Photon Acceleration Research Articles

Neutron radiation effects on microcomputers in radiation therapy environments

AbstractAim:Microcomputers play an increasingly important role in the delivery of radiation therapy. Exposure to neutron irradiation can produce undesirable effects in modern microcomputers. The objective of this study is to measure acute and cumulative effects of neutron exposure of Intel-based microcomputers in photon and proton therapy treatment environments.Materials and methods:Multiple computers were irradiated with neutrons produced from MEVION S250 passive scattering proton therapy and from Varian 21EX Linear Accelerator photon therapy systems. The energy of the proton beam was 232 MeV and the photon beam energies were 6 and 18 MV. Rates of fatal errors in computer processing unit (CPU) cores were measured.Results:Varying rates of fatal system errors due to upsets in the CPU cores were observed. Post-exposure routine stress testing revealed no permanent hardware defects in the random access memory (RAM) or hard disk drive (HDD) of any tested systems. Microchip manufacturers fit increasingly high numbers of transistors in the same volume and the susceptibility to radiation thus increases.Conclusions:This work explores if the process size of a microchip is the dominant factor and also looked at the short- and long-term effects of neutron irradiation on modern microprocessors in a clinical environment. Additionally, methods of effective shielding are proposed.

Initializing photonic feed-forward neural networks using auxiliary tasks

Photonics is among the most promising emerging technologies for providing fast and energy-efficient Deep Learning (DL) implementations. Despite their advantages, these photonic DL accelerators also come with certain important limitations. For example, the majority of existing photonic accelerators do not currently support many of the activation functions that are commonly used in DL, such as the ReLU activation function. Instead, sinusoidal and sigmoidal nonlinearities are usually employed, rendering the training process unstable and difficult to tune, mainly due to vanishing gradient phenomena. Thus, photonic DL models usually require carefully fine-tuning all their training hyper-parameters in order to ensure that the training process will proceed smoothly. Despite the recent advances in initialization schemes, as well as in optimization algorithms, training photonic DL models is still especially challenging. To overcome these limitations, we propose a novel adaptive initialization method that employs auxiliary tasks to estimate the optimal initialization variance for each layer of a network. The effectiveness of the proposed approach is demonstrated using two different datasets, as well as two recently proposed photonic activation functions and three different initialization methods. Apart from significantly increasing the stability of the training process, the proposed method can be directly used with any photonic activation function, without further requiring any other kind of fine-tuning, as also demonstrated through the conducted experiments.

CLEAR: A Holistic Figure-of-Merit for Post- and Predicting Electronic and Photonic-based Compute-system Evolution

Continuing demands for increased computing efficiency and communication bandwidth have pushed the current semiconductor technology to its limit. This led to novel technologies with the potential to outperform conventional electronic solutions such as photonic pre-processors or accelerators, electronic-photonic hybrid circuits, and neural networks. However, the efforts made to describe and predict the performance evolution of compute-performance fall short to accurately predict and thereby explain the actually observed development pace with time; that is all proposed metrics eventually deviate from their development trajectory after several years from when they were originally proposed. This discrepancy demands a figure-of-merit that includes a holistic set of driving forces of the compute-system evolution. Here we introduce the Capability-to-Latency-Energy-Amount-Resistance (CLEAR) metric encompassing synchronizing speed, energy efficiency, physical machine size scaling, and economic cost. We show that CLEAR is the only metric to accurately describe the historical compute-system development. We find that even across different technology options CLEAR matches the observed (post-diction) constant rate-of-growth, and also fits proposed future compute-system (prediction). Therefore, we propose CLEAR to serve as a guide to quantitatively predict required compute-system demands at a given time in the future.

Reservoir computing and decision making using laser dynamics for photonic accelerator

The photonic accelerator is a new paradigm of photonic technologies for artificial intelligence, wherein the systems can accelerate information processing in electronic computing. Schemes for reservoir computing and decision making are promising examples of the photonic accelerator. In this work, we describe recent advances in the architectures of reservoir computing and decision making based on complex photonics, wherein the physics of light, including ultrafast dynamics in semiconductor lasers with optical feedback, plays a crucial role in the photonic accelerator.

