DOCSLIB.ORG

Phase Noise Components (F > 1 Hz) - Expressed in Units of Spectral Densities Or Timing Jitter

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Phase Noise Components (F > 1 Hz) - Expressed in Units of Spectral Densities Or Timing Jitter RF Phase Reference Distribution Systems for Large Experiments Krzysztof Czuba Warsaw University of Technology Institute of Electronic Systems MDI Forum DESY, 05.07.2013 Objectives ● Give an overview of techniques used for RF synchronization (RF phase reference) systems working with high accuracy ● Show the most important problems and limitations of RF synchronization subsystems Krzysztof Czuba MDI Forum, DESY, 05.07.2013 2 Synchronization Signals There are various types of signals (frequently confused by users): ● Analog (RF phase reference, LO) ~ ● Clocks (digital circuits, ADC, DAC, CPU) ● Trigger signals (digital subsystems, CPU) ● Optical pulse trains (lasers, diagnostics, experiments) Krzysztof Czuba MDI Forum, DESY, 05.07.2013 3 Demand for Precise RF Phase Reference Signals ● LLRF systems of accelerators (LO for downconverters, klystron drive phase reference), diagnostics (BPM, HOM, ...) ● Particularly at Free-Electron Lasers (FEL) LINACs ● Required synchronization accuracy nowadays reaches sub 10fs levels (e.g. European-XFEL) Krzysztof Czuba MDI Forum, DESY, 05.07.2013 4 RF Synchronization System Components ● Master Oscillator ● Phase Reference Distribution (for harmonic RF signals) ● Diagnostics ● Interfaces to other subsystems Krzysztof Czuba MDI Forum, DESY, 05.07.2013 5 Generic RF Synchronization System Master Oscillator Distribution System Phase Reference and Timing Signals ~ RFS RFS RFS RFS Rack(s) Rack(s) Rack(s) Rack(s) RFS – Radio Frequency Station RACK CRATE CRATE Distrib. Distribution within within crates, racks CRATE PCBPCB between PCBs, PCB back-planes PCBPCB Local signal CRATE generation Krzysztof Czuba MDI Forum, DESY, 05.07.2013 6 Some Basic Definitions Before we go to RF Synch. sub-components Krzysztof Czuba MDI Forum, DESY, 05.07.2013 7 We Would Like to Distribute Pure Sine Wave BUT Krzysztof Czuba MDI Forum, DESY, 05.07.2013 8 Harmonic Signal With Noise Components In Time Domain Ideal Signal vt = V 0 sin 2 0t Noisy Signal vt = [V 0 t] sin [2 0t t] V0 - the nominal peak voltage amplitude ν0 - nominal frequency, called also instantaneous ε(t) - deviation of amplitude from nominal value φ(t) - deviation of phase from nominal value - noise component Krzysztof Czuba MDI Forum, DESY, 05.07.2013 9 Short and Long Term Instabilities The short-term instability refers to all phase/frequency changes about the nominal of less than a few second duration. - derives from a “fast” phase noise components (f > 1 Hz) - expressed in units of spectral densities or timing jitter The long-term instability refers to the phase/frequency variations that occur over time periods longer than a few seconds - derives from slow processes like long term frequency drifts, aging and susceptibility to environmental parameters like temperature - expressed in units of degree, second or ppm per time period (minute, hour, day ...) Krzysztof Czuba MDI Forum, DESY, 05.07.2013 10 Frequency, Time and Angle – Basic Relationships Why do we use “ps” when you talk about phase?? 1 Time domain T = measure 0 T → 360o in the angular domain T/2 T T t = Phase to time 360o conversion o Example: v0 = 1.3GHz → T = ~769ps, 1 → 2,13 ps Time domain measure is convenient for quantifying phase changes in distribution media (by means of propagation delay change) because it does not depend on the signal frequency. Krzysztof Czuba MDI Forum, DESY, 05.07.2013 11 Phase Noise It is a frequency domain measure of signal phase instabilities ф(t) Spectrum of harmonic signal with noise S(ν) r e dBc Is a Power Spectral Density w o P measured in dBc/Hz 1Hz ν 0 ν power densityin one phase noise modulation sideband , per Hz 1 ℒ ( f ) = = S ( f ) total signal power 2 ϕ f = v−v0 offset from the carrier frequency Krzysztof Czuba MDI Forum, DESY, 05.07.2013 12 Phase / Timing Jitter It is a time domain measure of signal phase instabilities ф(t) Phase jitter 2 is calculated in units of radian jitter Timing jitter t RMS is calculated in units of seconds RMS. Used frequently with digital signals Figure source: Corning Frequency Control Krzysztof Czuba MDI Forum, DESY, 05.07.2013 13 Phase Noise and Jitter Relationship Jitter is the integral of