-
Challenges in low losses and large acceptance ion beam transport
Authors:
F Osswald,
E Traykov,
T Durand,
M Heine,
J Michaud,
J C Thomas
Abstract:
A prototype of ion beam transport module has been developed at the Institut Pluridisciplinaire Hubert Curien (IPHC) and used as a test bed to investigate key issues related to the efficient transport of ion beams. This includes the reduction of the beam losses, the increase of the acceptance, and the definition of the instrumentation necessary to evaluate the performances. An experiment was perfor…
▽ More
A prototype of ion beam transport module has been developed at the Institut Pluridisciplinaire Hubert Curien (IPHC) and used as a test bed to investigate key issues related to the efficient transport of ion beams. This includes the reduction of the beam losses, the increase of the acceptance, and the definition of the instrumentation necessary to evaluate the performances. An experiment was performed on a full-scale beam line and following a standard beam analysis, steering, and focusing procedure. After a review of the developments carried out for some demanding facilities and for the design of the quadrupoles implemented in the transport module, the paper highlights the challenge of measuring the preservation of transverse phase-space distributions with large acceptance conditions, i.e. with the highest ratio of beam filling to quadrupole aperture. Then, the tolerance to the errors and mitigation of the risks are discussed, in particular by considering the electric stability of the transport module, beam trips, behavior of the tail and the halo, and misalignment errors.
△ Less
Submitted 13 March, 2023;
originally announced March 2023.
-
Green beam lines, a challenging concept
Authors:
F. Osswald,
E. Traykov,
T. Durand,
M. Heine,
J. Michaud,
J. C. Thomas
Abstract:
Due to increasing environmental and economic constraints, optimization of ion beam transport and equipment design becomes essential. The future should be equipped with planet-friendly facilities, that is, solutions that reduce environmental impact and improve economic competitiveness. The tendency to increase the intensity of the current and the power of the beams obliges us and brings us to new c…
▽ More
Due to increasing environmental and economic constraints, optimization of ion beam transport and equipment design becomes essential. The future should be equipped with planet-friendly facilities, that is, solutions that reduce environmental impact and improve economic competitiveness. The tendency to increase the intensity of the current and the power of the beams obliges us and brings us to new challenges. Installations tend to have larger dimensions with increased areas, volumes, weights and costs. A new ion beam transport prototype was developed and used as a test bed to identify key issues to reduce beam losses and preserve transverse phase-space distributions with large acceptance conditions.
△ Less
Submitted 17 November, 2022;
originally announced November 2022.
-
Training a Vision Transformer from scratch in less than 24 hours with 1 GPU
Authors:
Saghar Irandoust,
Thibaut Durand,
Yunduz Rakhmangulova,
Wenjie Zi,
Hossein Hajimirsadeghi
Abstract:
Transformers have become central to recent advances in computer vision. However, training a vision Transformer (ViT) model from scratch can be resource intensive and time consuming. In this paper, we aim to explore approaches to reduce the training costs of ViT models. We introduce some algorithmic improvements to enable training a ViT model from scratch with limited hardware (1 GPU) and time (24…
▽ More
Transformers have become central to recent advances in computer vision. However, training a vision Transformer (ViT) model from scratch can be resource intensive and time consuming. In this paper, we aim to explore approaches to reduce the training costs of ViT models. We introduce some algorithmic improvements to enable training a ViT model from scratch with limited hardware (1 GPU) and time (24 hours) resources. First, we propose an efficient approach to add locality to the ViT architecture. Second, we develop a new image size curriculum learning strategy, which allows to reduce the number of patches extracted from each image at the beginning of the training. Finally, we propose a new variant of the popular ImageNet1k benchmark by adding hardware and time constraints. We evaluate our contributions on this benchmark, and show they can significantly improve performances given the proposed training budget. We will share the code in https://github.com/BorealisAI/efficient-vit-training.
△ Less
Submitted 9 November, 2022;
originally announced November 2022.