Numerical Simulation of an InP Photonic Integrated Cross-Connect for Deep Neural Networks on Chip

We propose a novel photonic accelerator architecture based on a broadcast-and-weight approach for a deep neural network through a photonic integrated cross-connect. The single neuron and the complete neural network operation are numerically simulated. The weight calibration and weighted addition are reproduced and demonstrated to behave as in the experimental measurements. A dynamic range higher than 25 dB is predicted, in line with the measurements. The weighted addition operation is also simulated and analyzed as a function of the optical crosstalk and the number of input colors involved. In particular, while an increase in optical crosstalk negatively influences the simulated error, a greater number of channels results in better performance. The iris flower classification problem is solved by implementing the weight matrix of a trained three-layer deep neural network. The performance of the corresponding photonic implementation is numerically investigated by tuning the optical crosstalk and waveguide loss, in order to anticipate energy consumption per operation. The analysis of the prediction error as a function of the optical crosstalk per layer suggests that the first layer is essential to the final accuracy. The ultimate accuracy shows a quasi-linear dependence between the prediction accuracy and the errors per layer for a normalized root mean square error lower than 0.09, suggesting that there is a maximum level of error permitted at the first layer for guaranteeing a final accuracy higher than 89%. However, it is still possible to find good local minima even for an error higher than 0.09, due to the stochastic nature of the network we are analyzing. Lower levels of path losses allow for half the power consumption at the matrix multiplication unit, for the same error level, offering opportunities for further improved performance. The good agreement between the simulations and the experiments offers a solid base for studying the scalability of this kind of network.

Comparing fission-product yields from photon-induced fission of 240Pu and neutron-induced fission of 239Pu as a test of the Bohr hypothesis in nuclear fission

Fission product yields (FPYs) are a uniquely sensitive probe of the fission process, with well established dependence on the species of nucleus undergoing fission, its excitation energy and spin. Thus FPYs are well suited for testing Bohr’s hypothesis in the context of nuclear fission, which states that the decay of a compound nucleus with a given excitation energy, spin and parity is independent of its formation. Using FPYs, we have performed a new highprecision test of the combined effects of the entrance channel, spin and parity on the fission process from two of the most commonly used particles to induce fission neutrons and photons. The 239 Pu(n,f) reaction at En = 4.6 MeV and the 240 Pu(γ,f) reaction at Eγ = 11.2 MeV were used to produce a 240 Pu∗ compound nucleus with the same excitation energy. The FPYs from these two reactions were measured using quasimonoenergetic neutron beams from the TUNL’s FN tandem Van de Graaff accelerator and quasimonenergetic photon beams from the High Intensity γ-ray Source (HIγS) facility. The FPYs from these two reactions are compared quantitatively for the first time.

Comparison BIPM.RI(I)-K6 of the standards for absorbed dose to water of the KRISS, Korea and the BIPM in accelerator photon beams

A key comparison has been made between the absorbed dose to water standards of the KRISS, Korea and the BIPM in accelerator photon beams. The results show the standards to be in agreement within the expanded uncertainty (k = 2) of the comparison of 12 parts in 103. The results are analysed and presented in terms of degrees of equivalence, suitable for entry in the BIPM key comparison database.Main textTo reach the main text of this paper, click on Final Report. Note that this text is that which appears in Appendix B of the BIPM key comparison database kcdb.bipm.org/.The final report has been peer-reviewed and approved for publication by the CCRI, according to the provisions of the CIPM Mutual Recognition Arrangement (CIPM MRA).

A Winograd-Based Integrated Photonics Accelerator for Convolutional Neural Networks

Neural Networks (NNs) have become the mainstream technology in the artificial intelligence (AI) renaissance over the past decade. Among different types of neural networks, convolutional neural networks (CNNs) have been widely adopted as they have achieved leading results in many fields such as computer vision and speech recognition. This success in part is due to the widespread availability of capable underlying hardware platforms. In parallel, hardware specialization can expose us to novel architectural solutions, which can outperform general purpose computers for the tasks at hand. Although different applications demand for different performance measures, they all share speed and energy efficiency as high priorities. Meanwhile, photonics processing has seen a resurgence due to its inherited high speed and low power nature. Here, we investigate the potential of using photonics in CNNs by proposing a CNN accelerator design based on Winograd filtering algorithm. Our evaluation results show that while a photonic accelerator can compete with current state-of-the-art electronic platforms in terms of both speed and power, it has the potential to improve the energy efficiency by up to three orders of magnitude.