Sφ( f ) over the Fourier frequencies of application 1 S f 2 f 2 2 jitter =∫ S f df f 1 f 1 2 t = S f df rms 2 ∫ 0 f 1 Krzysztof Czuba MDI Forum, DESY, 05.07.2013 14 Phase Noise Contributions to Jitter 1.3 GHz Oscillator Phase Noise -80 -90 ] z -100 H / c B -110 d [ e s i -120 o n e -130 s a h P 92 fs -140 -150 17 fs 9 fs 4 fs -160 3 fs 10 100 1k 10k 100k 1M Offset from carrier frequency [Hz] Krzysztof Czuba MDI Forum, DESY, 05.07.2013 15 Generic RF Synchronization System - MO Master Oscillator Distribution System Phase Reference and Timing Signals ~ RFS RFS RFS RFS Rack(s) Rack(s) Rack(s) Rack(s) RFS – Radio Frequency Station RACK CRATE CRATE Distrib. Distribution within within crates, racks CRATE PCBPCB between PCBs, PCB back-planes PCBPCB Local signal CRATE generation Krzysztof Czuba MDI Forum, DESY, 05.07.2013 16 Master Oscillator This device is providing the reference signal for the entire synchronization system Single signal source But in practice ... Krzysztof Czuba MDI Forum, DESY, 05.07.2013 17 Master Oscillator System Example FLASH “MO” Scheme* 1.3 GHz PLL 1.3 GHz POWER 1.3 GHz DISTRI- AMPLIFIER BUTION BOX PLL PA 1.3 GHz x 144 44 dBm 81 MHz POWER 81 MHz DISTRI- The MO AMPLIFIER BUTION BOX PA 81 MHz 40 dBm The MO System OCXO LOW POWER PART ~ 81 MHz PLL Div. 27 MHz x 9 : 3 9 MHz Div. 13.5 MHz For convenience : 2 people call it MO Div. PA 9 MHz : 9 30 dBm Div. 1 MHz : 9 108 MHz PLL x 12 108 MHz DDS 50 Hz PA – Power Amplifier Krzysztof Czuba MDI Forum, DESY, 05.07.2013 18 FLASH MO System Racks and Crates Krzysztof Czuba MDI Forum, DESY, 05.07.2013 19 MO System ● MO reference generator ● Frequency generation scheme ● Signal level adjustment (power amplifiers) ● Splitters and interface to distribution links ● Power supply (very important issue!), sometimes must be unbreakable ● Diagnostics Krzysztof Czuba MDI Forum, DESY, 05.07.2013 20 Most Important Task of MO System ● Generate signal with sufficient stability. Usually short- term stability (phase noise) is of concern. ● Well known RF generation techniques (OCXO, PLL, RF Synthesis). ● Stability of ps to ns is easy to achieve (and very cheap) ● No problem to go down to 20 fs (but the last 100 fs gets expensive!) ● At FLASH we demonstrated source with 30 fs of jitter @ 1.3 GHz (10 Hz to 1MHz BW) Krzysztof Czuba MDI Forum, DESY, 05.07.2013 21 Laboratory Prototype of XFEL MO (1.3 GHz) Reference: OCXO 100MHz VCO: 1.3 GHz DRO from PSI Output power + 13 dBm Jitter 42 fsec RMS (10Hz to 1 MHz) by (Ł. Zembala, H. Weddig) Krzysztof Czuba MDI Forum, DESY, 05.07.2013 22 Stable Signal Distribution Main Distribution to Master Oscillator Distribution System Phase Reference and Timing Signals RF Stations and ~ other accelerator RFS RFS RFS RFS subsystems Rack(s) Rack(s) Rack(s) Rack(s) RFS – Radio Frequency Station RACK CRATE CRATE Distrib. Distribution within within crates, racks CRATE PCBPCB between PCBs, PCB back-planes PCBPCB Local Distribution: Local signal CRATE generation racks, crates, PCBs The importance of a local distribution is frequently underestimated Last 2 meters of a poor quality cable exposed to vibrations or a “wrong” track on a PCB can destroy the signal performance achieved over hundreds of meters of distribution! Krzysztof Czuba MDI Forum, DESY, 05.07.2013 23 Most Important Task of Signal Distribution ● Deliver the phase reference signal with specified degradation compared to the MO source and with specified output power level – Usually phase noise/jitter is not affected significantly in coax cable links – Cheap optical links limit jitter performance to few – few tens of ps – The biggest problem: long term phase drifts in distribution media Krzysztof Czuba MDI Forum, DESY, 05.07.2013 24 RF vs Optical Synchronization The MO and phase reference distribution can be realized either in RF technology or in optical (laser oscillators and synchronization) ● RF: – mature technology – well known subsystems – limited performance (but sufficient for many applications) – sensitive to EMI ● Optical: – low loss, easier installation (fiber as media) – promising performance (sub-fs accuracy estimated) – high performance systems still under development - reliability not proven Krzysztof Czuba MDI Forum, DESY, 05.07.2013 25 Distribution Media: Cable vs. Optical Fiber Parameter Coaxial Fiber Low at any RF Attenuation High frequency Distribution distance short long Temperature coefficient of phase ~10-5/oC ~10-5/oC lenghth Need of feedback controlling phase YES YES drifts Fiber – low but Price Relatively high Tx and Rx high The decision not obvious and usually some compromise is needed Krzysztof Czuba MDI Forum, DESY, 05.07.2013 26 Example: Phase Drift Requirements for European XFEL ● Injector and booster devices: 100 fs p-p drift within 1h. Distribution distance ~150 m ● Main linac: 1 ps (~0.5o) p-p, distribution distance ~1.7 km ● Within 100 fs light travels a distance of ~30μm ● We need to assure sub ~30μm of length stabilization of 150 m long link ● Is this easy to do? Krzysztof Czuba MDI Forum, DESY, 05.07.2013 27 Phase Drifts in Distribution Media Drifts caused mainly by temperature changes RF signal source Target ~ Device Electrical / optical length change 1.3 GHz signal phase change in 5km of fiber. 10 oC temperature change Reason of drifts: 1600 - In fiber: neff change 1400 1200 ] - In cable: physical dimension and g e 1000 d [ dielectric properties change e g n 800 a h c o e 1400 @ 1.3 GHz s 600 a h P 400 corresponds to ~2800 ps!! 200 Feedback on phase required!! 0 0 10 20 30 40 50 60 70 80 90 100 Time [min] Krzysztof Czuba MDI Forum, DESY, 05.07.2013 28 Temperature Measurements of Coax Cables Sometimes is a challenging task..
Load more

Recommended publications
  • Fast, High Quality Noise
    The Importance of Being Noisy: Fast, High Quality Noise Natalya Tatarchuk 3D Application Research Group AMD Graphics Products Group Outline Introduction: procedural techniques and noise Properties of ideal noise primitive Lattice Noise Types Noise Summation Techniques Reducing artifacts General strategies Antialiasing Snow accumulation and terrain generation Conclusion Outline Introduction: procedural techniques and noise Properties of ideal noise primitive Noise in real-time using Direct3D API Lattice Noise Types Noise Summation Techniques Reducing artifacts General strategies Antialiasing Snow accumulation and terrain generation Conclusion The Importance of Being Noisy Almost all procedural generation uses some form of noise If image is food, then noise is salt – adds distinct “flavor” Break the monotony of patterns!! Natural scenes and textures Terrain / Clouds / fire / marble / wood / fluids Noise is often used for not-so-obvious textures to vary the resulting image Even for such structured textures as bricks, we often add noise to make the patterns less distinguishable Ех: ToyShop brick walls and cobblestones Why Do We Care About Procedural Generation? Recent and upcoming games display giant, rich, complex worlds Varied art assets (images and geometry) are difficult and time-consuming to generate Procedural generation allows creation of many such assets with subtle tweaks of parameters Memory-limited systems can benefit greatly from procedural texturing Smaller distribution size Lots of variation
    [Show full text]
  • Synthetic Data Generation for Deep Learning Models
    32. DfX-Symposium 2021 Synthetic Data Generation for Deep Learning Models Christoph Petroll 1 , 2 , Martin Denk 2 , Jens Holtmannspötter 1 ,2, Kristin Paetzold 3 , Philipp Höfer 2 1 The Bundeswehr Research Institute for Materials, Fuels and Lubricants (WIWeB) 2 Universität der Bundeswehr München (UniBwM) 3 Technische Universität Dresden * Korrespondierender Autor: Christoph Petroll Institutsweg 1 85435 Erding Germany Telephone: 08122/9590 3313 Mail: [email protected] Abstract The design freedom and functional integration of additive manufacturing is increasingly being implemented in existing products. One of the biggest challenges are competing optimization goals and functions. This leads to multidisciplinary optimization problems which needs to be solved in parallel. To solve this problem, the authors require a synthetic data set to train a deep learning metamodel. The research presented shows how to create a data set with the right quality and quantity. It is discussed what are the requirements for solving an MDO problem with a metamodel taking into account functional and production-specific boundary conditions. A data set of generic designs is then generated and validated. The generation of the generic design proposals is accompanied by a specific product development example of a drone combustion engine. Keywords Multidisciplinary Optimization Problem, Synthetic Data, Deep Learning © 2021 die Autoren | DOI: https://doi.org/10.35199/dfx2021.11 1. Introduction and Idea of This Research Due to its great design freedom, additive manufacturing (AM) shows a high potential of functional integration and part consolidation [1]-[3]. For this purpose, functions and optimization goals that are usually fulfilled by individual components must be considered in parallel.
