8,212 research outputs found
Computing the fast Fourier transform on SIMD microprocessors
No full textThis thesis describes how to compute the fast Fourier transform (FFT) of a power-of-two length signal on single-instruction, multiple-data (SIMD) microprocessors faster than or very close to the speed of state of the art libraries such as FFTW (“Fastest Fourier Transform in the West”), SPIRAL and Intel Integrated Performance Primitives (IPP).
The conjugate-pair algorithm has advantages in terms of memory bandwidth, and three implementations of this algorithm, which incorporate latency and spatial locality optimizations, are automatically vectorized at the algorithm level of abstraction. Performance results on 2- way, 4-way and 8-way SIMD machines show that the performance scales much better than FFTW or SPIRAL.
The implementations presented in this thesis are compiled into a high-performance FFT library called SFFT (“Streaming Fast Fourier Trans- form”), and benchmarked against FFTW, SPIRAL, Intel IPP and Apple Accelerate on sixteen x86 machines and two ARM NEON machines, and shown to be, in many cases, faster than these state of the art libraries, but without having to perform extensive machine specific calibration, thus demonstrating that there are good heuristics for predicting the performance of the FFT on SIMD microprocessors (i.e., the need for empirical optimization may be overstated)
In-process fast surface measurement using wavelength scanning interferometry
Get PDFA wavelength scanning interferometry system for fast areal surface measurement of
micro and nano-scale surfaces which is immune to environmental noise is introduced in this paper.
It can be used for surface measurement of discontinuous surface profiles by producing phase shifts
without any mechanical scanning process. White light spectral scanning interferometry, together
with an acousto-optic tuneable filtering technique, is used to measure both smooth surfaces and
those with large step heights. An active servo control system is used to serve as a phase
compensating mechanism to eliminate the effects of environmental noise. The system can be used
for on-line or in-process measurement on a shop floor
Frequency-Dependent F-Number Increases the Contrast and the Spatial Resolution in Fast Pulse-Echo Ultrasound Imaging
Get PDFFixed -numbers reduce grating lobe artifacts in fast pulse-echo ultrasound
imaging. Such -numbers result in dynamic receive subapertures whose widths
vary with the focal position. These subapertures, however, ignore useful
low-frequency components in the excluded radio frequency (RF) signals and,
thus, reduce the lateral resolution. Here, we propose a frequency-dependent
-number to simultaneously suppress grating lobe artifacts and maintain the
lateral resolution. This -number, at high frequencies, reduces the receive
subaperture to remove spatially undersampled components of the RF signals and
suppress grating lobes. The -number, at low frequencies, enlarges the
receive subaperture to use the components of all RF signals and maintain the
lateral resolution. Experiments validated the proposed -number and
demonstrated improvements in the contrast and the widths of wire targets of up
to 3.2 % and 12.8 %, respectively.Comment: 5 pages, 3 figures, 1 table; corrected typo in Fig. 2 and legend of
Fig. 3, removed mathematical expression from title, added journal referenc
Business advice to fast growth small firms
Get PDFA small proportion of small and medium sized enterprises (SMEs) create most new employment in the sector (Storey et al., 1987). Thus UK Business Link’s remit to provide advice to small firms with the potential to grow should maximise the employment impact of small firm advice. Yet, the risks involved in fast growth and the perceived need for visionary leadership constrains advisors from proffering that advice. This research uses data from 29 transcribed semi-structured research interviews and a group interview of 10 business advisors in the UK’s West Midlands region collected in Autumn 1996 to Spring 1997. The interviewees respond to a prompt for advice for a firm contemplating fast growth. This research suggests that in the implementation stage strategy toward small firms is subtly altered. The research suggests that advisors tend to offer general advice and support to all firms, and focus toward support for all, rather than targeting, and support to help companies survive, rather than grow. Given the importance of these fast growing firms to local employment the findings suggest that present business advice might reduce insolvency rather than increase the number of fast growth firms
Now for the long term: the report of the Oxford Martin Commission for Future Generations
Get PDFThis report is the product of a year long process of research and debate undertaken by a group of eminent leaders on the successes and failures in addressing global challenges over recent decades.
As the world slowly emerges from the devastating Financial Crisis, it is time to reflect on the lessons of this turbulent period and think afresh about how to prevent future crises.