Photon Acceleration in a Flying Focus.

A high-intensity laser pulse propagating through a medium triggers an ionization front that can accelerate and frequency upshift the photons of a second pulse. The maximum upshift is ultimately limited by the accelerated photons outpacing the ionization front or the ionizing pulse refracting from the plasma. Here, we apply the flying focus-a moving focal point resulting from a chirped laser pulse focused by a chromatic lens-to overcome these limitations. Theory and simulations demonstrate that the ionization front produced by a flying focus can frequency upshift an ultrashort optical pulse to the extreme ultraviolet over a centimeter of propagation. An analytic model of the upshift predicts that this scheme could be scaled to a novel tabletop source of spatially coherent x rays.

Novel frontier of photonics for data processing—Photonic accelerator

In the emerging Internet of things cyber-physical system-embedded society, big data analytics needs huge computing capability with better energy efficiency. Coming to the end of Moore’s law of the electronic integrated circuit and facing the throughput limitation in parallel processing governed by Amdahl’s law, there is a strong motivation behind exploring a novel frontier of data processing in post-Moore era. Optical fiber transmissions have been making a remarkable advance over the last three decades. A record aggregated transmission capacity of the wavelength division multiplexing system per a single-mode fiber has reached 115 Tbit/s over 240 km. It is time to turn our attention to data processing by photons from the data transport by photons. A photonic accelerator (PAXEL) is a special class of processor placed at the front end of a digital computer, which is optimized to perform a specific function but does so faster with less power consumption than an electronic general-purpose processor. It can process images or time-serial data either in an analog or digital fashion on a real-time basis. Having had maturing manufacturing technology of optoelectronic devices and a diverse array of computing architectures at hand, prototyping PAXEL becomes feasible by leveraging on, e.g., cutting-edge miniature and power-efficient nanostructured silicon photonic devices. In this article, first the bottleneck and the paradigm shift of digital computing are reviewed. Next, we review an array of PAXEL architectures and applications, including artificial neural networks, reservoir computing, pass-gate logic, decision making, and compressed sensing. We assess the potential advantages and challenges for each of these PAXEL approaches to highlight the scope for future work toward practical implementation.

Large-Scale Optical Neural Networks Based on Photoelectric Multiplication

Recent success in deep neural networks has generated strong interest in hardware accelerators to improve speed and energy consumption. This paper presents a new type of photonic accelerator based on coherent detection that is scalable to large ($N \gtrsim 10^6$) networks and can be operated at high (GHz) speeds and very low (sub-aJ) energies per multiply-and-accumulate (MAC), using the massive spatial multiplexing enabled by standard free-space optical components. In contrast to previous approaches, both weights and inputs are optically encoded so that the network can be reprogrammed and trained on the fly. Simulations of the network using models for digit- and image-classification reveal a "standard quantum limit" for optical neural networks, set by photodetector shot noise. This bound, which can be as low as 50 zJ/MAC, suggests performance below the thermodynamic (Landauer) limit for digital irreversible computation is theoretically possible in this device. The proposed accelerator can implement both fully-connected and convolutional networks. We also present a scheme for back-propagation and training that can be performed in the same hardware. This architecture will enable a new class of ultra-low-energy processors for deep learning.

Photon acceleration and tunable broadband harmonics generation in nonlinear time-dependent metasurfaces

Time-dependent nonlinear media, such as rapidly generated plasmas produced via laser ionization of gases, can increase the energy of individual laser photons and generate tunable high-order harmonic pulses. This phenomenon, known as photon acceleration, has traditionally required extreme-intensity laser pulses and macroscopic propagation lengths. Here, we report on a novel nonlinear material—an ultrathin semiconductor metasurface—that exhibits efficient photon acceleration at low intensities. We observe a signature nonlinear manifestation of photon acceleration: third-harmonic generation of near-infrared photons with tunable frequencies reaching up to ≈3.1ω. A simple time-dependent coupled-mode theory, found to be in good agreement with experimental results, is utilized to predict a new path towards nonlinear radiation sources that combine resonant upconversion with broadband operation.