    [Show full text]
  • Digital to Analog, PWM, Stepper Motors
    Lecture #19 Digital To Analog, PWM, Stepper Motors 18-348 Embedded System Engineering Philip Koopman Monday, 28-March-2016 Electrical& Computer ENGINEERING © Copyright 2006-2016, Philip Koopman, All Rights Reserved Example System: HVAC Compressor Expansion Valve [http://www.solarpowerwindenergy.org] 2 HVAC Embedded Control Compressors (reciprocating & scroll) • Smart loading and unloading of compressor – Want to minimize motor turn on/turn off cycles – May involve bypassing liquid so compressor keeps running but doesn’t compress • Variable speed for better output temperature control • Diagnostics and prognostics – Prevent equipment damage (e.g., liquid entering compressor, compressor stall) – Predict equipment failures (e.g., low refrigerant, motor bearings wearing out) Expansion Valve • Smart control of amount of refrigerant evaporated – Often a stepper motor • Diagnostics and prognostics – Low refrigerant, icing on cold coils, overheating of hot coils System coordination • Coordinate expansion valve and compressor operation • Coordinate multiple compressors • Next lecture – talk about building-level system level diagnostics & coordination 3 Where Are We Now? Where we’ve been: • Interrupts, concurrency, scheduling, RTOS Where we’re going today: • Analog Output Where we’re going next: • Analog Input •Human I/O • Very gentle introduction to control •… • Test #2 and last project demo 4 Preview Digital To Analog Conversion • Example implementation • Understanding performance • Low pass filters Waveform encoding PWM • Digital way
    [Show full text]
  • Networking, Noise
    Final TAs • Jeff – Ben, Carmen, Audrey • Tyler – Ruiqi – Mason • Ben – Loudon – Jordan, Nick • David – Trent – Vijay Final Design Checks • Optional this week • Required for Final III Playtesting Feedback • Need three people to playtest your game – One will be in class – Two will be out of class (friends, SunLab, etc.) • Should watch your victim volunteer play your game – Don’t comment unless necessary – Take notes – Pay special attention to when the player is lost and when the player is having the most fun Networking STRATEGIES The Illusion • All players are playing in real-time on the same machine • This isn’t possible • We need to emulate this as much as possible The Illusion • What the player should see: – Consistent game state – Responsive controls – Difficult to cheat • Things working against us: – Game state > bandwidth – Variable or high latency – Antagonistic users Send the Entire World! • Players take turns modifying the game world and pass it back and forth • Works alright for turn-based games • …but usually it’s bad – RTS: there are a million units – FPS: there are a million bullets – Fighter: timing is crucial Modeling the World • If we’re sending everything, we’re modeling the world as if everything is equally important Player 1 Player 2 – But it really isn’t! My turn Not my turn Read-only – Composed of entities, only World some of which react to the StateWorld 1 player • We need a better model to solve these problems Client-Server Model • One process is the authoritative server Player 1 (server) Player 2 (client) – Now we
    [Show full text]
  • Procedural Generation of Infinite Terrain from Real-World Elevation
    Journal of Computer Graphics Techniques Vol. 3, No. 1, 2014 http://jcgt.org Designer Worlds: Procedural Generation of Infinite Terrain from Real-World Elevation Data Ian Parberry University of North Texas Figure 1.A 180◦ panorama of terrain generated using elevation data from Utah. Abstract The standard way to procedurally generate random terrain for video games and other applica- tions is to post-process the output of a fast noise generator such as Perlin noise. Tuning the post-processing to achieve particular types of terrain requires game designers to be reason- ably well-trained in mathematics. A well-known variant of Perlin noise called value noise is used in a process accessible to designers trained in geography to generate geotypical terrain based on elevation statistics drawn from widely available sources such as the United States Geographical Service. A step-by-step process for downloading and creating terrain from real- world USGS elevation data is described, and an implementation in C++ is given. 