The Oxford Martin Commission for Future Generations focuses on the increasing short-termism of modern politics and our collective inability to break the gridlock which undermines attempts to address the biggest challenges that will shape our future. In Now for the Long Term, they urge decision-makers to overcome their pressing daily preoccupations to tackle problems that will determine the lives of today’s and tomorrow’s generations. Dr James Martin, the founder of the Oxford Martin School, highlights that humanity is at a crossroads. This could be our best century ever, or our worst. The outcome will depend on our ability to understand and harness the extraordinary opportunities as well as manage the unprecedented uncertainties and risks.
The report identifies what these challenges are, explains how progress can be made, and provides practical recommendations. The Commission outlines an agenda for the long term. The case for action is built in three parts. The first, Possible Futures , identifies the key drivers of change and considers how we may address the challenges that will dominate this
century. Next, in Responsible Futures, the Commission draws inspiration from previous examples of where impediments to action have been overcome, and lessons from where progress has been stalled. We then consider the characteristics of our current national and global society that frustrate progress. The final part, Practical Futures, sets out the principles for action and offers illustrative recommendations which show how we can build a sustainable, inclusive and resilient future for all.
 
Fast identification of biological pathways associated with a quantitative trait using group lasso with overlaps.
Get PDFWhere causal SNPs (single nucleotide polymorphisms) tend to accumulate within biological pathways, the incorporation of prior pathways information into a statistical model is expected to increase the power to detect true associations in a genetic association study. Most existing pathways-based methods rely on marginal SNP statistics and do not fully exploit the dependence patterns among SNPs within pathways.We use a sparse regression model, with SNPs grouped into pathways, to identify causal pathways associated with a quantitative trait. Notable features of our "pathways group lasso with adaptive weights" (P-GLAW) algorithm include the incorporation of all pathways in a single regression model, an adaptive pathway weighting procedure that accounts for factors biasing pathway selection, and the use of a bootstrap sampling procedure for the ranking of important pathways. P-GLAW takes account of the presence of overlapping pathways and uses a novel combination of techniques to optimise model estimation, making it fast to run, even on whole genome datasets.In a comparison study with an alternative pathways method based on univariate SNP statistics, our method demonstrates high sensitivity and specificity for the detection of important pathways, showing the greatest relative gains in performance where marginal SNP effect sizes are small
Lung stereotactic body radiotherapy with an MR-linac – Quantifying the impact of the magnetic field and real-time tumor tracking
Get PDFAbstractBackground and purposeThere are concerns that radiotherapy doses delivered in a magnetic field might be distorted due to the Lorentz force deflecting secondary electrons. This study investigates this effect on lung stereotactic body radiotherapy (SBRT) treatments, conducted either with or without multileaf collimator (MLC) tumor tracking.Material and methodsLung SBRT treatments with an MR-linac were simulated for nine patients. Two different treatment techniques were compared: conventional, non-tracked deliveries and deliveries with real-time MLC tumor tracking, each conducted either with or without a 1.5T magnetic field.ResultsSlight dose distortions at air-tissue-interfaces were observed in the presence of the magnetic field. Most prominently, the dose to 2% of the skin increased by 1.4Gy on average. Regardless of the presence of the magnetic field, MLC tracking was able to spare healthy tissue, for example by decreasing the mean lung dose by 0.3Gy on average, while maintaining the target dose.ConclusionsAccounting for the magnetic field during treatment plan optimization allowed for design and delivery of clinically acceptable lung SBRT treatments with an MR-linac. Furthermore, the ability of MLC tumor tracking to decrease dose exposure of healthy tissue, was not inhibited by the magnetic field
Frequency Support of Fast-Multi-Energy Storage Systems in Low Rotational Inertia Scenarios
Get PDFModern power systems are increasing the connection of power electronic converters (PECs), and new inertia-less technologies displace the synchronous generation units. Therefore, the total rotational system inertia is reduced, creating new problems related to the system frequency control and stability. Several mechanisms to enable the PECs with frequency sensible control loops have been proposed in the scientific literature. This paper considers the use of Fast-Acting Power Injections (FAPI), where the frequency-sensible control uses a frequency-active power (f-P) based on proportional and derivative control. The FAPI is obtained from PECs installed in a Fast-Multi-Energy Storage System (F-MESS), it consists of a flywheel storage system and supercapacitor storage system. The objective is to assess the frequency support provided by an F-MESS considering low rotational inertia scenarios. One additional contribution is a full detailed model using a set of differential-algebraic equation (DAE) in order to ensure an appropriate representation of all devices at the time that ensure scalability of the model in order to include new PEC-based technologies. Simulation results demonstrate the positive effect of the use of FAPI controllers in F-MESS to provide frequency support in low inertia scenarios."Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public."Intelligent Electrical Power Grid
Heterogeneous Datasets for Federated Survival Analysis Simulation
Get PDFHeterogeneous Datasets for Federated Survival Analysis Simulation
This repo contains three algorithms for constructing realistic federated datasets for survival analysis. Each algorithm starts from an existing non-federated dataset and assigns each sample to a specific client in the federation. The algorithms are:
uniform_split: assigns each sample to a random client with uniform probability;
quantity_skewed_split: assigns each sample to a random client according to the Dirichlet distribution [3, 4];
label_skewed_split: assigns each sample to a time bin, then assigns a set of samples from each bin to the clients according to the Dirichlet distribution [3, 4].