-
Transverse emittance measurement in 2D and 4D performed on a Low Energy Beam Transport line: benchmarking and data analysis
Authors:
F Osswald,
T Durand,
M Heine,
J Michaud,
F Poirier,
J C Thomas,
E Traykov
Abstract:
2D and 4D transverse phase-space of a low-energy ion-beam is measured with two of the most common emittance scanners. The article covers the description of the installation, the setup, the settings, the experiment and the benchmark of the two emittance meters. We compare the results from three series of measurements and present the advantages and drawbacks of the two systems. Coupling between phas…
▽ More
2D and 4D transverse phase-space of a low-energy ion-beam is measured with two of the most common emittance scanners. The article covers the description of the installation, the setup, the settings, the experiment and the benchmark of the two emittance meters. We compare the results from three series of measurements and present the advantages and drawbacks of the two systems. Coupling between phase-space planes, correlations and mitigation of deleterious effects are discussed. The influence of background noise and aberrations of trace-space figures on emittance measurements and RMS calculations is highlighted, especially for low density beams and halos. A new data analysis method using noise reduction, filtering, and reconstruction of the emittance figure is described. Finally, some basic concepts of phase-space theory and application to beam transport are recalled.
△ Less
Submitted 9 November, 2022;
originally announced November 2022.
-
A versatile flexure-based 6-axes force/torque sensor and its application to tribology
Authors:
M Guibert,
C Oliver,
T Durand,
T Le Mogne,
A Le Bot,
D Dalmas,
J Scheibert,
J Fontaine
Abstract:
Six-axes force/torque sensors are increasingly needed in mechanical engineering. Here, we introduce a flexure-based design for such sensors, which solves some of the drawbacks of the existing designs. In particular, it is backlash-free, it can be wirelessly monitored, it exactly enforces 90 degrees angles between axes, and it enables visual inspection of the monitored system thanks to its hollow s…
▽ More
Six-axes force/torque sensors are increasingly needed in mechanical engineering. Here, we introduce a flexure-based design for such sensors, which solves some of the drawbacks of the existing designs. In particular, it is backlash-free, it can be wirelessly monitored, it exactly enforces 90 degrees angles between axes, and it enables visual inspection of the monitored system thanks to its hollow structure. We first describe the generic design, implementation and calibration procedure. We then demonstrate its capabilities through three illustration examples relevant to the field of tribology: low friction measurements under ultra-high vacuum, multi-directional friction measurements of elastomer contacts, and force/torque-based contact position monitoring.
△ Less
Submitted 21 May, 2021;
originally announced May 2021.
-
Variational Selective Autoencoder: Learning from Partially-Observed Heterogeneous Data
Authors:
Yu Gong,
Hossein Hajimirsadeghi,
Jiawei He,
Thibaut Durand,
Greg Mori
Abstract:
Learning from heterogeneous data poses challenges such as combining data from various sources and of different types. Meanwhile, heterogeneous data are often associated with missingness in real-world applications due to heterogeneity and noise of input sources. In this work, we propose the variational selective autoencoder (VSAE), a general framework to learn representations from partially-observe…
▽ More
Learning from heterogeneous data poses challenges such as combining data from various sources and of different types. Meanwhile, heterogeneous data are often associated with missingness in real-world applications due to heterogeneity and noise of input sources. In this work, we propose the variational selective autoencoder (VSAE), a general framework to learn representations from partially-observed heterogeneous data. VSAE learns the latent dependencies in heterogeneous data by modeling the joint distribution of observed data, unobserved data, and the imputation mask which represents how the data are missing. It results in a unified model for various downstream tasks including data generation and imputation. Evaluation on both low-dimensional and high-dimensional heterogeneous datasets for these two tasks shows improvement over state-of-the-art models.
△ Less
Submitted 24 February, 2021;
originally announced February 2021.
