gamma jet, with a power output comparable to the progenitor supernova (SN) or XRay pulsar, may explain these issues. The precessing jets may have a few hours characteristic decay time, while their decreasing intensity follow a power law for thousands of years. The orientation of the spinning and precessing beam respect to the line of sight plays a key role : the farthest GRB events in widest cosmic volumes correspond generally to a very narrow and on-axis beam, while for the nearest ones sources are mostly observable off-axis. Consequentely the far ones are the hardest and the most bright and viceversa nearest one are mostly softer and longest ones as in Amati correlation. We expect that nearby off-axis GRBs would be accompanied by a chain of OT and radio bumps as in GRB030329-SN2003 and latest GRB060218-SN2006 events. Delayed blazing jets are observable in the local universe as X-Ray Flahes (XRFs) or short GRBs often as orphan afterglow event, a
GeV beams of light ions and electrons are used for creating a high flux of real and virtual photons, with which some problems in Nuclear Astrophysics are studied. GeV 8B beams are used to study the Coulomb dissociation of 8B and thus the 7Be(p,gamma)8B reaction. This reaction is one of the major source of uncertainties in estimating the 8B solar neutrino flux and a critical input for calculating the 8B Solar neutrino flux. The Coulomb dissociation of 8B appears to provide a viable method for measuring the 7Be(p,gamma)8B reaction rate, with a weighted average of the RIKEN1, RIKEN2, GSI1 and MSU published results of S17(0) = 18.9 +/- 1.0 eV-b. This result however does not include a theoretical error estimated to be +/- 10 %. GeV electron beams on the other hand, are used to create a high flux of real and virtual photons at TUNL-HIGS and MIT-Bates, respectively, and we discuss two new proposals to study the 12C(alpha,gamma)16O reaction with real and virtual photons. The 12C(alpha,gamma)16O reaction is essential for understanding Type II and Type Ia supernova. It is concluded that virtual and real photons produced by GeV light ions and electron beams are useful for studying some proble
We present the generalization of our FEM-based topology optimization framework to 3D blazed metasurfaces operating in reflection over the visible and near-infrared range [400-1,500]nm. The design region is described through a density-based SIMP interpolation and optimized using the adjoint method, enabling the treatment of several tens of thousands degrees of freedom. A first approach directly applies topology optimization to the 3D Finite Element mesh (mesh-based), yielding a freeform structure that achieves an average diffraction efficiency of 62% in order -1 over two octaves under the targeted incidence. However, such patterns remain difficult to manufacture. We therefore introduce a pillar-based parameterization, embedding fabrication constraints within the optimization loop. The resulting binary metasurface, compatible with e-beam lithography and Reactive Ion Etching techniques, achieves an average efficiency of 57% over the same spectral band in s-polarization, with low polarization dependence. This work demonstrates that large-scale 3D topology optimization can bridge the gap between broadband optical performance and realistic nanofabrication constraints for blazed metasurfa
This article provides a pedagogical introduction to the Silver Blaze problem. This problem refers to the difficulty of reconciling to perspectives on QCD with a chemical potential. The first is the phenomenological fact that at $T=0$ QCD remains in its ground state -- the vacuum -- with all physical observables unchanged whenever the magnitude of a chemical potential is less than some critical value. The second is the fact that in functional integral treatments, the inclusion of any nonzero chemical potential changes all eigenvalues of the Dirac operator for every gauge configuration, leading to a natural expectation that the functional determinants also changes, which leads to the expectation that physical observables should be altered. The problem amounts to explaining why nothing happens below the critical chemical potential. By focusing on the eigenvalues of $γ_0$ times the Dirac operator rather than the Dirac operator itself, it is possible to show that for QCD with two flavors and identical quark masses, an isospin chemical potential with a magnitude less than $m_π$ (and no baryon chemical potential), or a baryon chemical potential of less than $\frac{3}{2} m_π$ (and no isosp
This study explores the use of synchrotron measurements as a nanometrology tool for blazed gratings. In grazing incidence geometry, one can measure both the conical diffraction and the diffuse scattering on the grating simultaneously in a single scattering pattern. The sensitivity of scattering patterns to the structure of the blazed gratings is evaluated. The diffraction component of the pattern is shown to be sensitive to the average groove profile of the gratings. Meanwhile, the diffuse scattering depends on the roughness morphology of the reflective surface of blazed gratings. These findings are supported by numerical simulations. The simulations were performed using several rigorous solvers for the Helmholtz equations, and with a perturbation theory. The analysis relies on synchrotron data, as well as data from atomic force microscopy and scanning electron microscopy. The aim of this article is to draw the attention of the optical community to the synchrotron measurements as a nanometrology tool for the modern optical elements.
Maintaining the highest quality and output of photon science in the VUV-, EUV-, soft- and tender-X-ray energy ranges requires high-quality blazed profile gratings. Currently, their availability is critical due to technological challenges and limited manufacturing resources. In this work we discuss the opportunity of an alternative method to manufacture blazed gratings by means of electron-beam lithography (EBL). We investigate the different parameters influencing the optical performance of blazed profile gratings produced by EBL and develop a robust process for the manufacturing of high-quality blazed gratings using polymethyl methacrylate (PMMA) as high resolution, positive tone resist and ion beam etching.
JSON Schemas provide useful guardrails for developers of Web APIs to guarantee that the semi-structured JSON input provided by clients matches a predefined structure. This is important both to ensure the correctness of the data received as input and also to avoid potential security issues from processing input that is not correctly validated. However, this validation process can be time-consuming and adds overhead to every request. Different keywords in the JSON Schema specification have complex interactions that may increase validation time. Since popular APIs may process thousands of requests per second and schemas change infrequently, we observe that we can resolve some of the complexity ahead of time in order to achieve faster validation. Our JSON Schema validator, Blaze, compiles complex schemas to an efficient representation in seconds to minutes, adding minimal overhead at build time. Blaze incorporates several unique optimizations to reduce the validation time by an average of approximately 10x compared existing validators on a variety of datasets. In some cases, Blaze achieves a reduction in validation time of multiple orders of magnitude compared to the next fastest valid
Software bugs require developers to exert significant effort to identify and resolve them, often consuming about one-third of their time. Bug localization, the process of pinpointing the exact source code files that need modification, is crucial in reducing this effort. Existing bug localization tools, typically reliant on deep learning techniques, face limitations in cross-project applicability and effectiveness in multi-language environments. Recent advancements with Large Language Models (LLMs) offer detailed representations for bug localization. However, they encounter challenges with limited context windows and mapping accuracy. To address these issues, we propose BLAZE, an approach that employs dynamic chunking and hard example learning. First, BLAZE dynamically segments source code to minimize continuity loss. Then, BLAZE fine-tunes a GPT-based model using challenging bug cases, in order to enhance cross-project and cross-language bug localization. To support the capability of BLAZE, we create the BEETLEBOX dataset, which comprises 26,321 bugs from 29 large and thriving open-source projects across five different programming languages (Java, C++, Python, Go, and JavaScript).