For more information, please take a look at our paper at https://arxiv.org/abs/2301.12166 [1].
Content
federated_survival_datasets.zip: the content of the repository at https://github.com/archettialberto/federated_survival_datasets
Heterogheneous_Datasets_for_Federated_Survival_Analysis_Simulation.pdf: the conference paper describing the work.
Installation
Federated Survival Datasets is built on top of numpy and scikit-learn. To install those libraries you can run pip install -r requirements.txt. To import survival datasets into your project, we strongly recommend SurvSet (https://github.com/ErikinBC/SurvSet) [2], a comprehensive collection of more than 70 survival datasets.
Usage
import numpy as np
import pandas as pd
from federated_survival_datasets import label_skewed_split
# import a survival dataset and extract the input array X and the output array y
df = pd.read_csv("metabric.csv")
X = df[[f"x{i}" for i in range(9)]].to_numpy()
y = np.array([(e, t) for e, t in zip(df["event"], df["time"])], dtype=[("event", bool), ("time", float)])
# run the splitting algorithm
client_data = label_skewed_split(num_clients=8, X=X, y=y)
# check the number of samples assigned to each client
for i, (X_c, y_c) in enumerate(client_data):
print(f"Client {i} - X: {X_c.shape}, y: {y_c.shape}")
We provide an example notebook in the zipped folder to illustrate the proposed algorithms. It requires scikit-survival, seaborn, and pandas.
References
[1] Archetti, A., Lomurno, E., Lattari, F., Martin, A., & Matteucci, M. (2023). Heterogeneous Datasets for Federated Survival Analysis Simulation. arXiv preprint arXiv:2301.12166.
[2] Drysdale, E. (2022). SurvSet: An open-source time-to-event dataset repository. arXiv preprint arXiv:2203.03094.
[3] Hsu, T. M. H., Qi, H., & Brown, M. (2019). Measuring the effects of non-identical data distribution for federated visual classification. arXiv preprint arXiv:1909.06335.
[4] Li, Q., Diao, Y., Chen, Q., & He, B. (2022, May). Federated learning on non-iid data silos: An experimental study. In 2022 IEEE 38th International Conference on Data Engineering (ICDE) (pp. 965-978). IEEE
Active toroidal field ripple compensation and MHD feedback control coils in FAST
No full textThe Fusion Advanced Study Torus (FAST) has been proposed as a high magnetic field, compact size tokamak providing a flexible integrated environment to study physics and technology issues in ITER and DEMO relevant conditions.
FAST has a quite large natural toroidal field ripple (around 1.5%) due to its compactness and to the number of access ports: this ripple must be lowered to an acceptable level to allow safe operations and a good confinement quality. An Active Ripple Compensating System (ARCS) has been designed, based on a set of poloidal coils placed between the plasma chamber and the Toroidal Field Coils (TFCs). These ARCS coils will be fed with adjustable currents, opposite in direction respect to the TFC currents, and will allow lowering the ripple up to zero and beyond.
The CAD model of FAST including the ARCS coils has been completed and preliminary electromagnetic and thermal analyses have been carried out.
Moreover, a Feedback Active Control System (FACS) composed of two arrays of in-vessel saddle coils has been designed to allow safe high plasma current, low safety factor operation and to mitigate possibly large ELMs effects in FAST. These FACS coils will be fed by a feedback system to control MHD modes: a first engineering assessment of the current requirements has been carried out. (C) 2013 Euratom-ENEA Association sulla Fusione. Published by Elsevier B.V. All rights reserved
- …