-
Design and characterization of an instrumented slider aimed atmeasuring local micro-impact forces between dry rough solids
Authors:
Camille Grégoire,
Bernard Laulagnet,
Joël Perret Liaudet,
Thibaut Durand,
Manuel Collet,
Julien Scheibert
Abstract:
Sliding motion between two rough solids under light normal loading involves myriad micro-impacts between antagonist micro-asperities. Those micro-impacts are at the origin of many emerging macroscopic phenomena, including the friction force, the slider's vibrations and the noise radiated in the surroundings. However, the individual properties of the micro-impacts (e.g. maximum force, position alon…
▽ More
Sliding motion between two rough solids under light normal loading involves myriad micro-impacts between antagonist micro-asperities. Those micro-impacts are at the origin of many emerging macroscopic phenomena, including the friction force, the slider's vibrations and the noise radiated in the surroundings. However, the individual properties of the micro-impacts (e.g. maximum force, position along the interface, duration) are essentially elusive to measurement. Here, we introduce an instrumented slider aimed at measuring the position and the normal component of the micro-impact forces during sliding against a rough track. It is based on an array of piezoelectric sensors, each placed under a single model asperity. Its dynamical characteristics are established experimentally and compared to a finite elements model. We then validate its measurement capabilities by using it against a track bearing simple, well-defined topographical features. The measurements are interpreted thanks to a simple multi-asperity contact model. Our slider is expected to be useful in future studies to provide local insights into a variety of tribological questions involving dry rough sliding interfaces.
△ Less
Submitted 8 January, 2021; v1 submitted 15 September, 2020;
originally announced September 2020.
-
Point Process Flows
Authors:
Nazanin Mehrasa,
Ruizhi Deng,
Mohamed Osama Ahmed,
Bo Chang,
Jiawei He,
Thibaut Durand,
Marcus Brubaker,
Greg Mori
Abstract:
Event sequences can be modeled by temporal point processes (TPPs) to capture their asynchronous and probabilistic nature. We propose an intensity-free framework that directly models the point process distribution by utilizing normalizing flows. This approach is capable of capturing highly complex temporal distributions and does not rely on restrictive parametric forms. Comparisons with state-of-th…
▽ More
Event sequences can be modeled by temporal point processes (TPPs) to capture their asynchronous and probabilistic nature. We propose an intensity-free framework that directly models the point process distribution by utilizing normalizing flows. This approach is capable of capturing highly complex temporal distributions and does not rely on restrictive parametric forms. Comparisons with state-of-the-art baseline models on both synthetic and challenging real-life datasets show that the proposed framework is effective at modeling the stochasticity of discrete event sequences.
△ Less
Submitted 22 December, 2019; v1 submitted 18 October, 2019;
originally announced October 2019.
-
LayoutVAE: Stochastic Scene Layout Generation From a Label Set
Authors:
Akash Abdu Jyothi,
Thibaut Durand,
Jiawei He,
Leonid Sigal,
Greg Mori
Abstract:
Recently there is an increasing interest in scene generation within the research community. However, models used for generating scene layouts from textual description largely ignore plausible visual variations within the structure dictated by the text. We propose LayoutVAE, a variational autoencoder based framework for generating stochastic scene layouts. LayoutVAE is a versatile modeling framewor…
▽ More
Recently there is an increasing interest in scene generation within the research community. However, models used for generating scene layouts from textual description largely ignore plausible visual variations within the structure dictated by the text. We propose LayoutVAE, a variational autoencoder based framework for generating stochastic scene layouts. LayoutVAE is a versatile modeling framework that allows for generating full image layouts given a label set, or per label layouts for an existing image given a new label. In addition, it is also capable of detecting unusual layouts, potentially providing a way to evaluate layout generation problem. Extensive experiments on MNIST-Layouts and challenging COCO 2017 Panoptic dataset verifies the effectiveness of our proposed framework.
△ Less
Submitted 1 June, 2021; v1 submitted 24 July, 2019;
originally announced July 2019.
