Investigation on the Ion Motion towards Clearing Electrodes in High-Brightness Electron Accelerators
Authors:
Gisela Pöplau,
Atoosa Meseck,
Fjodor Falkenstern,
Michael Markert,
Joachim Bartilla
Abstract:
High-brightness electron beams provided by modern accelerators require several measures to preserve their high quality and to avoid instabilities. The mitigation of the impact of residual ions is one of these measures. It is particularly important if high bunch charges in combination with high repetition rates are aimed for. This is because ions can be trapped in the strong negative electrical pot…
▽ More
High-brightness electron beams provided by modern accelerators require several measures to preserve their high quality and to avoid instabilities. The mitigation of the impact of residual ions is one of these measures. It is particularly important if high bunch charges in combination with high repetition rates are aimed for. This is because ions can be trapped in the strong negative electrical potential of the electron beam causing emittance blow-up, increased beam halo and longitudinal and transverse instabilities.
Over the last decades three ion-clearing strategies have been applied to counteract the degrading impact of the ions on the electron beam. These strategies are installation of clearing electrodes, introduction of gaps in filling pattern and beam shaking. Currently, their merit as clearing strategies for the next generation of high brightness accelerators such as energy recovery linacs (ERLs) are under intensive investigations by means of numerical and experimental studies. Of particular interest in this context is the performance of multi-purpose electrodes, which are designed such that they allow for a simultaneous ion-clearing and beam-position monitoring. Such electrodes will be installed in the bERLinPro facility.
In this paper, we present numerical studies for the behavior of ions generated by electron bunches while passing through the field of clearing electrodes. The objective is to investigate the ion motion towards the electrodes and to study under which circumstances and up to which ratio an equilibrium between ion-generation and ion-clearing is established. Hereby, several ion species and configurations of electrodes are considered in combination with typical beam parameters of high-brightness electron accelerators. Furthermore, we present detailed numerical studies of the performance of multi-purpose clearing electrodes planned for bERLinPro.
△ Less
Submitted 9 October, 2019;
originally announced October 2019.
Cross-Modal Data Programming Enables Rapid Medical Machine Learning
Authors:
Jared Dunnmon,
Alexander Ratner,
Nishith Khandwala,
Khaled Saab,
Matthew Markert,
Hersh Sagreiya,
Roger Goldman,
Christopher Lee-Messer,
Matthew Lungren,
Daniel Rubin,
Christopher Ré
Abstract:
Labeling training datasets has become a key barrier to building medical machine learning models. One strategy is to generate training labels programmatically, for example by applying natural language processing pipelines to text reports associated with imaging studies. We propose cross-modal data programming, which generalizes this intuitive strategy in a theoretically-grounded way that enables si…
▽ More
Labeling training datasets has become a key barrier to building medical machine learning models. One strategy is to generate training labels programmatically, for example by applying natural language processing pipelines to text reports associated with imaging studies. We propose cross-modal data programming, which generalizes this intuitive strategy in a theoretically-grounded way that enables simpler, clinician-driven input, reduces required labeling time, and improves with additional unlabeled data. In this approach, clinicians generate training labels for models defined over a target modality (e.g. images or time series) by writing rules over an auxiliary modality (e.g. text reports). The resulting technical challenge consists of estimating the accuracies and correlations of these rules; we extend a recent unsupervised generative modeling technique to handle this cross-modal setting in a provably consistent way. Across four applications in radiography, computed tomography, and electroencephalography, and using only several hours of clinician time, our approach matches or exceeds the efficacy of physician-months of hand-labeling with statistical significance, demonstrating a fundamentally faster and more flexible way of building machine learning models in medicine.
△ Less
Submitted 26 March, 2019;
originally announced March 2019.