Flying focus: Spatial and temporal control of intensity for laser-based applications

An advanced focusing scheme, called a “flying focus,” uses a chromatic focusing system combined with a broadband laser pulse with its colors arranged in time to propagate a high intensity focus over a distance that can be much greater than its Rayleigh length while decoupling the speed at which the peak intensity propagates from its group velocity. The flying focus generates a short effective pulse duration with a small diameter focal spot that co- or counter-propagates along the optical axis at any velocity. Experiments validating the concept measured subluminal (−0.09c) to superluminal (39c) focal spot velocities with a nearly constant peak intensity over 4.5 mm. Experiments that increased the peak intensity above the ionization threshold for gas demonstrated ionization waves propagating at the velocity of the flying focus. These ionization waves of any velocity overcome several laser-plasma propagation issues, including ionization-induced refraction. The flying focus presents opportunities to overcome current fundamental limitations in laser-plasma amplifiers, laser wakefield accelerators, photon accelerators, and high-order frequency conversion.

A Fokker–Planck Framework for Studying the Diffusion of Radio Burst Waves in the Solar Corona

Abstract Electromagnetic wave scattering off density inhomogeneities in the solar corona is an important process that determines both the apparent source size and the time profile of radio bursts observed at 1 au. Here we model the scattering process using a Fokker–Planck equation and apply this formalism to several regimes of interest. In the first regime the density fluctuations are considered quasi-static and diffusion in wavevector space is dominated by angular diffusion on the surface of a constant energy sphere. In the small-angle (“pencil-beam”) approximation, this diffusion further occurs over a small solid angle in wavevector space. The second regime corresponds to a much later time, by which scattering has rendered the photon distribution near-isotropic, resulting in a spatial diffusion of the radiation. The third regime involves time-dependent fluctuations and, therefore, Fermi acceleration of photons. Combined, these results provide a comprehensive theoretical framework within which to understand several important features of propagation of radio burst waves in the solar corona: emitted photons are accelerated in a relatively small inner region and then diffuse outward to larger distances. En route, angular diffusion results both in source sizes that are substantially larger than the intrinsic source and in observed intensity-versus-time profiles that are asymmetric, with a sharp rise and an exponential decay. Both of these features are consistent with observations of solar radio bursts.

Comparison of the standards for absorbed dose to water of the METAS, Switzerland and the BIPM in accelerator photon beams (BIPM.RI(I)-K6)

A key comparison has been carried out of the absorbed dose to water standards of the Federal Institute of Metrology (METAS), Switzerland, and the BIPM in accelerator photon beams. The results show the standards to be in agreement within the standard uncertainty of the comparison of 6.6 parts in 103. The results are analysed and presented in terms of degrees of equivalence, suitable for entry in the BIPM key comparison database.KEY WORDS FOR SEARCHIonizing radiation, dosimetry, key comparison, absorbed dose to waterMain textTo reach the main text of this paper, click on Final Report. Note that this text is that which appears in Appendix B of the BIPM key comparison database kcdb.bipm.org/.The final report has been peer-reviewed and approved for publication by the CCRI, according to the provisions of the CIPM Mutual Recognition Arrangement (CIPM MRA).

OA028] On the potential of direct MV calibration of ionization chambers by secondary standards laboratories

Purpose The IAEA protocol (TRS-398) for absorbed dose to water measurements in therapeutic MV photon beams, recommends that ionization chambers be calibrated directly in MV accelerator photon beams rather than in cobalt-60 combined with generic beam quality correction factors ( k Q ) tabulated in the protocol. The purpose of this work was to study the feasibility for secondary standards laboratories to provide direct MV calibrations for hospitals and to evaluate the benefit of such calibrations versus conventional cobalt calibrations. Methods A laboratory with the potential to provide direct MV calibrations was established. It includes both a cobalt irradiator and a linear accelerator with the following MV beams: 4, 6, 10, 15, and 18 MV with flattening filter and 6 and 10 MV without flattening filter. Procedures were designed for accurate positioning of ionization chambers, beam output monitoring and relevant ionization chamber correction factors. Over a two-month period, k Q -factors were measured in all beams for seven nominally identical chambers (IBA FC65G) covering a wide range of serial numbers. Traceability to primary standards were established using transfer chambers. The results were compared with published k Q -factors from TRS-398. Results The monitoring of beam output using an external monitor chamber was found to be essential for calibration purposes: The raw accelerator output changed systematically by 0.4% over the measurement period. However, with an external monitor chamber, the output variability was reduced to less than 0.05% (1 sd). For each beam quality, the relative standard deviation for the measured k Q factors for the seven ionization chambers was about 0.07%. Conclusions The study demonstrates the feasibility of highly reproducible MV calibrations at the level of a secondary standards laboratory, and the closeness of agreement (sd k Q -factors may be adequate for the tested ionization chamber model. Ongoing work focuses on characterization of additional ionization chambers models (PTW30013, PTW31021, and NE2571) and on an improved estimation of the uncertainty related to establishing traceability to primary standards. The latter work includes direct use of water calorimetry in the specific accelerator beams of the laboratory.