1. Introduction Terrain generation is an example of what is known in the game industry as procedu- ral content generation. Procedural content generation needs to have three important properties. First, it needs to be fast, meaning that it has to use only a fraction of the computing power on a current-generation computer. Second, it needs to be both ran- dom and structured so that it creates content that is varied and interesting. Third, it needs to be controllable in a natural and intuitive way. As noted in the excellent survey paper by Smelik et al. [2009], procedural terrain generation is often based on fractal noise generators such as Perlin noise [Perlin 1985; Perlin 2002] (for more details see, for example, the book by Ebert et al.
    [Show full text]
  • Coherent Noise
    Coherent Noise User manual Table of Contents Introduction...........................................................................................5 Chapter I. .............................................................................................6 Noise Basics........................................................................................6 Combining Noise. ................................................................................8 CoherentNoise library...........................................................................9 Chapter II............................................................................................12 Noise generators................................................................................12 Generator base class.......................................................................12 ValueNoise.....................................................................................13 GradientNoise................................................................................14 ValueNoise2D and GradientNoise2D...................................................16 Constant.......................................................................................16 Function........................................................................................17 Cache...........................................................................................17 Patterns........................................................................................18 Cylinders....................................................................................18
    [Show full text]
  • Interfacing to Analog World Sensor Interfacing
    Interfacing to Analog World Sensor Interfacing • Introduction to Analog to digital Conversion • Why Analog to Digital? • Basics of A/D Conversion. • A/D converter inside PIC16F887 • Related Problems Prepared By- Mohammed Abdul Kader Assistant Professor, EEE, IIUC Introduction to Analog to digital Conversion Signals in the real world are analog: light, sound, temperature, pressure, acceleration or other phenomenon. So, real-world signals must be converted into digital, using a circuit called ADC (Analog-to-Digital Converter), before they can be manipulated by digital equipment. When you scan a picture with a scanner what the scanner is doing is an analog-to-digital conversion: it is taking the analog information provided by the picture (light) and converting into digital. When you record your voice on your computer, you are using an analog-to-digital converter to convert your voice, which is analog, into digital information. When an audio CD is recorded at a studio, once again analog-to-digital is taking place, converting sounds into digital numbers that will be stored on the disc. Whenever we need the analog signal back, the opposite conversion – digital-to-analog, which is done by a circuit called DAC, Digital-to-Analog Converter – is needed. When you play an audio CD, what the CD player is doing is reading digital information stored on the disc and converting it back to analog so you can hear the audio. 2 Lecture Materials on "Interfacing to Analog World", By- Mohammed Abdul Kader, Assistant Professor, EEE, IIUC Why Analog to Digital? 1. Reducing Noise: Since analog signals can assume any value, noise is interpreted as being part of the original signal.