-
A Variational Auto-Encoder Model for Stochastic Point Processes
Authors:
Nazanin Mehrasa,
Akash Abdu Jyothi,
Thibaut Durand,
Jiawei He,
Leonid Sigal,
Greg Mori
Abstract:
We propose a novel probabilistic generative model for action sequences. The model is termed the Action Point Process VAE (APP-VAE), a variational auto-encoder that can capture the distribution over the times and categories of action sequences. Modeling the variety of possible action sequences is a challenge, which we show can be addressed via the APP-VAE's use of latent representations and non-lin…
▽ More
We propose a novel probabilistic generative model for action sequences. The model is termed the Action Point Process VAE (APP-VAE), a variational auto-encoder that can capture the distribution over the times and categories of action sequences. Modeling the variety of possible action sequences is a challenge, which we show can be addressed via the APP-VAE's use of latent representations and non-linear functions to parameterize distributions over which event is likely to occur next in a sequence and at what time. We empirically validate the efficacy of APP-VAE for modeling action sequences on the MultiTHUMOS and Breakfast datasets.
△ Less
Submitted 5 April, 2019;
originally announced April 2019.
-
Learning a Deep ConvNet for Multi-label Classification with Partial Labels
Authors:
Thibaut Durand,
Nazanin Mehrasa,
Greg Mori
Abstract:
Deep ConvNets have shown great performance for single-label image classification (e.g. ImageNet), but it is necessary to move beyond the single-label classification task because pictures of everyday life are inherently multi-label. Multi-label classification is a more difficult task than single-label classification because both the input images and output label spaces are more complex. Furthermore…
▽ More
Deep ConvNets have shown great performance for single-label image classification (e.g. ImageNet), but it is necessary to move beyond the single-label classification task because pictures of everyday life are inherently multi-label. Multi-label classification is a more difficult task than single-label classification because both the input images and output label spaces are more complex. Furthermore, collecting clean multi-label annotations is more difficult to scale-up than single-label annotations. To reduce the annotation cost, we propose to train a model with partial labels i.e. only some labels are known per image. We first empirically compare different labeling strategies to show the potential for using partial labels on multi-label datasets. Then to learn with partial labels, we introduce a new classification loss that exploits the proportion of known labels per example. Our approach allows the use of the same training settings as when learning with all the annotations. We further explore several curriculum learning based strategies to predict missing labels. Experiments are performed on three large-scale multi-label datasets: MS COCO, NUS-WIDE and Open Images.
△ Less
Submitted 25 February, 2019;
originally announced February 2019.
-
Effects of ion drift on time-over-threshold charge estimates using first-order integrators in thermal neutron MWPC
Authors:
Patrick Van Esch,
Thomas Durand
Abstract:
Time-over-threshold measurements of the output of a first order integration amplifier can be used as a cheap way to estimate the charge deposited on a wire in an MWPC. Even if the time scale of the first order system is seemingly much larger than the signal development time, we noticed experimentally a significant deviation from the relationship of time-over-threshold and the charge, as would be…
▽ More
Time-over-threshold measurements of the output of a first order integration amplifier can be used as a cheap way to estimate the charge deposited on a wire in an MWPC. Even if the time scale of the first order system is seemingly much larger than the signal development time, we noticed experimentally a significant deviation from the relationship of time-over-threshold and the charge, as would be naively expected from the exponential response of the amplifier. We strongly suspect the current induced by the residual ion movement to be at least partially responsible for this observation. It is of course well-known that the ion movement is responsible for the bulk of the signal generation in a wire chamber, this is not the point ; however, the surprise is rather that this movement has influences on time scales which are so long that the finite ion movement could be thought negligible. In this paper we will treat theoretically the influence of ion movement and of AC coupling on the expected time-over-threshold relationship and confront this to the measurements on a small test MWPC in thermal neutron detection.
△ Less
Submitted 3 November, 2006;
originally announced November 2006.