A proposed method for linear accelerator photon beam steering using EPID.

Beam steering is the process of calibrating the angle and translational position with which a linear accelerator's (linac's) electron beam strikes the x‐ray target with respect to the collimator rotation axis. The shape of the dose profile is highly dependent on accurate beam steering and is essential for ensuring correct delivery of the radiotherapy treatment plan. Traditional methods of beam steering utilize a scanning water tank phantom that makes the process user‐dependent. This study is the first to provide a methodology for both beam angle steering and beam translational position steering based on EPID imaging of the beam and does not require a phantom. Both the EPID‐based beam angle steering and beam translational steering methods described have been validated against IC Profiler measurement. Wide field symmetry agreement was found between the EPID and IC Profiler to within 0.06 ± 0.14% (1 SD) and 0.32 ± 0.11% (1 SD) for flattened and flattening‐filter‐free (FFF) beams, respectively. For a 1.1% change in symmetry measured by IC Profiler the EPID method agreed to within 0.23%. For beam translational position steering, the EPID method agreed with IC Profiler method to within 0.03 ± 0.05 mm (1 SD) at isocenter. The EPID‐based methods presented are quick to perform, simple, accurate and could easily be integrated with the linac, potentially via the MPC application. The methods have the potential to remove user variability and to standardize the process of beam steering throughout the radiotherapy community.

Elements of a dielectric laser accelerator

We experimentally demonstrate several physical concepts necessary for the future development of dielectric laser accelerators—photonic elements that utilize the inelastic interaction between electrons and the optical near fields of laser-illuminated periodic nanostructures. To build a fully photonic accelerator, concatenation of elements, large energy gains, and beam steering elements are required. Staged acceleration is shown using two spatio-temporally separated interaction regions. Further, a chirped silicon grating is used to overcome the velocity dephasing of subrelativistic electrons with respect to its optical near fields, and last, a parabolic grating geometry serves for focusing of the electron beam.

Ionizing radiation‐induced acoustics for radiotherapy and diagnostic radiology applications

Acoustic waves are induced via the thermoacoustic effect in objects exposed to a pulsed beam of ionizing radiation. This phenomenon has interesting potential applications in both radiotherapy dosimetry and treatment guidance as well as low-dose radiological imaging. After initial work in the field in the 1980s and early 1990s, little research was done until 2013 when interest was rejuvenated, spurred on by technological advances in ultrasound transducers and the increasing complexity of radiotherapy delivery systems. Since then, many studies have been conducted and published applying ionizing radiation-induced acoustic principles into three primary research areas: Linear accelerator photon beam dosimetry, proton therapy range verification, and radiological imaging. This review article introduces the theoretical background behind ionizing radiation-induced acoustic waves, summarizes recent advances in the field, and provides an outlook on how the detection of ionizing radiation-induced acoustic waves can be used for relative and invivo dosimetry in photon therapy, localization of the Bragg peak in proton therapy, and as a low-dose medical imaging modality. Future prospects and challenges for the clinical implementation of these techniques are discussed.

Cascaded chirped photon acceleration for efficient frequency conversion

A cascaded sequence of photon acceleration stages using the instantaneous creation of a plasma density gradient by flash ionization allows the generation of coherent and chirped ultraviolet and x-ray pulses with independently tunable frequency and bandwidth. The efficiency of the cascaded process scales with 1/ω in energy, and multiple stages produce significant frequency up-conversion with gas-density plasmas. Chirping permits subsequent pulse compression to few-cycle durations, and output frequencies are not limited to integer harmonics.