    [Show full text]
  • Fast Paralleled Removal Method of Impulsive Noise Using Edge Strength
    Proceedings of the 6th IIAE International Conference on Intelligent Systems and Image Processing 2018 Fast Paralleled Removal Method of Impulsive Noise Using Edge Strength Yusuke Koshimuraa*, Takashi Miyazakia, Yasuki Yokoyamaa, Yasushi Amaria, Hiroaki Yamamotob aNational Institute of Technology, Nagano College, 716 Tokuma, Nagano city, Nagano Prefecture and 381-8550, Japan bShinshu University, 4-17-1 Wakasato, Nagano city, Nagano Prefecture and 380-0928, Japan *Yusuke KOSHIMURA: [email protected] Takashi MIYAZAKI: [email protected] Abstract of the variation in sensitivity among light receiving cells, impulsive noise is superimposed on the images and The Median Filter is known as a method for effectively deterioration is likely to occur. removing to impulsive noise. However, this method occurs The Median Filter method (MF) is effective to remove deteriorating image quality since the filtering process is also such as these noises (1,2), but this method performs processing applied to non-noise pixels. Therefore, a method called the on all pixels, so degrades non-noise pixel. As a method Switching Median Filter, improved capable of distinguishing solving this problem, the Switching Median Filter method between noise pixels and non-noise pixels has been proposed. (SMF) using switching type noise detection filter has been Our Multi-directional Switching Median Filter is one of proposed. Therefore, degradation of non-noise pixels caused which has been improved such as same. As the first step in by MF is able to be suppressed and both denoising and edge this method, the detection and removal processing of pixels preservation are able to be suppressed and both denoising considered as noise are performed by using a 2x2 detection and edge preservation are able to be achieved (3-12).
    [Show full text]
  • BACHELOR's THESIS Implicit Procedural Textures
    2007:12 HIP BACHELOR'S THESIS Implicit Procedural Textures as a means of saving texture memory Joakim Lindqvist Luleå University of Technology BSc Programmes in Engineering BSc programme in Computer Engineering Department of Skellefteå Campus Division of Leisure and Entertainment 2007:12 HIP - ISSN: 1404-5494 - ISRN: LTU-HIP-EX--07/12--SE Implicit Procedural Textures as a means of saving texture memory Page | I Joakim Lindqvist LTU Skellefteå 6/4/07 Abstract This thesis explores the possibility of exchanging regular textures with procedural textures. It focuses on classic procedural textures like different types of rock and sand. Most time is spent on implicit textures generation on a GPU, this means a limit to the types of algorithms that can be used. As such, much of the work focuses on noise implementations. The goal is to lower video memory usage by changing regular sampled textures with procedural ones. In the end a few ways of creating textures with a satisfactory level of detail is suggested using a blend of regular textures and procedural ones. Sammanfattning Detta examensarbete undersöker möjligheterna att byta ut vanliga texturer mot procedurella texturer. Den fokuserar på klassiska procedurella texturer som exempelvis sten och sand. Jag jobbar mest med implicita texturer körda på GPUn vilket medför en del begränsningar i vilka algoritmer som kan köras. På grund av detta ligger väldigt mycket fokus på olika noise implementationer. Målet är att minska texturminnet som används igenom att ändra vanliga texturer med procedurella motsvarigheter. I slutändan presenterar jag några sätt att skapa texturer med tillräckliga detaljer där vi använder en blandning av vanliga texturer och procedurella.
    [Show full text]
  • Scene Classification Based on a Deep Random-Scale Stretched
    remote sensing Article Scene Classification Based on a Deep Random-Scale Stretched Convolutional Neural Network Yanfei Liu, Yanfei Zhong * ID , Feng Fei, Qiqi Zhu and Qianqing Qin State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, China; [email protected] (Y.L.); [email protected] (F.F.); [email protected] (Q.Z.); [email protected] (Q.Q.) * Correspondence: [email protected]; Tel./Fax: +86-27-6877-9969 Received: 10 January 2018; Accepted: 19 February 2018; Published: 12 March 2018 Abstract: With the large number of high-resolution images now being acquired, high spatial resolution (HSR) remote sensing imagery scene classification has drawn great attention but is still a challenging task due to the complex arrangements of the ground objects in HSR imagery, which leads to the semantic gap between low-level features and high-level semantic concepts. As a feature representation method for automatically learning essential features from image data, convolutional neural networks (CNNs) have been introduced for HSR remote sensing image scene classification due to their excellent performance in natural image classification. However, some scene classes of remote sensing images are object-centered, i.e., the scene class of an image is decided by the objects it contains. Although previous methods based on CNNs have achieved comparatively high classification accuracies compared with the traditional methods with handcrafted features, they do not consider the scale variation of the objects in the scenes. This makes it difficult to directly utilize CNNs on those remote sensing images belonging to object-centered classes to extract features that are robust to scale variation, leading to wrongly classified scene images.
    [Show full text]
  • Package 'Ambient'
    Package ‘ambient’ March 21, 2020 Type Package Title A Generator of Multidimensional Noise Version 1.0.0 Maintainer Thomas Lin Pedersen <[email protected]> Description Generation of natural looking noise has many application within simulation, procedural generation, and art, to name a few. The 'ambient' package provides an interface to the 'FastNoise' C++ library and allows for efficient generation of perlin, simplex, worley, cubic, value, and white noise with optional pertubation in either 2, 3, or 4 (in case of simplex and white noise) dimensions. License MIT + file LICENSE Encoding UTF-8 SystemRequirements C++11 LazyData true Imports Rcpp (>= 0.12.18), rlang, grDevices, graphics LinkingTo Rcpp RoxygenNote 7.1.0 URL https://ambient.data-imaginist.com, https://github.com/thomasp85/ambient BugReports https://github.com/thomasp85/ambient/issues NeedsCompilation yes Author Thomas Lin Pedersen [cre, aut] (<https://orcid.org/0000-0002-5147-4711>), Jordan Peck [aut] (Developer of FastNoise) Repository CRAN Date/Publication 2020-03-21 17:50:12 UTC 1 2 ambient-package R topics documented: ambient-package . .2 billow . .3 clamped . .4 curl_noise . .4 fbm .............................................6 fracture . .7 gen_checkerboard . .8 gen_spheres . .9 gen_waves . 10 gradient_noise . 11 long_grid . 12 modifications . 13 noise_cubic . 14 noise_perlin . 15 noise_simplex . 17 noise_value . 19 noise_white . 20 noise_worley . 21 ridged . 24 trans_affine . 25 Index 27 ambient-package ambient: A Generator of Multidimensional Noise Description Generation of natural looking noise has many application within simulation, procedural generation, and art, to name a few. The ’ambient’ package provides an interface to the ’FastNoise’ C++ library and allows for efficient generation of perlin, simplex, worley, cubic, value, and white noise with optional pertubation in either 2, 3, or 4 (in case of simplex and white noise) dimensions.
    [Show full text]
  • Training a Quantum Pointnet with Nesterov Accelerated Gradient Estimation by Projection
    Training a Quantum PointNet with Nesterov Accelerated Gradient Estimation by Projection Ruoxi Shi Hao Tang Xian-Min Jin Center for Integrated Quantum Information Technologies (IQIT) Shanghai Jiao Tong University Shanghai, 200240, China {eliphat, htang2015, xianmin.jin}@sjtu.edu.cn Abstract PointNet is a bedrock for deep learning methods on point clouds for 3D machine vision. However, the pointwise operations involved in PointNet is resource in- tensive, and the expressiveness of PointNet is limited by the size of its feature space. In this paper, we propose Quantum PointNet with a rectifed max pooling operation to achieve an exponential speedup performing the pointwise operations and meanwhile obtaining a quantum-enhanced feature space. We provide an im- plementation with quantum tensor networks and specify a circuit model that runs on near-term quantum computers. Meanwhile, we develop the NA-GEP (Nesterov Accelerated Gradient Estimation by Projection) optimization framework, together with a periodic batching scheme, to help train large-scale quantum networks more efficiently. We demonstrate that Quantum PointNet reaches competitive perfor- mance to its classical counterpart on a subset of the ModelNet40 dataset with 48x fewer operations required to process a point cloud. It is also shown that NA-GEP is robust under different kinds of noises. A mini Quantum PointNet is able to run on real quantum computers, achieving ∼100% accuracy classifying three kinds of shapes with a small number of shots. 1 Introduction We live in a 3-dimensional world. Machines, especially robots, need to interpret 3-dimensional sur- roundings precisely to interact rationally with their environment. Thus, understanding 3D information is of great importance.
    [Show full text]